MEASURING MICROBIAL WATER QUALITY RESPONSES TO LAND AND CLIMATE
USING FECAL INDICATOR BACTERIA AND MOLECULAR SOURCE TRACKING IN
RIVERS AND NEAR-SHORE SURFACE WATERS OF MICHIGAN
By
Marc Paul Verhougstraete

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Fisheries and Wildlife
2012

ABSTRACT
MEASURING MICROBIAL WATER QUALITY RESPONSES TO LAND AND CLIMATE
USING FECAL INDICATOR BACTERIA AND MOLECULAR SOURCE TRACKING IN
RIVERS AND NEAR-SHORE SURFACE WATERS OF MICHIGAN
By
Marc Paul Verhougstraete

Current recreational water quality science in the Great Lakes relies on measuring E. coli
concentrations via cultivation techniques to estimate the risk associated with swimming in a
particular waterbody. However, this dissertation showed such approaches inaccurately describe
water quality across the entire beach and fail to represent underlining water quality issues. Water,
sediment, and algae samples were collected in creeks, rivers, and beaches from multiple
watersheds around Michigan. Samples were analyzed for cultivation based E. coli, enterococci,
Clostridium perfringens, and coliphage as well as molecular markers for E. coli, Enterococcus
spp., enterococci surface protein gene (human), Bacteroides (human and bovine) and Bacteroides
thetaiotaomicron mannanase gene (human).
In the Saginaw Bay, microbial indicators at four beaches were generally highest to lowest
in stranded algae mats, sediment, shallow water, and deep water, respectively. Contamination in
algae mats and sediment was identified in part as human specific using the enterococci surface
protein gene. Higher concentrations of E. coli and enterococci in algae mats and sediment,
compared to shallow and deep waters, were attributed largely to sediment bound bacteria and
bacterial regrowth or persistence. Results demonstrated the potential for sediment and algae mats
to act as non-point sources of pollution in the nearshore zone.
Water and sediment samples collected from Mitchell Creek and Traverse City State Park
beach quantified fecal indicator bacteria across space and time. Fecal indicator bacteria

concentrations represented widespread, long-term, and recent fecal contamination in the Mitchell
Creek. Despite the close proximity of the Mitchell Creek discharge to the Traverse City State
Park beach, microbial concentrations were significantly lower (p < 0.01) in beach water which
was partially impacted by creek discharge. Assessment of land use type at the watershed scale
failed to identify consistent correlations with fecal indicator bacteria. However, Bacteroides
thetaiotaomicron detections in both waterbodies indicated fecal contamination was partially
human. Additional analysis on a subset of data identified significant disconnect between
molecular and cultivation based results in creek and beach water. However, across all
waterbodies, cultivated enterococci would have resulted in the greatest number of regulatory
actions compared to cultivated E. coli and molecular based Enterococcus spp.
A snapshot survey of 64 rivers discharging to the Great Lakes quantified E. coli and
Bacteroides thetaiotaomicron under baseflow conditions. Bacteroides thetaiotaomicron was
-1

detected in all samples (X = 5.1 log10 Cell Equivalents 100 ml ). The E. coli geometric mean
-1

across all rivers (1.4 log10 MPN 100 ml ) suggests a potential regional reference condition.
Classification And Regression Tree analysis indicated the total number of septic system in a
watershed significantly impacted Bacteroides thetaiotaomicron concentrations under baseflow
conditions. Land use characteristics better predicted microbial water quality than land use type.
This work coupled molecular tools and novel monitoring strategies of unique
environments (algae mats, sediments, beaches, small creek systems, and large river watersheds)
to better understand the impact of human activities on Great Lakes water quality. Land use
characteristics, not land use type, were related to qPCR markers in rivers which are transported
to nearby beaches. Finally, septic systems, algae mats, and sediments were identified as nonpoint sources of pollution in Michigan surface water systems.

Copyright
By
Marc Paul Verhougstraete
2012

This dissertation is dedicated to my family who unknowingly taught me the ability of water to
unite communities, my friends who supported me during all of my endeavors, and the Great
Lakes which generated a thirst for water science eclipsed only by a really thirsty person.

v

ACKNOWLEDGEMENTS

I would like to first thank Joan Rose, Rebecca Ives, Rachel McNinch, Erin Dreelin, and
Shannon Briggs who have been there since the beginning. Joan Rose was a model of dedication
and passion for water science who taught me the importance of perseverance, endurance, and
patience. Rebecca Ives helped develop my microbiology skills and was irreplaceable during
project development and implementation strategies. Starting with my first day, Rachel McNinch
and Erin Dreelin provided invaluable support through insightful, creative, and focused
recommendations. Had it not been for Shannon Brigg’s suggestion that I pursue a graduate
degree with Joan Rose and her support along the way, this Dissertation would be substantially
different than its current form.
Although this work is already dedicated to my family and friends, they deserve special
recognition here as they provided personal support during my graduate school carrier. They
stood beside me while I chased microorganisms across Michigan and allowed me to accomplish
far more than I could have on my own.
A number of graduate students provided vital sample collection and processing support
essential to the development of my research. Asli Aslan, Sibel Zeki, Rebecca Ives, Chris Wendt,
and Stephanie Longstaff helped immensely during the final leg of my Mitchell Creek research.
Anthony Kendall and Sherry Martin were instrumental in the development of chapters three and
five.
Finally, funding for this dissertation has come from a number of organizations, including
NOAA, Michigan Department of Environmental Quality, local health departments, and the
Watershed Center Grand Traverse Bay.
vi

TABLE OF CONTENTS

LIST OF TABLES……………………………………………………………….……….….…..x
LIST OF FIGURES……………………………………………………………….……….…..xiii
CHAPTER 1. THE GREAT LAKES BASIN: POLICY, POLLUTION, AND WATER
QUALITY...…………………..……………………………………………………………...1
1.1. The Laurentian Great Lakes..................................................................................................... 2
1.2. Indicators of Water Quality ................................................................................................... 12
1.2.1. Water quality and health studies ..................................................................................... 12
1.2.2. Fecal indicator bacteria ................................................................................................... 16
1.2.3. Molecular Source Tracking............................................................................................. 22
1.3. Nonpoint Source, Land, and Weather Impacts on Water Quality ......................................... 29
1.3.1. Algae ............................................................................................................................... 29
1.3.2. Sediment ......................................................................................................................... 33
1.3.3. Land-water interface ....................................................................................................... 36
1.3.4. Weather ........................................................................................................................... 41
1.4. Scientific Needs ..................................................................................................................... 43
1.5. Research Objectives ............................................................................................................... 45
Goal 1 ........................................................................................................................................ 45
Goal 2 ........................................................................................................................................ 46
Goal 3 ........................................................................................................................................ 46
REFERENCES ............................................................................................................................. 48
CHAPTER 2. MICROBIAL INVESTIGATIONS OF WATER, SEDIMENT, AND
ALGAE MATS IN A MIXED USE WATERSHED. ............................................................... 67
2.1. Introduction ............................................................................................................................ 68
2.2. Materials and Methods ........................................................................................................... 72
2.2.1. Description of study area ................................................................................................ 72
2.2.2. Sample collection and processing ................................................................................... 73
2.2.3. Water analysis ................................................................................................................. 74
2.2.4. Sediment and algae analysis ........................................................................................... 74
2.2.5. Molecular analysis .......................................................................................................... 75
2.2.6. Environmental and physical data .................................................................................... 77
2.2.7. Data analysis ................................................................................................................... 77
2.3. Results .................................................................................................................................... 78
2.3.1. Fecal indicator organisms in water and sediment ........................................................... 78
2.3.2. Occurrence of fecal indicator organisms in algae and source tracking markers............. 85
2.3.3. Fecal indicator associations with environmental parameters.......................................... 88
2.4. Discussion .............................................................................................................................. 91
2.5. Conclusions .......................................................................................................................... 100
REFERENCES ........................................................................................................................... 101
vii

CHAPTER 3. LINKING LAND-USE AND MICROBES IN A SMALL DIVERSE
WATERSHED USING INDICATOR BACTERIA AND MOLECULAR SOURCE
TRACKING............................................................................................................................... 108
3.1. Introduction .......................................................................................................................... 109
3.2. Materials and Methods ......................................................................................................... 112
3.2.1. Sampling strategy.......................................................................................................... 112
3.2.2. Environmental monitoring ............................................................................................ 117
3.2.3. Cultivation analyses ...................................................................................................... 118
3.2.4. Molecular analyses........................................................................................................ 119
3.2.5. Spatial analysis.............................................................................................................. 120
3.2.6. Statistical analysis ......................................................................................................... 121
3.3. Results .................................................................................................................................. 125
3.3.1. Land use and cover ....................................................................................................... 125
3.3.2. Spatial analyses of water and sediment quality ............................................................ 126
3.3.3. Temporal analysis ......................................................................................................... 132
3.3.4. Drivers of bacterial water quality ................................................................................. 135
3.3.5. Microbial responses to sources and drivers .................................................................. 138
3.4. Discussion ............................................................................................................................ 148
APPENDIX ................................................................................................................................. 155
REFERENCES ........................................................................................................................... 172
CHAPTER 4. FINE TUNING MICROBIAL WATER QUALITY CRITERIA FOR
LOCAL WATERSHEDS ......................................................................................................... 179
4.1. Introduction .......................................................................................................................... 180
4.2. Materials and methods ......................................................................................................... 184
4.2.1. Sampling location and collection.................................................................................. 184
4.2.2. Enumeration of bacteria using cultivation techniques .................................................. 186
4.2.3. Enumeration of bacteria using molecular techniques ................................................... 186
4.2.4. Statistical analysis ......................................................................................................... 189
4.3. Results .................................................................................................................................. 193
4.3.1. Comparing culture versus molecular results ................................................................. 193
4.3.2. Indicators and criteria ................................................................................................... 197
4.3.3. Implications for use historical data sets ........................................................................ 198
4.4. Discussion ............................................................................................................................ 202
APPENDIX ................................................................................................................................. 208
REFERENCES ........................................................................................................................... 212
CHAPTER 5. MICROBIAL RESPONSES TO LAND, PHYSICAL, CHEMICAL,
ENVIRONMENTAL, AND HYDROLOGICAL FACTORS .............................................. 218
5.1. Introduction .......................................................................................................................... 219
5.2. Materials and methods ......................................................................................................... 222
5.2.1. Study area...................................................................................................................... 222
5.2.2. Water sample collection ................................................................................................ 223
5.2.3. Water analysis ............................................................................................................... 224
5.2.4. Hydrometry ................................................................................................................... 227
viii

5.2.5. Land use ........................................................................................................................ 227
5.2.6. Statistical analysis ......................................................................................................... 229
5.3. Results .................................................................................................................................. 232
5.3.1. Microbial water quality ................................................................................................. 233
5.3.2. Land use ........................................................................................................................ 235
5.3.3. Physical, chemical, environmental, and hydrology ...................................................... 239
5.3.5. CART analysis of microbial water quality ................................................................... 240
5.4. Discussion ............................................................................................................................ 244
APPENDIX ................................................................................................................................. 250
REFERENCES ........................................................................................................................... 261
CHAPTER 6. CONCLUSION ................................................................................................. 268

ix

LIST OF TABLES

Table 1.1. Measurements of key physical parameters for each of the Laurentian Great Lakes
during modern times……………………………………………………………………………… 3
Table 1.2. History of U.S. water quality public laws……………...……………………………...6
Table 1.3. Michigan’s Great Lakes beach closure or advisory days per year………..…….……15
Table 1.4. Sequences and targeted organisms for select molecular source tracking
markers………………………………………………………………………………………..….26
Table 1.5. Studies and the microorganisms identified in Cladophora collected within the Great
Lakes..……...……...……………………………………………………………………………..31
Table 1.6. E. coli density ranges identified in Great Lakes sediment....………………………...34
Table 1.7. Key findings of studies attempting to link land use and microbial water quality…....40
Table 2.1. Site description, water quality exceedances, and potential pollution influences…….73
Table 2.2. PCR assays examined during source tracking in water, sediment, and algae…….….76
Table 2.3. Range and geometric mean concentrations of fecal indicator organisms in algae
samples at SB1 (n =2) and SB2 (n = 7) ………………………………………………………....87
Table 2.4. Significant correlations identified between fecal indicator organisms and
environmental parameters …………………………………………………………………….…89
Table 3.1. Description of Mitchell Creek watershed including land use and number of samples
collected for each site. Mitchell Creek flows from sites 8, 7, 6, and 5 (headwater catchments),
through sites 4 and 3 or 2, before discharging near site 1 (upstream of outlet)………………...115
Table 3.2. B. thetaiotaomicron results in Grand Traverse Bay and Mitchell Creek water
samples………………………………………………………………………………………….131
Table 3.3. Sites identified were water quality could be partially explained by sediment levels
(CART) and the associated correlations and significance………………………………..…….148
Table S.3.1. Number of water and sediment samples assayed for E. coli, enterococci, C.
perfringens, and coliphage CN-13………………………………………………………...……157
Table S.3.2. Water E. coli, enterococci, C. perfringens, and coliphage CN-13 recordings for each
event at TCSP and in the Mitchell Creek………………………………………………….……158

x

Table S.3.3. Sediment E. coli, enterococci, C. perfringens, and coliphage CN-13 recordings for
each event at TCSP and in the Mitchell Creek ……………………………………...…………162
Table S.3.4. Summary statistics of physical and environmental properties for all sites……….164
Table S.3.5. Spearman’s correlation matrix of environmental, weather, and microorganisms in
water (Mitchell Creek and TCSP)………………………………………………………….…..165
Table S.3.6. Spearman’s correlation matrix of environmental, weather, and microorganisms in
sediment (Mitchell Creek and TCSP)…………………………………………………………..166
3 -1

Table S.3.7. Average daily discharge for each Mitchell Creek site (m s )………...……..….167
Table S.3.8. Precipitation details for the Mitchell Creek watershed………………………...…168
th

Table S.3.9. Fecal indicator bacteria exceedances of the 95 percentile and associated discharge
percentile…………………………………………………………………………………..……170
Table S.3.10. Spearman’s correlation coefficients and significance levels between precipitation
and E. coli and enterococci in water and sediment from the Mitchell Creek…………………..171
Table 4.1. Log10 transformed bacterial concentrations from a Great Lakes water system and the
respective number of regulatory exceedances……………………………………………….....194
Table S.4.1. Spearman’s rank correlation matrix among microorganism detection
method………………………………………………………………………………………….209
Table S.4.2. CART results describing the ability of different methods to predict microorganism
concentrations. Model results include model target assay and interactions between assays
measured using the different method (i.e. molecular (target) associations with cultivation only
and vice versa). Refer to methods and Figure 4.2. for interpretation of CART analysis.
………………………………………………………………………..…………………….…..211
Table 5.1. Summary of chemical and nutrient methods with respective references…………...226
Table 5.2. Anderson level 1 land use classifications and descriptions……………………..….228
Table 5.3. Land use summary for full watersheds, reduced watersheds, and reduced watersheds
riparian buffers (60 m) …...…………………………………………………………………….238
Table 5.4. CART analyses for E. coli and B. thetaiotaomicron as dependent variables and land
use, nutrient, chemical, hydrological, and environmental as independent variables
………………………………………………………………………………………………..…241

xi

Table S.5.1. E. coli and B. theta levels measured in 64 Michigan rivers under baseflow, spring
thaw, and summer rain conditions……………………………….….………………………….251
Table S.5.2. Land use composition of defined river catchments using Anderson Land Use
Classification systems………………………………………………………………………..…254
Table S.5.3. Descriptive statistics of physical-chemical, and hydrological variables measured
during baseflow conditions at 64 rivers ………………………………………………………..257

xii

LIST OF FIGURES

Figure 2.1. Location of beach sites in the Saginaw Bay selected for deep water, shallow water,
sediment, and stranded algae mat investigation using fecal indicator bacteria and molecular
source tracking. For interpretation of the references to color in this and all other figures, the
reader is referred to the electronic version of this dissertation…………………………………..69
Figure 2.2. Geometric mean concentrations of fecal indicator organisms in shallow water (n =
32), deep water (n = 32), and sediment (n = 32) at SB1, SB2, SB3, and SB4. : Short error bars
result of small C. perfringens standard deviations;
: No standard deviation due to high
percentage of non-detects; When organisms were not detected they were assigned a value equal
to the lowest method detection limit; Water reported as log10 CFU or PFU 100 ml-1; Sediment
reported as log10 CFU or PFU 100 g-1 wet weight…………………………………………….....82
Figure 3.1. The Mitchell Creek watershed with sampling site locations. Upper inset image: The
Digital Elevation Model of the Mitchell Creek watershed. Bottom inset image: Location of the
Mitchell Creek watershed in Michigan and the Great Lakes…………………………………...114
Figure 3.2. Classification And Regression Trees are composed of root nodes that contain all
available data and are split into binary groups using recursive partitioning algorithm and 10-fold
cross validation with a complexity parameter value of 0.05. Primary splitting variables and
values are described for each child node. Terminal nodes (bottom of the tree) include the mean
concentration and number of target organism cases in each node. Each node was derived based
on mean value of each response variable, group size, and defining variables……………….....124
Figure 3.3. Box plots illustrating the ranges of fecal indicator organisms and B. thetaiotaomicron
concentrations measured in the water and sediment (E. coli and enterococci only) at each site.
th
th
The lower, middle, and top box edges correspond to the 25 , median, and75 percentiles of
each measurement. The whiskers indicate the 10
measurements outside the 5

th

and 95

th

th

and 90

th

percentile. The points indicate

percentiles. E. coli, enterococci, C. perfringens, and
-1

coliphage were measured in sediment and reported per 100 g dry weight. The number of
samples assayed for each cultivated microorganism at each site is presented in Table S.3.1. The
number of B. theta assays for each site is presented in Table 3.2……………………………....129
Figure 3.4. Temporal variation of E. coli in the Mitchell Creek. Water daily geometric means
-1
-1
reported as log10 MPN 100 ml and sediment reported as log10 MPN 100 g dry weight.
Enterococci trended similarly to E. coli. Minor temporal variations were identified using C.
perfringens and coliphage CN-13. Precipitation is shown as cumulative rainfall in 24 hours prior
to sample collection.
denotes dates when TCSP water samples exceeded Michigan E. coli
water quality standards. No sediment samples were collected in 2009………………………...134
Figure 3.5. Mitchell Creek water CART outputs for (A) E. coli, (B) enterococci, (C) C.
perfringens, (D) coliphage (CN-13), and (E) B. theta. Each split is labeled with splitting variable
xiii

and value. Terminal nodes (bottom rectangle) are labeled with means and cases of target
organism in each group. ………………………………………………………………………..139
Figure 3.6. TCSP water CART outputs for (A) E. coli, (B) enterococci, (C) C. perfringens, (D)
coliphage (CN-13), and (E) B. theta………………………………………………………...….144
Figure 3.7. Dendrogram showing clusters based on geometric mean indicator concentrations at
each site. TCSP is a designated swimming area with relatively low indicator organism
concentrations and root node splits on wind (direction and speed) and MC parameters (discharge
and water/sediment microbes). Clusters 2, 3, and 4 are sites within the Mitchell Creek watershed
and are not designated as primary contact recreation sites………………………………...…...145
Figure 3.8. Sediment CART outputs for (A) Mitchell Creek E. coli, (B) Mitchell Creek
enterococci, (C) Mitchell Creek C. perfringens, (D) Mitchell Creek coliphage (CN-13), (E)
TCSP E. coli, and (F) TCSP enterococci. Insufficient data was available to perform CART
analysis on C. perfringens and coliphage assays at TCSP……………………………………...147
Figure S.3.1. Land use patterns at multiple scales for each sampling location…………….….156
Figure S.3.2. WWTP daily discharge flows (A.) averaged per month and (B.) averaged per day
of week. *between January 1, 2009 and April 29, 2011; **whole year and tourist season (June to
October)………………………………………………………………………………………...169
Figure 4.1. Mitchell Creek watershed and sampling locations within Michigan and the Great
Lakes……………………………………………………………………………………………185
Figure 4.2. Example of a Classification And Regression Tree (CART) output. Root nodes
contain all available data and are split into binary groups using recursive partitioning algorithm
and 10-fold cross validation with a complexity parameter value of 0.05. Primary splitting
variables and values are described for each child node. Terminal nodes (bottom of the tree)
include the mean concentration and number of target organism cases in each node. Each node
was derived based on mean value of each response variable, group size, and defining
variables…………………………………………………………………………………..…….192
Figure 4.3. A spatial and temporal depiction of molecular assays (TOP FIGURE) B.
thetaiotaomicron (CIRCLE), E. coli (SQUARE), and Enterococci spp. (TRIANGLE) and
cultivation assays (BOTTOM FIGURE) E. coli (SQUARE) and enterococci (TRIANGLE)
concentrations. Lines are presented to discern between organisms (B. thetaiotaomicronDASHED, enterococci-SOLID, E. coli-DOTTED). Each collection date (n = 11) depicts results
from four creek sites (MC2-RED, MC3-GREEN, MC5-BLUE, MC6-PINK) and one beach
(TCSP-BLACK). On August 7, 2010 consecutive hourly samples (n = 12) were collected from
each site for a total of 60 samples. In total 111 samples were collected during this study….....196
Figure 4.4. A frequency distribution of cultivation E. coli measurements from a long term beach
monitoring database (2001-2011, n = 189), cultivation E. coli measurements during project
(2010, n = 23), and molecular E. coli measurements during project (2010, n = 23) at TCSP beach
xiv

during 2 time periods. Cultivation E. coli concentrations between time periods were within the
normal expected distribution (p = 0.116) while molecular based E. coli concentrations from the
project were outside expected normal distributions of long term cultivation E. coli (p <
0.001)…………………………………………………………………………………….……..199
Figure 4.5. Classification And Regression Tree (CART) analysis of (A) E. coli MPN, (B)
enterococci MPN, (C) E. coli CE, (D) Enterococci spp. CE, and (E) B. theta CE. Binary splitting
of variables identified best categories according to splitting criteria. The target organism is
bolded in the top rectangle. Independent splitting variables and splitting value are presented for
each branch of the tree. Target organism means and target organism cases (in parenthesis) are
described for each terminal node (bottom rectangle)………...…………………………...……201
Figure S.4.1. Analysis of regulatory based outcomes occurring during entire project according to
USEPA suggested criteria. Values presented represent number of cases (total n =
111)..……………………………………………………………………………………..……..210
Figure 5.1. Sampled river systems and catchment areas in Michigan (USA) and NLCD 2006
land use in Michigan …………………………………………………………………………...223
Figure 5.2. Classification And Regression Tree analysis output example. Root nodes contain all
available data and are split into binary groups using recursive partitioning algorithm and 10-fold
cross validation with a complexity parameter value of 0.05. Primary splitting variables and
values are described for each child node. Terminal nodes (bottom of the tree) include the mean
concentration and number of target organism cases in each node. Each node was derived based
on mean value of each response variable, group size, and defining variables. …….………….232
-1

-1

Figure 5.3. (A.) E. coli (log10 MPN 100 ml ) and (B.) B. thetaiotaomicron (log10 CE 100 ml )
concentrations measured at 64 river catchments under baseflow conditions. E. coli and B.
thetaiotaomicron categories were evenly split across the concentration range. Areas in black
-1
were not sampled. The USEPA health exposure criterion for E. coli is 2.37 log10 MPN 100 ml ,
shown as the two highest categories in the E. coli figure and was detected at nine rivers. No
single river sample had measurable concentrations of both microorganisms in the highest
concentration categories.…………………….…………………………………………………234
Figure S.5.1. CART tree outputs for E. coli and B. thetaiotaomicron at the (A) full and (B)
reduced watersheds developed with (1) all data, (2) land use variables only, and (3) nutrient,
chemical, and environmental variables only. …………………………………………….…….259

xv

CHAPTER 1.
THE GREAT LAKES BASIN: POLICY, POLLUTION, AND WATER QUALITY

1

1.1. The Laurentian Great Lakes
The Great Lakes, prior to European settlers, were largely separated from outside surface waters
but connected together through rivers, straits, and waterfalls. As population increased during the
American industrial revolution (Voth 2003), the Great Lakes became an important waterbody for
transportation and waste disposal (Page and Walker 1991). Sewage from the City of Chicago
polluted drinking water sources and led to the development of the Chicago Sanitary and Ship
Canal (Ashworth 1986). This canal opened the Mississippi river to the Great Lakes and
redirected the flow of surface water away from Lake Michigan, Chicago’s only source of
drinking water (completed in 1900) (Ashworth 1986). As the population of the Midwestern
region grew so did the need for faster and more efficient transportation. In response, the St.
Lawrence Seaway (1959) was opened to direct nautical shipping from the Atlantic Ocean to the
Great Lakes (Whitfield and Kolenosky 1978). As a result of the increased connectivity and
human population, the Great Lakes ecosystems became increasingly stressed.

The Laurentian Great Lakes have recently been described in terms of great flux (Karrow and
Calkin 1985; Larson and Schaetzl 2001). The land-water interface, lake level, percent ice
coverage, and surface water temperature are constantly changing due to natural and
anthropogenic influences. The most recent measurements of these parameters for each of the
lakes are detailed in Table 1.1. Land use in the Great Lakes basin has also undergone change in
recent decades with one study reporting a 9.8% decrease in agriculture land use since 1982 and
49.3% increase in urban developed land use (Wolter et al. 2006).

2

Table 1.1. Measurements of key physical parameters for each of the Laurentian Great Lakes
during modern times.
Great Lake

Lake Erie

Surface
area
2A
(km )
25,700

Average
depth
B
(m)
19

Lake Huron

59,600

59

Lake Michigan

57,800

85

Lake Ontario

18,960

86

Lake Superior

82,100

147

A

B

Lake
levels
C
(m)
174.0 174.3
176.3 176.6
176.3 176.6
74.5 75.1
183.3 183.6

C

Surface water
temperature
D
(range)
0.0 - 23.0 °C

Ice
coverage
E
(%)
87.5

Water retention
time
A
(years)
2.6

0.0 - 18.0 °C

68.1

22

0.0 - 25.5 °C

42.7

99

0.0 - 20.1 °C

32.5

6

0.0 - 15.5 °C

70.1

191
D

USEPA 2006; Grady 2007; United States Army Corp of Engineers 2011; Schwab et al.
E

1999. Ferris and Andrachuk 2009

Recommended beneficial uses for the Great Lakes include: aquatic life; fish and shellfish
consumption; drinking water supply; primary contact recreation; secondary contact recreation;
and agriculture (USEPA 1992). Designated beneficial uses are specific to a waterbody and are
the desired uses that water quality should support (USEPA 1992). At minimum, all waters of the
Great Lakes should meet the swimmable and fishable uses as set forth in the Clean Water Act.

One of the most valued beneficial uses and utilized areas in the Great Lakes is for recreation in
the nearshore zone, respectively, which includes beaches and shallow waters. The Great Lakes
include 8851 km (Dorfman and Rosselot 2011) of some of the world’s greatest sandy beaches
(Chrzastowski et al. 1994; Folger et al. 1994), but a growing trend of increasing beach closures
(Dorfman and Rosselot 2011) has plagued many coastal communities. Nationally, tourism has
become a primary factor driving economic activity, job creation, wealth, and investment
3

(Houston 2008) and the economic value gained from Great Lakes beach tourism can be seen in
the fiscal impacts of beach closures. Song et al. (2010) estimated closing all Lake Michigan
beaches located in the state of Michigan would result in an economic loss of $2.7 billion.
Another Great Lakes Basin study estimated beach closures cost the surrounding community
nearly $228,000 per beach closure (Murray et al. 2001). Austin et al. (2007) suggested a 20%
reduction in Great Lakes beach closures would result in an economic benefit of at least $130
million per year. Therefore, Great Lakes beaches are not only a treasured natural resource but
also a vital economic driver to the surrounding states which require protection against further
degradation.

Significant federal and state policies were required to address decades of misuse and degradation
th

of America’s water systems. Over the latter half of the 20 century, policy slowly emerged that
partially addressed the most significant stressors including invasive species (Mills et al. 1993;
Ricciardi 2006), toxins (Baumann and Whittle 1988), nutrient loads (Wiley et al. 2010), and
pathogens associated with human fecal pollution, mainly from wastewater infrastructure or
combined sewer overflows (Dreelin et al. 2007; Patz et al. 2008). Water quality policy
specifically, slowly developed, with each new act building upon previous legislation (Table 1.2.).
However, it was not until the 1970’s when a more unified water quality protection act addressing
ambient waters was enacted via the development of the Clean Water Act.

Initial laws governing United States water quality protection date back to the 1880s. In 1886,
Congress enacted the first federal environmental law known as the River and Harbor
Appropriations Act, later renamed the Rivers and Harbors Act (1899) which gave water quality
4

protection power to the Army Corp of Engineers. The Act aimed to reduce pollution of navigable
waters by classifying refuse discharges misdemeanors and requiring a permit for any alterations
to harbors or channels. The Act did not address most liquid wastes, an area that would later
affect many American waterways. Nonetheless, the Rivers and Harbors Act was an important
first step in protecting American water systems that would stand for over 50 years.

After five decades, the Rivers and Harbors Act had failed to reach environmentalists
expectations. During the late 19

th

and early 20

th

century, following lax enforcement and

widespread unmonitored discharges, America’s surface waters were increasingly stressed from
human derived pollution (Dempsey 2004). Between 1900 and 1970, rivers throughout the Great
Lakes watershed, including the Cuyahoga, Rouge, and Buffalo Rivers, had reportedly caught fire
on multiple occasions (Adler 2002). In the 1960s, American surface waters were perceived
unsafe for swimming or fishing and the current standards provided inadequate protection
(Copeland 2010). In addition to point sources, non-point pollution such as agricultural runoff
also resulted in the annual loss of soil and the deposition of nutrients into surface waters which
was estimated in the billions of tons (USDA-NRCS 1997). A lack of environmental protection
and enforcement required government assistance beyond the River and Harbors Act if the highly
polluted surface waters were to recover.

5

Table 1.2. History of U.S. water quality public laws.
Public Law
33-403
80-845

Official Title
Rivers and Harbors Appropriation Act of
1899
Water Pollution Control Act of 1948

84-660

Federal Water Pollution Control Act of
1956

87-88

Federal Water Pollution Control Act of
1961

89-234

Water Quality Act of 1965

89-753

Clean Water Restoration Act of 1966

91-190

National Environmental Policy Act of
1969

91-224

Water Quality Improvement Act of 1970

92-500

Federal Water Pollution Control Act
Amendments of 1972
Clean Water Act (CWA) of 1977

95-217

96-510
100-4

106-284

Comprehensive Environmental Response,
Compensation, and Liability Act (1980)
Water Quality Act of 1987

Beaches Environmental Assessment and
Coastal Health Act (2000)

6

Primary Goals
Addressed dumping of trash to
waterways
Cooperative programs between state
and federal agencies; limited federal
enforcement and financial assistance
Cooperative programs between state
and federal agencies; limited federal
enforcement and financial assistance
Strengthens federal enforcement and
support to states; Established
Department of Health, Education,
and Welfare
Directs States to develop standards
for interstate waters
Study the effects of sedimentation
and pollution on designated uses
Establish a national policy for
environment and the Council of
Environmental Quality
Established stricter limits on
pollutant discharge to waters
Established NPDES
Provides pollution control authority
to EPA; Funded wastewater
treatment plant construction;
Discharge without permit to
navigable waters became illegal
Addresses environmental issues from
accidental spills or releases
Required states to develop numeric
criteria for water body segments
where toxic pollutants were
negatively affect designated uses;
antidegradation policy
Criteria development by States;
Standard modification; Monitoring
notification; Appropriations

The Rivers and Harbors Act (1899) stood as the only federal legislation protecting American
waterways until the passage of the Federal Water Pollution Control Act (1948) which
strengthened the Federal government’s authority of pollution control and aimed to enhance water
quality through the creation of national pollution control and prevention policy. Federal authority
improved with the creation of the 1956 Federal Water Pollution Control Act, Federal Water
Pollution Control Act of 1961, the Water Quality Act (1965), the Clean Water Restoration Act
(1966), and the Water Quality Improvement Act (1970). These Acts allowed Federal government
to file lawsuits against polluting entities, established national enforceable standards, imposed
monetary fines for polluters, and mandated States to develop and adopt antidegradation
standards. All of these acts laid the foundation for water quality improvement, protection, and
enforcement but the multiple amendments hindered the Federal government’s ability to
effectively implement and enforce regulations.

Following decades of cobbled policies and a plethora of highly publicized environmental
problems, policymakers undertook the long process of cleaning, protecting, and rehabilitating
public surface waters and the surrounding environment. This process culminated with the
National Environmental Policy Act (NEPA) which established the Council of Environmental
Quality and paved the way for future measures aimed at protecting the environment. The Council
of Environmental Quality acts as a moderator between environmental federal agencies and
reports to the president of the United States on the progress of environmental conditions. NEPA
was instrumental in raising environmental awareness which eventually resulted in the creation of
the Federal Water Pollution Control Act Amendments (known as the Clean Water Act of 1972

7

and amended in 1977). After decades of gross misuse and impacts on human health, America’s
surface waters finally received substantial federal governmental attention.

Under the Clean Water Act (CWA), programs and funding sources were developed to establish,
implement, and enforce chemical, biological, and physical integrity for America’s surface
waters. Waterbodies that failed to meet designated use criteria were to be listed on an impaired
waters list (303(d)) and receive further control measures including Total Maximum Daily Loads
(TMDL) which established maximum pollutant levels that could be discharged daily to a
waterbody and still meet water quality standards. The National Pollutant Discharge Elimination
System (NPDES) permitting program enforced water quality rules limiting point source pollution
discharges. The Clean Water Act and associated programs sought to improve water quality to
swimmable and fishable standards by 1983 and eliminate the discharge of pollutants to navigable
waters by 1985. The key goals of the CWA were to provide financial assistance for wastewater
treatment plant construction; regulate pollution discharge (NPDES); and achieve water quality
safe for swimming and fishing.

Following the passage of the CWA, it became evident that additional environmental protection
was required. As, part of the superfund (Comprehensive Environmental Response,
Compensation, and Liability Act of 1980) program, the U.S. Environmental Protection Agency
(EPA) was authorized to address toxic spills which threatened water quality or human health. In
1987, Congress addressed the need to improve stormwater and required municipal treatment
plants to obtain NPDES permits.

8

The CWA established recreational water quality regulations, health based criteria, and state
standards, however many beaches remained heavily impacted by polluted water and required
additional protection. In response, Congress passed a CWA amendment, the Beach
Environmental Assessment and Coastal Health (BEACH) Act, in 2000. It required coastal States,
including those along the Great Lakes, to adopt recreational water quality standards equally
protective as established Federal criteria. The BEACH Act also established a monitoring
appropriation for beaches to identify acute and chronic pollution issues. The BEACH Act was
authorized at $30 million per year but annual appropriations are often less than $10 million.
Since its initiation, the BEACH Act has supported Great Lakes beach monitoring in Michigan
with awards totaling $2.82 million (USEPA 2011). Furthermore, the Act included a timeframe
for criteria revisions that must occur at least every five years and required beach monitoring
result to be reported in a timely fashion.

NEPA and subsequent water quality public laws have been instrumental in the protection and
enhancement of surface water quality throughout the United States. Since the enactment of the
CWA, publicly owned water treatment (POWT) plants have increased in number, resulting in
decreased biological oxygen demand (BOD) in surface waters across the country (Najjum 2009).
Increasing POWT plants reduced the number of point source pollution discharges to surface
waters and subsequently reduced nutrient and pathogen loads.

Separate from the CWA but equally important for Great Lakes protection were the International
Joint Commission, the Great Lakes Water Quality Agreement, and the Great Lakes Basin

9

Compact. These entities offered additional protection for the Great Lakes water and ecosystem at
a regional/basin wide scale, as described below.

The International Joint Commission (IJC) was established under the Canada-U.S. Boundary
Waters Treaty of 1909 to resolve water quality disputes between the two nations (IJC 2005). The
IJC was tasked with informing each national government of emerging Great Lakes issues that
have potential to escalate into disputes and providing recommendations on projects that may
alter the flow and levels of boundary waters (IJC 2005). In 1914, the IJC undertook an extensive
bacteriological survey of the Great Lakes to identify causation and sources of pollution in
boundary waters and to identify remediation actions for water quality improvement (Durfee and
Bagley 1997; Dreelin 2008). This survey identified sewage entering surface waters, crossing
geographic boundaries, and causing human illnesses (IJC 2008). Unfortunately, the results from
this study were never incorporated to policy and Great Lakes pollution was not addressed until
later in the century. Following World War II, the IJC again addressed stark pollution concerns in
the Great Lakes and eventually formed a binational agreement focused on improving the Great
Lakes water quality (IJC 2008). Since its inception, the IJC has worked on over 120 international
requests for boundary water applications (Clamens 2005). Over the last 100 years, the IJC has
been instrumental in protecting the transboundary waters of the Great Lakes through research,
intervention, and neutral political recommendations.

The Great Lakes Basin Compact is a regional organization representative of the eight US States
(Public Law 90-419) and 2 Canadian Provinces bordering the Great Lakes. The Compact is led
by a board of directors and commissioners composed of US representatives and associated

10

members from Canada. The Compact, originally established in 1955 between the US and
Canada, was intended to:


Direct the development and conservation of the Basin



Provide a Basin wide development plan



Provide Basin residents with lakes access and all of their associated benefits to the
maximum possible extent



Advise on the appropriate level of each beneficial use to maintain proper balance
between all uses



Establish a governmental organization to meet the Compact purposes and goals.

Later, the Great Lakes Compact (Great Lakes-St. Lawrence River Basin Water Resources
Compact) was created between the eight Great Lakes states and signed into law in 2008. This US
interstate Compact addresses water use, withdrawal, and management of the Great Lakes; its
impact will be measured over time as demand for clean water increases with national population
growth.

The Great Lakes Water Quality Agreement was first signed by the United States and Canada in
1972 and affirmed each countries commitment to protecting and enhancing the Great Lakes. The
Great Lakes Water Quality Agreement initially focused on phosphorus and toxic substance
reduction and aimed to improve the Great Lakes by taking an ecosystem based approach
(Krantzberg 2007). The Great Lakes Water Quality Agreement created a Great Lakes Water
Quality Board and a Research Advisory Board which advices the International Joint Commission
on progress of the Agreement’s objectives and programs. Under the Great Lakes Water Quality

11

Agreement, the IJC was tasked with examining land use impacts on water quality and analyzing,
assessing, and reporting water quality information to the United States and Canadian
governments every two years (IJC 2005). Since the inception of the original Agreement, the
Great Lakes have seen a reduction in phosphorus and toxic substance loads (Botts and Muldoon
2005). This critical binational agreement bridged political boundaries between Canada and the
United States addressed transboundary pollution and put in place measures that will protect the
Great Lakes for generations.

Water quality policy in the 1800’s was largely nonexistent and water quality degradation resulted
in waterborne diseases, beach closures, and new stressors for the Great Lakes. Slowly water
quality was protected and improved through a series of federal policy, epidemiological studies,
water criteria and standards, and local authoritative associations. Most notably, the CWA
improved water quality and protection by designating standards, funding surveillance, and
establishing pollution reduction measures. Great Lake’s water has been further protected by the
IJC, the Great Lakes Water Quality Agreement, and the Great Lakes Basin Compact.
Recreational health was addressed by the BEACH Act which significantly increased water
quality monitoring through annual appropriations. Surface waters will continue to improve as
long as political protection, science, and technology continue to evolve.

1.2. Indicators of Water Quality
1.2.1. Water quality and health studies
It is known that exposure to poor water quality at recreational beaches can result in acute human
illnesses. Between 1997 and 2006, 100 outbreaks and 3,021 cases of illness (e.g. gastrointestinal,
12

skin irritation, respiratory/ear/eye infection) were associated with ambient recreational waters of
the United States (Barwick et al. 2000; Lee et al. 2002; Yoder et al. 2004; Dziuban et al. 2006;
Yoder et al., 2008). Fleisher et al. (1998) reported that only up to 22.2% of bathers seek medical
attention for illness associated with contaminated recreational water and symptoms remain for up
to eight days.

Epidemiological studies performed since the 1950’s provided vital information for recreational
water quality criteria development. Stevenson (1953) first described that illness incidences
occurred more frequently in swimmers than in non-swimmers and suggested a fecal coliform
criterion. In 1982 and 1984, three epidemiological studies were undertaken in the United States
that investigated gastrointestinal illness rates in individuals at recreational beaches. Cabelli et al.
(1982) performed studies at marine bathing beaches in New York, Massachusetts, and Louisiana.
Cabelli et al. (1982) reported enterococci densities were related to gastroenteritis. Dufour et al.
(1984) performed epidemiological studies in Lake Erie (Pennsylvania) and Keystone Lake
(Oklahoma). They reported enterococci and E. coli densities were statistically related to
gastroenteritis and noted E. coli had a slightly greater correlation coefficient. All of these studies
illustrated the most commonly reported illnesses were associated with infections of the eyes,
ears, and the upper respiratory system (Favero 1985). These early epidemiological studies
established a foundation for future studies and protective health criteria.

Reviews of the multiple recreational epidemiological studies by Prüss (1998), Wade et al.
(2003), Zmirou et al. (2003), and Wade et al. (2006) found strongest correlations between the GI
illness and predictors (enterococci and E. coli) which supported the EPA’s total body contact
criteria (USEPA 2009a) below which no illness could be observed. Additional studies identified
13

young swimmers (children) were at an increased risk of illness compared to adult swimmers due
to less developed immune systems (Wade et al. 2008; Parkins et al. 2003; White and Fenner
1994). Associated risks with waters impacted by different fecal sources (e.g. animals, humans,
etc.) were not explicitly assessed during early studies, but are a concern for scientists and public
health officials. Calderon et al. (1991) was unable to identify a statistical difference between
risks of swimming in waters impacted by human versus animal fecal material. Pruss (1998)
identified water quality measured by bacterial indicators and exposure caused gastrointestinal
illness symptoms regardless of the apparent source. Using new molecular technology, a more
recent study conducted at Great Lakes beaches suggests a positive association between
enterococcus using rapid DNA detection and GI illness (Wade et al. 2006). This study also
reported time of day and time spent swimming (exposure time) increased correlation strength
between enterococci and illness (Wade et al. 2006). Contamination sources however remained
elusive and were seen as an important concern during risk assessment, for restoration and water
quality protection, but the epidemiological studies have improved the connection between water
quality conditions and human health risk.

Guided by epidemiological results, the EPA published recreational water quality criteria for
marine and freshwater. Marine water criteria were set at a single sample maximum of 104
-1

enterococci 100 ml . Freshwater criteria were set at a single sample maximum of 61 enterococci
100 ml

-1

and 235 E. coli 100 ml

-1

(USEPA 1986; Wade et al. 2008). Recreational water quality

criteria were proposed to states as a suggestion for protection of human health.

14

Each state is responsible for developing and adopting standards for “swimmable” waters under
the Clean Water Act. These standards must be as protective, based on the risk of illness, as the
EPA criteria. Michigan’s E. coli total body contact standard has been set at a daily maximum
(geometric mean of three individual samples spatial representative of a defined swim area) of
-1

300 colony forming unit (CFU) 100 ml . Michigan’s monthly total body contact standard was
-1

set at 130 CFU 100 ml

as a geometric mean of at least five sample dates. Unlike federal

criteria, State recreational water standards are enforceable and help protect human health from
exposure to contaminated water. In Michigan, beach managers conduct routine beach monitoring
for bacteria to assess water quality conditions during the swim season. Samples are collected in
waist-deep water at least once per week (May - September). Closures or advisories are issued as
a response to potential health risks from fecal indicator bacteria densities that exceed state water
quality standards. A yearly summary starting in 2001of Michigan Great Lakes beach
closures/advisories is presented in Table 1.3. (MDEQ 2007; MDEQ 2011).

Table 1.3. Michigan’s Great Lakes beach closure or advisory days per year.
Year
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011

Closures or advisory days
122
211
140
578
474
886
1568
697
1596
1003
789

15

1.2.2. Fecal indicator bacteria
The most widely used tools for assessing water quality and risk of illness in recreational waters,
as mentioned above, are the fecal indicator bacteria. Fecal indicator bacteria are measured as
surrogates of true waterborne pathogen presence in water (Griffin et al. 2001). Fecal indicators
are generally easier to detect than their pathogen counterpart and their respective detection
methods tend to be cheaper. Optimal indicator species are based on the following premises: 1.
They are present only when fecal pathogens are present; 2. They are present in greater
concentrations than pathogens; 3. They are more resistant to treatment and environmental
surroundings than pathogens; 4. They are easily detected; 5. They represent a specific group of
pathogens and; 6. They are evenly distributed and grow independently in samples (Bonde 1966;
Colford et al. 2007). Indicators are supposed to be present in greater concentrations, thus their
detection often requires a smaller sample volume. Currently no indicator system meets all
criteria.

Natural shortfalls of fecal indicators limit their usefulness in recreational water quality
monitoring. Increased understanding of pathogens, particularly viruses and protozoa,
identification of regrowth potential, and improved laboratory techniques have increased scrutiny
of the use of a single fecal indicator for environmental assessments (Schwab 2007). Survival
rates and regrowth potential of fecal indicator bacteria varies depending on water temperature,
sunlight, nutrient availability, turbidity and specific biological properties of each bacterium
(McLellan et al. 2007). Detection of cultivation based indicator bacteria involves incubation
periods of 18-26 hours and do not identify the source of pollution. Not one organism is capable

16

of showing the whole picture, requiring a combination of approaches (McLellan 2004). For such
reasons, scientists are beginning to use multiple fecal indicator bacteria and viruses.

Contact with polluted waters has been associated with illness for many years, as previously
described, but determining which microbial indicator best represents human health risk is still
under debate. From the 1950s to mid-1980s, the fecal indicator of choice in the US was
coliforms and fecal coliforms in both marine and freshwater (Stevenson 1953). In 1986, E. coli
and enterococci were adopted as the leading fecal indicators for freshwater and marine water,
respectively (USEPA 1986). Additionally, microbial indicators of fecal contamination have
included Clostridium perfringens (Cabelli 1978) and bacteriophages (Lipp et al. 2001). Each
suggested fecal indicator bacteria have strengths and weaknesses as a tool for water quality
assessment. A brief description of each indicator is provided in the following paragraphs.

Escherichia coli
E. coli are a subgroup of the fecal coliforms, gram negative and facultative anaerobic bacteria.
They are commensal in the lower intestine of warm blooded mammals and birds (Winfield and
Groisman 2003). E. coli are distinguishable from other fecal coliforms by their ability to grow at
45°C, an absence of urease, and the presence of B-D glucuronidase which catalyzes B-Dglucopyranosiduronic acid (Toranzos and McFeters 1997). Most strains of E. coli are harmless
but some can cause diseases. These strains include the enterhemorrhagic, enterotoxigenic,
enteropathogenic, enteroinvasive, enteroadherent, and enteroaggregative E. coli (Rice 2003).
Most infectious E. coli are spread via fecal-oral route (Bischoff et al. 2005) such as exposure to

17

fecal matter while swimming in contaminated water. E. coli has been and remains the workhorse
for water quality investigations despite several issues that fail the true fecal indicators premise.

Recent research has questioned the continued use of E. coli as a fecal indicator. The
disadvantages of E. coli include: their ability to grow in the water environment; their presence
only when pathogens are present; and their uneven distribution. E. coli has been shown to
replicate in water outside of its natural host (McLellan et al. 2001; Vital et al. 2008). In regards
to uneven distribution, beach monitoring samples are routinely taken in the waste deep water (i.e.
swimmable water), but waste deep water have significantly different E. coli densities than other
nearshore depths (Whitman and Nevers 2004). E. coli are generally found in lower densities in
deeper waters due to settling and sunlight inactivation (Thupaki et al. 2010). Furthermore,
current cultivation methods require an 18-24 hour incubation time, reducing the effectiveness of
protecting bathers in real-time.

Advantages of using E. coli as a fecal indicator are primarily due to its long term use in
monitoring including during epidemiological studies which correlated E. coli with gastroenteritis
in freshwater, as previously stated. Although E. coli can regrow, Winfield and Groisman (2003)
showed it does not survive as well in these secondary environments compared to inside
mammalian guts. Current methods for E. coli detection are fairly inexpensive and produce MPN
or CFU results (Rompre et al. 2002).

18

Enterococci
Enterococci are gram positive, non-spore forming members of the Streptococci bacteria
commonly found in the feces of warm blooded animals. Enterococci can be distinguished from
Streptococci by their ability to grow in sodium chloride (6.5%), at high pH (9.6), and at 45°C
(Toranzos et al., 2007). There are multiple strains of enterococci, many of which are not harmful
to humans. Compared to E. coli, enterococci are more resistant to chlorination and environmental
stress and persist longer in the environment (Gleeson and Gray 1997). Enterococci have
distinguishing characteristics which allow them to be isolated and used as a fecal indicator in
recreation water. The selective media mEI, uses Indoxyl-B-D-glucoside as a chromogen which
reacts with by B-D-glucosidase to produce a blue halo around positive enterococci colonies
(USEPA 2002).

Enterococcus plays an important role in recreational water quality monitoring, but like E. coli,
have significant shortfalls as a true fecal indicator. Enterococci can regrow in the environment
under favorable conditions (Desmarais et al. 2002). Enterococci concentrations rapidly fluctuated
by as much as 140 MPN 100 ml

-1

in a single minute or by as much as 345 MPN 100 ml

-1

in ten

minutes (Boehm 2007). Furthermore, streptococci are often less numerous than E. coli in human
fecal material (Pipes 1982) which may make them more difficult to detect in surface waters.
Regrowth and lower organism concentrations limited enterococci’s ability to be used as a fecal
indicator of recent fecal contamination.

Despite their shortfalls, using enterococci as a fecal indicator offer key advantages over E. coli.
Kinzelman et al. (2003) reported that when US EPA enterococci and E. coli total body contact
19

threshold levels were applied to beach monitoring more closures were identified using
enterococci than E. coli. Thus, indicating a more protective estimation of water quality compared
to E. coli and a better statistical relationship to human health. Another advantage is that many
groups have identified relationships between enterococci results using culture based methods and
rapid (< 3 hours) methods (Whitman et al. 2010; Byappanahalli et al. 2010; Haugland et al.
2005). These comparisons are crucial for integrating technology with monitoring to lower illness
risks from contaminated recreational water exposure.

Clostridium perfringens
Clostridium perfringens (C. perfringens) are obligate anaerobic, rod shaped, gram positive, spore
forming, and sulfate reducing organisms. They are opportunistic pathogens that produce
enterotoxins (Gleeson and Gray 1997). The spores do not regrow in the environment and are
resistant to high temperatures and disinfection treatments (Payment and Franco 1993). C.
perfringens are found in sewage and highly impacted waters (Lisle et al. 2004). Detection is
accomplished using phenolphthalein diphosphate which reacts with an acid phosphate enzyme
elaborated by C. perfringens. After the bacterium is exposed to ammonium hydroxide fumes, the
diphosphate bond is cleaved and the reaction becomes visible by the absorption of Indoxyl-B-DGlucoside (Sartory 1986; Bisson and Cabelli 1979).

Similar to E. coli and enterococci, C. perfringens has advantages and disadvantages as a fecal
indicator. C. perfringens spores can be an index parameter for persistent intestinal pathogens
(e.g. viruses and oocysts of protozoa) (Fujioka and Shizumura 1985). Since the spores are highly
resistant to disinfection (Cabelli 1978), suitable applications include the assessment of:

20

chlorinated waters; industrial waters which contain compounds lethal to non-spore forming
bacterial indicators; samples that cannot be processed within 12 hours; and the detection long
term inputs of fecal pollution. Conversely, C. perfringens are often found in low concentrations
(Garrido-Perez et al. 2008) making them more difficult to detect and the spores may provide a
far too conservative estimate for protecting human health. Furthermore, there has not been much
application of C. perfringens as a fecal indicator in recreational waters, except in Hawaii where
traditional fecal indicators are ubiquitous (Mahin and Pancorbo 1999), requiring significant
investment prior to their application in monitoring plans.

Viruses
Bacteriophages are viruses that infect only bacteria while the bacteriophage coliphage
specifically infects E. coli. Coliphage have a finite life, do not regrow in the environment, and
physically resembles many enteric viruses (Havelaar 1987). Two coliphage types (T and F),
defined by infection mechanisms, have been widely used in the water quality field (Guzman et
al. 2008; Donnison and Ross 1995). Coliphage serve a unique role in water quality monitoring
for potential enteric pathogens and show promise as a fecal indicator. Phages can be measured as
a lytic unit in a lawn of their respective bacterial host grown in tryptic soy agar (TSA) using a
double agar overlay according to EPA method 1601 (USEPA 2001) and reported as plaque
-1

forming units (PFU) 100 ml . Similar to other fecal indicators, current method for detecting
coliphage require 24 hour incubation to produce results.

Coliphage have been employed as fecal indicators during recreational water quality monitoring
with some success. Brion et al. (2002) identified one serotype of F+RNA coliphage was related
21

to untreated human fecal material. Another study identified coliphage as an indicator of
noroviruses in freshwater exhibiting similar seasonal variation, propensity for removal and
resistance to environmental stress (Allwood et al. 2003). Whitman et al. (2008) used FRNA
coliphage in a Great Lakes study and found their concentrations increased during wet weather
summer periods’, suggesting fecal contamination was impacting water quality following heavy
rain events. Viruses are increasingly being implicated in water related outbreaks (Hlavsa et al.
2011) and using coliphage as a virus indicator is becoming increasingly important to reduce the
risk of exposure to contaminated water.

Despite the promising results supporting coliphage as a fecal indicator, there are disadvantages to
its application including non-human sewage specific occurrence (Allwood et al. 2003).
Additionally, coliphage results are not consistent across studies (Ashbolt et al. 2001), leaving
their meaning and future use questionable. Ashbolt et al. (2001) further remark that such
differences are likely the results of inconsistent techniques, temperature, pH, and original
densities of coliphage and bacteria host. As methods become more standardized, coliphage
shortfalls may be improved upon and their use in water quality monitoring may grow.

1.2.3. Molecular Source Tracking
The inability of fecal indicator bacteria to identify pollution sources has emerged as a significant
gap in water quality microbiology, leading to the development of molecular source tracking.
Molecular source tracking (MST) is a field of study that seeks to identify the origin of fecal
waste. Current methods quantify species specific gene targets generally using Polymerase Chain
Reaction (PCR) approaches (Scott et al. 2002), known as library-independent host specific

22

source tracking. Host specific source markers are specific DNA sequences that can be extracted
from water and assayed with PCR to indicate the presence of feces from a single species.
Advantages of using PCR based MST over cultivation methods include a high sensitivity and
specificity, quicker result outputs, and in most methods, a more automated processes (Girones et
al. 2010). Current molecular approaches do no distinguish between viable and non-viable
organisms, producing mixed interpretations of molecular results for human health risk (Girones
et al. 2010). Source tracking has gained popularity throughout the Great Lakes region as fecal
source identification becomes more critical for maintaining or improving water quality.

Library-independent source tracking methods involve the detection of one species using one or
more of the following: chemicals, sterols, viruses, and bacterial genes (Santo Domingo et al.
2007). In the case of bacterial genes, the DNA sequences first are identified and then undergo
validation testing to evaluate the specificity of the sequence as unique to a particular species
(Walters and Field 2006), but no marker has been or is likely to be completely validated. Water
samples are assayed for specific source markers, generally through a non-culture based method
(Santo Domingo et al. 2007) and DNA amplification through polymerase chain reactions (PCR).
The library-independent method can use conventional or real-time PCR (quantitative method) to
detect the specific gene sequence. Library-independent methods return less false positive and
false negative results than the library dependent method (Griffith et al. 2003). Originally, few
specific markers were available (Field and Samadpour 2007) and these markers were not always
present in large quantities throughout the environment (Scott et al. 2005). However the field is
expanding dramatically. Discussed below are two popular MST markers for recreational water
quality monitoring.

23

One bacterial group that has become widely used and reported in water quality analysis is
Bacteroides. Bacteroides (Order Bacteroidales) are anaerobic bacteria found in high
concentration throughout the intestinal tract and feces (Field et al. 2003) of most animals.
Bacteroides are anaerobic and thus do not survive long in surface waters, but using PCR method
their presence can be detected in water for long periods of time (Okabe and Shimazu 2007). Bell
et al. (2009) suggests microorganism predation and temperature result in the greatest reduction of
Bacteroides (AllBac) marker densities in water. Balleste and Blanch (2010) confirmed that
temperature was a significant factor in Bacteroides DNA degradation. Dick et al. (2010) used
microcosm experiments to expose seeded water to varying simulated conditions and concluded
the BacHum source marker decayed faster, relative to cultivated E. coli, under sunlight,
sediment, and decreased predation treatments. All of these studies have predicted that the signal
can last in the environment at least 200 days. Due in part to molecular methods, the identification
of Bacteroides persistence in surface waters has led to the popular use of Bacteroides as a marker
of fecal contamination in water.

Water quality investigations are increasingly turning to molecular detection of Bacteroides for
improved understanding of water environments. Bernhard and Field (2000a, 2000b) identified
unique gene sequences (Bac32F, Bac708R, Bac303R) highly specific to human and ruminant
feces. Ahmed et al. (2008) confirmed two other markers (HF 183 and HF 134) were highly
specific to humans and nearly absent in a multitude of animals. Ahmed et al. (2009) used the
HF183 to differentiate human sewage and animal waste and found the marker had a host
specificity of 98%. Another marker, B. thetaiotaomicron, α-1-6 mannanase, described by

24

Yampara-Iquise et al. (2008) was found to be a good indicator of sewage impacted waters with a
low method detection limit (9.3 copies per reaction). Additionally, Srinivasan et al. (2011) found
significant correlations between this marker and traditional cultivation based E. coli and
enterococci concentrations throughout the wastewater treatment processes. Reischer et al. (2008)
applied a Bacteroides marker (BacR) to determine that ruminants were the leading cause of fecal
contamination in a large catchment. Additional Bacteroides source tracking research has targeted
specific gene markers including gulls (Jeter et al. 2009), pigs (Gourmelon et al. 2007), horses
(Dick et al. 2005), muskrat (Marti et al. 2011), and dogs (Kildare et al. 2007). Table 1.4.
describes popular MST sequences and their respective targeted organism. Cumulatively, these
studies represent the common development of the Bacteroides marker, each one developing,
refining, and applying molecular source tracking methods to improve our understanding of water
quality pollution in complex watersheds.

25

Table 1.4. Sequences and targeted organisms for select molecular source tracking markers.
Marker
ID

Gene sequence (5’-3’)

Target microorganism
(Specificity)

Reference

Bac32F

AACGCTAGCTACAGGCTT

Bacteroides-Prevotella
(General)

Bernhard and
Field 2000b

Bac303R

CCAATGTGGGGGACCTTC

Bac708R

CAATCGGAGTTCTTCGTG

HF183

ATCATGAGTTCACATGTCCG

BacR

GCGTATCCAACCTTCCCG

AllBac

GAGAGGAAGGTCCCCCAC

Bacteroides-Prevotella
(Cow)
Bacteroides-Prevotella
(Human)
Bacteroidales species
(Human)
Bacteroidales
(Ruminant)
Bacteroidales
(General)

Bernhard and
Field 2000a
Bernhard and
Field 2000b
Bernhard and
Field 2000a
Reischer et al.
2006
Layton et al.
2006

BacHum

TGAGTTCACATGTCCGCATGA

Bacteroidales
(Human)

Kildare et al.
2007

BtH

CATCGTTCGTCAGCAGTAACA
GTAATTGCTACACCTGCTGAA
ACCACTGTCCCT TTTTCTTGG

Bacteroides
thetaiotaomicron α-1-6,
mannanase (Human)

YamparaIquise et al.
2008

The Bacteroides have offered a substantial benefit to water quality investigators by targeting
specific source(s) of fecal contamination; however, there are disadvantages to their use in beach
monitoring. The Bac32F, Bac708R, Bac303R were shown to have little or no correlation with
fecal coliforms, E. coli, or enterococci in river water samples (McQuaig et al. 2006). Drawing
from this study, it is uncertain whether human health risk and Bacteroides levels are related. In
order to achieve an understanding of human health risk in recreational water, Bacteroides must
be monitored in conjunction with a health based fecal indicator in specialized studies as
suggested by Rose et al. (1997). This would add significant cost to beach monitoring and regress
from the concept of water quality monitoring using an inexpensive indicator. Improving

26

analytical methods will allow for increased use of Bacteroides during routine monitoring, but
until then their use remains limited.

Another MST marker is the enterococci surface protein (esp) gene. Esp is a potential virulence
gene involved in biofilm formation and possibly involved in intestinal colonization (Heikens et
al. 2007). The protein gene marker has been suggested as a useful tool for human fecal source
tracking (Scott et al. 2005). This reportedly human-specific marker has been consistently
detected in sewage and septage and inconsistently detected in animal feces (Ahmed et al. 2008,
Whitman et al. 2007). Given the ability of esp to colonize intestinal tracts and cause
gastrointestinal illness, the esp gene may be used to improve water quality and human health
protection.

Supporters of the esp gene as a human fecal marker claim a high degree of human specificity
which accurately describes human fecal contamination. Kim et al. 2010 claimed the E. faecium
esp gene had a specificity of 100%. The same group also assayed 237 enterococcus species
isolates from thirty-four human, chicken, pig, cow, and goose samples and found significant
genetic differences between human and animals. The esp gene was absent in all tested animals
by Scott et al. (2005) and Masago et al. (2011). Additionally, the marker was found in all or
nearly all sewage samples (Scott et al. 2005, Ahmed et al. 2008) and in 67-97% of septic samples
(Ahmed et al. 2008, Scott et al. 2005). With the esp, false negatives may be potentially
problematic (Goodwin et al. 2008).

27

Acceptance of esp as an indicator of human fecal contamination has been received with some
skepticism. Byappanahalli et al. (2008) concluded the esp gene was inconsistent and did not
accurately distinguish between animal and human sources. Furthermore, they argue the esp
marker has inconsistent findings between studies and a low specificity between animal and
human fecal sources. These conclusions were based on a study which detected the esp gene in
29% of seagulls, 14% mice, and 9% songbirds by swabbing fecal material then diluting and
processing (Byappanahalli et al. 2008). As new markers are developed, it is important that they
are critically reviewed and tested to ensure they accurately describe the source of contamination.
The esp gene may be a reliable indicator of human fecal contamination but first a standard
method is required to produce consistent results across all studies.

Fecal indicator bacteria have provided information on the wide spread and continued pollution of
the Great Lakes for over half a century. Improved scientific and molecular technology has
allowed for increased accuracy, precision, and source identification in the water quality field.
Although much work is still required to improve and fully understand the sources of fecal
contamination in the Great Lake and health risks, the foundation has been laid for long term
improvements. Sewage has often been the focus of pollution studies, but nonpoint sources (e.g.
contaminated urban, agriculture, or industrial runoff, atmospheric deposition, or seepage from
contaminated subsurface sites), have been a challenge to identify. While the use of geographic
mapping of nonpoint source pollution and land use has improved the understanding of water
quality degradation in the Great Lakes, the need for further research is still of great interest.

28

1.3. Nonpoint Source, Land, and Weather Impacts on Water Quality
Implementation and enforcement of the CWA for the past 40 years has provided considerable
improvements in the management of point source discharges, especially NPDES, CSOs, and
SSOs. As improvements continue to reduce contamination from identified sources, regulators
and stakeholders in the Great Lakes have begun the arduous task of addressing nonpoint sources
of pollution. Nonpoint source pollutants are now the leading cause of impaired waters (USEPA
2009b). One type of nonpoint source pollution is contaminated runoff or the overland flow of
water transporting pollution from the landscape to surface waters. Land use and land cover
(LULC) management can alter the natural percolation and runoff patterns along with the
movement of pollutants in water. Dreelin et al. (2007) stated land use decisions affect the sources
and transport of pathogens into environments. Understanding LULC and influences on
hydrology and nonpoint source pollutants is vital to protecting water quality in the Great Lakes,
but there are also key pollutant reservoirs such as sediment and algae. Sediment resuspension,
algae accumulation in the littoral zone, and contaminated runoff are among the most problematic
in the Great Lakes, generating significant concern for water quality managers (Kinzelman et al.
2004), but they do not have well-defined impacts on public health (Verhougstraete et al. 2010).

1.3.1. Algae
Algae mats act as a significant reservoir of pollutants for nearshore zones. Many studies have
discovered high levels of fecal indicator bacteria in nearshore algae mats (Garrido-Perez et al.
2008; Whitman and Nevers 2003; Englebert et al. 2008a; Englebert et al. 2008b; Ishii et al. 2006;
Olapade et al. 2006). Bacteria may regrow in algae and can influence surrounding water quality
when disturbed. This water quality change occurs in the absence of recent fecal contamination,

29

potentially generating false positive results regarding risk during beach monitoring. Science is
progressing to identify the potential risk posed by algae presence in the nearshore area on human
health and recreational water quality monitoring. Algae samples collected around the Great
Lakes have confirmed the presence of elevated fecal indicator bacteria in the mats (Englebert et
al. 2008b; Byappanahalli et al. 2007; Olapade et al. 2006; Ksoll et al. 2007; Whitman et al.
2003). In addition to fecal indicator bacteria, human pathogens, specifically Shiga Toxin
producing E. coli (STEC) and Shigella (Ishii et al. 2006a), Salmonella (Byappanahalli et al.
2009; Ishii et al. 2006), Clostridium botulinum type E (Byappanahalli and Whitman 2009), and
Campylobacter (Ishii et al. 2006) have been detected in algae. The microorganisms identified in
algae, the respective Great Lakes studies, and the detected concentrations are presented in Table
1.5. (Verhougstraete et al. 2010).

30

Table 1.5. Studies and the microorganisms identified in Cladophora collected within the Great Lakes*.
Study
E. coli
Enterococci
Salmonella
Campylobacter Shigella
Byappanahalli et al.
2007

2.55-3.09 log CFU/g
Laboratory

A

Byappanahalli et al.
2009
Whitman et al. 2003

0.16-89.46
A
MPN/g
Env.
5.3 ± 4.8 log CFU/g

A

Env.
Byappanahalli et al.
2003a

1.3-4.3 log CFU/ml
Laboratory

Englebert et al. 2008a

Englebert et al. 2008b
Olapade et al. 2006
Ishii et al. 2006
Ksoll et al. 2007

STEC

8.0-9.0 log CFU/ml
Microcosm
2.9-3.4 log MPN/100 ml
Env.
38,000 CFU/100 g
Env.

4.8 ± 4.5 log
A
CFU/g
Env.
1.3-3.3 log
CFU/ml
Laboratory
5.0-6.0 log
CFU/ml
Microcosm

7.0-8.0 log
CFU/ml
Microcosm

B

5.8 ± 5.5 log MPN/g
Env.

A

1.5 log MPN/g
Env.

2

A

39 cells/g
Env.

A

C

C

Env.

Env.

294,000 CFU/cm
Microcosm
A: dry weight; B: wet weight; C: organisms detected but not quantified. Laboratory: Bacterial analysis performed under laboratory
settings; Microcosm: Bacterial analysis performed under simulated conditions on a minute scale; Env. (environmental): Bacterial
analysis performed on natural samples; *Reproduced from Verhougstraete, M.P., Byappanahalli, M.N., Rose, J.B., and Whitman, R L.
(2010). ‘Cladophora in the Great Lakes: Impacts on beach water quality and human health’. Water Science and Technology, 62 (1), 68-76,
with permission from the copyright holders, IWA Publishing.
31

Bacteria not only accumulate in algae mats they likely persist longer in algae mats than in
surrounding water and can regrow. E. coli was shown to survive less than four days in lake water
(Olapade et al. 2006) but as long as 45 days in algae mats (Englebert et al. 2008a). Many theories
have been proposed for this occurrence including increased attachment points for bacteria in
algae mats (Brettar and Höfle 1992; Weinbauer and Höfle 1998; Signoretto et al. 2004),
increased substrates vital for bacterial growth in algae leachate (Byappanahalli et al. 2003a), and
increased protection from sunlight in algae mats thicker than 6mm (Whitman et al. 2003).
Additionally, Whitman and Byappanahalli (unpublished) demonstrated that E. coli can survive in
dried algal mats stored at 4 °C for over two years. Salmonella was shown to persist for 10 days
in algae manipulated in a laboratory microcosm environment (Englebert et al. 2008a). Englebert
et al. (2008a) showed, via viability tests, that Shigella survived up to 2 days in an algae
microcosm. The fate of fecal indicator bacteria in algae mats is becoming increasingly
understood, the next step is to identify how presence and regrowth in algae mats will impact
nearshore water quality.

Algae impact on nearshore water quality is not as clear as bacteria accumulation, persistence, and
regrowth in algae. Two studies have reported contrasting conclusions regarding the impact of E.
coli found in algae and water quality. Whitman et al. (2003) demonstrated floating algae mats
have the ability to release bacteria to surrounding waters and influence water quality when
present in the nearshore. In contrast, Englebert et al. (2008b) and Olapade at al. (2006) report
that water quality was not significantly influenced by the presence of algae, regardless of E. coli
concentrations in the algal mats or amount of algae material present. E. coli adherence to algae
material was suggested as the main reason that the impacts to water quality were minimal

32

(Englebert et al. 2008b) and it was demonstrated by Whitman et al. (2003) that algae required
multiple washes to remove nearly all E. coli cells. Despite the contrast in conclusions from these
cases, each identified bacteria in algae may be released to surrounding waters as a result of
environmental conditions. Whitman et al. (2003) identified that bacteria were released from
floating mats as a result of wave action while Englebert et al. (2008b) noted strong wind and
wave influences were required to dislodge the bacteria from algae mats. The contrast in
conclusions from each of these cases shows the highly variable interactions between algae
amounts and water quality that cannot be captured through a single study.

1.3.2. Sediment
For decades, scientists have known that sediment harbors chemicals, toxins, and nutrients which
have the potential to impact overlying water (Smart and Barko 1978; Mortimer 1971; MarvinDiPasquale and Agee 2003). The relationship between bacteria and sediment or the impact of
such bacteria on surrounding water quality is not as clear. But bacteria can accumulate, regrow
and potentially be released to the overlying water. Standards for bacteria analysis in sediments
do not exist which have resulted in multiple reporting formats including bacteria numbers per
gram of wet weight, gram of dry weight, or milliliters. Such reporting inconsistencies cannot
easily be compared across studies as illustrated in the summary below.

Fecal indicator bacteria concentrations observed in beach sand and sediment are often higher
than in surrounding water. Zehms et al. (2008) found E. coli concentrations as high as 21,670
CFU 100 g

-1

dry weight in the sand at one GL beach. Whitman and Nevers (2003) reported
-1

nearshore sand had higher E. coli concentrations (4000 CFU 100 ml ) than the surrounding
33

-1

water (43 CFU 100 ml ). Alm et al. (2003) found E. coli was up to 38 times higher in beach
sand than the water column and E. coli concentrations decreased with sediment depth. GarridoPerez et al. (2008) found proximity to contamination source has a significant influence on
sediment bacterial densities. They also reported beaches with low energy and circulation had a
high accumulation of sediment and high bacterial concentrations (Garrido-Perez et al. 2008). The
E. coli density range in Great Lakes sediment studies are provided in Table 1.6. Cumulatively
these results indicate sediment exhibit a higher density of fecal indicator bacteria compared to
surrounding waters.

Table 1.6. E. coli density ranges identified in Great Lakes sediment.
Study

E. coli density ranges

Alm et al. 2003

12-80 CFU GDW

Byappanahalli et al. 2003b

1-119 MPN GDW

Kinzelman et al. 2004

0-20000 CFU GDW

Ge et al. 2010

10-10 CFU 100 ml

Whitman and Nevers 2003

10 -10 CFU 100 ml

Byappanahalli et al. 2006

1-1657 MPN 100 GDW

Zehms et al. 2008

1800-21670 CFU GDW

5

2

6

-1
-1
-1

-1
-1
-1
-1

GDW: Gram dry weight

In addition to bacterial storage, sediments can provide a suitable habitat for some bacteria, and
potentially MST markers, to persist and/or regrow. Kinzelman et al. (2004) detected a relatively
low number of E. coli clonal patterns isolates from Lake Michigan beach sediment, indicating E.
coli was accumulating in the sediment and not regrowing. LaLiberte and Grimes (1982) used
34

reduction rates in sediment to conclude bacteria detected in surface waters may not always be the
result of fresh fecal contamination. Craig et al. (2002) found enterococci, E. coli, and coliphage
all persisted longer in sediment than in water. One group removed and replaced contaminated
sand at a Chicago beach with E. coli free sand and found that E. coli concentrations in the sand
returned to historical levels within two weeks (Whitman and Nevers 2003). Alm et al. (2006)
5

-1

found E. coli densities increased fivefold in two days and remained above 2 X 10 CFU g

wet

weight for 35 days in an in situ experiment. Garzio-Hadzick et al. (2010) found faster E. coli
decay rates in water than in sediment, regardless of temperature. In the same study, undertaken in
5

-1

a field experiment, E. coli reached a density of 7.5 X 10 CFU g for at least 48 days. Yamahara
et al. (2007; 2009) showed enterococci regrew in wetted beach sand and doubled in as little as
1.1 days in sediment exposed to diurnal tidal wetting. This same study identified the origin of
enterococci contamination in sediment was human based on the presence of the esp gene
(Yamahara et al. 2007). It was also demonstrated that moisture and sand temperature had a
significant impact on enterococci and E. coli decay rates (Mika et al. 2009).

The fate of molecular source tracking markers in sediment is not well understood. In fact, only
one study has attempted to investigate the persistence of source markers in water while taking
into account their presence in sediments (Dick et al. 2010). This study used Bacteroides markers
in microcosms controlled for light, sediment exposure, temperature, and predation. The authors
found, amongst other findings, that the decay rate of source markers in river water exposed to
sediments was slower than the decay rate of the same markers without sediment. One important
finding of this investigation was that the AllBac markers had a longer 2-log reduction time in the
sediment exposed water than any other condition or marker. Cumulatively, these studies
35

implicate sediment as a bacterial sink which encourage bacterial persistence and growth. The
accumulation and growth of bacteria can have significant impacts on surface waters, as described
below.

Fecal indicator bacteria present in nearshore sediment can result in significant water quality
impairments during resuspension. Whitman and Nevers (2003) found lake water quality is
significantly influenced by foreshore sand and may raise health concerns even in the absence of
recent human or animal fecal material. LeFevre and Lewis (2003) and Kinzelman et al. (2004)
identified fecal indicator bacteria accumulated in sediment significantly contributed to
surrounding water quality as a result of wave action. More specifically, wave action in the
nearshore swash zone was used to explain E. coli concentrations in knee deep water at a Lake
Michigan beach (Ge et al. 2010). Increased turbidity was shown to increase the survival of E.
coli (Pote et al. 2009; Garcia-Armisen and Servais 2009), again implicating sediment
resuspension as a source of impairment for water quality. Many studies have identified latent
sediments impact overlaying waters and sediments are now considered one of the primary
sources of bacteria in nearshore waters.

1.3.3. Land-water interface
The surface water hydrologic cycle involves multiple land systems and parameters that influence
water quality. A watershed is the total land area that drains to one point in a water body and is
often defined by ridges/drainage divides that define the direction of overland flows (Kalff 2002).
Within a watershed, overland flow (from precipitation) drains to a river, lake, or wetland. Rivers
can be gaining (groundwater percolates into river), losing (river water percolates into

36

groundwater), or intermediate (no continuous source of water) and are composed of headwaters
(groundwater and snow/glacier melt), tributaries, and confluences of tributaries before the mouth
of the river. Lakes are non-ocean bodies of water that exhibit little to no horizontal movement.
The shallow, nearshore zones of lakes, including beaches, are referred to as the littoral zones, the
open waters of lakes are referred to as the limnetic zone, and the lake bottom is referred to as the
profundal zone. The transport of fecal indicators between the land and water in these zones is of
scientific interest regarding links between pollution sources and impact.

Land use changes impact pollutant transport by modifying the surface water hydrologic cycle.
Water movement continuously resuspends and deposit sediments and pollutants throughout
hydrologic systems. Overland flow rates are determined by the type of the landscape (Lavee and
Poesen 1991), precipitation rates and land roughness (Katz et al. 1995), and vegetative cover
(Loch 2000). Modification to any of these parameters presents additional pollutant sources and
sinks and movement in surface waters. Walters et al. (2011) found E. coli, enterococci, and
pathogens (Salmonella and E. coli O157:H7) presence in waters change with land use type and
are magnified by rainfall. Landscape changes influence natural percolation/infiltration of
precipitation/melt off, causing changes to stream velocity and transport rates of pollutants to
surface waters (Desai et al. 2010; Allan 2004). Patz et al. (2000) reviewed the application of land
use changes on pathogen and diseases, noting forest removal increases habitat fragmentation and
allows for exchange and transmission of pathogens to new areas, including into surface waters.
Changes to land use create new sources of pollution, including bacterial sources, which modify
the transport and quality of surface water, but investigative scales remains uncertain.

37

Watershed and LULC investigations have traditionally been costly and required large spatial and
temporal scales. The current trend in ecosystem research is to investigate processes at multiple
spatial scales (Soranno et al. 2010; Chang 2008) which often require 3-5 years (Spooner and
Line 1993). Desai et al. (2010) described the need for longer time scale requirements in order to
identify bacterial (E. coli) changes in water quality resulting from LULC changes in the
watershed. Mehaffey et al. (2005) linked fecal coliform bacteria to urban and agriculture land
use, noting land use location relative to the surface water was more significant than watershed
percentages. Furthermore, the authors concluded large area watersheds can be evaluated for fecal
coliform bacteria using simple geographic information system (GIS) approaches (Mehaffey et al.
2005).

Landscape ecology is emerging in the water quality field as a useful concept that links spatial
terrestrial patterns (LULC) patterns and ecological processes. Using global information systems
(GIS), connections between LULC and water quality have been identified, often focusing on
physical parameters, nutrients, chemicals, and biological indices (e.g. fish and macrophytes) as
the metric of water quality (Wang and Yin 1997; Broussard and Turner 2009; Akasaka et al.
2010). For instance, Mattikalli and Richards (1996) found elevated nitrogen concentrations in
surface waters were related to changes in land use and fertilizer applications over five decades.
Wang et al. (2001) found significant connections with urbanized development proximity to
streams and fish communities. LeBlanc (1997) created a decision support tool that predicts the
impact of land use changes on surface water temperatures in lotic systems. Linking water quality
to landscape patterns has improved scientific understanding of environmental processes and

38

implications on aquatic health, but the connection between land use and microbial densities in
water remains unclear.

Few studies have successfully linked bacterial water quality to a specific land use type. A brief
summary of studies attempting to connect LULC with microbial water quality is provided in
Table 1.7. A few studies identified high concentrations of E. coli were characteristic of source
sheds dominated by urban landscapes (Mallin et al. 2000; Desai and Rifai 2010; Desai et al.
2010; Wu et al. 2011). Kang et al. (2010) linked bacteria (E. coli and enterococci) to urban and
industrial land uses. Desai et al. (2010) further noted that E. coli concentrations were highest in
-2

stream segments with high percent land development (40%), population density (1996 km ),
-2

dog density (513 km ), and low grassland percentage (20%). Similarly, Goto and Yan (2011)
found a distinct urban land use pattern influence on E. coli, enterococci, and C. perfringens
concentrations in Hawaii. Another study in Hawaii identified direct associations between
Salmonella and agriculture/forested land cover while Staphylococcus aureus was directly
associated with urban/agriculture land cover (Viau et al. 2011). Mehaffey et al. (2005) directly
linked fecal indicator bacteria to both agriculture and urban percent development in a watershed,
but noted correlations were stronger for land use located near the reservoir in the catchment than
total percentage of the catchment. Hunter et al. (1999) found fecal coliform concentrations
increased 100 fold as waters moved through agriculture lands. There is only one known study
that addresses and claims a relation exists between microbial source tracking markers (Bac708
and CF128) to LULC (agriculture) (Kirs et al. 2011). In the previously mentioned studies
bacteria appear to be ubiquitous in the environment influenced by any LULC change in a
watershed. Few studies have successfully linked one microbe to a specific type of land use. This
39

may be a result of the multiple influences not measured including soil characteristics, spatial
implications from source, or weighted calculations for sources closer to surface water.

Table 1.7. Key findings of studies attempting to link land use and microbial water quality.
Study location

Key findings

Reference

Derbyshire,
England

Fecal coliforms and fecal streptococci increased in
agriculture catchments as a result of sediment storage
and hydrological transport from landscape
New York,
Fecal coliforms directly related to percent urban and
USA
agriculture in catchment
North
Fecal coliforms linked to population, percent urban,
Carolina, USA and percent impervious surface in watershed
Oregon, USA Enterococci linked to urban and agriculture land use;
Urban land use change resulted in microbial water
quality exceedances
Yeongsan
E. coli and enterococci are predominately from urban
River, Korea
and industrial land use
Texas, USA
E. coli concentrations were higher, less temporally
variable, and more spatially variable in urban
dominated watersheds compared to grasslands
Oahu, Hawaii E. coli, enterococci, and C. perfringens higher in urban
streams compared to forested streams
Nelson, New
Human (HF183) marker found in stormwater drains;
Zealand
Ruminant (Bac709/CF128) marker in agriculture
catchments
O’ahu, Hawaii Salmonella directly related to forest and agriculture
land cover; Staphylococcus aureus directly related to
urban and agriculture land cover

Hunter et al. 1999

California,
USA

Walters et al. 2011

Ontario,
Canada

E. coli, enterococci, and Salmonella directly related to
percent urban in catchment; E. coli O157:H7 directly
related to percent agriculture in catchment
E. coli O157:H7 related to agriculture density
upstream

Massachusetts, Highest human derived E. coli (ribotyping) found
USA
when urban >30% of catchment; Wildlife sources
dominated when natural lands > 54% of catchment

40

Mehaffey et al.
(2005)
Mallin et al. 2000
Nash et al. 2009

Kang et al. 2010
Desai and Rifai
2010; Desai et al.
2010
Goto and Yan 2011
Kirs et al. 2011

Viau et al. 2011

Wilkes et al. 2011
Wu et al. 2011

1.3.4. Weather
Weather patterns and processes influence surface waters on shorter temporal scales and
groundwater on longer temporal scales (Johnson et al. 2004) and likely drive the movement of
th

pollutants from land to surface waters. Precipitation events above the 90 percentile were linked
to more than half of all waterborne disease outbreaks in the United States, thus implicating the
role of rainfall in the transport of waterborne pathogens (Curriero et al. 2001). Poor water quality
has resulted from contaminated runoff or combined sewer overflow discharges as precipitation
inundates soil or treatment systems (McLellan et al. 2007). In a Florida estuary, Lipp et al.
(2001) linked elevated fecal contamination with variability in precipitation, temperature, and
stream flow. Understanding weather influences on water quality is becoming more important in
the Great Lakes as nonpoint source pollution and the potential association with health risks
emerge.

Precipitation and snow melt have the potential to influence microbial water quality. The
connection between microbes and precipitation has been demonstrated by many studies
throughout the world and in the Great Lakes (Shehane et al. 2005; Cho et al. 2010;
Jayawickreme and Hyndman 2007; Dorner et al. 2007; Scopel et al. 2006). Whitman et al.
(2008) showed E. coli increased 10 and 100 fold following rainfall and snowmelt events,
respectively, in an artificial stream system. Haack et al. (2003) identified a 48-72 hour lag
between precipitation and elevated E. coli concentrations along the Grand Traverse Bay coast.
Another study found positive correlations between 72 hour cumulative rainfall and E. coli,
enterococci, and C. perfringens concentrations in urban dominated catchments (Goto and Yan
2011). Wu et al. (2011) showed water quality standards were exceeded more often following wet
41

weather events. Jamieson et al. (2005) found storm events increased sediment resuspension and
in turn elevated bacterial concentration in a Canadian stream. Furthermore, Nevers et al. (2007)
identified a significant correlation between E. coli and barometric pressure in Lake Michigan.
Collectively, these results demonstrate the impact precipitation has on natural (e.g. stream flow)
and human (e.g. wastewater treatment) systems which led to degraded surface water.

In the Great Lakes, wind direction and speed has also been implicated as a significant influential
factor of fecal indicator bacteria concentrations (Crowther et al. 2001; Haack et al. 2003; Nevers
et al. 2007). The role wind plays on nearshore water is evident causing both wave action and
sediment resuspension. This result has been repeated by the European Union which found
increased winds contributed to increased bacterial concentrations in the surrounding water at
recreational beaches (Roslev et al. 2008). Furthermore, wind (direction and speed) was shown to
dislodge bacteria from Saginaw Bay algae mats and resuspends them into the water column
(Verhougstraete and Rose, in prep.). Beach orientation, as opposed to wind direction, has also
been suggested as a significant environmental parameter effecting nearshore water quality
(Nevers and Whitman 2005; Nevers and Whitman 2008; Wong et al. 2009).

Temperature can affect a microorganism’s ability to survive and grow in water. Pip and Allegro
(2010) showed higher concentrations of total coliform bacteria were associated with increased
temperature in Lake Winnipeg. Wilkes et al. (2011) detected Campylobacter more often when
mean air temperatures were cooler. One study addressed temperature (air and water) on
Bacteroides using molecular and cultivation approaches and found Bacteroides organisms were
detected more often in lower temperatures and high temperatures were associated with greater

42

organism die-off rates (Balleste and Blanch 2010). Schulz and Childers (2011) also identified
Bacteroidales decay rates were faster at higher temperatures using molecular approaches
targeting Bacteroidales 16S rRNA gene. A growing concern in the Great Lakes is a shift in long
term air and water temperature which may present water quality managers with significant
challenges for beach monitoring and human health protection.

In summary, changes to land use land cover alter natural hydrologic process which affect surface
runoff and water quality. Precipitation, wind, and temperature are often implicated as factors
influencing water quality. Furthermore, sediment and algae mats have been shown to harbor
bacteria and even foster regrowth of some fecal indicator bacteria. Increased energy from
precipitation or wind can cause such bacteria to be released to surrounding water. Although
scientists are investigating the implications of nonpoint source pollution on water quality, their
associated human health risks remain unclear. Such associations will require further
investigations at multiple spatial and temporal scales throughout the Great Lakes.

1.4. Scientific Needs
The current state of science for recreational water quality in the Great Lakes primarily rely on
one fecal indicator microbe (E. coli) to inform the public of health risks from swimming in deep
water. A trend in recent years to use multiple parameters, microbes, and source specific
molecular analysis has been suggestive of a greater understanding of ecosystem and human
health risks. Such methods allow scientists to explore important links between fecal indictors and
true pathogens and enteric bacteria. Advancements in the tools available for source identification

43

and survey (microbial, spatial, and sanitarian) efforts are needed to remediate fecal
contamination impacting beaches.

Algae mat accumulation in the nearshore zone has posed significant challenges to beach
managers protecting water quality. A number of pathogens have been identified within
Cladophora mats, however the relationships between algal mat formation and a variety of fecal
indicators, including MST, and water quality along beaches requires further investigation.

Currently, scientists understand that sediment can influence microbial and molecular persistence
in the environment. Bacteria and molecular source markers have been shown to exist in higher
concentrations and survive longer in sediments than in surrounding water. However, the linking
of microbial accumulation and persistence characteristics in environmental sediments to water
quality change at various watershed scales are not well studied particularly given both natural
and anthropogenic influences.

Protection of surface waters requires an understanding of microbial water quality, nonpoint
source pollution, and terrestrial/aquatic processes. Currently there is a significant knowledge gap
between microbial contamination and terrestrial landscape patterns. Landscape ecology uses a
variety of tools and scales to improve upon the knowledge of relationships between landscape
patterns and ecosystem processes, but this approach has not been widely applied to the microbial
water quality/landscape interactions. Scientists need to identify the appropriate landscape scale at
which significant relationships between one or multiple fecal indicators can be inferred.

44

Water quality can be inferred by studying microbial (occurrence, fate, survival, and life cycles)
and environmental (precipitation, wind, solar radiation, etc.) processes. However, further
research focused on the quantification of microbial responses to environmental processes is
required in nearshore waters. The scientific queries presented below expand upon current
knowledge of parameters that influence non-point source pollution in the Great Lakes. These
advancements will feed hydrologic models, address land use/cover impacts on water quality, and
describe ambient water quality conditions useful for measuring changes over time. Developing
process based water quality models will provide a framework for development of more effective
water quality policies and improve the connection between recreational waters and human
health.

1.5. Research Objectives
Goal 1
Multiple sources of pollution threaten the Great Lakes and one waterbody that has become
dangerously threatened is the Saginaw Bay. This ecosystem is threatened by industrial, urban,
and agricultural runoff, wastewater discharge, and algae blooms. Few studies have focused on
microbial water quality in Saginaw Bay. A survey of four Saginaw Bay beaches was undertaken
to quantitatively describe the fecal contamination in beach sediment, stranded algae mats, and
surrounding water using fecal indicator bacteria and source specific DNA markers.
The objectives of this goal were to: 1) determine the occurrence and relationship of fecal
indicator bacteria in water, sediment, and algae at beaches in a mixed use watershed; 2)

45

identify human and bovine sources of fecal contamination using microbial source tracking
and; 3) identify the environmental parameters influencing water quality degradation.

Goal 2
Smaller watershed landscape scales may provide better associations between water and sediment
quality. However, small scale systems may be subject to rapid temporal changes that may not be
identifiable using current methods. Mitchell Creek is a small, flashy system draining mixed use
watershed and discharging to the Grand Traverse Bay (Michigan). Mitchell Creek is a known
contributor of E. coli to the Bay, but the sources of contamination are not well described. Water
and sediment samples were collected over multiple seasons from Mitchell Creek sites
representative of multiple land use types and assayed for fecal indicator bacteria and molecular
source markers.
The objectives were to 1) examine the spatial and temporal distribution of traditional and
alternative fecal indicators in a watershed influenced by non-point source pollution, 2) use
a quantitative PCR marker to measure human sources associated with fecal bacteria, and
3) assess land use patterns effect on bacterial water quality.

Goal 3
“The valley rules the river” is a phrase ecologists have used for decades. Land use has been
reported as a significant factor for non-microbial based water quality measures at multiple scales.
However, land use/cover impacts on microbial water quality parameters, at any scale, is not

46

understood. A broad spatial and temporal survey of bacterial water quality across Michigan was
undertaken to quantitatively describe fecal pollution concentrations entering the Great Lakes
during distinct hydrologic conditions and across multiple landscape scales.
The objectives of this goal were to: 1) Examine the occurrence of Escherichia coli (E. coli)
and a human specific source marker (B. thetaiotaomicron) in river systems under baseflow
conditions; 2) identify specific land uses that modify reference levels of fecal contamination
in rivers; and 3) determine key chemical, physical, environmental, and hydrological
variables driving water quality of rivers draining to the Great Lakes.

47

REFERENCES

48

REFERENCES

Adler, J.H. (2002). Fables of the Cuyahoga: Reconstructing a history of environmental
protection. Fordham Environmental Law Journal, 14, 89-146.
Ahmed, W., Stewart, J., Powell, D., and Gardner, T. (2008). Evaluation of Bacteroides markers
for the detection of human faecal pollution. Letters in Applied Microbiology, 46, 237-242.
Ahmed, W., Goonetilleke, A, Powell, D., Chauhan, K., and Gardner, T. (2009). Comparison of
molecular markers to detect fresh sewage in environmental waters. Water Research, 43, 49084917.
Akasaka, M., Takamura, N., Mitsuhashi, H., and Kadono, Y. (2010). Effects of land use on
aquatic macrophyte diversity and water quality of ponds. Freshwater Biology, 55, 909-922.
Allan, J.D. (2004). Landscapes and riverscapes: The influence of land use on stream ecosystems.
Annual Review of Ecology, Evolution, and Systematics, 35, 257-284.
Allwood, P.B., Malik, Y., Hedberg, C., and Goyal, S. (2003). Survival of F-specific RNA
coliphage, feline Calicivirus, and Escherichia coli in water: A comparative study. Applied and
Environmental Microbiology, 69, 5707-5710.
Alm, E.W., Burke, J., and Spain, A. (2003). Fecal indicator bacteria are abundant in wet sand at
freshwater beaches. Water Research, 37, 3978-3982.
Alm, E.W., Burke, J., and Hagan, E. (2006). Persistence and potential growth of the fecal
indicator bacteria, Escherichia coli, in shoreline sand at Lake Huron. Journal of Great Lakes
Research, 32, 401-405.
Ashbolt, N.J., Grabow, W.O., and Snozzi, M. (2001). Indicators of microbial water quality. In L.
Fewtrell and J. Bartram (Eds), Development, 289-316. IWA Publishing. Retrieved from
(http://www.csa.com/partners/viewrecord.php?requester=gsandcollection=ENVandrecid=530543
9).
Ashworth, W. (1986). The Late, Great Lakes: An environmental history. Random House: New
York.
Austin, J., Anderson, S., Courant, P., and Litan, R. (2007). Healthy waters, strong economy: The
benefits of restoring the Great Lakes ecosystem. The Brookings Institution.

49

Ballesté, E. and Blanch, A.R. (2010). Persistence of Bacteroides species populations in a river as
measured by molecular and culture techniques. Applied and Environmental Microbiology, 76,
7608-7616.
Barwick, R.S., Levy, D.A., Craun, G.F., Beach, M.J., and Calderon, R.L. (2000). Surveillance
for waterborne-disease outbreaks, United States, 1997-1998. Morbidity and Mortality Weekly
Report (MMWR), CDC Surveillance Summaries. 49; 1-35. Centers for Disease Control and
Prevention (CDC), Atlanta, Georgia.
Baumann, P. and Whittle, M. (1988). The status of selected organics in the Laurentian Great
Lakes: An overview of DDT, PCBs, dioxins, furans, and aromatic hydrocarbons. Aquatic
Toxicology, 11, 241-257.
Bell, A., Layton, A.C., McKay, L., Williams, D., Gentry, R., and Sayler, G.S. (2009). Factors
influencing the persistence of fecal Bacteroides in stream water. Journal of Environmental
Quality, 38, 1224-1232.
Bernhard, A. and Field, K. (2000a). Identification of nonpoint sources of fecal pollution in
coastal waters by using host-specific 16S ribosomal DNA genetic markers from fecal anaerobes.
Applied and Environmental Microbiology, 66, 1587-1594.
Bernhard, A. and Field, K. (2000b). A PCR assay to discriminate human and ruminant feces on
the basis of host differences in Bacteroides-Prevotella genes encoding 16S rRNA. Applied and
Environmental Microbiology, 66, 4571-4574.
Bischoff, C., Lüthy, J., Altwegg, M., and Baggi, F. (2005). Rapid detection of diarrheagenic E.
coli by real-time PCR. Journal of Microbiological Methods, 61, 335-241.
Bisson, J.W. and Cabelli, V.J. (1979). Membrane filter enumeration method for Clostridium
perfringens . Applied and Environmental Microbiology, 37, 55-66.
Boehm, A. (2007). Enterococci concentrations in diverse coastal environments exhibit extreme
variability. Environmental Science and Technology, 41, 8227-8232.
Bonde, G. (1966). Bacteriological methods for estimation of water pollution. Healthy Laboratory
Science, 3, 124-128.
Botts, L. and Muldoon, P. (2005). Evolution of the Great Lakes Water Quality Agreement. East
Lansing, Michigan: Michigan State University Press.
Brettar, I. and Höfle, M.G. (1992). Influence of ecosystamatic factors on survival of E. coli after
large-scale release into lake water mesocosms. Applied and Environmental Microbiology, 58,
2201-2210.
Brion, G.M., Meschke, J.S., and Sobsey, M.D. (2002). F-specific RNA coliphages: Occurrence,
types, and survival in natural waters. Water Research, 36, 2419-2425.
50

Broussard, W. and Turner, R.E. (2009). A century of changing land-use and water-quality
relationships in the continental US. Frontiers in Ecology and the Environment, 7, 302-307.
Byappanahalli, M.N., Shively, D.A., Nevers, M.B., Sadowsky, M.J., and Whitman, R.L. (2003a).
Growth and survival of Escherichia coli and enterococci populations in the macro-alga
Cladophora (Chlorophyta). Federation of European Microbiological Societies, 46, 203-211.
Byappanahalli, M.N., Fowler, M., Shively, D., and Whitman, R. (2003b). Ubiquity and
persistence of Escherichia coli within a Midwestern stream. Applied and Environmental
Microbiology, 69, 4549-4555.
Byappanahalli, M.N., Przybyla-Kelly, K., Shively, D.A., and Whitman, R.L. (2008).
Environmental occurrence of the Enterococcal Surface Protein (esp) gene is an unreliable
indicator of human fecal contamination. Environmental Science and Technology, 42, 8014-8020.
Byappanahalli, M.N., Whitman, R.L., Shively, D.A., Ferguson, J., Ishii, S., and Sadowsky, M.J.
(2007). Population structure of Cladophora-borne E. coli in nearshore water of Lake Michigan.
Water Research, 41, 3649-3654.
Byappanahalli, M.N., Sawdey, R., Ishii, S., Shively, D.A., Ferguson, J.A., Whitman, R.L., and
Sadowsky, M.J. (2009). Seasonal stability of Cladophora-associated Salmonella in Lake
Michigan watersheds. Water Research, 43, 806-814.
Byappanahalli, M.N. and Whitman, R.L. (2009). Clostridium botulinum type E occurs and grows
in the alga Cladophora glomerata. Canadian Journal of Fisheries and Aquatic Sciences, 66,
879-882.
Byappanahalli, M.N., Whitman, R.L., Shively, D.A., and Nevers, M.B. (2010). Linking nonculturable (qPCR) and culturable enterococci densities with hydrometeorological conditions. The
Science of the Total Environment, 408, 3096-3101.
Cabelli, V.J. (1978). Obligate anaerobic bacterial indicators. In Indicators of Viruses in Water
and Food (ed. G. Berg), Page 171-200, Ann Arbor Science, Ann Arbor, Michigan.
Cabelli, V.J., Dufour, A.P., McCabe, L.J., and Levin, M.A. (1982). Swimming-associated
gastroenteritis and water quality. American Journal of Epidemiology, 115, 606-616.
Calderon, R., Mood, E., and Dufour, A. (1991). Health effects of swimmers and nonpoint
sources of contaminated water. International Journal of Environmental Health Research, 1, 2131.
Chang, H. (2008). Spatial analysis of water quality trends in the Han River basin, South Korea.
Water Research, 42, 3285-304.

51

Cho, K.H., Cha, S.M., Kang, J.H., Lee, S.W., Park, Y., Kim, J.W., and Kim, J.H. (2010).
Meteorological effects on the levels of fecal indicator bacteria in an urban stream: A modeling
approach. Water Research, 44, 2189-2202.
Chrzastowski, M.J., Thompson, T.A., and Trask, C.B. (1994). Coastal geomorphology and
littoral cell divisions along the Illinois-Indiana coast of Lake Michigan. Journal of Great Lakes
Research, 20, 27-43.
Clamens, M. (2005). The International Joint Commission: A model for inter-American
cooperation. Vertigo, 2. Retrieved from (http://vertigo.revues.org/1885#quotation).
Colford, J.M., Wade, T.J., Schiff, K.C., Wright, C.C., Griffith, J.F., Sandhu, S.K., Burns, S.,
Sobsey, M., Lovelace, G., and Weisberg, S.B. (2007). Water quality indicators and the risk of
illness at beaches with nonpoint sources of fecal contamination. Epidemiology, 18, 27-35.
Copeland, C. (2010). Clean Water Act: A summary of the Law. CRS Report for Congress:
Prepared
for
members
and
committees
of
Congress.
Retrieved
from
(http://www.nationalaglawcenter.org/assets/crs/RL30030.pdf).
Craig, D., Fallowfield, H., and Cromar, N. (2002). Decay rates of faecal indicator organisms and
pathogens: Use of microcosm and in situ studies for the estimation of exposure risk in
recreational waters. American Water Works Association Annual Conference Proceedings.
Crowther, J., Kay, D., and Wyer, M.D. (2001). Relationships between microbial water quality
and environmental conditions in coastal recreational waters: The Fylde coast, UK. Water
Research, 35, 4029-4038.
Curriero, F.C., Patz, J.A, Rose, J.B., and Lele, S. (2001). The association between extreme
precipitation and waterborne disease outbreaks in the United States, 1948-1994. American
Journal of Public Health, 91, 1194-1199.
Dempsey, D. (2004). On the Brink: The Great Lakes in the 21st Century. East Lansing,
Michigan: Michigan State University Press.
Desai, A.M., Rifai, H., Helfer, E., Moreno, N., and Stein, R. (2010). Statistical investigations
into indicator bacteria concentrations in Houston metropolitan watersheds. Water Environment
Research, 82, 302-318.
Desai, A.M. and Rifai, H.S. (2010). Variability of Escherichia coli concentrations in an urban
watershed in Texas. Journal of Environmental Engineering, (December), 1347-1359.
Desmarais, T.R., Solo-Gabriele, H.M., and Palmer, C.J. (2002). Influence of soil on fecal
indicator organisms in a tidally influenced subtropical environment. Applied and Environmental
Microbiology, 68, 1165-1172.

52

Dick, L.K., Bernhard, A.E., Brodeur, T.J., Domingo, W.S., Simpson, J.M., Walters, S.P., and
Field, K.G. (2005). Host distributions of uncultivated fecal Bacteroidales bacteria reveal genetic
markers for fecal source identification. Applied and Environmental Microbiology, 71, 31843191.
Dick, L.K., Stelzer, E.A, Bertke, E.E., Fong, D.L., and Stoeckel, D.M. (2010). Relative decay of
Bacteroidales microbial source tracking markers and cultivated Escherichia coli in freshwater
microcosms. Applied and Environmental Microbiology, 76, 3255-3262.
Donnison, A.M. and Ross, C.M. (1995). Somatic and F-specific coliphages in New Zealand
waste treatment lagoons. Water Research, 29, 1105-1110.
Dorfman, M. and Rosselot, K. (2011). A guide to water quality at vacation beaches: Twenty-first
annual report. Natural Resources Defense Council (NRDC).
Dorner, S., Anderson, W., Gaulin, T., Candon, H., Slawson, R., Payment, P., and Huck, P.M.
(2007). Pathogen and indicator variability in a heavily impacted watershed. Journal of Water and
Health, 5, 241-257.
Dreelin, E., McNinch, R., and Rose, J. (2007). Surface water summary of E. coli and pathogens
in Michigan. Report prepared for the Michigan Department of Environmental Quality Water
Bureau.
Dreelin, E.A. (2008). Bacteriological monitoring in the great lakes: A historical perspective to
inform the present. In Effective Cross-Border Monitoring Systems for Waterborne Microbial
Pathogens: A plan for action. Edt. E. Dreelin and J. Rose. IWA Publishing, London, UK.
Dufour, A.P. (1984). Health effects criteria for fresh recreational waters. EPA-600/1-84-004,
Office of Research and Development, US Environmental Protection Agency, Cincinnati, OH.
Durfee, M. and Bagley, S.T. (1997). Bacteriology and diplomacy in the Great Lakes 1912-1920.
Paper prepared for the 1997 meeting of the American Society for Environmental History,
Baltimore, MD March 6-9, 1997.
Dziuban, E.J., Liang, J.L., Craun, G.F., Hill, V., Yu, P.A. , Painter, J., Moore, M.R., Calderon,
R.L., Roy, S.L. and Beach, M.J. (2006). Surveillance for waterborne disease and outbreaks
associated with recreational water - United States, 2003-2004. Morbidity and Mortality Weekly
Report (MMWR), CDC Surveillance Summaries. 55(SS12); 1-24. Centers for Disease Control
and Prevention (CDC), Atlanta, Georgia.
Englebert, E.T., McDermott, C., and Kleinheinz, G.T. (2008a). Impact of alga Cladophora on
the survival of E. coli, Salmonella, and Shigella in laboratory microcosm. Journal of Great Lakes
Research, 34, 377-382.

53

Englebert, E.T., McDermott, C., and Kleinheinz, G.T. (2008b). Effects of the nuisance algae,
Cladophora, on Escherichia coli at recreation beaches in Wisconsin. Science of the Total
Environment, 404, 10-17.
Favero, M. (1985). Microbiological indicators of health risks associated with swimming.
American Journal of Public Health, 75, 1051-1054.
Ferris, G. and Andrachuk, H. (2009). State of the Great Lakes 2009; Climate change: Ice
duration on the Great Lakes. (EPA 950 K 09 001).
Field, K., Bernhard, A., and Brodeur, T. (2003). Molecular approaches to microbiological
monitoring: Fecal source detection. Environmental Monitoring and Assessment, 81, 313-326.
Field, K.G. and Samadpour, M. (2007). Fecal source tracking, the indicator paradigm, and
managing water quality. Water Research, 41, 3517-3538.
Fleisher J.M., Kay D., Wyer, M.D., and Godfree, A.F. (1998). Estimates of the severity of
illnesses associated with bathing in marine waters contaminated with domestic sewage.
International Journal of Epidemiology, 27, 722-726.
Folger, D.W., Colman, S.M., and Barnes, P.W. (1994). Overview of the southern Lake Michigan
coastal erosion study. Journal of Great Lakes Research, 20, 2-8.
Fujioka, R.S. and Shizumura, L.K. (1985). Clostridium perfringens : A reliable indicator of
stream water quality. Journal of the Water Pollution Control Federation, 57, 986-992.
García-Armisen, T. and Servais, P. (2009). Partitioning and fate of particle associated
E. coli in river waters. Water Environment Research, 81, 21-28.
Garrido-Pérez, M.C., Anfuso, E., Acevedo, A., and Perales-Vargas-Machuca, J.A. (2008).
Microbial indicators of faecal contamination in waters and sediments of beach bathing zones.
International Journal of Hygiene and Environmental Health, 211, 510-517.
Garzio-Hadzick, A., Shelton, D.R., Hill, R.L., Pachepsky, Y., Guber, K., and Rowland, R.
(2010). Survival of manure-borne E. coli in streambed sediment: Effects of temperature and
sediment properties. Water Research, 44, 2753-2762.
Ge, Z., Nevers, M.B., Schwab, D.J., and Whitman, R.L. (2010). Coastal loading and transport of
Escherichia coli at an embayed beach in Lake Michigan. Environmental Science and
Technology, 44, 6731-6737.
Girones, R., Ferrús, M.A., Alonso, J.L., Rodriguez-Manzano, J., Calgua, B., Corrêa, A.,
Hundesa, A., Carratala, A., and Bofill-Mas, S. (2010). Molecular detection of pathogens in
water--the pros and cons of molecular techniques. Water Research, 44, 4325-4339.

54

Gleeson, C. and Gray, N. (2003). The coliform index and waterborne disease: Problems of
microbial drinking water assessment. New York, NY: Taylor and Francis e-Library.
Goodwin, K., Matragrano, L., Wanless, D., Sinigalliano, C., and LaGier, M. (2008). The
possibility of false negative results hampers the ability of elucidate the relationship between fecal
indicator bacteria and human pathogens and source tracking markers in beach water and sand. In
Tobias Hofer (Eds.), Marine Pollution: New research (255-277). New York: Nova Science
Publishers.
Goto, D.K. and Yan, T. (2011). Effects of land uses on fecal indicator bacteria in the water and
soil of a tropical watershed. Microbes and Environments, 26, 254-260.
Gourmelon, M., Caprais, M.P., Ségura, R., LeMennec, C., Lozach, S., Piriou, J.Y., and Rincé, A.
(2007). Evaluation of two library-independent microbial source tracking methods to identify
sources of fecal contamination in French estuaries. Applied and Environmental Microbiology,
73, 4857-4866.
Grady, W. (2007). The Great Lakes: The natural history of a changing region. Vancouver, BC:
Greystone Books.
Griffin, D.W., Lipp, E.K., Mclaughlin, M.R., and Rose, J.B. (2001). Marine recreation and
public health microbiology : Quest for the ideal indicator. BioScience, 51, 817-826.
Griffith, J.F., Weisberg, S.B., and McGee, C.D. (2003). Evaluation of microbial source tracking
methods using mixed fecal sources in aqueous test samples. Journal of Water and Health, 1,
141-151.
Guzman, C., Moce-Llivina, L., Lucena, F., and Jofre, J. (2008). Evaluation of Escherichia coli
host strain CB390 for simultaneous detection of somatic and F-specific coliphages. Applied and
Environmental Microbiology, 74, 531-534.
Haack, S.K., Fogarty, L.R., and Wright, C. (2003). Escherichia coli and enterococci at beaches
in the Grand Traverse Bay, Lake Michigan: Sources, characteristics, and environmental
pathways. Environmental Science and Technology, 37, 3275-3282.
Haugland, R.A, Siefring, S.C., Wymer, L.J., Brenner, K.P., and Dufour, A P. (2005).
Comparison of enterococcus measurements in freshwater at two recreational beaches by
quantitative polymerase chain reaction and membrane filter culture analysis. Water Research, 39,
559-568.
Havelaar, AH. (1987). Bacteriophages as model organisms in water treatment. Microbiological
Science, 4, 362-364.
Heikens, E., Bonten, M.J.M., and Willems, R.J.L. (2007). Enterococcal surface protein Esp is
important for biofilm formation of Enterococcus faecium E1162. Journal of Bacteriology, 189,
8233-8240.
55

Hlavsa, M.C., Roberts, V.A., Anderson, A.R., Hill, V.R., Kahler, A.M., Orr, M., Garrison, L.E.,
Hicks, L.A., Newton, A., Hilborn, E.D., Wade, T.J., Beach, M. J., and Yoder, J.S. (2011).
Morbidity and mortality weekly report: Surveillance for waterborne disease outbreaks and other
health events associated with recreational water - United States , 2007-2008. Prevention, 60,
2007-2008.
Houston, J.R. (2008). The economic value of beaches – A 2008 update. Shore and Beach, 76,
22-26.
Hunter, C., Perkins, J., Tranter, J., and Gunn, J. (1999). Agricultural land-use effects on the
indicator bacterial quality of an upland stream in the Derbyshire peak district in the U.K. Water
Research, 33, 3577-3586.
International Joint Commission (IJC). (2005). A guide to the Great Lakes Water Quality
Agreement: Background for the 2006 Governmental Review.
International Joint Commission (IJC). (2008). Annual report for 2008: Boundary Waters Treaty
centennial edition. Retrieved from (http://www.ijc.org/php/publications/pdf/ID1629.pdf).
Ishii, S., Yan, T., Shively, D.A., Byappanahalli, M.N., Whitman, R.L., and Sadowsky, M.J.
(2006). Cladophora (Chlorophyta) spp. Harbor human bacterial pathogens in nearshore water of
Lake Michigan. Applied and Environmental Microbiology, 72, 4545-4553.
Jamieson, R.C., Joy, D.M., Lee, H., Kostaschuk, R., and Gordon, R.J. (2005). Resuspension of
sediment-associated Escherichia coli in a natural stream. Journal of Environmental Quality, 34,
581-589.
Jayawickreme, D.H. and Hyndman, D.W. (2007). Evaluating the influence of land cover on
seasonal water budgets using Next Generation Radar (NEXRAD) rainfall and stream flow data.
Water Resources Research, 43, 1-11.
Jeter, S.N., McDermott, C.M., Bower, P.A, Kinzelman, J.L., Bootsma, M.J., Goetz, G.W., and
McLellan, S.L. (2009). Bacteroidales diversity in ring-billed gulls (Laurus delawarensis)
residing at Lake Michigan beaches. Applied and Environmental Microbiology, 75, 1525-1533.
Johnson, W.C., Boettcher, S.E., Poiani, K.A., and Guntenspergen, G. (2004). Influence of
weather extremes on the water levels of glaciated prairie wetlands. Wetlands, 24, 385-398.
Kalff, J. (2002). Limnology: Inland water ecosystems. Upper Saddle River, NJ: Prentice-Hall.
Kang, J.H., Lee, S.W., Cho, K.H., Ki, S.J., Cha, S.M., and Kim, J.H. (2010). Linking land-use
type and stream water quality using spatial data of fecal indicator bacteria and heavy metals in
the Yeongsan River basin. Water Research, 44, 4143-4157.

56

Karrow, P.F. and Calkin, P.E. (1985). Quaternary evolution of the Great Lakes. Special Paper
30, Geological Association of Canada, Canada.
Katz, D.M., Watts, F.J., and Burroughs, E.R. (1995). Effects of surface roughness and rainfall
impact on overland flow. Journal of Hydraulic Engineering, 121, 546-553.
Kildare B.J., Leutenegger C.M., McSwain B.S., Bambic D.G., Rajal V.B., and Wuertz S. (2007).
16S rRNA-based assays for quantitative detection of universal, human-, cow-, and dog-specific
fecal Bacteroidales: A Bayesian approach. Water Research, 16, 3701-3715.
Kim, S.Y., Lee, J.E., Lee, S., Lee, H.T., Hur, H.G., and Ko, G. (2010). Characterization of
Enterococcus spp. from human and animal feces using 16S rRNA sequences, the esp gene, and
PFGE for microbial source tracking in Korea. Environmental Science and Technology, 44, 34233428.
Kinzelman, J., Ng, C., Jackson, E., Gradus, S., and Bagley, R. (2003). Enterococci as indicators
of Lakes Michigan recreational water quality: Comparison of two methodologies and their
impacts on public health regulatory events. Applied and Environmental Microbiology, 69, 92-96.
Kinzelman, J., McLellan, S.L., Daniels, A.D., Cashin, S., Singh, A., Gradus, S., Bagley, R.
(2004). Non-point source pollution: determination of replication versus persistence of
Escherichia coli in surface water and sediments with correlation of levels to readily measurable
environmental parameters. Journal of Water and Health, 2, 103-114.
Kirs, M., Harwood, V.J., Fidler, A., Gillespie, P., Fyfe, W., Blackwood, A., and Cornelisen, C.
(2011). Source tracking faecal contamination in an urbanized and a rural waterway in the
Nelson-Tasman region, New Zealand. New Zealand Journal of Marine and Freshwater
Research, 45, 43-58.
Krantzberg, G. (2007). Commentary: The ongoing review of the Great Lakes Water Quality
Agreement. Journal of Great Lakes Research, 33, 699-703.
Ksoll, W.B., Ishii, S., Sadowsky, M.J., and Hicks, R.E. (2007). Presence and sources of fecal
coliform bacteria in epilithic periphyton communities of Lake Superior. Applied and
Environmental Microbiology, 73, 3771-3778.
LaLibertet, P. and Grimes, D.J. (1982). Survival of Escherichia coli in lake bottom sediment.
Applied and Environmental Microbiology, 43, 623-628.
Larson, G. and Schaetzl, R. (2001). Review: Origin and Evolution of the Great Lakes. Journal of
Great Lakes Research, 27, 518-546.
Lavee, H. and Poesen, J.W. (1991). Overland flow generation and continuity on stone-covered
soil surfaces. Hydrological Processes, 5, 345-360.

57

Layton, B., McKay, L., Williams, D., Garrett, V., Gentry, R., and Sayler, G. (2006).
Development of Bacteroides 16S rRNA gene TaqMan-based real-time PCR assays for estimation
of total, human, and bovine fecal pollution in water. Applied and Environmental Microbiology,
72, 4214-4224.
LeBlanc, R. (1997). Modeling the effects of land use change on the water temperature in
unregulated urban streams. Journal of Environmental Management, 49, 445-469.
Lee, S.H., Levy, D.A., Craun, G.F., Beach, M.J., and Calderon, R.L. (2002). Surveillance of
waterborne-disease
outbreaks,
United
States,
1999-2000.
CDC.
Retrieved
from (http://www.cdc.gov/mmwr/preview/mmwrhtml/ss5108a1.htm).
LeFevre, N.M. and Lewis, G.D. (2003). The role of resuspension in enterococci distribution in
water at an urban beach. Water Science and Technology, 47, 205-210.
Lipp, E.K., Farrah, S.A., and Rose, J.B. (2001). Assessment and impact of microbial fecal
pollution and human enteric pathogens in a coastal community. Marine pollution bulletin, 42,
286-293.
Lisle, J.T., Smith, J.J., Edwards, D.D., and Mcfeters, G.A. (2004). Occurrence of microbial
indicators and Clostridium perfringens in wastewater, water column samples, sediments,
drinking water, and Weddell seal feces collected at Mcmurdo station, Antarctica. Applied and
Environmental Microbiology, 70, 7269-7276.
Loch, R.J. (2000). Effects of vegetation cover on runoff and erosion under simulated rain and
overland flow on a rehabilitated site on the Meandu Mine, Tarong, Queensland. Australian
Journal of Soil Research, 38, 299-312.
Mahin, T. and Pancorbo, O. (1999). Waterborne pathogens. Water Environment and Technology,
11, 51-55.
Mallin, M.A., Williams, K.E., Esham, E.C., and Lowe, R.P. (2000). Effect of human
development on bacteriological water quality in coastal qatersheds. Ecological Applications, 10,
1047-1056.
Marti, R., Zhang, Y., Lapen, D.R., and Topp, E. (2011). Development and validation of a
microbial source tracking marker for the detection of fecal pollution by muskrats. Journal of
Microbiological Methods, 87, 82-88.
Marvin-DiPasquale, M. and Agee, J.L. (2003). Microbial mercury cycling in sediments of the
San Francisco Bay-Delta. Estuaries, 26, 1517-1528.
Masago, Y., Pope, J.M., Kumar, L.S., Masago, A., Omura, T., and Rose, J.B. (2011). Prevalence
and survival of Enterococcus faecium populations carrying the esp gene as a source tracking
marker. Journal of Environmental Engineering, 137, 315-321.

58

Mattikalli, N. and Richards, K.S. (1996). Estimation of surface water quality changes in response
to land use change: Application of the export coefficient model using remote sensing and
geographical information system. Journal of Environmental Management, 48, 263-282.
McLellan, S.L., Daniels, A.D., and Salmore, A.K. (2001). Clonal populations of thermotolerant
enterobacteriaceae in recreational water and their potential interference with fecal Escherichia
coli counts. Applied and Environmental Microbiology, 67, 4934-4938.
McLellan, S. (2004). Genetic diversity of E. coli isolated from urban rivers and beach water.
Applied and Environmental Microbiology, 70, 4658-4665.
McLellan, S.L., Hollis, E.J., Depas, M.M., Dyke, M.V., Harris, J., and Scopel, C.O. (2007).
Distribution and fate of Escherichia coli in Lake Michigan following contamination with urban
stormwater and combined sewer overflows. Journal of Great Lakes Research, 33, 566-580.
McQuaig, S.M., Scott, T.M., Harwood, V.J., Farrah, S.R., and Lukasik, J.O. (2006). Detection of
human-derived fecal pollution in environmental waters by use of a PCR-based human
polyomavirus assay. Applied and Environmental Microbiology, 72, 7567-7574.
Mehaffey, M.H., Nash, M.S., Wade, T.G., Ebert, D.W., Jones, K.B., and Rager, A. (2005).
Linking land cover and water quality in New York City’s water supply watersheds.
Environmental Monitoring and Assessment, 107, 29-44.
Michigan Department of Environmental Quality (MDEQ). (2007). Michigan beach monitoring
year 2006 annual report. Lansing, MI. MI/DEQ/WB-07/068
Michigan Department of Environmental Quality (MDEQ). (2011). Michigan Beach Monitoring
Year 2010 Annual Report. Lansing, MI. MI/DEQ/WRD-11/009.
Mika, K.B., Imamura, G., Chang, C., Conway, V., Fernandez, G., Griffith, J.F., Kampalath, R.A,
Lee, C.M., Lin, C.C., Moreno, R., and Thompson, S. (2009). Pilot- and bench-scale testing of
faecal indicator bacteria survival in marine beach sand near point sources. Journal of Applied
Microbiology, 107, 72-84.
Mills, E.L., Leach, J.H., Carlton, J.T., and Secor, C.L. (1993). Exotic species in the Great Lakes:
A history of biotic crises and anthropogenic introductions. Journal of Great Lakes Research, 19,
1-54.
Mortimer, C.H. (1971). Chemical exchanges between sediments and water in the Great LakesSpeculations on probable regulatory mechanisms. Limnology and Oceanography, 2, 387-404.
Murray, C., Sohngen, B., and Pendleton, L. (2001). Valuing water quality advisories and beach
amenities in the Great Lakes. Water Resources Research, 37, 2583-2590.
Najjum, W.T. (2009). The Clean Water Act after 37 years : Recommitting to the Protection of the
Nation’s Waters. Environmental Protection Agency. Washington, DC.
59

Nash, M.S., Heggem, D.T., Ebert, D., Wade, T.G., and Hall, R.K. (2009). Multi-scale landscape
factors influencing stream water quality in the state of Oregon. Environmental Monitoring and
Assessment, 156, 343-360.
Nevers, M.B., Whitman, R.L., Frick, W.E., and Ge, Z. (2007). Interaction and influence of two
creeks on Escherichia coli concentrations of nearby beaches: Exploration of predictability and
mechanisms. Journal of Environmental Quality, 36, 1338-1345.
Nevers, M.B. and Whitman, R.L. (2008). Coastal strategies to predict Escherichia coli
concentrations for beaches along a 35 km stretch of Southern Lake Michigan. Environmental
Science and Technology, 42, 4454-4460.
Nevers, M.B. and Whitman, R.L. (2005). Nowcast modeling of Escherichia coli concentrations
at multiple urban beaches of southern Lake Michigan. Water Research, 39, 5250-5260.
Okabe, S. and Shimazu, Y. (2007). Persistence of host-specific Bacteroides-Prevotella 16S
rRNA genetic markers in environmental waters: Effects of temperature and salinity. Applied
Microbiology and Biotechnology, 76, 935-944.
Olapade, O.A., Depas, M.M., Jenson, E.T., and McLellan, S.L. (2006). Microbial communities
and fecal indicator bacteria associated with Cladophora mats on beach sites along Lake
Michigan shores. Applied and Environmental Microbiology, 72, 1932-1938.
Page, B. and Walker, R. (1991). From settlement to Fordism: The agro-industrial revolution in
the American Midwest. Economic Geography, 67, 281-315.
Parkins, R.T., Soller, J.A., and Olivieri, A.W. (2003). Incorporating susceptible subpopulations
in microbial risk assessment: Pediatric exposures to enteroviruses in river water. Journal of
Exposure Analysis and Environmental Epidemiology, 13, 161-168.
Patz, J., Graczyk, T., Geller, N., and Vittor, A. (2000). Effects of environmental change on
emerging parasitic diseases. International Journal for Parasitology. 20, 1395-1405.
Patz, J.A, Vavrus, S.J., Uejio, C.K., and McLellan, S.L. (2008). Climate change and waterborne
disease risk in the Great Lakes region of the U.S. American Journal of Preventive Medicine, 35,
451-458.
Payment, P. and Franco, E. (1993). Clostridium perfringens and somatic coliphages as indicators
of the efficiency of drinking water treatment for viruses and protozoan cysts. Applied and
Environmental Microbiology, 59, 2418-2424.
Pip, E. and Allegro, E. (2010). Nearshore fluctuations in water chemistry, microcystins and
coliform bacteria during the ice-free season in Lake Winnipeg, Manitoba, Canada. Ecohydrology
and Hydrobiology, 10, 35-43.

60

Pipes, W.O. (1982). Indicators and water quality, in Bacterial Indicators of Pollution, (ed. W.O.
Pipes), page 83-96, CRC Press, Boca Raton, FL.
Pote, J., Haller, L., Kottelat, R., Sastre, V., Arpagaus, P., and Wildi, W. (2009). Persistence and
growth of faecal culturable bacterial indicators in water column and sediments of Vidy Bay,
Lake Geneva, Switzerland. Journal of Environmental Science, 21, 62-69.
Prüss, A. (1998). Review of epidemiological studies on health effects from exposure to
recreational water. International Journal of Epidemiology, 27, 1-9.
Reischer, G.H., Kasper, D.C., Steinborn, R., Mach, R.L., and Farnleitner, A.H., (2006).
Quantitative PCR method for sensitive detection of ruminant fecal pollution in freshwater and
evaluation of this method in Alpine karstic regions. Applied and Environmental Microbiology,
72, 5610-5614.
Reischer, G.H., Haider, J.M., Sommer, R., Stadler, H., Keiblinger, K.M., Hornek, R., Zerobin,
W., Mach, R.L., and Farnleitner, A.H. (2008). Quantitative microbial faecal source tracking with
sampling guided by hydrological catchment dynamics. Environmental Microbiology, 10, 25982608.
Ricciardi, A. (2006). Patterns of invasion in the Laurentian Great Lakes in relation to changes in
vector activity. Diversity Distributions, 12, 425-433.
Rice, E.W. (2003). Escherichia coli: Pathogenic strains in Encyclopedia of Environmental
Microbiology. John Wiley and Sons, NY.
Rompré, A., Servais, P., Baudart, J., De-Roubin, M.R., and Laurent, P. (2002). Detection and
enumeration of coliforms in drinking water: current methods and emerging approaches. Journal
of Microbiological Methods, 49, 31-54.
Rose, J.B., Zhou, X., Griffin, D.W., and Paul, J.H. (1997). Comparison of PCR and plaque assay
for detection and enumeration of coliphage in polluted marine waters. Applied and
Environmental Microbiology, 63, 4564-4566.
Roslev, P., Bastholm, S., and Iverson, N. (2008). Relationship between fecal indicators in
sediment and recreational waters in a Danish estuary. Water, Air, and Soil Pollution, 194, 13-21.
Santo Domingo, J.W., Bambic, D.G., Edge, T., and Wuertz, S. (2007). Quo vadis source
tracking? Towards a strategic framework for environmental monitoring of fecal pollution. Water
Research, 41, 3539-3552.
Sartory, D.P. (1986). Membrane filtration enumeration of faecal clostridia and Clostridium
perfringens in water. Water Research, 20, 1255-1260.
Schulz, C.J., and Childers, G.W. (2011). Fecal Bacteroidales diversity and decay in response to
variations in temperature and salinity. Applied and Environmental Microbiology, 77, 2563-2572.
61

Schwab, D.J., Leshkevich, G.A., and Muhr, G.C. (1999). Automated mapping of surface water
temperature in the Great Lakes. Journal of Great Lakes Research, 25, 468-481.
Schwab, K.J. (2007). Are existing bacterial indicators adequate for determining recreational
water illness in waters impacted by nonpoint pollution? Epidemiology, 18, 21-22.
Scopel, C.O., Harris, J., and Mclellan, S.L. (2006). Influence of nearshore water dynamics and
pollution sources on beach monitoring outcomes at two adjacent Lake Michigan beaches.
Journal of Great Lakes Research, 32, 543-552.
Scott, T.M., Rose, J.B., Jenkins, T.M., Farrah, S.R., and Lukasik, J. (2002). Microbial source
tracking: Current methodology and future directions. Applied and Environmental Microbiology,
68, 5796-5803.
Scott, T.M., Jenkins, T.M., Lukasik, J., and Rose, J.B. (2005). Potential use of a host associated
molecular marker in Enterococcus faecium as an index of human fecal pollution. Environmental
Science and Technology, 39, 283-287.
Shehane, S.D., Harwood, V.J., Whitlock, J.E., and Rose, J.B. (2005). The influence of rainfall on
the incidence of microbial faecal indicators and the dominant sources of faecal pollution in a
Florida river. Journal of Applied Microbiology, 98, 1127-1136.
Signoretto, C., Burlacchini, G., Del Mar Lleo, M., Pruzzo, C., Zampini, M., Pane, L., Franzini,
G., and Caepari, P. (2004). Adhesion of Enterococcus faecalis in the nonculturable state to
plankton is the main mechanism responsible for persistence of this bacterium in both lake and
seawater. Applied and Environmental Microbiology, 70, 6892-6896.
Smart, R. and Barko, J. (1978). Influence of sediment salinity and nutrients on the physiological
ecology of selected salt marsh plants. Estuarine and Coastal Marine Science, 7, 487-495.
Soranno, P.A, Cheruvelil, K.S., Webster, K.E., Bremigan, M.T., Wagner, T., and Stow, C.A.
(2010). Using landscape limnology to classify freshwater ecosystems for multi-ecosystem
management and conservation. BioScience, 60, 440-454.
Song, F., Lupi, F., and Kaplowitz, M. (2010). Valuing Great Lakes beaches. Agriculture and
Applied Economics Association Annual Meeting.
Spooner, J. and Line, D. (1993). Effective monitoring strategies from demonstrating water
quality changes from nonpoint source controls on a watershed scale. Water Science and
Technology, 28, 143-148.
Stevenson, A.H. (1953). Studies of bathing water quality and health. Journal of American Public
Health Association, 43, 529-538.

62

Thupaki, P., Phanikumar, M.S., Beletsky, D., Schwab, D.J., Nevers, M.B., and Whitman, R.L.
(2010). Budget analysis of Escherichia coli at a southern Lake Michigan beach. Environmental
Science and Technology, 44, 1010-1016.
Toranzos, G.A. and McFeters, G.A. (1997). Detection of microorganisms in environmental
freshwaters and drinking waters, p249-264. In Hurst, C.J., Knudsen, G., McInerney, M.,
Stetzenbach, L., and Walter, M. (Editors) Manual of Environmental Microbiology. Washing,
D.C.: ASM Press.
United States Army Corp of Engineers. (2011). Great Lakes Basin Hydrology: October 2010.
Retrieved from
(http://www.lre.usace.army.mil/_kd/Items/actions.cfm?action=Showanditem_id=7153anddestina
tion=ShowItem).
United States Department of Agriculture-Natural Resource Conservation Service (USDANRCS). (1997). National resources inventory summary report. Retrieved from
(http://www.nrcs.usda.gov/Internet/FSE_DOCUMENTS/nrcs143_012094.pdf).
United States Environmental Protection Agency (USEPA), O. of W. (1986). Ambient water
quality criteria for bacteria - 1986, (EPA440/5-8).
United States Environmental Protection Agency (USEPA), O. of W. (1992). The quality of our
Nation’s water, 305, 1-38.
United States Environmental Protection Agency (USEPA), O. of W. (2001). Method 1601:
Male-specific (F+) and somatic coliphage in water by two-step enrichment procedure. (EPA 821R-01-030).
United States Environmental Protection Agency (USEPA), O. of W. (2002). Method 1600:
Enterococci in water by membrane filtration using membrane-Enterococcus indoxyl-B-DGlucoside agar (mEI). (EPA 821-R-02-022).
United States Environmental Protection Agency (USEPA). (2006). Great Lakes fact sheet.
Retrieved from (http://www.epa.gov/greatlakes/factsheet.html).
United States Environmental Protection Agency (USEPAb), O. of W. (2009). National water
quality inventory: Report to Congress 2004 reporting cycle, (EPA 841-R-08-001).
United States Environmental Protection Agency (USEPA). (2009a). Review of published studies
to characterize relative risks from different sources of fecal. (EPA 822-R-09-001).
United States Environmental Protection Agency (USEPA). (2011). Beach grants: Fact sheets
from 2002-2011. Retrieved from (http://water.epa.gov/grants_funding/beachgrants/).

63

Verhougstraete, M.P., Byappanahalli, M.N., Rose, J.B., and Whitman, R.L. (2010). Cladophora
in the Great Lakes: Impacts on beach water quality and human health. Water Science and
Technology, 62, 68-76.
Verhougstraete, M.P. and Rose, J. (in prep.). Microbial investigations of water, sediment, and
algae mat in a mixed use watershed. Environmental Monitoring and Assessment.
Viau, E., Goodwin, K., Yamahara, K., Layton, B., Sassoubre, L., Burns, S., Tong, H., Wong, S.,
Lu, Y., and Boehm, A. (2011). Bacterial pathogens in Hawaiian coastal streams--associations
with fecal indicators, land cover, and water quality. Water Research, 45, 3279-3290.
Vital, M., Hammes, F., and Egli, T. (2008). Escherichia coli O157 can grow in natural
freshwater at low carbon concentrations. Environmental Microbiology, 10, 2387-2396.
Voth, H. (2003). Living standards during the Industrial Revolution: An economist’s guide.
American Economic Assoc., 93, 221-226.
Wade, T.J., Pai, N., Eisenberg, J.N.S., and Colford J.M. (2003). Do US EPA water quality
guidelines for recreational waters prevent gastrointestinal illness? A systematic review and metaanalysis. Environmental Health Perspectives, 111, 1102-1109.
Wade, T.J., Calderon, R.L., Sams, E., Beach, M., Brenner, K.P., Williams, A.H., and Dufour,
A.P. (2006). Rapidly measured indicators of recreational water quality are predictive of
swimming-associated gastrointestinal illness. Environmental Health Perspectives, 114, 24-28.
Wade, T., Calderon, R., Brenner, K., Sams, E., Beach, M., Haugland, R., Wymer, L., and
DuFour, A. (2008). High sensitivity of children to swimming-associated gastrointestinal illness:
Results using a rapid assay of recreational water quality. Epidemiology, 19, 375-383.
Walters, S.P. and Field, K.G. (2006). Persistence and growth of fecal Bacteroidales assessed by
bromodeoxyuridine immunocapture. Applied and Environmental Microbiology, 72, 4532-4539.
Walters, S.P., Thebo, A.L., and Boehm, A.B. (2011). Impact of urbanization and agriculture on
the occurrence of bacterial pathogens and stx genes in coastal waterbodies of central California.
Water Research, 45, 1752-1762.
Wang, X. and Yin, Z.Y. (1997). Using GIS to assess the relationship between land use and water
quality at a watershed level. Environment International, 23, 103-114.
Wang, L., Lyons, J., Kanehl, P., and Bannerman, R. (2001). Impacts of urbanization on stream
habitat and fish across multiple spatial scales. Environmental management, 28, 255-266.
Weinbauer, M.G. and Höfle, M.G. (1998). Distribution and life strategies of two bacterial
populations in a eutrophic lake. Applied Environmental Microbiology, 64, 3776-3783.
White, D.O. and Fenner, F.J. (1994). Medical Virology, 4th edition. New York, NY, USA:
64

Academic Press.
Whitfield, R.E. and Kolenosky, D.P. (1978). The progressive fish-culturist prototype eel ladder
in the St. Lawrence river. The Progressive Fish Culturist, 40, 152-154.
Whitman, R.L. and Nevers, M.B. (2003). Foreshore sand as a source of Escherichia coli in
nearshore water of a Lake Michigan beach. Water, 69, 5555-5562.
Whitman, R.L. and Nevers, M.B. (2004). Policy analysis Escherichia coli sampling reliability at
a frequently closed Chicago beach: Monitoring and management implications. Environmental
Science and Technology, 38, 4241-4246.
Whitman, R., Nevers, M.B., Boehm, A.B., Eunice, C., Haugland, R.A., Lukasik, A.M., Molina,
M., Przybyla-kelly, K., Shively, D.A., White, E.M., Zepp, R.G., and Byappanahalli, M.N.
(2010). Relationship and variation of qPCR and culturable enterococci estimates in ambient
surface waters are predictable. Environmental Science and Technology, 44, 5049-5054.
Whitman, R.L., Przybyla-Kelly, K., Shively, D., Nevers, M.B., and Byappanahalli, M.N. (2008).
Sunlight, season, snowmelt, storm, and source affect E. coli populations in an artificially ponded
stream. The Science of the Total Environment, 390, 448-455.
Whitman, R.L., Przybyla-Kelly, K., Shively, D., and Byappanahalli, M.N. (2007). Incidence of
the enterococcal surface protein (esp) gene in human and animal fecal sources. Environmental
Science and Technology, 41, 6090-6095.
Whitman, R.L., Shivel, D.A., Pawlik, H., Nevers, M.B., and Byappanahalli, M.N. (2003).
Occurrence of E. coli and enterococci in Cladophora (Chlorophyta) in nearshore water and
beach sand of Lake Michigan. Applied and Environmental Microbiology, 69, 4714-4719.
Wiley, M.J., Hyndman, D.W., Pijanowski, B.C., Kendall, A.D., Riseng, C., Rutherford, E.S.,
Cheng, S.T., Carlson, M.L., Tyler, J.A., Stevenson, R.J., Steen, P.J., Richards, P.L., Seelbach,
P.W., Koches, J.M., and Rediske, R.R. (2010). A multi-modeling approach to evaluating climate
and land use change impacts in a Great Lakes river basin. Hydrobiologia, 657, 243-262.
Wilkes, G., Edge, T., Gannon, V., Jokinen, C., Lyautey, E., Neumann, N., Ruecker, N., Scott, A.,
Sunohara, M., Topp, E., and Lapen, D. (2011). Associations among pathogenic bacteria,
parasites, and environmental and land use factors in multiple mixed-use watersheds. Water
Research, 45, 5807-5825.
Winfield, M.D. and Groisman, E.A. (2003). Role of nonhost environments in the lifestyles of
Salmonella and Escherichia coli. Applied and Environmental Microbiology, 69, 3687-3694.
Wolter, P.T., Johnston, C.A., and Niemi, G.J. (2006). Land use land cover change in the U.S.
Great Lakes Basin 1992 to 2001. Journal of Great Lakes Research, 32, 607-628.

65

Wong, M., Kumar, L., Jenkins, T.M., Xagoraraki, I., Phanikumar, M.S., and Rose, J.B. (2009).
Evaluation of public health risks at recreational beaches in Lake Michigan via detection of
enteric viruses and a human-specific bacteriological marker. Water Research, 43, 1137-1149.
Wu, J., Rees, P., and Dorner, S. (2011). Variability of E. coli density and sources in an urban
watershed. Journal of Water and Health, 9, 94-106.
Yamahara, K.M., Layton, B.A., Santoro, A.E., and Boehm, A.B. (2007). Beach sands along the
California coast are diffuse sources of fecal bacteria to coastal waters. Environmental Science
and Technology, 41, 4515-4521.
Yamahara, K.M., Walters, S.P., and Boehm, A.B. (2009). Growth of enterococci in unaltered,
unseeded beach sands subjected to tidal wetting. Applied and Environmental Microbiology, 75,
1517-1524.
Yampara-Iquise, H., Zheng, G., Jones, J.E., and Carson, C.A. (2008). Use of a Bacteroides
thetaiotaomicron-specific alpha-1-6, mannanase quantitative PCR to detect human faecal
pollution in water. Journal of Applied Microbiology, 105, 1686-1693.
Yoder, J.S., Blackburn, B.G., Craun, G.F., Hill, V., Levy, D.A., Chen, N., Lee, S.H., Calderon,
R.L., and Beach, M.J. (2004). Surveillance of waterborne-disease outbreaks associated with
recreational water - United States, 2001-2002. Morbidity and Mortality Weekly Report
(MMWR), CDC Surveillance Summaries. 53 (SS08); 1-22. Centers for Disease Control and
Prevention (CDC), Atlanta, Georgia.
Yoder, J.S., Hlavsa, M.C., Craun, G.F., Hill, V., Roberts, V., Yu, P.A., Hicks, L.A., Alexander,
N.T., Calderon, R.L., Roy, S.L., and Beach, M.J. (2008). Surveillance for waterborne disease and
outbreaks associated with recreational water use and other aquatic facility-associated health
events - United States, 2005-2006. Morbidity and Mortality Weekly Report (MMWR), CDC
Surveillance Summaries 57 (SS09); 1-29. Centers for Disease Control and Prevention (CDC),
Atlanta, Georgia.
Zehms, T.T., Mcdermott, C.M., and Kleinheinz, G.T. (2008). Microbial concentrations in sand
and their effect on beach water in Door County, Wisconsin. Journal of Great Lakes Research,
34, 524-534.
Zmirou, D., Pena, L., Ledrans, M., and Letertre, A. (2003). Risks associated with the
microbiological quality of bodies of fresh and marine water used for recreational purposes:
Summary estimates based on published epidemiological studies. Archives of Environmental
Health, 53, 703-711.

66

CHAPTER 2.
MICROBIAL INVESTIGATIONS OF WATER, SEDIMENT, AND ALGAE MATS IN A
MIXED USE WATERSHED

67

2.1. Introduction
Saginaw Bay, situated on the western shore of Lake Huron is an example of a mixed land use
watershed (Figure 2.1.). Like many bays across the United States, water is economically vital to
the region. The Bay averages 9.8 m deep, drains over 22,000 km2 (USEPA 2011), and hosts 43
public beaches. This Bay has numerous key pollution sources including four combined sewer
overflow systems (Saginaw Bay Science Committee Pathogen Work Group 2007) and an
unknown number of septic systems within the Saginaw Bay coastal zone. The Bay has been
heavily stressed by toxins (Yun and Kannan 2011), fish contamination (Jude et al. 2010),
nutrient loading (Cha et al. 2010), invasive species (Fahnenstiel et al. 1995), and changes in
natural phytoplankton populations (Fishman et al. 2009). In 1986, the Bay was added to the
United States Environmental Protection Agency’s (USEPA) Areas of Concern and the
Remediation Action Plan process began to address the impairment of beneficial uses (including
recreation) eutrophication/nuisance algae, degradation of aesthetics, and beach closures (USEPA
2011). Since 2002, 36 Saginaw Bay beaches have been monitored for E. coli weekly between
June and September, resulting in 894 closure/advisory days (MDEQ 2011). The causative agents
for these closures were categorized as unknown (93%), runoff (3%), and agriculture (4%). The
nearshore zone receives significant inputs of pollution including sediment contamination and
algae mat accumulations which produce highly visible impacts in the swimming area.

68

Figure 2.1. Location of beach sites in the Saginaw Bay selected for deep water, shallow water, sediment, and stranded algae mat
investigation using fecal indicator bacteria and molecular source tracking. For interpretation of the references to color in this and all
other figures, the reader is referred to the electronic version of this dissertation.
69

Algae mat accumulation on beaches is a growing concern for Great Lakes shorelines and the
Saginaw Bay. In the Great Lakes, algae grow rapidly during the warm summer months (May September). Following environmental forces, algae detaches from substrate to form free
floating mats. Many studies have shown algae mats provide a suitable habitat for enteric
bacteria such as Escherichia coli (E. coli), enterococci, Shigella, Campylobacter, and
Salmonella, allowing waterborne pathogens to persist and potentially regrow (Verhougstraete et
al. 2010; Byappanahalli et al. 2009; Ishii et al. 2006). As a result of wind and wave action, these
microorganisms can detach and enter surrounding waters which influence water quality
(Englebert et al. 2008b). In the last decade, algae mat occurrence in the Saginaw Bay has raised
concern about possible pathogen occurrence at beaches. Knowledge of pathogen occurrence in
algae mats or the impact on surrounding water quality and public health is not well understood
(Verhougstraete et al. 2010). To date, there have been no published studies investigating algae
mats for molecular source tracking markers to identify the origin of fecal pollution and potential
pathogens.

In addition to algae, sediments are being increasingly implicated as a cause of water quality
impairments at beaches. Whitman and Nevers (2003) reported nearshore sand had higher E. coli
-1

concentrations (4000 Colony Forming Units (CFU) 100 ml ) than the surrounding water (43
-1

CFU 100 ml ). Alm et al. (2003) found E. coli was up to 38 times higher in beach sand than the
water column and E. coli concentrations decreased with sediment depth. Garrido-Perez et al.
(2008) found proximity to contamination source had a significant influence on sediment bacterial
densities. Despite evidence linking beach water impacts and bacteria in sediment and algae mats,
recreational water quality monitoring continues to focus on deep water.
70

It is often not possible to directly test for multiple pathogenic microorganisms during routine
beach monitoring. A single fecal indicator is generally used as pathogen surrogates to
characterize water quality and protect human health (Griffen et al. 2001), limiting the ability to
fully characterize pollution and sources of fecal contamination. The Michigan Department of
Environmental Quality has set recreational water quality criteria for fresh water at 300 E. coli
100 ml

-1

as a single sample maximum or 130 E. coli 100 ml

-1

as a geometric mean of five or

more samples. The USEPA has also suggested enterococci as an indicator for evaluation of
public health risks for recreational waters (Wade et al. 2006; Wade et al. 2010). Clostridium
perfringens (C. perfringens) and coliphage viruses have also been used as fecal indicator
organisms. C. perfringens form spores that do not regrow in the environment and are resistant to
high temperatures and disinfection treatments (Payment and Franco 1993). Hawaii adopted a C.
perfringens regulatory standard of 50 CFU 100 ml

-1

(Mahin and Pancorbo 1999). Although no

regulatory standards exist for coliphage, a virus that infects E. coli, they are used to indicate the
presence of enteric pathogens in water (Allwood et al. 2003). Monitoring multiple fecal
indicators has been shown to provide better recreational water protection (Kinzelman et al.
2003), but routine beach monitoring continues to largely rely on a single indicator.

Molecular source tracking is used to define the cause of fecal pollution. Scott et al. (2005)
demonstrated Enterococci faecium surface protein (esp) gene was specific to human sewage and
septic systems while absent in swine, poultry, and cattle feces using cultivatable enterococci.
Bernhard and Field (2000) developed two Bacteroides - Prevotella ribosomal DNA markers
specific to cow and human feces. The Bacteroides species has proven useful for molecular
71

source tracking and in recent years improvements in specificity and analytical techniques allow
for more definitive results (Yampara-Iquise et al. 2008). In addition to routine monitoring and
source tracking, environmental surveys capture surrounding parameters not identified during
microbial and molecular analysis (Field and Samadpour 2007).

Current standards focus only on water and fail to address the multiple other types of pollution
threatening the nearshore including sediments and algae mats. Thus, there is a need to better
characterize microbial occurrence in the nearshore zones as no microbial standards exist for such
matrices. Using multiple emerging analyses across beach transects, this investigation aimed to:
1) determine the occurrence and relationship of fecal indicator bacteria in water, sediment, and
algae at beaches in a mixed use watershed; 2) identify human and bovine sources of fecal
contamination using microbial source tracking and; 3) identify the environmental parameters
influencing water quality degradation.

2.2. Materials and Methods
2.2.1. Description of study area
Four beaches were chosen based on their proximity to rivers with large drainage basins,
historically poor beach water quality, and summer popularity for swimming (Table 2.1. and
Figure 2.1.). Furthermore, nearshore algae mass accumulation has been recorded at SB1 and
SB2. Three equally spaced sites parallel to the shoreline were monitored at each beach. Beach
transects samples (sediment, shallow water, deep water, and algae mats) were collected
perpendicular to each of the three equally spaced sites.

72

Table 2.1. Site description, water quality exceedances, and potential pollution influences.
Site Location
ID

Site description
(Latitude/longitude)

SB1 Caseville
County Park
beach

(Huron County) Adjacent to the Pigeon
River which has been subject to fish kills
from sediment and agricultural runoff
(43.98964, -83.27540)

Water quality
standard
A
exceedances
5

Suspect
pollution
source(s)
Sewage
Agriculture
Algae

SB2 Bay City State (Bay County) North of the Saginaw River
Recreation
which receives significant inputs from
Area beach
CSOs and urban runoff
(43.67407, -83.90903)

9

Sewage
Algae

SB3 Whites Beach

21

Septage

(Arenac County) Surrounded by dense
residential homes relying on septic
systems for wastewater management
(43.92861, -83.89051)

SB4 Port Crescent (Huron County) Southwest of the
3
Agriculture
State Park-day Pinnebog River which receives inputs
use
from multiple agricultural drains
(44.00246, -83.06981)
A. Closures reported since the creation of Michigan Beach Guard database (circa 2001)

2.2.2. Sample collection and processing
Each beach was sampled eight times between June-September, 2008. Each event included
collection of shallow water, waist deep water, sediment, and, when present, stranded algae mats.
Using sterile one liter bottles, water column grab samples were collected at depths of 15-20 cm
(shallow) and 100 cm (deep) above lake bottom. Sediment and algae samples were collected in
the swash zone by inverting a Whirl-Pak®, grabbing a handful of material from three horizontal
points along the beach, and then compositing all subsamples in one bag. All Samples were
placed on ice (4 °C), stored in a cooler, transported to the Michigan State University Water

73

Quality, Environmental, and Molecular Microbiology Laboratory (East Lansing, Michigan,
USA) and processed within 24 hours.

2.2.3. Water analysis
Microbial analysis of water and sediment included E. coli, enterococci, Clostridium perfringens
(C. perfringens), and coliphage (CN-13 and F+amp). Undiluted water samples were filtered
through 0.45 µm hydrophilic mixed cellulose esters filters (Pall Corporation 66278). E. coli,
enterococci, and C. perfringens were analyzed using cultivation and selective media mTEC
(USEPA 2005), mEI (USEPA 2002), and mCP (USEPA 1995; Bisson and Cabelli 1979),
respectively. E. coli, enterococci, and C. perfringens were reported as colony forming units 100
-1

ml . Double agar layers were utilized to detect two coliphage strains following EPA methods
1601 (USEPA 2001). The two selected bacterial hosts were E. coli F+amp and E. coli CN-13.
-1

Clearings in the host lawn were counted and reported as plaque forming units (PFU) 100 ml .
Escherichia coli C-3000 (ATCC 15597), Enterococci faecium (ATCC 35667), Clostridium
perfringens (ATCC 3624), ΦX-174 coliphage were used as a positive controls for verification of
media integrity. Sterile water was used as negative controls for verification of method integrity.

2.2.4. Sediment and algae analysis
Sediment and algae samples were diluted with sterile Phosphate Buffered Water (PBW) to a final
weight/volume ratio of 10% and 1%, respectively, to obtain countable results. Algae mat samples
underwent an initial pulse blending in a sterile blender until homogenized. Each sample and
PBW solution was vigorously hand shaken (10 cm radius) for two minutes, allowed to settle for
2 minutes, and the eluent was decanted into a sterile bottle (Shibata et al. 2004; Boehm et al.,
74

2009). An additional volume of PBW was added to the sediment or algae sample, swirled for 10
seconds, allowed to settle for 30 seconds, and the eluent was added to the sterile bottle from the
first rinse. E. coli, Enterococci, C. perfringens, and coliphage were assayed from the diluted
sediment or algae solution using the same methods described in the water analysis section and
reported as colony or plaque forming units per gram wet weight of material.

2.2.5. Molecular analysis
Water, sediment, and algae samples were analyzed for enterococci surface protein (esp) gene and
Bacteroides human and bovine specific markers using PCR. The marker, primer sequence, and
product size for each assay are described in Table 2.2. The esp analysis was performed from the
CFU membranes grown on mEI during enterococci cultivation (described above). For the
Bacteroides markers, water and eluted sediment and algae were filtered and DNA was extracted
directly from filters containing non-cultivated cells. For all molecular assays, each filter was
placed into a 50 ml centrifuge tube containing 20 ml of sterile PBW, vortexed to recover the
CFU and cells from the membrane, and then centrifuged (30 minutes; 4000 x g) to pellet the
cells. Eighteen ml were decanted from the tube and the remaining eluent and pellet were stored
at -80°C until DNA extraction.

75

Table 2.2. PCR assays examined during source tracking in water, sediment, and algae.
Marker name

Primer sequence

Product Reference
size (bp)
680
Scott et al. 2005

Enterococci
surface protein

F 5ʹ-TATGAAAGCAACAGCACAAGTT-3ʹ
R 5ʹ-ACGTCGAAAGTTCGATTTCC-3ʹ

Human
Bacteroides

F 5’-ATCATGAGTTCACATGTCCG-3’
R 5’-CAA TCG GAG TTC TTC GTG-3’

116

Bernhard and
Field 2000

Bovine
Bacteroides

F 5’-CCAACY TTCCCG WTACTC-3’
R 5’-CAATCGGAGTTCTTCGTG-3’

100

Bernhard and
Field 2000

After thawing, approximately 50 μl of DNA was extracted from 200 μl of the pellet using
QIAamp® DNA mini kit. Bacteroides were assayed using polymerase chain reaction (PCR)
amplification according to Bernhard and Field (2000). Briefly, analysis were carried out using
forward and references primers (0.4 μM), MgCl2 (3 mM), HotStarTaq Master Mix, and
molecular grade water to make up a final volume of 22 μl. The bovine and human Bacteroides
assays were processed using a Bio-Rad PCR thermocycler (iCycler) with a 15 min initial
warming step (95°C), followed by 30 cycles of the amplification step (94°C for 30 s, 58°C for 30
s, and 72°C for 60 s), and a final extension step of 8 min (72°C).

Esp was assayed using PCR amplification according to Scott et al. (2005). Briefly, master mix is
prepared with forward and reverse primers (3 μM), HotStarTaq Master Mix (Qiagen 203443),
and molecular grade water to make up a final volume of 19 μl. Using a Bio-Rad PCR
thermocycler (iCycler), esp analysis were carried out under the following conditions: 15min at
95°C, 35 cycles each consisting of 1min at 94°C, 1min at 58°C, and 1min at 72°C, followed by a
final extension step for 5min at 72°C.

76

DNA from sewage, cow manure, and E. faecium EL1 templates were used as positive controls
for human and bovine Bacteroides and esp, respectively. Molecular grade water was used as a
negative control for each analysis. Gel electrophoresis for each assay was performed on the PCR
product in duplicates, run on a 1.2% w/v agarose gel at 95 V for one hour. Samples with bands at
116 base pairs (bp), 100 bp, and 680 bp were recorded as positives for human Bacteroides,
bovine Bacteroides, and esp, respectively.

2.2.6. Environmental and physical data
During sample collection, wave height was measured as the distance from the trough to the crest.
Visual counts of birds on the beach or in the swimming area were also made during sample
collection. Precipitation (24, 48, and 72 hour cumulative totals prior to the sampling date) and
wind direction and speed at the date and time of sampling were collected from Enviro-weather
(www.agweather.geo.msu.edu/mawn/) and NOAA National Weather Service (weather.gov)
observation and forecast reports. Observational data for SB2 and SB3 were collected from
Enviro-weather station LIN (43.7199, -84.0275) and data for SB1 and SB4 were collected from
PIG (43.8992, -83.2667). Daily and hourly observations of total precipitation (mm), wind
-1

direction (degress, north = 0°), and wind speed (m s ) were available at these stations
throughout the duration of the project.

2.2.7. Data analysis
Non-detects and concentrations below method detection limits for water and sediment/algae
were calculated using equation 1 and 2, respectively, and recorded as the lowest detection limit.
77

1.

(

)

2.

Where Vt is the total volume used for analysis, Vd is the dilution volume, and M is the mass of
sediment or algae used for dilution and analysis. Results that exceeded the upper detection limit
of applied methods were recorded as the upper method detection limit. Microorganism
concentrations underwent log-transformation to fit the data to a normal distribution, but not one
transformation satisfied all parameters. All water, sediment, and algae microbial results were
evaluated for intra-site relatedness and associations with environmental parameters using
independent samples-Kruskal-Wallis 1 way ANOVA and bivariate Spearman Rank (r)
correlation coefficients. All tests were performed using SPSS Statistic 17.0 software with
significance set at α = 0.05.

2.3. Results
2.3.1. Fecal indicator organisms in water and sediment
Ninety-six samples were collected from three zones (deep and shallow water, sediment) at four
Saginaw Bay beaches (9 stranded algae mat samples were collected from two Saginaw Bay
beaches when present, see section 2.3.2). Each sample was assayed for the fecal indicators E.
coli, enterococci, C. perfringens, and coliphage CN-13 and F+amp. At least one fecal indicator
was detected during every sample event at each beach and in each zone. Figure 2.2. compares the
geometric mean concentrations of each indicator for the three zones at each beach.
78

Each beach was ranked based on the geometric means, single sample maximum, and percent
positive detections for each of the five fecal indicators in sediment, shallow water, and deep
water. Based on this ranking scheme, SB1 was determined the most contaminated site. All
sediment, deep, and shallow water samples collected at SB1 were positive for E. coli,
enterococci, and C. perfringens. At SB1, the E. coli, enterococci, and C. perfringens geometric
-1

means were 0.41, 0.36, 0.75 log10 CFU 100 g (sediment); 2.0, 1.7, and 0.75 log10 CFU 100 ml
1

(shallow water); and 0.72, 0.51, and 0.27 log10 CFU 100 ml

-1

-

(deep water), respectively. The

-1

-1

highest coliphage F+amp (2.56 log10 PFU 100 g ) and CN-13 (3.00 log10 PFU 100 g ) single
sample measurements were recorded in the sediment at SB1. The SB1 geometric means for
coliphage F+amp in the sediment was 0.34 log10 PFU 100 g

-1

and 1.0 log10 PFU 100 ml

-1

in

shallow and deep water samples. The SB1 geometric means for coliphage CN-13 in the sediment
-1

-1

was 0.37 log10 PFU 100 g and 1.3 log10 PFU 100 ml shallow and deep water samples.

The site ranking second most impacted was SB3. At SB3, the E. coli, enterococci, and C.
perfringens geometric means were 0.86, 0.73, and 0.50 log10 CFU 100 g
1.60, and 0.51 log10 CFU 100 ml
1

-1

-1

(sediment); 1.64,

(shallow water); and 1.01, 0.57, and 0.13 log10 CFU 100 ml

-

(deep water), respectively. This beach had the highest percentage of samples positive for CN-

13 (70.6%) as well as the highest single sample concentration of C. perfringens (2.82 log10 CFU
-1

100 g ) measured in the sediment. The SB3 geometric means for coliphage F+amp in the

79

sediment was 0.28 log10 PFU 100 g

-1

and 1.0 log10 PFU 100 ml

-1

in shallow and deep water,

respectively. The SB3 geometric means for coliphage CN-13 in the sediment was 0.41 log10
PFU 100 g

-1

-1

and in the shallow and deep water samples 1.6 and 1.1 log10 PFU 100 ml ,

respectively.

The site ranking third most impacted was SB2. At this site E. coli, enterococci, and C.
-1

perfringens geometric means were 0.87, 1.1, and 0.31 log10 CFU 100 g
and 0.99 log10 CFU 100 ml

-1

(sediment); 2.2, 1.5,

(shallow water); and 0.65, 0.35, and 0.38 log10 CFU 100 ml

-1

(deep water), respectively. The SB2 coliphage F+amp geometric mean in the sediment was 0.31
-1

log10 PFU 100 g and in the shallow and deep water samples the means were 1.1 and 1.0 log10
-1

PFU 100 ml , respectively. The SB2 geometric means for coliphage CN-13 in the sediment, was
0.31 log10 PFU 100 g

-1

and in the shallow and deep water samples 1.3 and 1.0 log10 PFU 100

-1

ml , respectively. In comparison to the other three beaches and their respective zones, SB2 had
the highest geometric means for E. coli (sediment), enterococci (shallow water), and C.
perfringens (shallow and deep water). Furthermore, the highest single sample measurements of
-1

E. coli and enterococci of the entire project (> 4.34 log10 CFU 100 g ) were recorded in the
sediment at SB2.

The site ranking fourth and least impacted was SB4. At this site E. coli, enterococci, and C.
perfringens geometric means were 0.09, 0.07, and 0.01 log10 CFU 100 g
80

-1

(sediment); 1.06,

0.47, 0.32 log10 CFU 100 ml

-1

(shallow water); and 0.99, 0.35, and 0.23 log10 CFU 100 ml

-1

(deep water), respectively. The lowest E. coli, enterococci, and C. perfringens geometric means
in shallow water and sediment were measured at SB4. However, this site had the highest
percentage of samples positive for F+amp (35.3%) across all zones. The SB4 geometric means
for coliphage F+amp in the sediment was 0.32 log10 PFU 100 g

-1

and in the shallow and deep

-1

water samples 1.1 and 1.3 log10 PFU 100 ml , respectively. The geometric means for coliphage
CN-13 in the sediment was 0.42 log10 PFU 100 g

-1

and in the shallow and deep water samples

-1

1.4 and 1.3 log10 PFU 100 ml , respectively.

In summary, SB1 was the most polluted site followed by SB3, SB2, and SB4 based on geometric
means, total number of positive samples, and single sample maximum of all indicators. In
comparing beaches, E. coli, enterococci, and C. perfringens concentrations were statistically
different amongst all beaches in sediment and deep water (p ≤ 0.017 and p ≤ 0.024, respectively).
However, enterococcus was the only indicator different across all beaches in the shallow water
(p ≤ 0.026). Although as a whole, coliphage means were different between zones, they were not
statistically different between beaches (p > 0.05); likely due to the detection limits, small assay
volumes (~10 ml), and the number of non-detects (72.7%).

81

SB 1
SB 2
SB 3
SB 4

2.5
2.0

3.0

E. coli

Log10 concentration
(CFU 100 ml-1 or CFU 100 g-1)

Log10 concentration
(CFU 100 ml-1 or CFU 100 g-1)

3.0

1.5
1.0
0.5
0.0

2.5
2.0

2.5
2.0
1.5
1.0
0.5
0.0

Deep

Shallow

Sediment

3.0

C. perfringens

Log10 concentration
(PFU 100 ml-1 or PFU 100 g-1)

Log10 concentration
(CFU 100 ml-1 or CFU 100 g-1)

3.0

Enterococci

SB 1
SB 2
SB 3
SB 4

1.5
1.0
0.5
0.0

Deep

Shallow

Sediment
Coliphage F+amp

2.5
2.0
1.5
1.0
0.5
0.0

Deep

Shallow

Sediment

Deep

Shallow

Sediment

Figure 2.2. Geometric mean concentrations of fecal indicator organisms in shallow water (20 cm; n = 32), deep water (1 meter; n =
32), and sediment (n = 32) at SB1, SB2, SB3, and SB4. : Short error bars result of small C. perfringens standard deviations;
:
No standard deviation due to high percentage of non-detects; Non-detects were recorded at the detection limit; Water reported as log10
-1

-1

CFU or PFU 100 ml ; Sediment reported as log10 CFU or PFU 100 g wet weight.
82

Figure 2.2. (cont’d)

Log10 concentration
(PFU 100 ml-1 or PFU 100 g-1)

3.0

Coliphage CN-13

2.5
2.0
1.5
1.0
0.5
0.0
Deep

Shallow

Sediment

The first objective was to determine the occurrence and relationship of fecal indicator bacteria in
water, sediment, and algae. E. coli, enterococci, and C. perfringens were detected in 99, 88, and
87 percent of all (i.e. water, sediment, and algae) tested samples, respectively. With respect to all
water, sediment, and algae samples, coliphage F+amp and CN-13 were detected in 15 and 45
percent of samples, respectively. Not surprisingly, deep water had the lowest number of
enterococci, C. perfringens, and coliphage CN-13. Kruskal-Wallis tests confirmed all
microorganisms were statistically different across zones (p ≤ 0.005). Based on E. coli,
enterococci, and C. perfringens concentrations, the most contaminated zone was sediment
followed by shallow water and finally deep water. However, E. coli concentration ranges in each
zone showed shallow water quality was the most variable (ranged over 2.78 log10 CFU 100 ml
1

-

-1

), followed by deep water (ranged over 1.75 log10 CFU 100 ml ), and sediment (ranged over
-1

0.50 log10 CFU 100 g ). This trend was also found with enterococci, C. perfringens, and
83

coliphage CN-13. However, coliphage F+amp deviated from this trend and were most variable in
deep water followed by sediment and then shallow water, but the large percentage of F+amp
non-detects likely influenced such results.

Comparisons of microorganism concentrations across zones were undertaken to better
characterize each zone and identify potential interactions. Sediment-shallow water pairwise
comparisons indicated that 70% of sediment samples had higher E. coli levels compared to the
shallow water. This percentage increased with the examination of enterococci (81%) and C.
perfringens (90%). The high number of F+amp (100%), and CN-13 non-detects produced less
meaningful results and were not considered statistically significant. Shallow-deep water pairwise
comparisons found higher levels of E. coli, enterococci, and C. perfringens in shallow water
during 90%, 67%, and 83% of samples, respectively. Using CN-13 and F+amp, shallow water
was more contaminated than deep water in 50% and 8% of paired samples (n = 12), respectively.
The low F+amp percentage occurred because of the significant number of non-detects in both
zones (83% deep and 94% shallow). In comparison, coliphage CN-13 was not detected in the
38% and 75% of shallow and deep water samples, respectively. Cumulatively, these results
indicate sediment acts as a sink for fecal indicator bacteria in the nearshore of Saginaw Bay
which can occasionally be released to the shallow water with some eventually entering deep
water.

Correlation analysis attempted to define associations of bacteria to each other across the beach
transect. All the bacteria were positively correlated with each other in the sediments and shallow
water but not in the deep water. The correlation coefficient between E. coli and enterococci in

84

the sediment was r = 0.760 (p = 0.001) and in shallow water r = 0.585 (p = 0.001). Correlations
between E. coli and C. perfringens in the sediment was r = 0.640 (p = 0.002) and r = 0.379 (p =
0.032) in the shallow water. Between enterococci and C. perfringens in the sediment r = 0.617 (p
= 0.004) and r = 0.453 (p = 0.009) in the shallow water. The greatest correlations were found
between the two gram negative bacteria (E. coli and enterococci) in sediments followed by
shallow water.

F+amp and CN13 were positively correlated with each other in deep water (r = 0.704, p = 0.011),
but showed no relationships to each other in shallow water, sediments, nor were they correlated
to any other bacterial indicators.

2.3.2. Occurrence of fecal indicator organisms in algae and source tracking markers
Algae mats were present and collected at SB1 (n = 2) and SB2 (n = 7), but absent at sites SB3
and SB4. Higher levels of all indicator bacteria were found in algae from SB2 compared to SB1.
All algae mat samples were reported in grams wet weight. At SB1, algae mats exhibited
geometric means of E. coli , enterococci, C. perfringens, coliphage F+amp, and coliphage CN-13
-1

-1

-1

of 1.23 CFU 100 g wet weight, 2.11 CFU 100 g wet weight, -0.91 CFU 100 g wet weight,
-1

-1

0.07 PFU 100 g wet weight, and 0.07 PFU 100 g wet weight, respectively. At SB1, algae mat
geometric mean concentrations of E. coli, enterococci, C. perfringens, colipohage F+amp, and
coliphage CN-13 were 2.20 CFU 100 g
100 g

-1

-1

wet weight, 2.57 CFU 100 g

-1

wet weight, 2.02 PFU 100 g

-1

wet weight, 1.90 CFU
-1

wet weight, and 1.97 PFU 100 g

wet weight,

respectively. When algae mats were present, 85% of all microbial measurements (including
85

coliphage) at that site were higher in the algae mat than the underlying sediment. Likewise, when
algae mats were present, 55% of all microbial measurements at that site were higher in the algae
mats than shallow water; demonstrating a shift in microbe concentration trends compared to sites
and dates when algae were not present during sample collection. E. coli, enterococci, and C.
perfringens were detected in every algae mat sample (Table 2.3.). No statistical correlations were
identified between fecal indicators in algae mats as statistical analysis was likely limited by
sample size.

86

Table 2.3. Range and geometric mean concentrations of fecal indicator organisms in algae samples at SB1 (n =2) and SB2 (n = 7).
Site

Zone

E. coli
(log10)
CFU 100 g

-1

-1

SB1 Algae

Sediment
Shallow water
Deep water***
SB2 Algae

Sediment
Shallow water
Deep water

or 100 ml
Range
0.71-1.85
Mean**
1.15
% positive
100
Mean**
Mean**
Mean**
Range
Mean**
% positive
Mean**
Mean**
Mean**

Enterococci
(log10)

C. perfringens Coliphage F+amp
(log10)
(log10)

Coliphage CN-13
(log10)

CFU 100 g

CFU 100 g

PFU 100 g or

-1

-1

or 100 ml
1.98-2.25
2.11
100

or 100 ml
0.05
100

-1

-1

PFU 100 g or

-1

100 ml

-1

Esp
(+/-)

-1

100 ml

-1

0.33
0

0.33
0

0.22
1.46
0.59

0.40
0.83

0.16
0.75

0.28
1.04

0.28
1.04

1.08-2.92
2.17
100
0.96
2.16
0.78

0.97-3.38
2.52
100
1.23
1.80
0.33

0.73-4.04
1.62
100
0.31
1.09
0.68

1.04-3.52
1.89
60
0.31
1.07
1.04

1.04-3.52
1.84
60
0.31
1.25
1.04

1/1*

-1

1/6

Sediment and algae reported in wet weight; coliphage detection limit: 90 PFU 100 g wet weight; *ESP also detected in one sediment
sample at SB1 (9-30-2008). **Geometric mean. ***Deep water samples were collected by the local health department on the two
days when algae mats were present and processed for E. coli. Sediment, shallow water, and deep water geometric mean concentrations
when algae mats were present are shown for comparison purposes.

87

Another goal was to identify the source(s) of pollution. Thus, human and bovine Bacteroides and
enterococci surface protein gene markers were employed. Twenty-seven samples were tested for
the Bacteroides marker and 48 samples were assayed for the esp gene. The esp gene was
identified in three samples, all non-water samples (i.e. algae and sediment) at SB1 and SB2.
Human and bovine Bacteroides were not identified in any of the samples collected in 2008.
However, bovine and human feces were identified in water and algae mats sampled from
Saginaw Bay beaches in 2007 using the same markers and methods (Singh and Rose 2007).

2.3.3. Fecal indicator associations with environmental parameters
As previously described, non-parametric correlation analysis were used to determine significant
relationships between microbial concentrations and environmental conditions (objective 3). The
most recurrent variables related to indicator concentrations were wave height and wind
speed/direction. Precipitation (24 and 72 hour totals) was the next most recurring variable.
Occasionally, bird populations (at the beach or in the water) and temperature (water, mean daily
air temperature) were significant parameters associated with fecal indicator densities. The top
two most influential parameters at each beach and zone are presented in Table 2.4.

88

Table 2.4. Significant correlations identified between fecal indicator organisms and environmental parameters.
Site

Zone

Fecal
Parameters
indicator

Fecal
indicator

Parameters

SB 1

Shallow

Entero.

0.028

C. per.

0.862

0.006

0.798
0.845
0.849

0.018
0.033
0.008

C

0.810

0.015

A

0.768

0.026

Precip.

r

p

B

0.762

A

Sediment

E. coli

Precip.

SB 2

Deep
Shallow
Sediment

Entero.
E. coli
E. coli

Wind speed
Wave height
A
Precip.

SB 3

Shallow

CN-13

Temp.

Sediment

C. per.

Precip.

Deep
Shallow
Sediment

r

p

Birds

0.732

0.039

E. coli

Wave height

0.764

0.027

C. per.
Entero.
F+amp/CN13

Wind speed
Wind direction
B
Precip.

0.833
0.831
0.898

0.010
0.011
0.002

E. coli

Wind speed

-0.708

0.050

E. coli
Entero.
CN-13

SB 4

Wind speed
-0.812 0.050
Wave height -0.864 0.006 E. coli
Wave height
-0.832 0.010
A
0.788 0.020 CN13
Wind speed
-0.791 0.019
Precip.
A. 24 hour precipitation; B. 72 hour precipitation; C. Mean daily air temperature; D. Water temperature at time of sampling. SB1:
Caseville County Park; SB2: Bay City State Recreation Area; SB3: Whites Beach; SB4: Port Crescent State Park. r: bivariate
Spearman Rank correlation coefficient and p values ≤ 0.05. No significant associations were found in deep water samples at SB1 and
SB3.

89

Microbial concentrations were influenced by wind or waves in at least one zone of every beach.
However, the association strength between these parameters and microbes varied throughout the
Saginaw Bay depending on the beach, which will be described in more detail below. Wave
height ranged from 0.0 to 0.46 m with an average of 0.14 m. The highest wave heights were all
recorded at SB4 on July 22 and September 30, 2008. When wave height was above average (>
0.14m), deep water enterococci concentrations were statistically lower (mean = 0.38 log10 CFU
-1

-1

100 ml ) than when wave height was below average ( X = 0.88 log10 CFU 100 ml ). Wind
A

E

A

speed ranged from 0.0 to 14.5 kmh with an average of 3.89 kmh out of the south-southwest
(212.6°). The highest wind speed measurements were recorded at SB2 on July 15 and September
30, 2008. The lowest wind speeds (0.0 kmh) were recorded at SB1 and SB3 on 7 different dates.

At the start of the project a wet weather threshold was set at 6.4 mm of cumulative rainfall in the
48 hours prior to sample collection, per local health department recommendations. Wet weather
events above this thresholds accounted for 40% of all sampling dates. The greatest variation
between wet and dry weather concentrations were seen in the shallow water of SB1 (range 2.82
-1

log10 Enterococci 100 ml ). During wet weather monitoring, deep water enterococci
concentrations averaged 0.31 log10 CFU 100 ml

-1

and were statistically higher (p = 0.015) than
-1

enterococci measured during dry weather in the same zone (mean of 0.05 log10 CFU 100 ml ).
Interestingly, twenty-four hour precipitation totals were related to an increase in one or more
fecal indicators in the sediment at all sites. Specifically, coliphage F+amp and CN-13
concentrations were statistically different between wet and dry conditions in the sediment (p <

90

0.001). Outside of enterococci in deep water and both coliphage in the sediment, no other
indicators were statistically different between wet and dry conditions (p ≥ 0.102) in any zone.

Sites SB1 and SB2 showed similar responses to environmental parameters. At SB1, precipitation
(72 hour) influenced microbe concentrations in the shallow water, while precipitation (24 hour)
and wave height were the most influential parameters on sediment microbes. At SB2, wind speed
was the primary influence on enterococci and C. perfringens in deep water. In the shallow water,
wave height, precipitation (24 hour), and wind direction were directly correlated with E. coli. In
algae mats, E. coli concentrations were indirectly correlated with wind direction (r = -0.937, p =
0.002). At site SB3, shallow water coliphage concentrations were positively influenced by daily
mean air temperature. C. perfringens and coliphage CN-13 concentrations in sediment at SB3
and SB4 were directly influenced by precipitation (24 hour).

SB4 showed very difference responses to the environmental parameters associated with wind and
the E. coli and enterococci (as well as coliphage in sediments) concentrations were inversely
influenced by wind speed and wave height. At SB4, deep water E. coli concentrations were
inversely associated with wind speed and shallow water enterococci and E. coli were inversely
associated with wave height.

2.4. Discussion
This study aimed to determine the occurrence of fecal indicators and define their relationships
between various areas across the beach. As expected, and previously shown (Whitman et al.

91

2003; Ishii et al. 2007), E. coli, enterococci, C. perfringens, and coliphage (CN-13 and F+amp)
were highest in algae mats and sediment. In this study, E. coli, enterococci, C. perfringens, and
coliphage levels were routinely 1 log greater in sediments than shallow water, also shown by
Whitman and Nevers (2003), Alm et al. (2006), and Ishii et al. (2007) regardless of the presence
of algal mats. The results from the current study supported previous studies attributing elevated
bacteria in shallow waters to sediment and algae mats occurrence (Whitman and Nevers 2003;
Boehm 2007; Engelbert et al. 2008b; Whitman et al. 2011) and found bacteria diminished with
increasing water depth (Whitman and Nevers 2004). Measured levels of C. perfringens indicate
chronic pollution in algae and sediment zones since C. perfringens persists and do not readily
regrow in water environments (Fujioka and Shizumura 1985; Davies et al. 1995; Desmarais et al.
2002). On the other hand, coliphage was detected in only about 23% of all sediment, shallow
water, and deep water samples, indicating sporadic fecal contamination at selected beaches.

Although E. coli, enterococci, and C. perfringens were consistently present in algae mats and
sediments, recent fecal contamination was not always suspected. Previous findings demonstrate
bacterial regrowth, accumulation, and persistence in algae mats absent of fresh fecal material
(Byappanahalli et al. 2003; Ishii et al. 2006a; Englebert et al. 2008a; Byappanahalli et al. 2009).
Detection of the esp gene in algae mats (SB2) when good water quality was expected (i.e. no
recorded rainfall in previous 5 days, above average air temperatures, and generally calm wind
conditions) supports the concept that bacteria are growing in the Saginaw Bay.

Fecal indicator bacteria in these media create a conundrum for beach managers and researchers
attempting to identify and remediate pollution sources. For instance, Whitman and Nevers (2003)

92

removed and replaced sand at one beach only to find that E. coli had recolonized within two
weeks. Eliminating bacteria in sediments or algae mats is not feasible or practical at any natural
beach. However, using best management practices including moderate beach grooming, and
removing stranded algae mat can improve water quality. Based on identified drivers it is
recommended that beach grooming measures be undertaken in the Saginaw Bay during minimal
wind and wave action, in absence of recent or near future precipitation, and in the morning to
allow for sunlight inactivation of bacteria prior to peak bather loads.

Overall, microorganism correlation coefficients indicated microbial levels in sediments were
influencing shallow water microorganism concentrations. However, most of the previous studies
focused on E. coli and enterococci which continuously fail to represent the true presence of fecal
contamination or pathogens as shown by Yamahara et al. (2012) who identified weak
associations between fecal indicator organisms (E. coli, enterococci, and F+amp coliphage) and
pathogen (Salmonella spp., Campylobacter spp., Staphylococcus aureus, and methicillinresistant S. aureus) presence in sand from 53 California beaches. In the current study two types
of coliphage (somatic and F+) were used because they have been suggested as surrogates for
human enteric viruses (Wiedenmann et al. 2006; Krometis et al. 2010) and have exhibited strong
associations with pathogens (Borrego et al. 1987), noroviruses (Allwood et al. 2003), and illness
outcomes following exposure in water (Colford et al. 2007). Krometis et al. (2010) illustrated
that somatic coliphage were less associated with particles and were more likely to remain in
suspension for longer periods of time compared to other measured microorganisms including F+
coliphage. This may be the reason coliphage were not found readily in the sediments of Saginaw
Bay.

93

Collectively, these results indicate sediment and algae mats acted as non-point sources of fecal
indicator bacteria and influence shallow water in the Great Lakes. However, coupling detected
bacteria results (E. coli, enterococci, and C. perfringens) with the overwhelming number of
coliphage non-detects, and the weak correlations between indicators and pathogens in sand
(Yamahara et al. 2012), does not explicitly imply that pathogens were consistently present
throughout the Saginaw Bay. Therefore, we recommend future pathogen testing be included in
parallel with beach monitoring across the beachscape.

-1

No beach exceeded 300 E. coli 100 ml , Michigan’s single sample maximum standard for
recreational water, in deep water. Additionally, there were no samples collected from deep water
-1

that exceeded 61 enterococci 100 ml , EPA’s single sample maximum criterion for recreational
water. However, in shallow water, seven samples exceeded E. coli standard and 12 samples
exceeded enterococci criteria. In total, 27% and 40% of shallow water samples exceeded E. coli
standard and enterococci criteria, respectively. Seventy-one percent of samples exceeding the E.
coli standard were associated with at least one measurement of wave height, wind speed, and
precipitation (24 hour) above average. Eighty-three percent of samples exceeding the enterococci
criteria were associated with at least one measurement of wave height, wind speed, and
precipitation (24 hour) above average. Interestingly, six (86%) E. coli exceedances were
associated with offshore wind direction relative to each beach with the one exception occurring
at SB1 during above average wind speed (5.6 kmh) directed onshore. E. coli and enterococci
concentrations agreed with regulatory outcomes during 76% of samples with respect to their
individual criterion (i.e. both indicators either meet or exceeded thresholds). An additional five
94

exceedances would have occurred if monitoring included only enterococci in respect to
monitoring only for E. coli. Coliphage F+amp were not detected in the same shallow water
samples collected during the E. coli and enterococci exceedances. However, coliphage CN-13
measurements were above the coliphage CN-13 geometric mean during four E. coli and six
enterococci exceedances. There were no shallow or deep water quality exceedances based on the
-1

Hawaii surface water standard for C. perfringens (50 CFU 100 ml ). In summary, all
exceedances for E. coli and enterococci were reported in the shallow water, currently an
unmanaged source of contamination which evades current recreational water regulations. In
order to adequately protect bathers, samples should be collected in the shallow water (15-20 cm).

Molecular source tracking methods employed at each beach failed to routinely identify the
source of fecal contamination. The enterococci surface protein (esp) marker was detected in
sediments and algae, suggestive of human fecal material present in a small percentage of samples
(6%). However, it is not clear if this was due to Enterococci spp. regrowth, as described above,
or the addition of recent fecal contamination. This method detects the esp gene present in
cultivated enterococci which are absent in chlorinated wastewater and in water with generally
less than 100 enterococci CFU per 100 ml (Masago et al. 2011). Therefore, the esp marker may
be better suited for point source dominated watersheds where examinations focus on disinfected
versus non-disinfected wastewaters.

The fact that bovine and human specific markers were regularly absent throughout this project
indicates either poor method approaches or other significant contributing sources in the Saginaw
Bay. Identified limitations of molecular methods include: reduced assay volumes during
95

filtration from excess suspended solids which were exacerbated by blending algae samples and
likely increased inhibiting substances (Girones et al. 2010; Toze 1999); method results used in
this study were presence/absence and gave no quantitative measurements (Villari et al. 1998);
and methods likely produced a high number of false negatives (Toze 1999; Yang and Rothman
2004). Previous sample collections at the same beaches were successful at detecting human and
bovine sources of contamination in water and algae mats. However, results were not duplicated
in the current study using similar processing methods. This may suggest an intermediate
presence of bovine feces in the Saginaw Bay. Wildlife, domesticated pets, cattle, and endemic
waterfowl populations have been recorded in the Saginaw Bay watershed (Johnson et al. 1997;
Singh and Rose 2007; Kraus et al. 2009), suggesting other animals are significant contributors to
fecal pollution in the Bay and signaling the need for additional source markers. Fecal source
tracking is further complicated by constantly changing source inputs, hydrology, environmental
influences, and algae mat source/occurrence (Fishman et al. 2009). Given the large amount of
variability in the Saginaw Bay dynamics and source tracking techniques (Girones et al. 2010),
detecting a single source of contamination illustrates the need for more frequent monitoring
using improved source tracking methods (i.e. qPCR) and markers including B. thetaiotaomicron
(Yampara-Iquise et al. 2008) and Norovirus (Wolf et al. 2010).

The final objective of this project focused on identifying environmental factors associated with
water quality in the Saginaw Bay. The predominant mechanisms driving microorganism
concentrations and water quality degradation in the Saginaw Bay were associated with wind and
waves. These dependent variables represent a source of energy often associated with impaired
beaches (Frick et al. 2008). In the shallow water of Saginaw Bay, surface water currents respond

96

rapidly to wind (within minutes) and persist for approximately eight hours before returning to
normal current patterns dominated by water entering and leaving the bay (Danek and Saylor
1977). During the current study, average wind direction at each beach followed similar patterns
described by Danek and Saylor (1977) with average wind directed offshore at SB3, parallel to
shore at SB1 and SB2, and onshore at SB4. Although wind and waves were influential at all
beaches, they were most visible at SB2 between algae mats and shallow water since SB2
routinely had the highest wind speeds ( X = 7.4 KPH) and stranded algae mats were consistently
A

E

A

present in the shallow water. Additionally, E. coli and coliphage CN-13 levels in the sediment at
SB3 and SB4 were inversely related to wind speed and wave height, respectively, indicating
bacteria settle out of the water column during low energy conditions. This cyclical process of
deposition, accumulation, and resuspension explains the abundance of fecal indicator bacteria in
shallow waters even in absence of fresh fecal inputs. Due to the inconsistent presence of algae
mats at SB1 and the continuous presence at SB2, it was difficult to determine what role wind and
wave action have on algae presence or associated bacteria concentrations. Wind induced surface
water currents drive the movement of pollution throughout the waterbody until the polluted
water becomes detected in the nearshore zone, but the specific current dynamics (i.e. parallel
perpendicular to the shore) could not be identified under the current project design. Additional
analysis focused on measuring microbial concentrations and nearshore currents at hourly
intervals (or potentially shorter) is required to define such movement in the nearshore.

Precipitation was another recurrent influential environmental factor throughout the Saginaw Bay.
Overall monthly precipitation totals in the Saginaw Bay during the project (July – 81 mm;
August – 69 mm; September – 105 mm) were similar to long-term (1899 and 2011) monthly

97

averages (July – 76 mm; August – 91 mm; September – 97 mm) (retrieved on July 2, 2012 from
weather.com). However, 24 and 72 hour precipitation totals prior to sample collection averaged
just 2.5 mm and 12.4 mm, respectively, indicating a significant portion of rainfall events were
not captured during this project. Precipitation transports pollution from the land to surfaces
waters which eventually enters the Bay primarily via rivers. Increased fecal indicator levels in
water were seen 24-72 hour after rainstorms, indicating transport mechanisms take 1-3 days to
transfer pollution from land to beach. Investigations aimed at rainfall intensity may provide
further insight into precipitation effect on microbial water quality, especially important under
climate change predictions for the Great Lakes which include increases in precipitation intensity
and dry periods between rainstorm events (Mortsch et al. 2003). If Saginaw Bay beach managers
continue using E. coli cultivation methods, these results suggest a sample collection shift of at
least one day following rain storms and illustrate the need for predictive beach water quality
models that incorporate wind and precipitation (Ge et al. 2012) to improve bather protection.

Addressing beach orientation and specific land use impacts on water quality is beyond the scope
of this manuscript, but it is important to mention their potential for influencing water quality.
Sites SB1 and SB2 are located near the Pigeon and Saginaw Rivers mouths (respectively), each
has a large urban composition in their lower reaches. Interestingly, these beaches were the only
sites closed during the project as a result of elevated E. coli levels (per local government
monitoring) and the only locations where human markers were detected. At SB4, relationships
between fecal indicators and wind/wave associates were inversely related, contrary to the other
sites. This beach, situated near the Saginaw Bay/Lake Huron boundary, has virtually no
protective coastal barriers (e.g. piers or peninsulas). It is suspected that the exposed

98

characteristics of this beach increases wind fetch and allow lake currents to continuously dilute
and circulate water, resulting in lower fecal indicator bacteria concentrations in water. Together
these results demonstrate beaches situated near rivers and in areas with low circulation/renewal
exhibit elevated fecal indicator organisms. It is evident that surrounding land use, including
upstream areas drained by nearby rivers, and beach orientation influence water quality. Future
Saginaw Bay investigations should focus on these characteristics during water quality
monitoring by coupling microbial surveys at more spatial transects on the beach and upstream
river sites with GIS based land use composition and empirical based current models.

Microbial indicators, previously linked to human health, were described across multiple beach
zones and partially identified as human specific using the esp gene. This project demonstrated
the potential for sediment and algae mats to act as non-point sources of pollution in the nearshore
zone. Higher concentrations of traditionally monitored indicators were found in shallow waters
than in deep water in large part due to sediment bound bacteria levels and potentially regrowth of
the bacteria. Despite such evidence, governing bodies and beach managers continue to focus
beach monitoring efforts in deep water. The following suggestions are based on findings
presented in this manuscript:
1. The USEPA should: reevaluate recreational criteria with a nearshore health focus aimed at
defining the potential for this zone to influence the traditionally monitored deep water zone;
2. State governments should: investigate shallow water and sediment at a regional scale, taking
into consideration wind, waves, precipitation, and temperature as primary drivers;
3. Local health officials should: utilize newer molecular source tracking methods during routine
beach monitoring to pinpoint pollution sources and focus remediation efforts for long-term

99

water quality improvements; and adjust routine beach monitoring to coincide with conditions
known to produce the greatest bather risk such as precipitation, wind and wave action;
4. Scientists should: continue to evaluate coliphage as a potential indicator of pathogen
presence at Great Lakes beaches; develop and improve source tracking marker techniques to
a greater suite of sources; and measure microbe concentrations at incremental distances from
known pollution sources (e.g. wastewater treatment outfall) down river and to nearby
beaches.

2.5. Conclusions
After assessing multiple fecal indicators, molecular source tracking markers, and environmental
surveys across four beachscapes, this study was able to conclude that: 1) stranded algae mats and
sediment harbor the highest levels of fecal indicator organisms and can act as localized non-point
sources of bacteria; 2) human sewage is partially contributing to the fecal contamination in the
Saginaw Bay; and 3) Saginaw Bay water quality is significantly impacted by wind, waves, and
precipitation.

Acknowledgments
Partial funding of this project was provided by the Michigan Department of Environmental
Quality with support from Bay County Health Department, Huron County Health Department,
and Central Michigan District Health Department. Special thanks to Arun Kumar Nayak and
Sangeetha Shrinivasan for sample processing support.

100

REFERENCES

101

REFERENCES

Allwood, P.B., Malik, Y., Hedberg, C., and Goyal, S. (2003). Survival of F-specific RNA
coliphage, feline Calicivirus, and Escherichia coli in water: A comparative study. Applied and
Environmental Microbiology, 69, 5707-5710.
Alm, E.W. and Burke, J. (2003). Fecal indicator bacteria are abundant in wet sand at freshwater
beaches. Water Research, 37, 3978-3982.
Alm, E. W., Burke, J., and Hagan, E. (2006). Persistence and potential growth of the fecal
indicator bacteria , Escherichia coli , in shoreline sand at Lake Huron. Journal of Great Lakes
Research, 32, 401-405.
Bernhard, A.E. and Field, K.G. (2000). A PCR assay To discriminate human and ruminant feces
on the basis of host differences in Bacteroides-Prevotella genes encoding 16S rRNA. Applied
and Environmental Microbiology, 66, 4571-4574.
Bisson, J.W. and Cabelli, V.J. (1979). Membrane filter enumeration method for Clostridium
perfringens . Applied and Environmental Microbiology, 37, 55-66.
Boehm, A.B. (2007). Enterococci concentrations in diverse coastal environments exhibit extreme
variability. Environmental Science and Technology, 41, 8227-8232.
Boehm, A.B., Griffith, J., McGee, C., Edge, T.A., Solo-Gabriele, H.M., Whitman, R., Cao, Y.,
Getrich, M., Jay, J.A., Ferguson, D., Goodwin, K.D., Lee, C., Madison, M., and Weisberg, S.B.
(2009). Faecal indicator bacteria enumeration in beach sand: A comparison study of extraction
methods in medium to coarse sands. Journal of Applied Microbiology, 107, 1740-1750.
Borrego, J.J., Moriinigo, M.A., Vicente, A.D., Cornax, R., and Romero, P. (1987). Coliphages as
an indicator of faecal pollution in water. Its relationship with indicator and pathogenic
microorganisms. Water Research, 21, 1473-1480.
Byappanahalli, M.N., Sawdey, R., Ishii, S., Shively, D.A., Ferguson, J.A., Whitman, R.L. and
Sadowsky, M.J. (2009). Seasonal stability of Cladophora-associated Salmonella in Lake
Michigan watersheds. Water Research, 43, 806-814.
Byappanahalli, M.N., Shively, D.A., Nevers, M.B., Sadowsky, M.J., and Whitman, R.L. (2003).
Growth and survival of Escherichia coli and enterococci populations in the macro-alga
Cladophora (Chlorophyta). Federation of European Microbiological Societies, 46, 203-211.

102

Cha, Y., Stow, C., Reckhow, K.H., DeMarchi, C., and Johengen, T.H. (2010). Phosphorus load
estimation in the Saginaw River, MI using a Bayesian hierarchical/multilevel model. Water
Research, 44, 3270-3282.
Colford, J., Wade, T., Schiff, K., Wright, C., Griffith, J., Sandhu, S., Burns, S., Sobsey, M.,
Lovelace, G., and Weisberg, S. (2007). Water quality indicators and the risk of illness at beaches
with nonpoint sources of fecal contamination. Epidemiology, 18, 27-35.
Danek, L.J. and Saylor, J.H. (1977). Measurements of the summer currents in Saginaw Bay,
Michigan. Journal of Great Lakes Research, 3, 65-71
Davies, C.M., Long, J.A, Donald, M., and Ashbolt, N.J. (1995). Survival of fecal
microorganisms in marine and freshwater sediments. Applied and Environmental Microbiology,
61, 1888-1896.
Desmarais, T.R., Solo-Gabriele, H.M., Carol, J., and Palmer, C.J. (2002). Influence of soil on
fecal indicator organisms in a tidally influenced subtropical environment. Applied and
Environmental Microbiology, 68, 1165-1172.
Englebert, E.T., McDermott, C., and Kleinheinz, G.T. (2008b). Effects of the nuisance algae,
Cladophora, on Escherichia coli at recreation beaches in Wisconsin. Science Total Environment,
404, 10-17.
Englebert, E. T., McDermott, C., and Kleinheinz, G. T. (2008a). Impact of alga Cladophora on
the survival of E. coli, Salmonella, and Shigella in laboratory microcosm. Journal of the Great
Lakes Research, 34, 377-382.
Fahnenstiel, G.L., Lang, G.A., Nalepa, T.F., and Johengen, T.H. (1995). Effects of zebra mussel
(Dreissena polymorpha) colonization on water quality parameters in Saginaw Bay, Lake Huron.
Journal of Great Lakes Research, 21, 435-448.
Field, K.G., and Samadpour, M. (2007). Fecal source tracking, the indicator paradigm, and
managing water quality. Water Research, 41, 3517-3538.
Fishman, D. B., Adlerstein, S., Vanderploeg, H., Fahnenstiel, G.L., and Scavia, D. (2009).
Causes of phytoplankton changes in Saginaw Bay, Lake Huron, during the zebra mussel
invasion. Journal of the Great Lakes Research, 35, 482-495.
Frick, W.A., Ge, Z., and Zepp, R.G. (2008). Nowcasting and forecasting concentrations of
biological contaminants at beaches: A feasibility and case study. Environmental Science and
Technology, 42, 4218-4824.
Fujioka, R.S. and Shizumura, L.K. (1985). Clostridium perfringens , a reliable indicator of
stream water quality. Water Pollution Control, 57, 986-992.

103

Garrido-Pérez, M.C., Anfuso, E., Acevedo, A., and Perales-Vargas-Machuca, J.A. (2008)
Microbial indicators of faecal contamination in waters and sediments of beach bathing zones.
International Journal of Hygiene and Environmental Health, 211, 510-517.
Ge, Z., Whitman, R.L., Nevers, M.B., and Phanikumar, M.S. (2012). Wave-Induced mass
transport affects daily Escherichia coli fluctuations in nearshore water. Environmental Science &
Technology, 46, 2204-2411.
Girones, R., Ferrús, M.A., Alonso, J.L., Rodriguez-Manzano, J., Calgua, B., Corrêa, A.D.A.,
Hundesa, A., Carratala, A., and Bofill-Mas, S. (2010). Molecular detection of pathogens in water
- The pros and cons of molecular techniques. Water Research, 44, 4325-4339.
Griffin, D.W., Lipp, E.K., Mclaughlin, M.R., and Rose, J.B. (2001). Marine recreation and
public health microbiology: Quest for the ideal indicator. BioScience, 51, 817-826.
Ishii, S., Yan, T., Shively, D.A., Byappanahalli, M.N., Whitman, R.L., and Sadowsky, M.J.
(2006). Cladophora (Chlorophyta) spp. harbor human bacterial pathogens in nearshore water of
Lake Michigan. Applied and Environmental Microbiology, 72, 4545-4553.
Ishii, S., Hansen, D.L., Hicks, R.E., and Sadowsky, M.J. (2007). Beach sand and sediments are
temporal sinks and sources of Escherichia coli in Lake Superior. Environmental Science and
Technology, 41, 2203-2209.
Johnson, L., Richards, C., Host, G., and Arthur, J. (1997). Landscape influences on water
chemistry in Midwestern stream ecosystems. Freshwater Biology, 37, 192-208.
Jude, D.J., Rediske, R., O’Keefe, J., Hensler, S., and Giesy, J.P. (2010). PCB concentrations in
walleyes and their prey from the Saginaw River, Lake Huron: A comparison between 1990 and
2007. Journal of the Great Lakes Research, 36, 267-276.
Kinzelman, J., Ng, C., Jackson, E., Gradus, S., and Bagley, R. (2003). Enterococci as indicators
of Lake Michigan recreational water quality: Comparison of two methodologies and their
impacts on public health regulatory events. Applied and Environmental Microbiology, 69, 92-96.
Kraus, A., Henson, B., and Ewert, D. (2009). Biodiversity and conservation of Lake Huron’s
islands. Aquatic Ecosystem Health and Management, 12, 90-100.
Krometis, L.A.H., Characklis, G.W., Drummey, P.N., and Sobsey, M.D. (2010) Comparison of
the presence and partitioning behavior of indicator organisms and Salmonella spp. in an urban
watershed. Journal of Water and Health, 8, 44-59.
Mahin, T. and Pancorbo, O. (1999). Waterborne pathogens. Water Environment and Technology,
11, 51-55.

104

Masago, Y., Pope, J. M., Kumar, L. S., Masago, A., Omura, T., and Rose, J. B. (2011).
Prevalence and survival of Enterococcus faecium populations carrying the esp gene as a sourcetracking marker. Journal of Environmental Engineering, 137, 315-321.
Michigan Department of Environmental Quality (MDEQ). (2011). Michigan Beach Guard
System. Retrieved from (http://www.deq.state.mi.us/beach/).
Mortsch, L., Alden, M., and Scheraga, J.D. (2003). Climate change and water quality in the
Great Lakes Region: Risks, opportunities, and responses. Environment Canada and the US EPA.
Payment, P. and Franco, E. (1993). Clostridium perfringens and somatic coliphages as indicators
of the efficiency of drinking water treatment for viruses and protozoan cysts. Applied and
Environmental Microbiology, 59, 2418-2424.
Saginaw Bay Science Committee Pathogen Work Group. (2007). Potential public health risks
associated with pathogens in detritus material (“muck”) in Saginaw Bay. Retrieved from
(http://www.baycounty-mi.gov/Docs/Executive/SBCI%20%20Potential%20Public%20Health%20Risks%20....pdf).
Scott, T.M., Jenkins, T.M., Lukasik, J., and Rose, J.B. (2005). Potential use of a host associated
molecular marker in Enterococcus faecium as an index of human fecal pollution. Environmental
Science and Technology, 39, 283-287.
Shibata, T., Solo-Gabriele, H.M., Fleming, L., and Elmir, S. (2004). Monitoring marine
recreational water quality using multiple microbial indicators in an urban tropical environment.
Water Research, 38, 3119-3131.
Singh, S. and Rose, J. (2007). Further investigation of water quality and muck at Saginaw Bay
parks and beaches.
http://cws.msu.edu/projects/documents/Saginaw_bayReportwithmaps2007.pdf
Toze, S. (1999). PCR and the detection of microbial pathogens in water and wastewater. Water
Research, 33, 3545-3556.
United States Environmental Protection Agency (USEPA), O. of W. (2001). Method 1601:
Male-specific (F+) and somatic coliphage in water by two-step enrichment procedure. (EPA 821R-01-030).
United States Environmental Protection Agency (USEPA), O. of W. (2002). Method 1600:
Enterococci in water by membrane filtration using membrane-Enterococcus indoxyl-B-DGlucoside agar (mEI). (EPA 821-R-02-022).
United States Environmental Protection Agency (USEPA). (2005). Method 1603: Escherichia
coli (E. coli) in water by membrane filtration using modified membrane-Thermotolerant
Escherichia coli agar (modified mTEC). EPA 821-R-04-025. Office of Water, Washing D.C.

105

United States Environmental Protection Agency (USEPA). (2011). Saginaw River and Bay area
of concern. Retrieved from (http://epa.gov/glnpo/aoc/sagrivr.html).
Verhougstraete, M.P., Byappanahalli, M.N., Rose, J.B., and Whitman, R.L. (2010). Cladophora
in the Great Lakes: Impacts on beach water quality and human health. Water Science and
Technology, 62, 68-76.
Villari, P., Motti, E., Farullo, C., and Torre, I. (1998). Comparison of conventional culture and
PCR methods for the detection of Legionella pneumophila in water. Letters in Applied
Microbiology, 27, 106-110.
Wade, T.J., Sams, E., Brenner, K.P., Haugland, R., Chern, E., Beach, M., Wymer, L., Rankin,
C., Love, D., Li, Q., Noble, R., and Dufour, A. (2010). Rapidly measured indicators of
recreational water quality and swimming-associated illness at marine beaches: A prospective
cohort study. Environmental Health, 9, 66.
Wade, T.J., Calderon, R.L., Sams, E., Beach, M., Brenner, K.P., Williams, A.H., and Dufour,
A.P. (2006). Rapidly measured indicators of recreational water quality are predictive of
swimming-associated gastrointestinal illness. Environmental Health Perspectives, 114, 24-28.
Whitman, R.L., Nevers, M.B., Przybyla-kelly, K., and Byappanahalli, M.N. (2011). Physical and
biological factors influencing environmental sources of fecal indicator bacteria in surface water.
In M.J. Sadowsky and R.L. Whitman (Eds.), The Fecal Bacteria (pp. 111-134). Washington,
DC: American Society for Microbiology.
Yamahara, K.M., Sassoubre, L.M., Goodwin, K.D., and Boehm, A.B. (2012). Occurrence and
persistence of bacterial pathogens and indicator organisms in beach sand along the California
coast. Applied and Environmental Microbiology, 78, 1733-45.
Yampara-Iquise, H., Zheng, G., Jones, J.E., and Carson, C.A. (2008). Use of a Bacteroides
thetaiotaomicron-specific alpha-1-6, mannanase quantitative PCR to detect human faecal
pollution in water. Journal of Applied Microbiology, 105, 1686-1693.
Yang, S. and Rothman, R.E. (2004). PCR-based diagnostics for infectious diseases: Uses,
limitations, and future applications in acute-care settings. The Lancet, 4, 337-348.
Yun, S.H., and Kannan, K. (2011). Distribution of mono- through hexa-chlorobenzenes in
floodplain soils and sediments of the Tittabawassee and Saginaw Rivers, Michigan.
Environmental Science and Pollution Research International, 18, 897-907.
Whitman, R.L. and Nevers, M.B. (2004). E. coli sampling reliability at a frequently closed
Chicago beach: Monitoring and management implications. Environmental Science and
Technology, 38, 4241-4246.

106

Whitman, R.L. and Nevers, M.B. (2003). Foreshore sand as a source of Escherichia coli in
nearshore water of a Lake Michigan beach. Applied and Environmental Microbiology, 69, 55555562.
Whitman, R.L., Shively, D.A., Pawlik, H., Nevers, M.B., and Byappanahalli, M.N. (2003).
Occurrence of Escherichia coli and enterococci in Cladophora (Chlorophyta) in nearshore water
and beach sand of Lake Michigan. Applied and Environmental Microbiology, 69, 4714-4719.
Wiedenmann, A., Kruger, P., Dietz, K., Lopez-Pila, J. M., Szewzyk, R. and Botzenhart, K.
(2006). A randomized control trial assessing infectious disease risks from bathing in fresh
recreational waters in relation to the concentration of Escherichia coli, intestinal enterococci,
Clostridium perfringens , and somatic coliphages. Environmental and Health Perspectives, 114,
228-236.
Wolf, S., Hewitt, J., and Greening, G.E. (2010). Viral multiplex quantitative PCR assays for
tracking sources of fecal contamination. Applied and Environmental Microbiology, 76, 13881394.

107

CHAPTER 3.
LINKING LAND-USE AND MICROBES IN A SMALL DIVERSE WATERSHED USING
INDICATOR BACTERIA AND MOLECULAR SOURCE TRACKING

108

3.1. Introduction
Small stream systems and their associated discharges and catchment areas, are often overlooked
as significant sources of pollution to larger receiving bodies. However, previous studies have
shown small systems to be significant sources of microbes, nutrients, and sediments (Kistemann
et al. 2012; Edwards et al. 2012; Wilkes et al. 2011; Nadal-Romero et al. 2008). Bacteria that
enter small systems attached to particles settle into underlying sediment when stream velocity is
low (Bai and Lung 2005). As stream velocities increase, bacteria are resuspended and
transported downstream and eventually enter larger receiving bodies (Rehmann and Soupir 2009;
Muirhead et al. 2004), in essence acting as nonpoint sources of bacterial pollution.

Understanding microbial water quality of small systems is complicated by anthropogenic
pressure in watersheds. It was first reported in the 1970’s that watershed impervious surface
coverage exceeding 10% resulted in a rapid decline of water quality and biodiversity (Klein
1979). Nearly forty years later, connections between land use and water quality primarily focus
on chemicals and nutrients as the metric of water quality (Wang and Yin 1997; Mehaffey et al.
2005; Broussard and Turner 2009; Akasaka et al. 2010), but linking land use with
microorganisms has proven particularly difficult. Hunter et al. (1999) and Boyer and Pasquarell
(1999) linked fecal and total coliforms in water to agriculture land use. Desai and Rifai (2010)
found noticeably higher concentrations of E. coli in urban dominated sites compared to grassland
sites. For instance, Kang et al. (2010) used constrained least squares models and a comparative
statistics to link E. coli and enterococci concentrations with urban and industrial land uses and
added that E. coli and enterococci concentrations attributed to land use decreased as the size of
the watershed increased. Additionally, Mehaffey et al. (2005) used regression analysis to directly

109

correlate fecal coliform bacteria to urban and agriculture land use, but noted that land use
position relative to the waterbody was more important than percent land use in the entire
watershed. These studies implicate the potential for all land use types to contribute bacteria to
water. Variable findings of microbe-land use relationships can be partially explained by the
spatial scale at which investigations occur but also show the difficulty of separating multiple
sources acting simultaneously in a watershed with general bacteria. Accounting for appropriate
land use scale during microbial water quality assessment is vital, especially in watersheds with
mixed land use patterns.

Land use and sediment pollution require transport mechanisms to influence water quality.
Pollution is primarily transported from the landscape and sediment to waterbodies via weather
and environmental forces. Specific driving forces are highly variable between watersheds and
depend on landscape characteristics (Lavee and Poesen 1991), vegetative cover (Loch 2000), and
precipitation rates and land roughness (Katz et al. 1995). The primary transport mechanism for
landscape pollution is precipitation and runoff, defined by surface characteristics, landscape
slope, and soil hydraulic conductivity. Previous studies have reported fecal indicator bacteria
typically exhibited proportionally greater concentrations during large storms (based on rainfall
intensity, total volume, and/or discharge rates) compared to baseflow conditions (Cho et al.
2010; Traister and Anisfeld 2006; Schilling et al. 2009). Likewise, the majority of E. coli
movement to downstream waters occurred during storm events and was attributed to
resuspension of sediment-bound bacteria (McKergow and Davies-Colley 2010). More
specifically, Stumpf et al. (2010) found E. coli loads were 30 times greater during storm events
than during baseflow conditions with statistically different concentrations between each

110

condition. Wilkes et al. (2011) detected multiple pathogens on dates when total rainfall of the
previous week exceeded the 62

nd

percentile (~27 mm in the previous 7 days) and also showed a

positive correlation between pathogen (Cryptosporidium and Giardia) densities and surface
water discharge.

Eutrophication, beach closures, algae blooms, and sediment loading were common water quality
responses to landscape-associated pollution (Stoermer et al. 1978; MacGregor et al. 2001; U’Ren
2005; Rediske 2010). Traditional statistical approaches for modeling ecosystem responses have
failed to define explicit links between pollution and source, scale, and driving force of microbial
water quality. An emerging tool called Classification and Regression Tress (CART) has proven
useful for environmental exploration in complex systems (De’ath and Fabricius 2000). CART
models use multiple explanatory variables to correlate the variation of one response variable in
relationship to multiple parameters by repeatedly splitting the data into two groups based on
defined splitting criteria (Breiman et al. 1984). CART has been used to describe the source of
fecal pollution in water via chemical indicators (Gregor et al. 2002), to link E. coli O157:H7 with
relatively high pasture density (Wilkes et al. 2011), and to define significant physical-chemical
variables (including dissolved oxygen and turbidity) influencing fecal and total coliforms in
beach water (Bae et al. 2010). To date, no studies have reported on CART’s ability to link fecal
indicator bacteria, molecular source tracking markers, environmental conditions, climate
variables, and land use in small complex systems.

This project aims to determine the dynamics of water quality change and identify factors
influencing microorganism transportation throughout the watershed. Water quality was assessed
111

using fecal indicator bacteria and molecular source tracking markers in a diverse watershed
(Mitchell Creek) draining to Lake Michigan. The objectives were to 1) examine the spatial and
temporal distribution of traditional and alternative fecal indicators in a watershed influenced by
non-point source pollution, 2) use a quantitative PCR marker to measure human sources
associated with fecal bacteria, and 3) assess land use pattern effects on bacterial water quality.

3.2. Materials and Methods
3.2.1. Sampling strategy
The Grand Traverse Bay, located in northwest Lower Michigan, is currently facing water quality
concerns following beach closures and pollutant loading from recreational, urban, industrial, and
agricultural stormwater runoff (U’Ren 2005). Water quality degradation concerns were raised
following a 2008 local health department survey of the Mitchell Creek that detected multiple
sites with elevated E. coli levels. The Mitchell Creek accounts for 1.6% of the Grand Traverse
Bay watershed area, but is considered a major source of polluted stormwater input to the Bay
(U’Ren 2005). The Creek discharges into the southern end of East-Grand Traverse Bay, 300 m
west of Traverse City State Park’s designated swimming area. The Traverse City State Park
(TCSP) beach is heavily utilized for recreation during summer months. This beach has exceeded
Michigan water quality standards fifteen times since 2001, the majority of which were attributed
to stormwater runoff (MDEQ 2012; U’Ren 2005).

Water samples were collected from eight Mitchell Creek sites and one Grand Traverse Bay
beach between: June and November (2009); June and August (2010); November (2010); and

112

March (2011) (Table 3.1. and Figure 3.1.). Additionally, sediment samples were collected
between June and August (2010) and March (2011) at Mitchell Creek sites 2, 3, 4, 5, 6, 8, and
TCSP. No sediment samples were collected from MC1 (rocks composed the entire creek bottom)
or MC7 (sampling locations was at a metal culvert). Beach samples were collected in ankle deep
water (15 cm) in triplicates (left, center, and right) from the designated swim area. Creek samples
were collected 2 m from shore using an extendable arm. Sediment samples were collected from
the top 3 cm of benthos immediately after water sample collection. All samples were collected
using aseptic techniques and sterile Nalogene bottles, placed on ice (4°C), and brought to
Michigan State University’s Water Quality, Environmental, and Molecular Microbiology
Laboratory. Samples were kept at 4°C and processed for fecal indicator organisms within 12
hours of collection.

An intensive ten day study began on August 7, 2010 at sites MC2, MC3, MC4, MC5, MC6,
MC8, and TCSP. Water and sediment samples were collected from each site hourly for 12 hours
on day one followed by one sample from each site at hours 24, 72, 120, 168, and 240 (n = 17 per
site). Each water sample was assayed for E. coli, enterococci, and Bacteroides thetaiotaomicron
α-1-6 mannanase (B. theta), while sediment analysis included E. coli and enterococci.

113

N
W

E
S

Figure 3.1. The Mitchell Creek watershed with sampling site locations. Upper inset image: The Digital Elevation Model of the
Mitchell Creek watershed. Bottom inset image: Location of the Mitchell Creek watershed in Michigan and the Great Lakes.
114

Table 3.1. Description of Mitchell Creek watershed including land use and number of samples collected for each site. Mitchell Creek
flows from sites 8, 7, 6, and 5 (headwater catchments), through sites 4 and 3 or 2, before discharging near site 1 (upstream of outlet).
Site name
(Site ID)

Physical description

Traverse City Swimming beach; sandy
State Park
shoreline and lake
A
(TCSP )
bottom; 500 m long

Basin
Size
2
(km )
0.6

Site location
Lat.
Long.

44.749

-85.555

Mitchell
Creek 1
(MC1)

Sample number
Water Sediment Urban

45

29

19.4

Land use (%)
Ag. Open/ Wetland Water
Forest
0.8

26.7

12.0

41.2

Modified rocky
embankments and
39.7 44.745 -85.560
28
0
23.4 37.7 24.7
14.0
0.1
bottom; 120 m upstream
of creek mouth
Mitchell
Sandy embankments
Creek 2
and bottom; 570 m
37.9 44.748 -85.558
44
29
20.6 39.3 25.4
14.7
0.1
(MC2)
upstream of creek mouth
Mitchell
Modified rocky
Creek 3
embankments and
39.6 44.748 -85.559
44
29
23.3 37.8 24.8
14.0
0.1
(MC3)
bottom; 567 m upstream
of creek mouth
Mitchell
Steep eroded
Creek 4
embankments and sandy
(MC4)
bottom; 1.1 km
36.7 44.743 -85.560
44
19
22.0 40.4 24.0
13.5
0.1
upstream from Creek
mouth
Mitchell
Concrete embankments
Creek 5
and sandy bottom; 2.5
35.3 44.735 -85.556
44
29
20.9 41.6 23.7
13.6
0.1
(MC5)
km upstream of creak
mouth
A. Reference condition due to it close proximity to the mouth of the Mitchell Creek (MC1) and its popularity as a public swimming
beach during summer months
115

Table 3.1. (cont’d)
Site name
(Site ID)

Physical description

Mitchell
Creek 6
(MC6)

Deciduous woodland
surrounding and
organic rich bottom;
2.6 km upstream of
creek mouth
Residential wetland
outlet at culvert; rocky
bottom; 2.6 km
upstream of creek
mouth
Grassy embankments
and modified rocky
embankments; 3.2 km
upstream of creek
mouth

Mitchell
Creek 7
(MC7)

Mitchell
Creek 8
(MC8)

Basin
Size
2
(km )

Site location
Lat.
Long.

Sample number
Water Sediment Urban

Land use (%)
Ag. Open/ Wetland Water
Forest

9.2

44.735

-85.554

44

29

8.1

35.7

40.9

15.1

0.3

3.3

44.733

-85.559

24

0

33.3

36.7

14.2

15.3

0.5

14.6

44.733

-85.566

44

29

25.1

48.2

16.4

10.3

0.0

116

3.2.2. Environmental monitoring
Bather load, bird presence, wave height, and water and air temperature were recorded at time of
sample collection. Wind speed/direction (0° = north, 180° = south), barometric pressure, and
relative humidity were collected from NOAA’s National Weather Service (Traverse City, Cherry
Capital Airport) (NOAA 2012). Hourly Precipitation data were extracted from the Gaylord,
Michigan Next Generation Radar (NEXRAD) through the National Climate Data Center
(http://www.ncdc.noaa.gov/nexradinv/). This station has a base reflectivity 0.50 degree with an
elevation range of 124 nautical miles. Samples were collected during dry and wet periods
throughout the summer. Sampling events were considered wet weather when 48 hour
precipitation totals were equal to or greater than 5.1 mm, following local beach manager
recommendations and Haack et al. (2003).

Streamflow was measured at each site during four sampling events using an Acoustic Doppler
Current Profiler (ADCP) or current-meter via wading following USGS protocol (Rantz 1982).
3 -1

River discharge was calculated from flow velocities and reported as m s . For all other events,
stream discharge at the nearby Boardman River from U.S. Geological Survey (USGS) daily
records (gage 04126970) was used. A statistically related dependent factor was calculated using
gage recorded discharges on the Boardman River and measured discharges in the Mitchell Creek
on the same day. Scaling of the Boardman River mean daily discharge to estimate daily mean
flow at each Mitchell Creek site was performed using this dependent factor (Fulcher 1991).

Daily flows (million gallons) were collected from the Traverse City wastewater treatment plant.
Wastewater Treatment Plant (wwtp) discharge was used as a proxy for human population
117

density, which is considered a stressor of water quality (Smith et al. 2003; Nobre 2009).
Although not perfect, this method was used in lieu of outdated and delayed census results that
fail to grasp seasonal and tourist populations. Other sewage based indicators of human
population have been suggested including caffeine and coprostanol (Daughton 2012), but such
methods were beyond the scope of this project.

3.2.3. Cultivation analyses
Microbial analysis included E. coli, enterococci, Clostridium perfringens (C. perfringens), and
coliphage CN-13. Undiluted water samples were filtered through 0.45 µm hydrophilic mixed
cellulose esters filters (Pall Corporation). Sediment samples were assayed by weighing 100 g wet
weight and mixing with 600 mL sterile Phosphate Buffered Water (PBW) using shaker arm for 2
minutes. The samples were allowed to settle for 30 seconds and the eluent was poured into a
sterile bottle using caution not to mix sediment into eluent. An additional volume of PBW (400
mL) was added to the sediment, swirled for 10 seconds, and allowed to settle for 30 seconds. The
eluent was added to the first rinse to achieve a 10% weight/volume dilution. Microbial
assessment was made directly from final eluent.

E. coli were assayed using IDEXX Colilert defined substrate method and reported as Most
Probable Number (MPN) 100 ml

-1

(water) or MPN 100 g

-1

dry weight (sediment). Enterococci

were analyzed using membrane filtration and cultivation with selective media mEI (USEPA
2002) and reported as CFU (colony forming unites) 100 ml

-1

(water) or CFU 100 g

-1

dry weight

(sediment). C. perfringens were assayed via membrane filtration (no pretreatment), cultivated
using selective media mCP (USEPA 1995; Bisson and Cabelli 1979) and reported as CFU 100
118

-1

ml (water) or CFU 100 g

-1

dry weight (sediment). Double agar layers were utilized to detect

coliphage strains following USEPA method 1601 (USEPA 2001) using E. coli F+amp (malespecific coliphage) and E. coli CN-13 (somatic coliphage) as host bacteria. Clearings in the host
-1

lawn were counted and reported as plaque forming units (PFU) 100 ml (water) or PFU 100 g

-1

dry weight (sediment). Escherichia coli C-3000 (ATCC 15597), Enterococci faecium (ATCC
35667), C. perfringens (ATCC 3624), ΦX-174 coliphage were used as a positive controls for
verification of media integrity. Sterile water was used as negative controls for verification of
method integrity.

3.2.4. Molecular analyses
Samples were analyzed for the human specific marker B. thetaiotaomicron α-1-6 mannanase (5’
CATCGTTCGTCAGCAGTAACA 3’) following a modified procdure from Yampara-Iquise et
al. (2008) as described by Srinivasan et al. (2011). Analysis was performed by filtering 1 L of
sample through a 0.45 µm hydrophilic mixed cellulose esters filter. The filter was placed into a
50 mL centrifuge tube containing 20 mL of sterile PBW, vortexed, and centrifuged (30 minutes;
4000 x g; 21°C). Eighteen mL were decanted from the tube, and the remaining eluent and pellet
were stored at -80°C. DNA was extracted from 200 µL (10%) of the thawed suspended pellet via
QIAamp® DNA mini kit protocol. Quantitative Polymerase Chain Reaction (qPCR) for B. theta
was performed on a Roche Light-Cycler® 2.0 Instrument (Roche Applied Sciences) according to
Yampara-Iquise et al. (2008) and a primer modification (below) from Srinivasan et al. (2011).
Each B. theta assay was carried out with 10 µL of LightCycler 480 Probe Mastermix (Roche
Applied Sciences), 0.4 µL forward and reverse primers, 0.2 µL probe #62 (Roche Applied
Sciences Universal Probe Library), 1.0 µL Bovine Serum Albumin, 3.0 µL nuclease free water,
119

and 5.0 µL of extracted DNA and processed in triplicates. The qPCR analyses included a 15
minute, 95°C pre-incubation cycle, followed by 50 amplification cycles, and a 0.5 minute, 40°C
cooling cycle. A diluted plasmid standard was included during each qPCR run as a positive
control and molecular grade water was used in place of DNA template for negative controls. B.
-1

theta results were reported as copies 100 ml .

3.2.5. Spatial analysis
ArcMap 9.2 was used to delineate catchments for sampling location and to quantify the
landscape patterns in each watershed. In this project, land cover was defined as the cover of a
landscape, physical or biological, and land use was defined as the anthropogenic activities and
changes implemented on a specific land cover type (Di Gregorio and Jansen 1997). The shape
files and layers employed for this proposal were obtained from NOAA’s Coastal Change
Analysis Program (C-CAP) Regional Land Cover dataset (NOAA Coastal Services Center 2001;
http://www.csc.noaa.gov/digitalcoast/data/ccapregional/) and Michigan’s Center for Geographic
Information Library (MDTMB 2002). Digital elevation models were obtained from NASA’s
Land Processes Distributed Active Archive Center (USGS 2012) and USGS’s Seamless data
warehouse National Elevation Dataset (USGS 2010).

Catchment delineation
Catchments were delineated for each sampling point using ArcMap’s spatial analyst watershed
tool based on 1/3 Arc-Second (NED 1/3) resolution contour lines on a GCS North American
coordinate system. Each catchment includes the entire upstream landscape that contributes water
(and subsequently pollution) to each sampling point. Each catchment was further refined to
120

include distal buffers upstream of each sampling point at distances of 500 m, 1000 m, and 5000
m.

Land use delineation
Land use in each catchment was determined using C-CAP land use imagery at a 5 meter
resolution (see section 3.2.5). The land use contained in each catchment was reclassified into the
following

categories:

Anthropogenic

(residential,

commercial,

industrial

landscape

classifications), Agriculture (orchard, fields, and agriculture landscape classifications), Natural
(Forest, grassland, and prairie landscape classification, beaches), Wetlands, or Water as
described in Michigan Land cover/use classification (MSU 2010).

3.2.6. Statistical analysis
The percent moisture content by mass was determined for each sediment sample by measuring
approximately 10 g wet weight sediment in a pre-weighed aluminum dish. Samples were placed
in a 45°C incubator for 24 hours and reweighed. Percent dry weight was calculated by
subtracting the dish weight from the dry and wet weights then dividing the dry weight by the wet
weight. Mean daily air temperatures were calculated as an average of hourly observations
recorded at Traverse City, Cherry Capital Airport (NOAA’s National Weather Service;
http://w1.weather.gov/data/obhistory/KTVC.html) over a 24 hour period.

When microbial and molecular analysis results were below method detection, a value equal to
the method detection limit was reported. Microorganism concentrations underwent log10
transformation to fit the data to a normal distribution, but not all data met normality tests. Sample
121

sets meeting normality assumptions were assayed for relatedness using Levenne’s test for
equality of variance, Pearson correlation, and one-way ANOVA with Bonferroni post hoc tests.
When normality tests were not met, measurements were evaluated for relatedness using
independent samples-Kruskal-Wallis one-way ANOVA and bivariate Spearman Rank correlation
tests. These tests were performed using SPSS Statistic 17.0 software with significance set at (α)
0.05.

An agglomeration hierarchical cluster analysis was applied to surface water geometric means of
E. coli, enterococci, C. perfringens, and coliphage at each site. Sites were grouped into clusters
based on the linkage between-groups measured by Euclidean distance. Clusters were then used to
classify sites and illustrated with a dendrogram.

Examination of microbial water quality associations with independent parameters was achieved
using Classification And Regression Tree (CART) analysis following Martin et al. (2011).
CART is a trial and error method that attempts to split dependent variables into homogeneous
categories based on independent variables that influence the dependent variable (target
organism). All CART analyses were performed using R software system (R foundation for
Statistical Computing). CART has been previously used to investigate pathogenic bacteria and
parasite relationships with environmental and land use factors (Wilkes et al. 2011), to classify
lakes based on chemistry and clarity (Martin et al. 2011), and to predict the occurrence of
total/fecal coliforms and enterococci with respect to physiochemical variables (Bae et al. 2010).

122

Models start out with a parent or root node which contains all available date. CART then looks at
all independent variables (splitting variables) and selects the single variable that produces the
two most different groups of dependent variables based on predefined splitting criterion and
regression analysis. In this study, splitting criteria were developed using recursive partitioning
algorithm and a 10-fold cross validation. A 10-fold cross validation breaks all data into 10
subsets and calculates the split based on nine of the ten subsets. Each time a group is split per
above criteria the binary splits are called child nodes. This method is used for each group until a
stopping rule is reached. For this project, the stopping criterion was set at a minimum of five
observations per subgroup (Martin et al. 2011). A terminal node is defined as a child node which
has met the defined stopping rule.

Fully grown trees often require pruning to ensure significant variable associations are not missed
as a result of following the splitting and stopping criteria (Lemon et al. 2003). Pruning is the
process of growing trees until they reach stopping criteria and then cutting less statistically
significant results back. Trees were pruned according to the 1-standard error rule (Breiman et al.
1984; Venables and Ripley 1999; De’ath and Fabricous 2000). This rule minimizes the crossvalidated error of the model which has been shown to produce optimal sized trees and produce
more stable tree sizes across replications compared to the 0-SE pruning method and (Breiman et
al. 1984; Questier et al. 2005).

Investigating detailed CART outputs, competitor and surrogate variables can be identified for
each node. Competitor splits are those variables that have similar complexity parameter values
compared to the primary split. A complexity parameter compares the complexity (number of

123

terminal nodes) to the cross-validated error for each group. For this project the complexity
parameter was set at 0.05. Surrogate splits are alternative variables that split the subgroup into
very similar groups.

An example of a CART output is presented below (Figure 3.2.). At the top of the tree is the
parent or root node with the primary splitting variables and values described for each child node.
At the bottom of the tree, terminal nodes include the mean concentration and number of target
organism cases in each node.

Figure 3.2. Classification And Regression Trees are composed of root nodes that contain all
available data and are split into binary groups using recursive partitioning algorithm and 10-fold
cross validation with a complexity parameter value of 0.05. Primary splitting variables and
values are described for each child node. Terminal nodes (bottom of the tree) include the mean
concentration and number of target organism cases in each node. Each node was derived based
on mean value of each response variable, group size, and defining variables.

124

3.3. Results
3.3.1. Land use and cover
Land use composition of each catchment included multiple land cover types categorized as
urban, agricultural, natural (forest and shrub), wetland, and water (Table 3.1. and Figure S.3.1.).
Land use trends illustrate urban development primarily near the lower reaches (MC1, MC2,
MC3, and MC4), wetlands in the middle reaches (MC5, MC6, MC7, and MC8) and agriculture
near the headwaters of the watershed. Water quality from the agriculturally dominated
headwaters was captured by sites in the middle reaches as they represent the drain point for the
upstream land use. Spatial variation within sub-catchments was identified by applying distal
buffers to upstream catchments of each sampling point and comparing land use composition
across scales (Figure S.3.1.).

All Mitchell Creek sites, except MC6, exceeded 20% urban development at the catchment, with
an average of 22.1% urban development amongst all Mitchell Creek sites. Overall, the dominate
catchment land use was agriculture (39.7%), with a high of 48.2% at MC8. MC6 was composed
of 40.9% natural land use, the highest of all sites at the catchment scale. MC7 had the highest
urban (33.3%), wetland (15.3%), and aquatic (0.5%) compositions at the catchment scale.

In the Mitchell Creek, urban development had the highest land use composition average across
all sites at the 5000 m (30.0%), 1000 m (38.9%), and 500 m (35.3%) scales. MC8 had the highest
urban (34.7%) and agriculture (19.3%) compositions at the 5000 m scale. MC6, MC7, and MC1
had the highest natural (23.3%), wetland (12.0%), and water (30.3%) coverage, respectively, at
the 5000 m scale. Urban coverage was highest at the 1000 and 500 m scales around MC4 (63.0%

125

and 65.0%, respectively). At the 1000 and 500 m scales, MC8 had the highest agriculture (16.9%
and 15.5%, respectively) and natural (26.3% and 32.5%, respectively) land use coverage. MC6
and MC1 had the highest wetland (45.5%) and water (42.3%) coverage at the 1000 m scale,
respectively. Spatial contrasts between the upper and lower catchments were exhibited at the 500
m buffer as urban composition exceeded 43% near MC1 but decreased to 18% at MC8.

Comparing urban development in the Mitchell Creek at the catchment scale (22.1%) to the 5000
m (28.9%), 1000 m (36.7%), and 500 m (33.5%) scales, confirms greater development near the
river, more so in lower reaching sites. Agriculture (cropland and orchard) was the dominant land
use at MC8 across all scales. Natural land use was highest in the MC6 catchment at the larger
scales (i.e. catchment and 5000 m) but highest in the 1000 m and 500 m surrounding MC8.
Wetland composition at all sites ranged between 10.3% and 15.3% with an average of 13.8%.
TCSP had the greatest water coverage (41.2%) which was attributed to the beach orientation in
the landscape and the surrounding topography. A Chi-square comparison of water coverage
between Mitchell Creek sites generated a Chi value below 0.001 (7 degrees of freedom, p =
0.05), resulting in acceptance of the null hypothesis that water coverage is similar at all Mitchell
Creek sites.

3.3.2. Spatial analyses of water and sediment quality
In total, 361 water samples and 193 sediment samples were collected from eight Mitchell Creek
sites and one beach site, the Traverse City State Park beach (TCSP). E. coli, enterococci, C.
perfringens, and coliphage CN-13 results for each site are shown in Figure 3.3. MC6 was the
only Mitchell Creek site distinguishable from other sites based on E. coli and enterococci

126

concentrations in water (p < 0.05). Coliphage CN-13 was not statistically different between any
Mitchell Creek sites (p > 0.05).

Fecal indicator concentrations in water
-1

E. coli densities ranged between 0.30 and 3.5 log10 MPN 100 ml and had an overall geometric
-1

mean of 2.4 log10 MPN 100 ml . Enterococci ranged between 0.29 and 4.5 log10 CFU 100 ml

-1

-1

with an overall geometric mean of 2.5 log10 CFU 100 ml . C. perfringens ranged from 0.05 to
-1

-1

2.5 log10 CFU 100 ml and exhibited an overall geometric mean of 1.1 log10 CFU 100 ml .
-1

Coliphage CN-13 ranged from 1.0 to 3.9 log10 PFU 100 ml and had an overall geometric mean
-1

of 1.8 log10 PFU 100 ml . The highest E. coli, enterococci, C. perfringens, and coliphage
individual measurements were found in samples collected at MC2, MC3, MC3, and MC5,
respectively. The highest site specific E. coli, enterococci, C. perfringens, and coliphage
geometric means were all found at MC5.

The B. theta marker was detected at every assayed site except MC7 (Figure 3.3 and Table 3.2.).
In total, 118 Mitchell Creek water samples were tested for B. theta with an overall geometric
-1

mean of 4.1 log10 copies 100 ml . B. theta concentrations in the Mitchell Creek ranged from 2.9
-1

to 6.5 log10 copies 100 ml . The highest B. theta geometric mean concentration was recorded at
MC2 and the highest single sample concentration was measured at MC3 (6.5 log10 copies 100
-1

ml ). No MC4 samples were assayed for B. theta as it was thought that any human

127

contamination present at this site would be captured at MC3 (~500 m downstream). More
samples were tested at TCSP, MC2, MC3, MC5, and MC6 compared to MC1, MC4, MC7, and
MC8 because the former were included in additional qPCR analysis stemming from the intensive
study.

TCSP was selected as a reference location because it is a popular designated swimming area next
to the mouth of the Mitchell Creek. The lowest single water measurements of E. coli (< 0.30
-1

-1

log10 MPN 100 ml ), enterococci (< 0.30 log10 MPN 100 ml ), C. perfringens (0.05 log10 CFU
-1

-1

100 ml ), and coliphage CN-13 (< 1.0 log10 PFU 100 ml ) were recorded at TCSP.
-1

Furthermore, E. coli and enterococci geometric means (1.12 log10 MPN 100 ml and 1.11 log10
-1

MPN 100 ml , respectively) amongst all sites were significantly lower than those in all Mitchell
-1

Creek sites (p > 0.05). The TCSP C. perfringens geometric mean (0.246 log10 CFU 100 ml )
was statistically lower than individual geometric means computed at each Mitchell Creek site
-1

-1

except MC6 (0.797 log10 CFU 100 ml ) and MC7 (0.679 log10 CFU 100 ml ). The TCSP
-1

coliphage mean (1.13 log10 PFU 100 ml ) was lower than each Mitchell Creek site except MC6
-1

(1.41 log10 PFU 100 ml ). Interestingly, in contrast to the other indicators, the TCSP B. theta
-1

geometric mean (4.2 log10 copies 100 ml ; n = 35) was slightly higher than the overall Mitchell
-1

Creek watershed geometric mean (4.1 log10 copies 100 ml ), however this was not statistically
significant (p = 0.262).

128

Enterococci log10 concentrations
(MPN 100 ml-1 or MPN 100 g-1)

7
6

4

5
4

3

3

2

2
1

Water
Sediment

M
C8

M
C7

M
C6

M
C5

M
C4

M
C3

M
C2

SP

M
C8

M
C7

M
C6

M
C5

M
C4

M
C3

M
C2

TC

M
C1

0

M
C1

0

TC

1

SP

E. coli log10 concentrations
(MPN 100 ml-1 or MPN 100 g-1)

5

and75

th

6

Coliphage log10 concentrations
(PFU 100 ml-1 or PFU 100 g-1)

C. perfringens log10 concentrations
(CFU 100 ml-1 or CFU 100 g-1)

Figure 3.3. Box plots illustrating the ranges of fecal indicator organisms and B. thetaiotaomicron concentrations measured in the
5
7
th
water and sediment (E. coli and enterococci only) at each site. The lower, middle, and top box edges correspond to the 25 , median,
th

percentiles of each measurement. The whiskers indicate the 104 and 90

th

percentile. The points indicate measurements

th
5 th
outside the 5 and 95 percentiles. E. coli, enterococci, C. perfringens, and coliphage were measured in sediment and reported per
-1
4
100 g dry weight. The number of samples assayed for each cultivated microorganism at each site is presented in Table S.3.1. The
3

number of B. theta assays for each site is presented in Table 3.2.

3
2
1
0

2
1

129

M
C8

M
C7

M
C6

M
C5

M
C4

M
C3

M
C2

M
C1

SP
TC

M
C8

M
C7

M
C6

M
C5

M
C4

M
C3

M
C2

M
C1

TC

SP

0

TC
S
P

7

6

5

4

3

2

M
C2
M
C3
M
C4
M
C5
M
C6
M
C7
M
C8

M
C1
M
C2
M
C3
M
C4
M
C5
M
C6
M
C7
M
C8

C. perfringens log10 concentrations
(CFU 100 ml-1 or CFU 100 g-1)
7

6

5

4

3

2

1

0

Coliphage log10 concentrations
(PFU 100 ml-1 or PFU 100 g-1)

Figure 3.3. (Cont’d)

SP

130

M
C8

M
C7

M
C6

M
C5

M
C4

M
C3

M
C2

M
C1

SP

(MPN 10

Enteroco

0

TC

M
C8

M
C7

M
C6

M
C5

M
C4

M
C3

M
C2

M
C1

TC

(MPN 10

E. coli

Water
Sediment

TC
SP
M
C1
M
C2
M
C3
M
C4
M
C5
M
C6
M
C7
M
C8

M
C1

SP

TC

B. thetaiotaomicron log10 concentrations
(CE 100 ml-1 or CE 100 g-1)

1
1

0

5

4

3

2

1

0

Table 3.2. B. thetaiotaomicron results in Grand Traverse Bay and Mitchell Creek water samples.
Site
ID

Number
of
Samples

Log mean
concentrations
-1
(log10 copies 100 ml )

Concentration range
-1
(log10 copies 100 ml )

Number samples
positive ( > 2.9 log10
copies 100 ml

-1

)

TCSP
35
2.9 – 6.0
4.2
29
MC1
9
2.9 – 4.8
3.1
2
MC2
26
2.9 – 6.4
4.6
22
MC3
23
4.3
2.9 – 6.5
15
MC4
0
NA
NA
NA
MC5
21
2.9 – 5.8
4.6
17
MC6
25
2.9 – 5.9
4.0
16
MC7
3
2.9
0
MC8
7
2.9 – 3.8
3.1
2
NA: No samples assessed for B. theta as it was assumed water quality would be captured at MC3
situated 500 m downstream from MC4.

Fecal indicator concentrations in sediment
Fecal indicators were routinely recovered in the sediment of each sampling site. At TCSP, the
geometric means for E. coli, enterococci, C. perfringens, and coliphage in the sediment were 1.9
-1

-1

-1

-1

log10 MPN 100 g , 2.0 log10 CFU 100 g , 1.3 CFU 100 g , and 1.2 PFU 100 g , respectively.
E. coli and enterococci in the sediment of all tested Mitchell Creek sites were statistically higher
than TCSP sediment (p ≤ 0.003). In general, E. coli and enterococci were not statistically
different between Mitchell Creek sites (p > 0.05). In the Mitchell Creek sediment, E. coli ranged
-1

from 1.7 (MC2 and MC4) to 5.9 (MC6) log10 MPN 100 g

and exhibited an overall geometric

-1

mean of 3.4 log10 MPN 100 g . On average amongst TCSP and Mitchell Creek sites, E. coli
concentrations in the sediment were 1.5 times higher than E. coli concentrations in the water
column. Specifically, E. coli averaged 1.4 times higher in sediment than in water in the Mitchell
Creek and 1.8 times higher at TCSP. Enterococci ranged from 1.2 (MC6) to 6.3 (MC6) log10
131

CFU 100 g

-1

in the Mitchell Creek with an overall geometric mean average of 3.0 log10 CFU

-1

100 g . When combining the TCSP and Mitchell Creek sites, enterococci concentrations in the
sediment were 1.4 times higher than enterococci concentrations in the water column.
Specifically, sediment enterococci averaged 1.3 times higher than water in the Mitchell Creek
and 1.9 times higher at TCSP. C. perfringens had an overall geometric mean of 2.5 log10 CFU
100 g

-1

-1

in the Mitchell Creek and ranged from 1.1 to 6.5 log10 CFU 100 g . Overall TCSP and

Mitchell Creek sites, C. perfringens concentrations in the sediment were 3.6 times higher than C.
perfringens concentrations in the water column. Specifically, sediment C. perfringens averaged
3.4 times higher than water in the Mitchell Creek and 4.5 times higher at TCSP. CN-13 ranged
-1

between 0.4 and 3.9 log10 PFU 100 g

-1

with a geometric mean of 1.6 log10 PFU 100 g . On

average amongst TCSP and Mitchell Creek sites, coliphage CN-13 concentrations in the
sediment were 1.0 log higher than CN-13 concentrations in the water column. Specifically, CN13 averaged 1.0 log higher in sediment than water in the Mitchell Creek and 1.1 times higher at
TCSP. Preliminary B. theta analysis returned a significant number of non-detects (n = 116; nondetect = 91%) in sediment samples suggesting the need for further method developments and
potentially a high inhibition, thus further analysis were discontinued due to timing constraints.

3.3.3. Temporal analysis
Water and sediment quality changes
Daily microbial averages in water and sediment of the creek were examined over time. Daily E.
coli geometric means over time in sediment and water amongst all Mitchell Creek sites are
presented in Figure 3.4, along with precipitation (24 hour totals) which will be discussed in the
132

next section. All E. coli, Enterococci, C. perfringens, and coliphage CN-13 measurements in
water and sediment from all sites are presented in Table S.3.2. and Table S.3.3., respectively.

133

20

24 hr precipitation
3
15
2

10

0

3/19/11
3/20/11

0

6/7/10
6/10/10
6/14/10
6/24/10
6/27/10
7/1/10
7/5/10
7/12/10
7/15/10
7/18/10
8/7/10
8/8/10
8/10/10
8/12/10
8/14/10
8/17/10

5

6/24/9
7/1/9
7/8/9
7/15/9
7/22/9
7/29/9
8/3/9
8/9/9
8/12/9
8/17/9
8/26/9
8/31/9
9/2/9
9/9/9
9/16/9
11/2/9

1

Precipitation (mm)

E. coli daily mean

4

25

Water (CFU 100 ml-1)
Sediment (CFU 100 g-1)

-1

Figure 3.4. Temporal variation of E. coli in the Mitchell Creek. Water daily geometric means reported as log10 MPN 100 ml and
-1

sediment reported as log10 MPN 100 g dry weight. Enterococci trended similarly to E. coli. Minor temporal variations were
identified using C. perfringens and coliphage CN-13. Precipitation is shown as cumulative rainfall in 24 hours prior to sample
collection. denotes dates when TCSP water samples exceeded Michigan E. coli water quality standards. No sediment samples were
collected in 2009.

134

Sediment showed less variability in the bacterial concentrations over time and in general did not
change dramatically between morning and afternoon, collection hour, or day of month. In
regards to water quality, E. coli, enterococci, and C. perfringens, averaged across all sites, were
significantly higher in pre-noon samples (n = 262) compared to post-noon samples (n = 99; p =
0.002) likely due to inactivation in the water column, but coliphage was not significantly
different (p = 0.241). Daily and monthly variations of microorganism concentrations were also
identified and suggestive of E. coli and enterococci regrowth during warm summer months.
Monthly differences of C. perfringens and coliphage CN-13 densities were generally
insignificant.

3.3.4. Drivers of bacterial water quality
Environment and weather conditions
During sample collection, Mitchell Creek water temperature ranged from -0.5 °C to 23.5 °C and
air temperatures ranged from -0.8 °C to 34.5 °C. Air and water temperature were highly related
to each other (r = 0.87, p = 0.01). Descriptive statistics of environmental parameters recorded
throughout the study period are presented in Table S.3.4. Correlation between independent
parameters (air and water temperature at time of sampling, mean daily air temperature,
cumulative precipitation totals [1, 2, 3, 4, 6, 8, 12, 16, 24, 48, and 72 hour] prior to sample
collection, barometric pressure, relative humidity, daily mean solar radiation, and wastewater
treatment plant daily discharge) and fecal indicators were generally positive but small or
insignificant. Correlation results for independent variables and microbes in water and sediment
are presented in TableS.3.5. and Table S.3.6, respectively.

135

Hydrology
Discharge in each catchment was directly associated with 24-72 hour precipitation totals. Total
2

72 hour precipitation described most of the discharge-precipitation relationship (R = 0.539, p <
0.001). On June 14, 2010, the largest discharges were recorded at each site which was preceded
by the largest 72 hour cumulative precipitation total. Eleven sampling events occurred after 48
hour cumulative precipitation exceeded 5.1 mm, the predetermined wet weather threshold. The
wettest month of the project was June 2010 (169 mm total monthly precipitation), far greater
than the long term monthly average of 84.3 mm (Midwest Regional Climate Center 2012).
Discharge and precipitation results recorded throughout the project period are detailed in Table
S.3.7 and Table S.3.8., respectively.

An intensive study began on August 7, 2010 and included hourly sample collection for 12
consecutive hours. On hour 10, it began to rain and at hour 12, the 95

th

percentile of the project

E. coli distribution was exceeded at TCSP. The following day (8-8-2010; hour 24), precipitation
th

totals for the previous 4-24 hours exceeded the 99 percentile and microbial responses included:
th

1) Mitchell Creek water E. coli exceeded the 95 percentile, 2) Mitchell Creek water enterococci
exceeded the 90
th

95

th

percentile, 3) Mitchell Creek sediment E. coli and enterococci exceeded the

percentile, 4) TCSP water quality failed Michigan’s single sample E. coli standard (300 E.
-1

coli 100 ml ), and 5) site MC2 exceeded water E. coli and enterococci 95

th

percentiles. On

August 10 (hour 72), cumulative B. theta averages for all sites reached a project high and MC4

136

th

exceeded water E. coli and enterococci 90 percentiles. On August 12 (hour 120) MC5 exceeded
th

water E. coli 95 percentile.

Human population and WWTP
The wastewater treatment facility discharges to the Boardman River and drains to the west
branch of the Grand Traverse Bay (Latitude: 44.75754, Longitude: -85.61270). The TCSP and
mouth of the Mitchell Creek are located in the east branch of the Grand Traverse Bay (separated
from the west branch by the ~27 km long Old Mission Peninsula) and thus the WWTP discharge
does not affect the Mitchell Creek watershed. Therefore, WWTP discharge was used as a
surrogate for short term human population presence in the study area which partially relies on
septic tanks. Daily discharge volumes from the WWTP were statistically higher from June
through October compared to December through May (p < 0.001). Daily effluent flows averaged
over each month and per day of week are illustrated in Figure S.3.2 (A and B, respectively).
Weekend flows were statistically lower than weekday flows (p < 0.001), illustrating the
significant presence of commerce in the watershed. Increases of daily flow existed during the
first week of July each year, attributed to population increase for an annual weeklong festival. In
the Mitchell Creek, statistically significant (p ≤ 0.002) and direct correlations were shown to
exist between daily WWTP discharges and concentrations of E. coli (r = 0.242), enterococci (r =
0.527), and C. perfringens (r = 0.237) in water. WWTP discharges were not statistically related
to coliphage CN-13 (p = 0.871) or B. theta (p = 0.993) levels in the Mitchell Creek. At TCSP,
WWTP discharges were significantly related to coliphage CN-13 in water (r = 0.361; p = 0.015),
but not to E. coli, enterococci, C. perfringens, or B. theta (p ≥ 0.194).

137

3.3.5. Microbial responses to sources and drivers
Environmental impacts on water quality
Mitchell Creek water quality is influenced rapidly by precipitation (Table S.3.9.) and E. coli and
enterococci measurements from sediment and water displayed weak positive correlations to
precipitation with the most significant responses generally associated with 16 hour cumulative
precipitation (Table S.3.10.). At TCSP, significant associations between rainfall and fecal
indicators in water and sediment were not identified. However, during three of the 11 wet
weather sampling events (June 10, 2010, June 24, 2010, and August 8, 2010), TCSP water
-1

quality exceeded Michigan’s E. coli standard (300 E. coli 100 ml ) as shown in Figure 3.4.
There were no exceedances of the same standards during dry weather sampling events (n = 34).

Associations between the four fecal indicator bacteria and independent variables were tested
using CART. The independent variables temperature (water and air), precipitation, collection
timing, humidity, and wastewater flows were considered indicators of temporal variation while
land use and scale represented of spatial variability. Temporal variables dominated the root node
splits in three of the four Mitchell Creek water models targeting fecal indicator bacteria (Figure
3.5). CART results explained 65% of E. coli, 74% of enterococci, 62% of C. perfringens, and
59% of coliphage CN-13 variations in Mitchell Creek water samples.

138

Figure 3.5. Mitchell Creek water CART outputs for (A) E. coli, (B) enterococci, (C) C.
perfringens, (D) coliphage (CN-13), and (E) B. theta. Each split is labeled with splitting variable
and value. Terminal nodes (bottom rectangle) are labeled with means and cases of target
organism in each group.

139

Figure 3.5. (cont’d)

140

Figure 3.5. (cont’d)

141

2

E. coli was best explained by mean daily air temperature (R = 0.457) and 12 hour cumulative
2

precipitation (R = 0.136). Further investigation of CART outputs indicated E. coli levels were
equally explained by mean daily air temperature, 12, 16, and 24 hour precipitation totals, percent
agriculture in the watershed and 1000 m buffer, and percent urban coverage in the watershed.
Enterococci was best explained by human population using Traverse City’s WWTP daily
2

2

discharge (R = 0.280) and water temperature (R = 0.122). Watershed wetland coverage best
2

explained C. perfringens levels (R = 0.273) while coliphage CN-13 was almost wholly
explained by precipitation (16 and 48 hour totals) and discharge. When Mitchell Creek sediment
-1

-1

E. coli was above 4.5 log10 MPN 100 ml , the highest B. theta levels (5.9 log10 CE 100 ml )
were also expected, although this variable explained just 14% of B. theta variation, suggesting at
least a portion of B. theta occurrence in water is coming from the sediments. For E. coli,
enterococci, and C. perfringens, a vast majority of land use nodes also included competitor
variables of other land use types and scales (agriculture, urban, wetlands, and natural).

The highest means of fecal indicator bacteria in the Mitchell Creek (i.e. highest contamination
and human health risk) were associated with a mixture of spatial and temporal parameters. The
highest fecal indicator means, as defined by CART, were explained by average daily air
temperature, water temperature, and wastewater discharge. Precipitation partially explained the
highest concentrations of each fecal indicator in water. The only land use variable identified as a
root node split for any bacteria was wetland coverage in the entire catchment. Wetland coverage
-1

below 15% was the primary driver of C. perfringens above 1.24 log10 CFU 100 ml .

142

Microbes in TCSP water were compared to independent variables using CART (Figure 3.6).
Land use was excluded from TCSP analysis because of the small catchment size and
homogeneous land use patterns across all scales (Figure S.3.1.). CART models explained
between 11.9% and 36.9% of fecal indicator bacteria detections at TCSP water, far less than
Mitchell Creek water. However, significant explanatory parameters were identified including
discharge from Mitchell Creek and wind direction. The root nodes of E. coli and coliphage CN13 were split by wind direction and wind speed, respectively. Enterococci and C. perfringens in
TCSP water were primarily explained by enterococci at MC3 water and discharge from MC1,
respectively. B. theta was the only microorganism influenced by both the environment (wind
direction) and the Mitchell Creek (MC1 discharge). Varying the number of Mitchell Creek
parameters input to TCSP CART analysis resulted in similar explanatory variables at TCSP with
overall explanatory powers slightly shifted. After analyzing multiple tree variations, it was
determined that C. perfringens and B. theta at TCSP were highly influenced by the Mitchell
Creek while E. coli, enterococci, and coliphage CN-13 were only moderately affected.

To better understand the timing of Mitchell Creek’s influence on swimming at TCSP, time lags
were imposed during statistical analysis for the 12 consecutive hourly samples collected as part
of the intensive study. Water quality comparisons between TCSP and Mitchell Creek
measurements from the previous hour produced the strongest relationships, specifically with
MC2 (p < 0.05), MC3, MC4, and MC5 (p < 0.01), but not with the upstream sites MC6 and MC8
p > 0.05).

143

Figure 3.6. TCSP water CART outputs for (A) E. coli, (B) enterococci, (C) C. perfringens, (D)
coliphage (CN-13), and (E) B. theta.

Sites were grouped into clusters using an agglomeration hierarchical analysis. Four distinct
groupings were identified between the clusters (Figure 3.7.). TCSP (Cluster 1) was the least
144

contaminated, followed by MC1 and MC7 (Cluster 2), and MC6 (Cluster 3). Cluster 4 was the
most contaminated group and consisted of MC5, MC8, MC2, MC3, and MC4. Each cluster was
defined by surface water geometric means of fecal indicators, but also shared common and
unique root node splitting variables in CART. Temperature was a common root node split in the
Mitchell Creek (clusters 2, 3, and 4) but was not important at TCSP. E. coli and enterococci
splitting variables in Cluster 2 were all variants of temperature (air, water, or mean daily air
temperature). Splitting variables at Cluster 3 included temperature (air and water), precipitation
(72 hour), and wastewater discharge. Cluster 4 split significantly on wastewater discharge as
well as air temperature and precipitation (48 and 72 hour).

Figure 3.7. Dendrogram showing clusters based on geometric mean indicator concentrations at
each site. TCSP is a designated swimming area with relatively low indicator organism
concentrations and root node splits on wind (direction and speed) and MC parameters (discharge
and water/sediment microbes). Clusters 2, 3, and 4 are sites within the Mitchell Creek watershed
and are not designated as primary contact recreation sites.
145

Sediment implications for water quality
CART analysis was again used to investigate associations between independent variables and
microbes occurrences, this time in the sediment. In regards to the Mitchell Creek, CART results
explained a lower percentage of fecal indicator presence in sediments compared to water models,
likely due to the steady state of microbial concentrations found in the sediments which were
more resistant to change from outside forces. Sample collection month explained 28.0% of
sediment E. coli detections in Creek sites. Precipitation (24 hour), wastewater discharge, and
wetland coverage explained 15.1%, 25.8%, and 19.1% of enterococci, C. perfringens, and
coliphage CN-13 detection in Mitchell Creek sediments, respectively. In the sediment of
Mitchell Creek, the highest fecal indicator bacteria means were associated with mean daily air
temperature above 5.3 °C, 24 hour precipitation totals above 0.07 mm, small wastewater
discharge volumes, and wetland coverage greater than 14.5% in the watershed. CART results for
Mitchell Creek sediment are presented in Figure 3.8. (A-D).

In the sediment at TCSP, CART analysis were slightly better than Mitchell Creek models at
explaining E. coli (37.9%) and enterococci (29.5%) levels. Insufficient data were available to
perform CART analysis on C. perfringens and coliphage assays at TCSP. Wave height explained
approximately 30% of E. coli occurrences. Enterococci concentrations in water explained 30% of
enterococci in sediment, but wind direction was a strong competitor variable. Results for each
fecal indicator at TCSP sediment are presented in Figure 3.8 (E and F). Multiple models were
again developed by varying the number of Mitchell Creek parameters input to CART. All model
variations produced similar results for E. coli and enterococci in TCSP sediment, indicating
sediment fecal indicator organisms act independent of the Mitchell Creek sediment.

146

Figure 3.8. Sediment CART outputs for (A) Mitchell Creek E. coli, (B) Mitchell Creek
enterococci, (C) Mitchell Creek C. perfringens, (D) Mitchell Creek coliphage (CN-13), (E)
TCSP E. coli, and (F) TCSP enterococci. Insufficient data was available to perform CART
analysis on C. perfringens and coliphage assays at TCSP.

Comparisons of water and sediment pairwise samples identified 95% of E. coli and 87% of
enterococci results were greater in sediment than in the overlying water column, but the majority
of differences were less than 1 order of magnitude. The largest difference between sediment and
147

water E. coli and enterococci concentrations occurred at MC6. To identify sediment
contributions to water, water microbe concentrations from each site were compared to microbial
concentrations in sediment, discharge, and precipitation. Significant positive correlations
between sediment and water were identified at MC2, MC5, and MC8 (Table 3.3.). Sediment
impacts on water were better explained using E. coli than enterococci.

Table 3.3. Sites identified were water quality could be partially explained by sediment levels
(CART) and the associated correlations and significance.
Site

Water Fecal indicator

Sediment fecal indicator

MC2* E. coli
Enterococci
MC3
E. coli
E. coli
MC4
E. coli
Enterococci
MC5* E. coli
E. coli
MC8** E. coli
E. coli
MC8** Enterococci
E. coli
* Significance level ≤ 0.05; ** Significance level ≤ 0.01

p
0.022
0.267
0.177
0.011
0.001
0.005

2

R
0.174
0.012
0.030
0.196
0.120
0.017

3.4. Discussion
This project aimed to quantify fecal contamination in a small mixed use watershed through
spatial and temporal monitoring of E. coli, enterococci, C. perfringens, and coliphage CN-13.
-1

Based on suggested geometric mean thresholds of E. coli (2.10 log10 MPN 100 ml , USEPA
-1

1986), enterococci (1.52 log10 MPN 100 ml , USEPA 1986), C. perfringens (1.70 log10 CFU
-1

-1

100 ml ), and coliphage CN-13 (1.56 log10 PFU 100 ml ), the Mitchell Creek was deemed
unsafe for swimming. The C. perfringens threshold comes from the Hawaii standard (Mahin and

148

Pancorbo 1999) and the coliphage threshold was developed by Love et al. (2010) following a
statistical equivalence comparison of coliphage to USEPA freshwater enterococci criteria.

E. coli and enterococci occurrences represented widespread fecal contamination in the Mitchell
Creek. C. perfringens levels indicated a long-term chronic input of fecal contamination and the
-1

lower concentrations of coliphage CN-13 (geometric mean = 1.83 log10 PFU 100 ml ) were
indicative of fresh fecal contamination. It was speculated, and further described below, that
wastewater infrastructure was the leading source of the overall fecal pollution in the Mitchell
Creek. Water quality at TCSP beach was generally regarded safe for recreational activities. E.
coli, enterococci, C. perfringens, and coliphage met the suggested geometric mean criteria
described above. Greater than 82% of samples met suggested E. coli, enterococci, and CN-13
single sample criterion, indicating TCSP was generally clean but for sporadic incidences of
elevated bacteria.

There are more than 1600 on-site septic systems in the Mitchell Creek watershed but a portion of
residents rely on a wastewater treatment plant which discharges outside of the watershed.
Therefore, this study used daily wastewater treatment plant (WWTP) discharge as a surrogate for
shorter term human population presence in the study area. This method is not completely
accurate for estimating small area populations since wastewater infrastructure is inherently leaky
and transient changes are not captured, but these gaps play into this project’s goals of identifying
human fecal contamination, regardless of input paths. While poor correlations existed between
WWTP discharge volume and B. theta (p = 0.993) and coliphage CN-13 (p = 0.871) in Mitchell
Creek, direct associations (p ≤ 0.002) were identified between WWTP discharge volumes and E.
149

coli, enterococci, and C. perfringens concentrations in Mitchell Creek water. Interestingly, there
are only four reported discharge points in Mitchell Creek watershed, none of which involve
sewage, suggestive of non-point sources of pollution (i.e. leaking wastewater infrastructure or
illicit discharges) as the primary sources of human fecal material.

In the current study, land use and microbe concentration analysis indicated the most appropriate
spatial scale for investigating E. coli, enterococci, C. perfringens, and coliphage CN-13 in the
Mitchell Creek was generally at the catchment scale. Specifically, sites with agriculture cover
above 38% in the catchment exhibited higher E. coli concentrations (> 2.49 log10 MPN 100 ml

-

1

) in water compared to sites with less than 38% agriculture cover. However, thorough analysis

indicated wetlands, urban, agriculture, and natural land types appear to be acting as sources of E.
coli, enterococci, C. perfringens, and coliphage in the Mitchell Creek. Differentiating between
specific land uses at any scale does not seem plausible using E. coli and enterococci alone, as
supported by Kang et al. (2010). Additional monitoring with a larger suite of molecular source
markers coupled with land use analysis at the catchment scale is recommended.

While land use was not helpful for identifying water quality microbial impairments, cluster
analysis showed that the most polluted cluster (MC2, MC3, MC4, MC5, and MC8) had the
highest average of agricultural cover (41%) at the catchment scale. Each site in this cluster was
composed of less than 40% wetland and forested cover combined and more than 20% urban
development in the catchment. Of special note in this cluster, MC5 had the highest overall E.
coli, C. perfringens, and B. theta geometric means recorded amongst all sites and coliphage was
detected in 100% of water samples, suggesting a chronic input of human feces entering Mitchell
150

Creek near MC5. In fact, a spatial survey of the surrounding area identified an old wastewater lift

station with structural integrity concerns. The lower river flow rates observed at MC6 may be
allowing time for C. perfringens spores and coliphage viruses to settle out of the water column
and accumulate in the sediment, resulting in the stark differences between sediment and water
concentrations not observed at any other site. These results demonstrate, and Goto and Yan
(2011) support, sediment in forested creek stretches contain more bacteria than creek sediments
in urban areas.

Characterization of environmental and physical parameters in parallel with microbial analysis
revealed precipitation significant relationships between E. coli, enterococci, C. perfringens,
coliphage CN-13, and B. theta levels in Mitchell Creek and TCSP water (p < 0.01). Furthermore,
temporal analysis indicated precipitation had a more rapid effect on water quality (E. coli and
enterococci responded to rainfall in 1 hour) than sediment quality (E. coli and enterococci
responded rainfall in 12-24 hours). These results were not surprising given a number of studies
have previously demonstrated precipitation driven microbe concentrations in water (Fong et al.
2

2007; Converse et al. 2011; Walters et al. 2011). In a small watershed (16 km ) with river flow
rates similar to those seen in the Mitchell Creek, Goto et al. (2011) reported that 72 hour
cumulative rainfall was a strong positive driver of E. coli, enterococci, and C. perfringens in
urban and agriculture dominated catchments, but not in catchments dominated by forests.

It is likely that the slower microbial response to precipitation seen in sediments was the result of
either creek flow rates, particularly the slowing of flows, and the settling of E. coli and
enterococci out of the water column following initial loading of runoff and bacteria or the
151

introduction from shallow groundwater. As indicated by enterococci and coliphage
concentrations, precipitation impacts on water quality were more pronounced in the creeks lower
reaches which were directly and significantly related to creek discharge rates (p < 0.002). A
discharge threshold for Mitchell Creek was identified (86

th

percentile; range = 0.025 to 0.509

3 -1

m s ) which was observed when enterococci (33%) and coliphage CN-13 (89%) samples
th

exceeded 95 percentiles of this project data set distribution. This threshold is recommended to
guide future sampling as the rate represents an important statistical threshold, above which water
samples should be assessed for enterococci, C. perfringens, coliphage CN-13, and B. theta.
Analysis should also measure Salmonella spp. which was detected more often when river
discharge exceeded the 83

rd

percentile in an agriculturally dominated stream (Wilkes et al.

2011). Similar discharge thresholds may be applied to other small, flashy, and highly mixed
watersheds throughout the Great Lakes.

Transport mechanisms at the TCSP beach were different than those observed in the Mitchell
Creek. Specifically, E. coli and coliphage CN-13 levels at TCSP were transported by wind
(direction and speed) while enterococci and C. perfringens were influenced by the Mitchell
Creek (water quality and discharge). Interestingly, B. theta was driven by both wind and Mitchell
Creek. It was speculated that C. perfringens and B. theta were more influenced by the Mitchell
Creek because their associated methods detected persistent targets that do not readily grow in
water (Tallon et al. 2005; Desmarais et al. 2002; Ballesté, and Blanch 2010; Converse et al.
2009), whereas E. coli and enterococci associations may have been masked by growth and
coliphage by inactivation. Dwight et al. (2002) reported correlations between discharge and
152

enterococci were highest at beaches next to river outlets. Results presented here strengthen such
findings as C. perfringens, B. theta, and enterococci from the Mitchell Creek negatively
impacted the nearby TCSP water.

It is recommended that in addition to discharge threshold sampling, E. coli and coliphage CN-13
monitoring at TCSP should be conducted when 12 hour cumulative precipitation exceeds 0.23
mm, wind is out of the southwest, and wave height exceeds 0.13 m to coincide with the greatest
potential risk to bathers and E. coli levels. It is strongly suggested that TCSP beach monitoring
expand to include analysis near the mouth of the Mitchell Creek and include molecular source
tracking and multiple fecal indicators (E. coli, enterococci, and coliphage CN-13). Eventually,
this type of data should feed into mechanistic and predictive models at TCSP beach to better
understand the timing and quantity of pollution entering the TCSP designated swimming area.

Results indicated both die-off and regrowth of bacteria need to be considered when developing
sampling strategy. Daily fecal indicator concentrations (E. coli and enterococci) were lower
when sampled in the afternoon, potentially from increased temperatures or solar radiation. Thus,
it is recommended that morning is the most appropriate time to collect the samples. On the other
hand, during peak recreational activity, bacteria were likely regrowing as observed temperature
ranges were consistent with previously reported growth thresholds for E. coli (18 - 44.5 °C) and
enterococci (9 - 47.8 °C) (Johnson and Lewin 1946; Fisher and Phillips 2009; Borrego et al.
2002). This is another reason why coliphage and C. perfringens should be included as part of the
sampling in watersheds that are supporting high levels of traditional enteric bacteria.

153

Local concerns surrounding the Mitchell Creek’s influence on TCSP bathing water was the
initially reason for this study. Regression analysis showed a portion of TCSP water quality could
be explained by the previous hours Mitchell Creek water quality while birds and contaminated
runoff from outside the Mitchell Creek may also be contributing fecal contamination, as
previously suggested by Haack et al. (2003), this should be further investigated particularly with
the coliphage markers. However, the Creek water may be more influential then test results
indicate as great uncertainty was associated with the number and timing of sampling events
which were not frequent enough to define associates between the creek and beach due to their
close proximity. Long-term water quality improvement in the Mitchell Creek, and subsequently
at TCSP, should focus on 1) the remediation of wetlands throughout the watershed while
adopting minimum catchment wetland coverage (14.5%) policy to minimize the effects of runoff
and reduce water flow rates and 2) investigate wastewater and stormwater infrastructure,
including on-site septic systems, for structural integrity and illicit connections as human fecal
contamination was identified throughout the watershed.

Acknowledgements
I would like to acknowledge the contributions of those who assisted in the research project,
including Sarah U’Ren and Maureen McManus (Watershed Center Grand Traverse Bay) and
Sibel Zeki, Rebecca Ives, Chris Wendt, and Stephanie Longstaff, (Michigan State University).
We would like to thank Kendra Spence Cheruvelil, Gary Roloff, Tao Zhang, Shannon Briggs,
Anthony Kendall, and Sherry Martin who were instrumental in the development of this chapter.
Partial funding of this research was provided by the Michigan Department of Environmental
Quality.

154

APPENDIX

155

Source shed composition
Figure S.3.1. Land use patterns at multiple scales for each sampling location.

156

Table S.3.1. Number of water and sediment samples assayed for E. coli, enterococci, C.
perfringens, and coliphage CN-13.
Site
TCSP MC1 MC2 MC3 MC4 MC5 MC6 MC7 MC8
Water
43
26
42
42
42
42
42
22
42
E. coli
Sediment
29
0
29
28
19
29
29
0
28
Water
41
25
41
41
41
41
41
21
41
Enterococci
Sediment
28
0
28
27
19
28
28
0
27
Water
21
23
22
22
22
22
21
19
22
C. perfringens
Sediment
9
0
9
8
0
9
9
0
8
Water
22
22
22
22
22
22
22
20
22
CN-13
Sediment
9
0
9
8
0
9
9
0
8
Fewer sediment samples were collected in comparison to water samples as sediment analysis
was added to the project in 2010 only. No sediment samples collected at MC1and MC7 because
the bottom substrate was cobbled rock or a culvert, respectively. MC1 and MC7 were also not
included in the intensive study, explaining the lower number of water and sediment samples at
these two sites. MC7 was also an intermediate stream site and water was not always available for
sample collection at this site. MC4 was included in the intensive study, however, prior to that
event sediment samples were not collected.
Microorganism Medium

157

Table S.3.2. Water E. coli, enterococci, C. perfringens, and coliphage CN-13 recordings for each
event at TCSP and in the Mitchell Creek.
Date

Indicator
E. coli

TCSP MC1 MC2 MC3 MC4 MC5 MC6 MC7 MC8
0.48
2.86 2.24 2.94 2.89 3.38 1.99 2.19 3.24

Enterococci
6/24/2009

2.69

2.37

2.71

2.51

2.70

1.96

2.51

2.65

C. perfringens

1.11

0.75

1.27

1.32

0.90

0.64

0.93

1.33

1.91

2.15

1.96

2.12

1.32

1.04

1.04

CN-13

2.34

E. coli

1.44

3.38

Enterococci

1.50

3.62

C. perfringens

0.25

1.68

E. coli

1.41

2.71

1.79

2.76

2.86

2.99

1.78

2.10

2.51

Enterococci

2.39

2.84

2.43

3.12

3.06

3.19

1.95

2.33

2.29

C. perfringens

0.53

1.16

0.91

1.24

1.31

1.45

1.15

0.59

1.24

CN-13

1.04

1.71

1.71

1.71

2.21

1.61

1.96

1.71

1.04

E. coli

0.71

2.74

1.90

2.66

2.79

3.02

2.02

2.04

2.79

Enterococci

2.65

3.01

2.71

3.24

3.12

3.33

2.36

2.68

3.12

C. perfringens

0.05

1.13

1.03

1.23

1.24

1.67

1.00

1.37

1.16

CN-13

1.13

1.61

2.05

1.32

1.61

1.04

1.04

1.32

1.49

E. coli

0.48

3.24

3.24

3.24

3.30

2.94

2.20

3.15

Enterococci

1.29

3.52

3.74

3.52

3.49

3.05

2.77

3.19

C. perfringens

0.39

1.67

1.35

1.67

2.15

2.04

0.29

1.38

CN-13

1.04

1.32

2.00

1.32

1.32

2.05

1.04

1.85

E. coli

0.97

2.76

2.27

2.81

2.84

2.84

2.04

2.32

2.59

Enterococci

0.29

3.08

2.78

3.16

2.95

3.14

2.39

2.53

2.91

C. perfringens

0.29

2.03

1.03

2.06

1.99

2.26

1.04

0.74

1.85

CN-13

1.32

1.85

1.04

1.80

1.49

1.96

1.32

1.85

2.08

E. coli

0.30

3.24

2.79

3.19

3.19

3.11

3.05

3.08

3.08

Enterococci

0.51

3.49

3.21

3.58

3.54

3.71

2.94

3.28

3.49

C. perfringens

0.05

1.53

1.16

1.63

2.16

2.50

1.03

0.87

2.38

CN-13

7/1/2009

1.04

1.04

3.18

2.05

3.18

3.15

3.54

2.52

3.20

3.27

CN-13

7/8/2009

7/15/2009

7/22/2009

7/29/2009

8/3/2009

♦ mCP auger not available; *Average of 12 samples collected hourly starting at 07:00 EST; E.
-1
coli reported as MPN 100 ml-1 dry sediment weight; Enterococci reported as CFU 100 ml dry
sediment weight; Due to time constraints and processing logistics, C. perfringens and coliphage
CN-13 were not assayed during intensive study beginning August 7, 2010.
158

Table S.3.2. (cont’d)
Date

TCSP MC1 MC2 MC3 MC4 MC5 MC6

MC7

MC8

E. coli

1.23

2.81

3.08

2.64

2.76

2.84

2.61

2.74

2.86

Enterococci
8/9/2009

Indicator

0.79

3.20

3.20

3.21

3.23

3.40

3.01

3.09

3.27

1.87

0.87

2.06

2.47

2.30

0.64

2.20

C. perfringens
CN-13
E. coli

2.29

2.44

2.61

2.59

1.76

2.29

2.41

2.07

3.32

3.21

3.36

3.28

3.49

2.73

3.12

3.39

C. perfringens

0.31

1.56

1.07

1.34

1.26

1.58

0.80

0.83

1.30

CN-13

1.04

2.18

2.60

1.85

1.91

2.08

1.32

2.62

1.96

E. coli

1.03

3.11

3.38

2.69

2.69

2.81

1.80

2.20

2.59

Enterococci
8/17/2009

2.56

Enterococci
8/12/2009

1.11

0.30

3.15

3.38

3.02

3.02

3.08

2.38

3.11

2.99

1.04

2.21

2.97

2.30

2.05

1.96

1.32

1.85

1.71

C. perfringens♦
CN-13
E. coli

2.81

2.81

3.08

2.99

Enterococci

1.56

3.20

3.13

4.50

3.28

3.36

2.72

2.94

3.25

C. perfringens

0.19

1.67

1.70

1.05

1.79

1.90

0.82

1.18

1.73

1.04

1.61

2.34

1.49

2.15

2.26

1.85

2.57

2.61

0.48

2.30

3.48

2.42

2.37

2.32

2.07

2.15

2.34

Enterococci

1.32

2.84

3.64

2.81

2.63

3.11

2.39

2.91

3.02

1.04

2.08

2.51

2.15

2.21

2.71

1.49

3.00

2.08

0.79

2.24

2.86

2.14

2.16

2.38

1.92

2.29

2.27

Enterococci

0.85

2.49

3.26

2.54

2.39

2.57

2.28

2.44

2.51

C. perfringens

0.10

1.41

1.04

1.35

1.71

1.97

1.15

1.66

1.67

CN-13

1.04

1.04

1.71

1.04

1.04

1.32

1.04

2.05

1.32

E. coli

1.74

2.29

2.99

2.05

2.21

2.51

1.79

2.49

2.27

Enterococci

1.30

2.50

3.45

2.57

2.49

2.74

2.18

2.61

2.60

C. perfringens

0.62

1.30

1.57

1.41

1.56

1.69

1.23

1.45

1.63

CN-13

1.04

1.32

1.04

1.04

1.04

1.04

1.04

2.68

1.04

E. coli
9/16/2009

2.61

E. coli

9/9/2009

3.38

CN-13

9/2/2009

2.74

E. coli
8/31/2009

3.02

CN-13

8/26/2009

1.75

1.76

2.40

0.48

2.34

2.18

2.36

1.82

2.42

Enterococci

1.56

2.83

2.56

2.84

2.74

2.86

2.63

2.80

C. perfringens

0.18

1.45

1.18

1.37

1.60

1.92

1.14

1.89

C. perfringens♦

CN-13
159

Table S.3.2. (cont’d)
Date

Indicator

TCSP MC1 MC2 MC3 MC4

MC5

MC6

MC7

MC8

E. coli

1.42

1.39

1.65

1.56

1.08

1.23

1.37

Enterococci

1.80

2.10

1.69

1.82

1.64

1.58

1.71

1.87

C. perfringens

1.43

1.17

0.93

1.00

0.70

0.83

0.64

0.71

CN-13

11/2/2009

1.66

1.61

2.28

1.49

1.32

1.71

1.04

1.61

1.91

E. coli

2.69

2.61

3.05

2.16

2.79

2.66

Enterococci

1.23

2.21

3.10

2.54

2.66

2.82

1.95

2.67

2.30

C. perfringens

0.43

1.52

1.08

1.25

1.68

1.97

0.94

0.70

1.59

1.04

2.23

3.12

1.71

2.15

2.23

2.00

1.71

2.18

2.61

2.40

2.00

2.42

2.42

2.69

1.94

2.35

2.21

Enterococci

2.26

2.49

1.95

2.43

2.39

2.42

1.97

2.71

2.08

C. perfringens

0.95

1.21

1.00

1.20

1.29

1.26

0.85

0.56

1.30

CN-13

1.04

1.91

2.72

2.00

1.71

2.18

1.32

1.49

2.00

E. coli

2.17

2.76

2.51

2.69

2.66

2.71

2.16

2.69

2.59

Enterococci

1.86

2.72

2.32

2.69

2.71

2.77

2.25

2.44

2.71

C. perfringens

0.83

0.56

0.83

1.02

0.85

0.83

0.50

0.61

1.05

E. coli

2.89

3.30

2.47

3.38

3.38

3.38

2.99

3.38

3.38

Enterococci

2.74

3.24

2.63

3.31

3.41

3.47

2.71

3.27

3.28

C. perfringens

0.48

0.64

0.62

0.41

1.00

1.90

0.43

0.26

0.67

CN-13

3.11

3.71

3.85

3.85

3.73

3.60

3.56

3.74

3.57

E. coli

2.01

2.76

2.81

2.50

2.61

2.72

2.81

2.12

2.79

Enterococci

2.70

2.72

2.55

2.58

2.56

2.68

2.11

2.43

2.57

C. perfringens

0.24

1.45

0.56

1.58

1.88

2.30

0.85

0.66

1.86

CN-13

1.04

1.91

2.46

1.91

2.08

2.05

1.85

2.43

2.05

E. coli

1.11

2.54

2.24

2.44

2.40

2.76

2.22

2.51

2.64

Enterococci

1.18

2.53

2.31

2.43

2.41

2.48

2.46

2.51

2.44

C. perfringens

0.34

1.22

1.13

0.74

1.12

1.90

0.72

0.44

1.42

CN-13

6/14/2010

2.84

E. coli
6/10/2010

2.59

CN-13

6/7/2010

1.59

1.13

1.32

1.91

1.04

1.49

1.61

1.04

2.23

1.32

E. coli

1.31

2.74

2.69

2.51

2.79

2.94

2.47

3.02

2.84

CN-13

1.13

1.13

2.00

1.95

2.10

2.10

1.13

2.69

1.95

CN-13

6/24/2010

6/27/2010

7/1/2010

7/5/2010

160

Table S.3.2. (cont’d)
Date

Indicator

TCSP MC1 MC2 MC3 MC4 MC5

MC6

MC7

MC8

E. coli

8/10/2010
8/12/2010
8/14/2010
8/17/2010
3/19/2011
3/20/2011

2.81

2.49

3.24

2.94

Enterococci

1.01

3.13

3.26

2.83

2.85

2.84

2.57

3.17

2.71

C. perfringens

0.18

1.01

1.23

1.00

0.85

1.37

0.74

0.62

1.17

1.13

1.04

1.61

1.49

1.49

1.79

1.04

2.15

1.85

1.32

3.15

3.38

2.84

2.91

2.99

2.81

3.38

2.84

Enterococci

0.85

3.47

3.16

2.94

2.94

2.41

2.79

3.41

3.06

C. perfringens

0.18

0.94

1.25

0.91

0.99

1.58

0.99

0.12

0.48

1.13

2.21

2.12

1.49

1.91

2.59

1.71

2.42

2.59

0.61

2.79

2.51

2.61

2.71

2.61

2.66

2.71

Enterococci

0.30

2.76

2.40

2.64

2.54

2.69

2.12

2.61

C. perfringens

0.18

1.11

0.82

1.00

1.01

1.32

0.56

0.87

CN-13

8/8/2010

2.71

E. coli

8/7/2010**

2.64

CN-13

7/18/2010

3.08

E. coli
7/15/2010

2.66

CN-13

7/12/2010

1.49

1.13

1.96

2.81

1.61

1.71

1.61

1.04

2.12

E. coli

1.12

2.44

2.45

2.47

2.67

2.17

2.71

Enterococci

1.15

2.27

2.39

2.36

2.55

1.95

2.47

E. coli

2.61

3.38

3.38

3.38

3.38

3.38

3.24

Enterococci

2.26

3.36

3.33

3.11

3.46

3.27

3.31

E. coli

1.47

2.27

2.74

2.69

2.96

2.17

2.81

Enterococci

1.16

2.31

2.68

2.68

2.78

2.17

2.60

E. coli

1.48

2.74

2.42

2.51

2.71

2.64

2.76

Enterococci

1.16

2.59

2.61

2.60

2.80

2.29

2.73

E. coli

1.61

2.49

2.81

2.99

3.05

2.24

2.96

Enterococci

1.44

2.69

2.70

2.80

2.94

2.28

2.83

E. coli

1.34

2.43

2.79

2.61

2.91

2.18

2.84

Enterococci

0.88

2.62

2.77

2.81

2.92

2.30

2.91

E. coli

0.30

1.12

0.30

0.86

0.61

1.13

1.03

0.48

1.45

Enterococci

0.30

1.26

1.08

1.03

0.98

1.89

1.21

1.03

1.56

E. coli

0.30

1.83

0.79

1.82

1.72

1.94

0.61

0.61

2.05

Enterococci

0.30

1.95

1.32

2.32

2.46

2.71

1.32

1.19

2.81

161

Table S.3.3. Sediment E. coli, enterococci, C. perfringens, and coliphage CN-13 recordings for
each event at TCSP and in the Mitchell Creek.
Date
6/7/2010

6/10/2010

6/14/2010

6/24/2010

6/27/2010

7/1/2010

7/12/2010

Indicator
E. coli
Enterococci
C. perfringens
CN-13
E. coli
Enterococci
C. perfringens
CN-13
E. coli
Enterococci
C. perfringens
CN-13
E. coli
Enterococci
C. perfringens
CN-13
E. coli
Enterococci
C. perfringens
CN-13
E. coli
Enterococci
C. perfringens
CN-13
E. coli
Enterococci
C. perfringens
CN-13

TCSP
2.50
2.20
2.10
< 2.10
2.20
2.70
< 1.11
< 1.11
2.30
2.30
< 1.14

MC2
3.20
3.10
3.50
< 2.11
2.70
2.90
3.00
< 1.15
4.50
3.70
3.47

MC3

2.50
2.90
1.57
< 1.11
1.90
2.20
< 1.15
< 1.15
< 1.1
1.40
< 1.13
< 1.13
1.90
1.10
1.13
< 1.13

3.80
3.30
3.31
2.67
3.60
2.90
2.93
2.59
3.80
3.30
2.57
2.39
3.80
3.00
2.92
1.97

MC4

MC6
4.00
4.00
4.97
< 2.90
5.90
6.30
6.52
< 3.90
4.10
3.30
3.84

MC8

3.10
3.10
2.94
< 1.14
4.30
3.40
3.10

MC5
3.80
3.60
3.57
2.14
3.70
3.50
3.78
1.18
4.00
3.10
3.58

3.80
3.70
2.76
2.19
5.40
4.50
4.71
3.10
3.50
3.20
2.88
1.43
3.20
3.20
2.89
< 1.15

3.70
3.40
3.46
1.49
4.10
3.20
2.80
1.17
3.60
2.70
2.39
< 1.18
3.50
3.00
2.18
< 1.16

4.50
4.00
3.45
2.39
4.30
3.40
3.95
< 1.94
4.10
3.40
3.60
< 1.85
2.90
2.70
2.57
< 1.18

4.00
3.80
1.97
1.61
3.90
2.90
3.48
< 1.15
4.60
4.30
2.98
2.75
4.70
3.70
3.44
1.63

3.20
2.80
2.87
< 1.14
3.40
3.00
1.81

*Average of 12 samples collected hourly starting at 07:00 EST; E. coli reported as log10 MPN
100 g

-1

dry sediment weight; Enterococci and C. perfringens reported as log10 CFU 100 g
-1

-1

dry

sediment weight; Coliphage CN-13 reported as log10 PFU 100 g dry sediment weight; Due to
time constraints and processing logistics, C. perfringens and coliphage CN-13 were not assayed
during intensive study beginning August 7, 2010.

162

Table S.3.3. (cont’d)
Date

Indicator

TCSP

MC2

MC3

MC4

MC5

MC6

MC8

E. coli

1.40

3.80

3.70

3.10

4.50

4.40

Enterococci

2.10

3.20

3.30

2.80

4.50

3.50

C. perfringens 1.59

2.63

2.72

1.63

3.69

2.09

CN-13

< 1.13

1.84

1.61

1.17

< 1.97

2.31

E. coli

2.40

3.70

3.00

3.90

3.40

3.40

Enterococci

1.10

3.50

2.80

3.20

3.00

3.20

C. perfringens 1.14

2.51

2.70

2.30

2.11

2.67

CN-13

< 1.14

1.62

< 1.14

1.47

< 1.18

< 1.18

E. coli

2.00

3.50

3.70

3.10

3.20

2.80

3.60

Enterococci

2.10

2.90

3.10

2.80

2.50

2.30

2.80

E. coli

1.10

4.60

3.70

3.40

3.60

3.50

4.40

Enterococci

1.80

4.00

3.20

2.60

3.10

3.00

3.70

7/15/2010

7/18/2010

8/7/2010*

8/8/2010
8/10/2010
8/12/2010
8/14/2010
8/17/2010
3/19/2011
3/20/2011

E. coli

2.00

3.70

3.60

4.40

3.90

3.50

4.50

Enterococci

1.80

2.80

3.30

3.30

3.00

2.60

3.50

E. coli

2.20

4.00

3.40

3.50

4.20

3.50

4.10

Enterococci

1.60

3.00

3.00

3.00

2.90

2.70

3.30

E. coli

1.70

3.80

3.50

2.60

4.00

3.40

4.40

Enterococci

2.80

2.90

3.60

3.10

3.50

3.40

3.70

E. coli

2.00

3.80

3.30

4.00

3.60

3.10

4.50

Enterococci

2.00

3.30

2.70

2.70

2.80

3.00

3.50

E. coli

< 1.2

1.70

2.10

2.10

2.20

2.20

2.70

Enterococci

1.60

2.40

3.40

2.70

2.80

3.60

4.00

E. coli

< 1.2

1.80

2.30

1.70

2.20

3.20

2.60

Enterococci

1.20

1.70

2.80

2.40

2.80

3.20

3.40

163

Table S.3.4. Summary statistics of physical and environmental properties for all sites.
Unit

n

Minimum

Mean

Air temp.

°C

361

-0.80

18.00

34.50

Standard
Deviation
8.02

Water temp.

°C

362

-0.50

14.43

26.00

4.70

Air temp. (day mean)

°C

361

0.00

18.47

27.70

5.87

Barometric pressure

Hg in.

360

0.00

28.88

32.67

3.72

%

359

30.45

79.99

97.50

17.76

360

1981.30

18284.95

28038.00

5758.52

378

3.75

4.46

5.03

0.32

378

0.00

5.55

24.14

6.22

Wind direction

Km h
Degree

378

-

154.65

-

87.75

Wave height*

m

45

0.00

6.55

1.83

0.33

Precipitation 1 hrs

mm

336

0.00

0.13

4.13

0.65

Precipitation 2 hrs

mm

336

0.00

0.24

5.63

1.00

Precipitation 3 hrs

mm

336

0.00

0.42

8.42

1.46

Precipitation 4 hrs

mm

336

0.00

0.87

18.08

3.11

Precipitation 6 hrs

mm

336

0.00

0.96

18.08

3.23

Precipitation 8 hrs

mm

336

0.00

1.01

18.08

3.24

Precipitation 12 hrs

mm

336

0.00

1.62

23.29

4.19

Precipitation 16 hrs

mm

336

0.00

2.06

23.29

4.77

Precipitation 24 hrs

mm

336

0.00

2.08

23.29

4.77

Precipitation 48 hrs

mm

336

0.00

5.31

29.21

6.48

Precipitation 72 hrs
*Measured at TCSP only

mm

336

0.00

9.42

73.88

13.87

Variable name

Relative humidity
Solar radiation (day mean)
WWTP
Wind speed

-2

kJ m
MGD

-1

164

Maximum

Coliphage

Air temp.

Water temp.

Daily air temp.

Barometric P.

Humidity

Solar radiation

Discharge

1 hour precip.

2 hour precip.

4 hour precip.

8 hour precip.

12 hour precip.

48 hour precip.

C. perfringens

24 hour precip.

Enterococci

Enterococci
C. perfringens
Coliphage
Air temp.
Water temp.
Daily air temp.
Barometric P.
Humidity
Solar radiation
Discharge
1 hour precip.
2 hour precip.
4 hour precip.
8 hour precip.
12 hour precip.
16 hour precip.
24 hour precip.
48 hour precip.
72 hour precip.

16 hour precip.

E. coli

Table S.3.5. Spearman’s correlation matrix of environmental, weather, and microorganisms in water (Mitchell Creek and TCSP).

0.80
0.37
0.54
0.25
0.05
0.34
0.21
0.20
0.01
0.20
0.17
0.25
0.27
0.30
0.45
0.46
0.44
0.20
0.15

0.46
0.47
0.05
0.06
0.22
0.05
0.34
0.05
0.15
0.14
0.20
0.26
0.27
0.43
0.44
0.39
0.15
0.13

0.18
0.11
0.34
0.02
0.13
0.03
0.06
0.02
0.01
0.01
0.09
0.03
0.06
0.01
0.02
0.08
0.06

0.22
0.10
0.07
0.06
0.04
0.03
0.28
0.02
0.02
0.02
0.03
0.29
0.33
0.37
0.47
0.46

0.74
0.64
0.64
0.40
0.32
0.05
0.03
0.03
0.08
0.13
0.22
0.23
0.16
0.26
0.16

0.56
0.47
0.13
0.21
0.20
0.07
0.09
0.19
0.19
0.27
0.28
0.22
0.25
0.22

0.49
0.04
0.16
0.12
0.17
0.23
0.30
0.42
0.27
0.26
0.13
0.08
0.01

0.47
0.12
0.11
0.14
0.04
0.06
0.17
0.17
0.17
0.10
0.08
0.03

0.29
0.03
0.17
0.20
0.15
0.14
0.18
0.18
0.19
0.16
0.07

0.01
0.37
0.29
0.34
0.36
0.30
0.29
0.24
0.23
0.24

0.04
0.02
0.03
0.01
0.04
0.04
0.07
0.06
0.08

0.89
0.77
0.68
0.54
0.50
0.47
0.26
0.13

0.89
0.78
0.64
0.60
0.56
0.35
0.22

0.87
0.72
0.68
0.64
0.45
0.32

0.80
0.78
0.72
0.35
0.20

0.99
0.93 0.94
0.48 0.49 0.58
0.31 0.32 0.40 0.77

Values italics indicate correlations significant at the 0.05 level; Values in bold indicate correlations significant at the 0.01 level;
Shaded boxes represent inverse (negative) correlation coefficients; Refer to Table S.3.4. for variable units.

165

C. perfringens

Coliphage

Air temp.

Water temp.

Daily air temp.

Barometric P.

Humidity

Solar radiation

Discharge

1 hour precip.

2 hour precip.

4 hour precip.

8 hour precip.

12 hour precip.

48 hour precip.

Enterococci

24 hour precip.

E. coli
Enterococci
C. perfringens
Coliphage
Air temp.
Water temp.
Daily air temp.
Barometric P.
Humidity
Solar radiation
Discharge
1 hour precip.
2 hour precip.
4 hour precip.
8 hour precip.
12 hour precip.
16 hour precip.
24 hour precip.
48 hour precip.
72 hour precip.

16 hour precip.

Table S.3.6. Spearman’s correlation matrix of environmental, weather, and microorganisms in sediment (Mitchell Creek and TCSP).

0.72
0.70
0.76
0.03
0.23
0.22
0.01
0.32
0.05
0.42
0.14
0.15
0.15
0.16
0.20
0.20
0.18
0.08
0.05

0.68
0.75
0.12
0.26
0.04
0.02
0.27
0.07
0.29
0.13
0.15
0.15
0.15
0.24
0.24
0.29
0.04
0.06

0.59
0.18
0.47
0.24
0.27
0.01
0.09
0.35
0.04
0.04
0.04
0.01
0.01
0.02
0.06
0.24
0.30

0.04
0.36
0.15
0.25
0.13
0.21
0.56
0.08
0.08
0.08
0.01
0.08
0.11
0.09
0.23
0.21

0.74
0.64
0.64
0.40
0.31
0.05
0.03
0.03
0.08
0.13
0.22
0.23
0.16
0.26
0.16

0.56
0.47
0.13
0.21
0.20
0.07
0.09
0.19
0.19
0.27
0.28
0.21
0.25
0.22

0.49
0.04
0.16
0.12
0.17
0.23
0.30
0.42
0.27
0.26
0.13
0.08
0.01

0.47
0.12
0.11
0.14
0.04
0.06
0.17
0.17
0.17
0.09
0.08
0.03

0.29
0.03
0.17
0.20
0.15
0.14
0.18
0.18
0.18
0.16
0.07

0.01
0.37
0.29
0.34
0.36
0.30
0.29
0.24
0.23
0.23

0.04
0.02
0.03
0.01
0.04
0.04
0.07
0.06
0.08

0.89
0.77
0.68
0.54
0.50
0.47
0.26
0.13

0.89
0.78
0.64
0.60
0.56
0.35
0.22

0.87
0.72
0.68
0.64
0.45
0.32

0.80
0.78
0.72
0.35
0.20

0.99
0.93 0.94
0.48 0.49 0.58
0.31 0.32 0.40 0.77

Values italics indicate correlations significant at the 0.05 level; Values in bold indicate correlations significant at the 0.01 level;
Shaded boxes represent inverse (negative) correlation coefficients; Refer to Table S.3.4. for variable units.

166

3 -1

Table S.3.7. Average daily discharge for each Mitchell Creek site (m s ).
Date
MC1
MC2
MC3
MC4
MC5
MC6
MC7
MC8
6/24/2009
0.623
0.041
0.583
0.587
0.345
0.244
0.082
0.256
7/1/2009
0.743
0.049
0.695
0.699
0.411
0.291
0.098
0.305
7/8/2009
0.605
0.04
0.566
0.570
0.335
0.237
0.080
0.248
7/15/2009
0.593
0.039
0.555
0.558
0.328
0.233
0.078
0.243
7/22/2009
0.623
0.041
0.583
0.587
0.345
0.244
0.082
0.256
7/29/2009
0.581
0.038
0.543
0.547
0.321
0.228
0.077
0.238
8/3/2009
0.641
0.042
0.599
0.604
0.355
0.251
0.084
0.263
8/9/2009
0.605
0.04
0.566
0.570
0.335
0.237
0.080
0.248
8/12/2009
0.665
0.044
0.622
0.626
0.368
0.261
0.088
0.273
8/17/2009
0.629
0.041
0.588
0.592
0.348
0.247
0.083
0.258
8/26/2009
0.611
0.04
0.571
0.575
0.338
0.240
0.080
0.251
8/31/2009
0.641
0.042
0.599
0.604
0.355
0.251
0.084
0.263
9/2/2009
0.695
0.046
0.650
0.654
0.384
0.273
0.091
0.285
9/9/2009
0.623
0.041
0.583
0.587
0.345
0.244
0.082
0.256
9/16/2009
0.563
0.037
0.527
0.530
0.311
0.221
0.074
0.231
11/2/2009
0.502
0.209
0.534
0.035
0.499
0.295
0.070
0.219
6/7/2010
0.893
0.058
0.835
0.841
0.494
0.350
0.118
0.366
6/10/2010
0.079
0.599
0.039
0.560
0.564
0.331
0.235
0.246
6/14/2010
0.635
0.042
0.594
0.598
0.351
0.249
0.084
0.260
6/24/2010
0.137
1.043
0.068
0.975
0.982
0.577
0.409
0.427
6/27/2010
0.683
0.045
0.639
0.643
0.378
0.268
0.090
0.280
7/1/2010
0.647
0.042
0.605
0.609
0.358
0.254
0.085
0.265
7/5/2010
0.563
0.037
0.527
0.530
0.311
0.221
0.074
0.231
7/12/2010
0.522
0.034
0.487
0.491
0.288
0.204
0.069
0.214
7/15/2010
0.486
0.032
0.454
0.457
0.268
0.190
0.064
0.199
7/18/2010
0.480
0.031
0.448
0.451
0.265
0.188
0.063
0.197
8/7/2010
0.468
0.031
0.062
0.437
0.440
0.258
0.183
0.192
8/8/2010
0.456
0.03
0.426
0.429
0.252
0.179
0.060
0.187
8/10/2010
0.456
0.03
0.426
0.429
0.252
0.179
0.060
0.187
8/12/2010
0.456
0.03
0.426
0.429
0.252
0.179
0.060
0.187
8/14/2010
0.456
0.03
0.426
0.429
0.252
0.179
0.060
0.187
8/17/2010
0.456
0.03
0.426
0.429
0.252
0.179
0.060
0.187
3/19/2011
0.456
0.03
0.426
0.429
0.252
0.179
0.060
0.187
3/20/2011
0.456
0.03
0.426
0.429
0.252
0.179
0.060
0.187
Values in bold represent actual discharge measurements; All other values were estimated using a
statistically related dependent factor based on the Boardman River gage (04126970) and
measured Mitchell Creek discharges.

167

Table S.3.8. Precipitation details for the Mitchell Creek watershed.
Month

Year Project daily
average
(mm)

Project
monthly total
(mm)

Long term
monthly average
(mm)

March

2009 1.35*

6.8*

50.3

April

2009 0.10

72.2

69.1

May

2009 0.10

72.1

58.4

June

2009 0.10

73.6

84.3

July

2009 0.08

59.9

79.8

August

2009 0.14

103.4

86.1

September 2009 0.07

50.2

90.9

October

2009 0.16

117.8

74.7

March

2010 0.00*

0.0*

50.3

April

2010 0.14

100.2

69.1

May

2010 0.09

69.1

58.4

June

2010 0.23

168.6

84.3

July

2010 0.12

92.3

79.8

August

2010 0.11

81.7

86.1

125.6

90.9

October
2010 0.11*
16.8*
*Average calculated for partial month

74.7

September 2010 0.17

168

5.5

Monthly discharge mean (MG)

A.*
5.0

4.5

4.0

3.5

5.5
Whole year
June-October

December

November

October

September

August

July

June

May

April

March

February

January

3.0

Month

B.**

Flow (MGD)

5.0

4.5

4.0

3.5

3.0
Sunday

Monday

Tuesday Wednesday Thursday

Friday

Saturday

Figure S.3.2. WWTP daily discharge flows (A.) averaged per month and (B.) averaged per day
of week. *between January 1, 2009 and April 29, 2011; **whole year and tourist season (June to
October)
169

th

Table S.3.9. Fecal indicator bacteria exceedances of the 95 percentile and associated discharge
percentile.
Date
E. coli
-1 8/31/2009
(MPN 100 ml )
Enterococci
7/1/2009
-1
(CFU 100 ml ) 7/22/2009
8/3/2009
8/3/2009
8/26/2009
8/31/2009
C. perfringens
NA
Coliphage
8/3/2009
-1
(PFU 100 ml ) 6/24/2010
6/24/2010
6/24/2010
6/24/2010
6/24/2010
6/24/2010
6/24/2010
6/24/2010
th

Site

Concentration
(log10)

Discharge
percentile

MC2

3.5

69.6

MC1
MC2
MC5
MC3
MC3
MC2

3.6
3.7
3.7
3.6
4.5
3.6

87.1
71.7
65.2
65.2
93.6
69.6

MC5
MC1
MC2
MC3
MC4
MC5
MC6
MC7
MC8

3.5
3.7
3.8
3.9
3.7
3.6
3.6
3.7
3.6

65.2
87.1
91.3
91.3
91.3
91.3
91.3
86.2
91.3

-1

th

E. coli 95 percentile: 3.4 log10 MPN 100 ml ; Enterococci 95 percentile: 3.5 log10 CFU 100
-1

th

-1

th

ml ; C. perfringens 95 percentile: 2.5 log10 CFU 100 ml ; Coliphage CN-13 95 percentile:
-1

th

3.3 log10 PFU 100 ml ; NA: No samples exceeded 95 percentile

170

Table S.3.10. Spearman’s correlation coefficients and significance levels between precipitation
and E. coli and enterococci in water and sediment from the Mitchell Creek.
Total precipitation
Water (r, p)
time (hour)
E. coli
Enterococci

Sediment (r, p)
E. coli

Enterococci

1

0.242, 0.000

0.175, 0.003

0.137, 0.070

0.132, 0.084

2

0.325, 0.000

0.249, 0.000

0.150, 0.050

0.149, 0.052

3

0.321, 0.000

0.242, 0.000

0.150, 0.046

0.149, 0.052

4

0.350, 0.000

0.324, 0.000

0.147, 0.051

0.148, 0.053

6

0.364, 0.000

0.322, 0.000

0.175, 0.020

0.148, 0.053

8

0.384, 0.000

0.349, 0.000

0.158, 0.036

0.148, 0.054

12

0.553, 0.000

0.539, 0.000

0.200, 0.008

0.235, 0.002

16

0.559, 0.000

0.543, 0.000

0.202, 0.007

0.243, 0.001

24

0.523, 0.000

0.474, 0.000

0.182, 0.015

0.284, 0.000

48

0.224, 0.000

0.183, 0.000

-0.081, 0.284

-0.039, 0.617

72

0.179, 0.000

0.175, 0.000

0.053, 0.483

0.064, 0.409

Data in bold indicates the highest correlation coefficients. Individual Mitchell Creek sites,
responded to precipitation on similar time frames. Other significantly influential parameters
included: wind direction, air temperature, water temperature, barometric pressure, relative
humidity, river discharge, and turbidity.

171

REFERENCES

172

REFERENCES

Akasaka, M. Takamura, Mitsuhashi, N., and Kadono, Y. (2010). Effects of land use on aquatic
macrophyte diversity and water quality of ponds. Freshwater Biology, 55, 909-922.
Bae, H.K., Olson, B., Hsu, K.L., and Sorooshian, S. (2010). Classification and Regression Tree
(CART) analysis for indicator bacterial concentration prediction for a Californian coastal area.
Water Science and Technology, 61, 545-553.
Bai, S. and Lung, W.S. (2005). Modeling sediment impact on the transport of fecal bacteria.
Water Research, 39, 5232-5240.
Ballesté, E. and Blanch, A.R. (2010). Persistence of Bacteroides species populations in a river as
measured by molecular and culture techniques. Applied and Environmental Microbiology, 76,
7608-7616.
Bisson, J.W. and Cabelli, V.J. (1979). Membrane filter enumeration method for Clostridium
perfringens . Applied and Environmental Microbiology, 37, 55-66.
Borrego, J., Castro, D, and Figueras, M. (2002). Fecal streptococci/Enterococci in aquatic
environments. In G. Bitton (Ed.), Encyclopedia of Environmental Microbiology (1264-1278).
New York: John Wiley and Sons, Inc.
Boyer, D.G. and Pasquarell, G.C. (1999). Agricultural land use impacts on bacterial water
quality in karst groundwater aquifer. Journal of American Water Resources Association, 35, 291300.
Breiman, L., Friedman, J., Olshen, R., and Stone C. (1984). Classification and Regression Trees.
Chapman and Hall, New York.
Broussard, W. and Turner, R. (2009). A century of changing land use and water quality
relationships in the continental US. Frontiers in Ecology and Environment, 7, 302-307.
Cho, K.H., Cha, S.M., Kang, J.H., Lee, S.W., Park, Y., Kim, J.W., and Kim, J.H. (2010).
Meteorological effects on the levels of fecal indicator bacteria in an urban stream: A modeling
approach. Water Research, 44, 2189-2202.
Converse, R.R., Blackwood, A.D., Kirs, M., Griffith, J.F., and Noble, R.T. (2009). Rapid QPCRbased assay for fecal Bacteroides spp. as a tool for assessing fecal contamination in recreational
waters. Water Research, 43, 4828-4837.

173

Converse, R.R., Piehler, M.F., and Noble, R.T. (2011). Contrasts in concentrations and loads of
conventional and alternative indicators of fecal contamination in coastal stormwater. Water
Research, 45, 5229-5240.
Daughton, C.G. (2012). Real-time estimation of small-area populations with human biomarkers
in sewage. The Science of the Total Environment, 414, 6-21.
De’ath, G. and Fabricius, K. (2000). Classification And Regression Trees: A powerful yet simple
technique for ecological data analysis. Ecology, 81, 3178-3192.
Desai, A.M., and Rifai, H.S. (2010). Variability of Escherichia coli Concentrations in an Urban
Watershed in Texas. Journal of Environmental Engineering, 136, 1347-1359.
Desmarais, T.R., Solo-Gabriele, H.M., Carol, J., and Palmer, C.J. (2002). Influence of soil on
fecal indicator organisms in a tidally influenced subtropical environment. Applied and
Environmental Microbiology, 68, 1165-1172.
Di Gregorio, A., and Jansen, L.J.M. (1997). A new concept for a land cover classification
system. Proceedings of the Earth Observation and Environmental Information 1997
Conference. Alexandria, Egypt, 13-16 October 1997.
Dwight, R.H., Semenza, J.C., Baker, D.B., and Olson, B.H. (2002). Association of urban runoff
with coastal water quality in Orange County, California. Water Environment Research, 74, 8290.
Edwards, A.C., Watson, H.A., and Cook, Y.E. (2012). Source strengths, transport pathways and
delivery mechanisms of nutrients, suspended solids and coliforms within a small agricultural
headwater catchment. The Science of the Total Environment,
DOI:10.1016/j.scitotenv.2011.10.055.
Fisher, K. and Phillips, C. (2009). The ecology, epidemiology and virulence of Enterococcus.
Microbiology, 155, 1749-1757.
Fong, T.-T., Mansfield, L.S., Wilson, D.L., Schwab, D.J., Molloy, S.L., and Rose, J.B. (2007).
Massive microbiological groundwater contamination associated with a waterborne outbreak in
Lake Erie, South Bass Island, Ohio. Environmental Health Perspectives, 115, 856-864.
Fulcher, J. (1991). Mitchell Creek hydrologic investigation: Incorporating both water quantity
and quality consideration in urbanizing watersheds. Land and Water Management Division,
Michigan Department of Natural Resources.
Goto, D.K., and Yan, T. (2011). Effects of Land Uses on Fecal Indicator Bacteria in the Water
and Soil of a Tropical Watershed. Microbes and Environments, 26, 254-260.

174

Gregor, J., Garrett, N., Gilpin, B., Randall, C., and Saunders, D. (2002). Use of Classification
and Regression Tree (CART) analysis with chemical faecal indicators to determine sources of
contamination. New Zealand Journal of Marine and Freshwater Research, 36, 387-398.
Haack, S.K., Fogarty, L.R., and Wright, C. (2003). Escherichia coli and enterococci at beaches
in the Grand Traverse Bay, Lake Michigan: sources, characteristics, and environmental
pathways. Environmental Science and Technology, 37, 3275-3282.
Hunter, C., Perkins, J., Tranter, J., and Gunn, J. (1999). Agricultural land-use effects on the
indicator bacterial quality of an upland stream in the Derbyshire Peak district in the U.K. Water
Research, 33, 3577-3586.
Johnson, F.H. and Lewin, I. (1946). The growth rate of E. coli in relations to temperature,
qunine, and coenzyme. Journal of Cellular Physiology, 28, 47-75.
Kang, J.H., Lee, S.W., Cho, K.H., Ki, S.J., Cha, S.M., and Kim, J.H. (2010). Linking land-use
type and stream water quality using spatial data of fecal indicator bacteria and heavy metals in
the Yeongsan river basin. Water Research, 44, 4143-4157.
Katz, D.M., Watts, F.J., and Burroughs, E.R. (1995). Effects of surface roughness and rainfall
impact on overland flow. Journal of Hydraulic Engineering, 121, 546-553.
Kistemann, T., Rind, E., Koch, C., Claßen, T., Lengen, C., Exner, M., and Rechenburg, A.
(2012). Effect of sewage treatment plants and diffuse pollution on the occurrence of protozoal
parasites in the course of a small river. International Journal of Hygiene and Environmental
Health, DOI:10.1016/j.ijheh.2011.12.008
Klein, R. (1979). Urbanization and stream quality impairment. Water Resources Bulletin, 15,
948-963.
Lavee, H. and Poesen, J.W. (1991). Overland flow generation and continuity on stone-covered
soil surfaces. Hydrological Processes, 5, 345-360.
Lemon, S.C., Roy, J., Clark, M.A., Friedmann, P.D., and Rakowski, W. (2003). Classification
and Regression Tree Analysis in public health: methodological review and comparison with
logistic regression. The Society of Behavioral Medicine, 26, 172-181.
Loch, R.J. (2000). Effects of vegetation cover on runoff and erosion under simulated rain and
overland flow on a rehabilitated site on the Meandu Mine, Tarong, Queensland. Australian
Journal of Soil Research, 38, 299-312.
Macgregor, B.J., Moser, D.P., Baker, B.J., Alm, E.W., Maurer, M., Nealson, K.H., and Stahl,
D.A. (2001). Seasonal and spatial variability in Lake Michigan sediment small-subunit rRNA
concentrations. Applied and Environmental Microbiology, 67, 3908-3922.

175

Mahin, T. and Pancorbo, O. (1999) Waterborne pathogens. Water Environment and Technology,
11, 51-55.
Martin, S.L., Soranno, P., Bremigan, M.T., and Cheruvelil, K.S. (2011). Comparing
hydrogeomorphic approaches to lake classification. Environmental Management, 48, 957-974.
McKergow, L.A. and Davies-Colley, R.J. (2010). Stormflow dynamics and loads of Escherichia
coli in a large mixed land use catchment. Hydrological Processes, 24, 276-289.
Mehaffey, M.H., Nash, M.S., Wade, T.G., Ebert, D.W., Jones, K.B., and Rager, A. (2005).
Linking land cover and water quality in New York City’s water supply watersheds.
Environmental Monitoring and Assessment, 107, 29-44.
Michigan Department of Environmental Quality (MDEQ). (2012). Michigan Beach Guard.
Retrieved from (http://www.deq.state.mi.us/beach/Default.aspx).
Michigan Department of Technology, Management, and Budget (MDTMB). (2002). Center for
shared solutions and technology partnerships (April 1, 2010). Retrieved from
(http://www.mcgi.state.mi.us/mgdl/).
Michigan State University (MSU). (2010). Michigan land cover/use classification system. D.P.
Lusch and R.F. Goodwin (Eds.) Michigan State University: Remote sensing and GIS research
and outreach services. Retrieved from
(http://www.rsgis.msu.edu/pdf/lclu/MI_LCLU_Classif_2010.pdf).
Midwest Regional Climate Center. (2012). Climate Summaries (April 30, 2012). Retrieved from
(http://mrcc.isws.illinois.edu/climate_midwest/maps/mi_mapselector.htm).
Muirhead, R.W., Davies-Colley, R.J., Donnison, A.M., and Nagels, J.W. (2004). Faecal bacteria
yields in artificial flood events: quantifying in-stream stores. Water Research, 38, 1215-1224.
Nadal-Romero, E., Regüés, D., and Latron, J. (2008). Relationships among rainfall, runoff, and
suspended sediment in a small catchment with badlands. Catena, 74, 127-136.
NOAA Coastal Services Center. (2001). Coastal change analysis program regional land cover
(April 1, 2010). Retrieved from (http://www.csc.noaa.gov/digitalcoast/data/ccapregional/).
NOAA. (2012). Weather observations for the past three days: Traverse City, Cherry Capital
Airport. Retrieved from (http://www.weather.gov/data/obhistory/KTVC.html).
Nobre, A.M. (2009). An ecological and economic assessment methodology for coastal
ecosystem management. Environmental Management, 44, 185-204.
Questier, F., Put, R., Coomans, D., Walczak, B., and Heyden, Y.V. (2005). The use of CART
and multivariate regression trees for supervised and unsupervised feature selection.
Chemometrics and Intelligent Laboratory Systems, 76, 45-54.
176

Rantz, S.E. (1982). Measurements and computation of streamflow: Volume 1. Measurements of
stage and discharge. USGS Water-Supply paper 2175. US Govt. Printing Office, Washington
D.C.
Rediske, R. (2010). Assessment of E. coli and microcystins in Cladophora mats in the nearshore
waters of Grand Traverse Bay, Little Traverse Bay, and Saginaw Bay [White paper]. Retrieved
from
(http://www.gvsu.edu/wri/envchem/assessment-of-em-e-coli-em-andmicrocystins-inCladophora-mats-in-the-nearshore-waters-of-grand-traverse-bay-littletraverse-bay-and-saginawbay--17.htm).
Rehmann, C.R. and Soupir, M.R. (2009). Importance of interactions between the water column
and the sediment for microbial concentrations in streams. Water Research, 43, 4579-4589.
Schilling, K.E., Zhang, Y.K., Hill, D.R., Jones, C.S., and Wolter, C.F. (2009). Temporal
variations of Escherichia coli concentrations in a large Midwestern river. Journal of Hydrology,
365, 79-85.
Smith, S.V., Swaney, D.P., Talaue-McManus, L., Bartley, J.D., Sandhei, P.T., McLaughlin, C.J.,
Dupra, V.C., Crossland, C.J., Buddemeier, R.W., Maxwell, B.A., and Wulff, F. (2003). Humans,
hydrology, and the distribution of inorganic nutrient loading to the ocean. BioScience, 53, 235245.
Srinivasan, S., Aslan, A., Xagoraraki, I., Alocilja, E., and Rose, J.B. (2011). Escherichia coli,
enterococci, and Bacteroides thetaiotaomicron qPCR signals through wastewater and septage
treatment. Water Research, 45, 2561-2572.
Stoermer, E.F., Ladewski, B.G., and Schelske, C.L. (1978). Population responses of Lake
Michigan Phytoplankton to nitrogen and phosphorus enrichment. Hydrobiologia, 57, 249-265.
Stumpf, C.H., Piehler, M.F., Thompson, S., and Noble, R.T. (2010). Loading of fecal indicator
bacteria in North Carolina tidal creek headwaters: hydrographic patterns and terrestrial runoff
relationships. Water Research, 44, 4704-4715.
Tallon, P., Magajna, B., Lofranco, C., and Leung, K.T. (2005). Microbial indicators of faecal
contamination in water: A current perspective. Water, Air, and Soil Pollution, 166, 139-166.
Traister, E., and Anisfeld, S.C. (2006). Variability of indicator bacteria at different time scales in
the upper Hoosic River watershed. Environmental Science and Technology, 40, 4990-4995.
United States EPA. (1995). Method for detection and enumeration of Clostridium perfringens in
water and sediments by membrane filtration. EPA/600/R-95/030/ Office of Research and
Development, Washington D.C.
United States Environmental Protection Agency (USEPA), O. of W. (1986). Ambient water
quality criteria for bacteria - 1986, (EPA440/5-8).
177

United States Environmental Protection Agency (USEPA). (2002). Method 1600: Enterococci in
water by membrane filtration using membrane-Enterococcus indoxyl-b-D-Glucoside agar (mEI).
EPA-821-R-02-022. Office of Water, Washington D.C.
United States Environmental Protection Agency (USEPA). (2001). Method 1601: Male specific
(F+) and somatic coliphage in water by two-step enrichment procedure. EPA 821-R-01-030.
U’Ren, Sarah. (2005). Grand Traverse Bay Watershed Protection Plan, The Watershed Center
Grand Traverse Bay (Feb 18, 2011). Retrieved from (http://www.gtbay.org/ourprograms/
watershedprotectionplan/).
USGS. (2012). Land processes distributed active archive center (April1, 2010). Retrieved from
(https://lpdaac.usgs.gov/).
USGS. (2010). Seamless
(http://seamless.usgs.gov/).

data

warehouse

(April

1,

2010).

Retrieved

from

Venables, W. and Ripley, B. (1999). Modern applied statistics with S-PLUS, 3rd edn. Springer,
New York
Yampara-Iquise, H., Zheng, G., Jones, J.E., and Carson, C.A. (2008). Use of a Bacteroides
thetaiotaomicron-specific alpha-1-6, mannanase quantitative PCR to detect human faecal
pollution in water. Journal of Applied Microbiology, 105, 1686-1693.
Walters, S.P., Thebo, A.L., and Boehm, A.B. (2011). Impact of urbanization and agriculture on
the occurrence of bacterial pathogens and stx genes in coastal waterbodies of central California.
Water Research, 45, 1752-62.
Wang, X. and Yin, Z. (1997). Using GIS to assess the relationship between land use and water
quality at a watershed level. Environmental International, 23, 103-114.
Wilkes, G., Edge, T., Gannon, V., Jokinen, C., Lyautey, E., Neumann, N., Ruecker, N., Scott, A.,
Sunohara, M., Topp, E., and Lapen, D. (2011). Associations among pathogenic bacteria,
parasites, and environmental and land use factors in multiple mixed-use watersheds. Water
Research, 45, 5807-5825.

178

CHAPTER 4.
FINE TUNING MICROBIAL WATER QUALITY CRITERIA FOR LOCAL
WATERSHEDS

179

4.1. Introduction
Molecular microbiological methods provide a promise of improved water quality assessment.
However, cultivation based detection of fecal indicator bacteria continue to be used for
recreational water quality assessment throughout the world (WH0 2003; USEPA 2011; EC
2006). Current scientific literature focusing on indicator organisms results in mixed conclusions,
calling into question the continued use of indicator systems for human health protection.
Multiple reports identify inconsistencies between culture and molecular based concentrations,
indicator and pathogen associations, and human health implications (Lavender and Kinzelman
2009; Byappanahalli et al. 2010; Wilkes et al. 2011).

Approved methods for beach monitoring include both cultivation and molecular technologies
(USEPA 2011; USEPA 2002a). The most commonly used standard method for routine beach
monitoring relies on the detection of cultivatable fecal indicator bacteria requiring incubation for
18-48 hours. Incubation periods delay water quality advisories and do not provide real-time
water quality information, and may result in unnecessary beach closures (Rabinovici et al. 2004;
Colford et al. 2012). To compensate, fecal indicator bacteria are increasingly being detected
using quantitative polymerase chain reaction (qPCR) which eliminates the cultivation process
(Siefring et al. 2008; Field and Samadpour 2007). This method can be used to detect most
microorganisms and produce quantitative results in a few hours (Girones et al. 2010). However,
qPCR detects both viable and non-viable organisms, potentially limiting its usefulness as a
method for assessing risk.

180

Discrepancies between methods have prompted a search for a common numerical factor
connecting cultivation and molecular methods. Whitman et al. (2010) suggests an empirical
relationship can be developed between cultivation and molecular Enterococci spp. following a
water survey from 37 US states. The authors do not provide a specific numerical factor; rather,
they suggest site specific function based on colony forming units (CFU) and CFU-CCE
(calibrated cell equivalents) coefficients reflective of background CCE persistence and CFU
variance (Whitman et al. 2010). Byappanahalli et al. (2010) and Haugland et al. (2005) report
strong positive correlations between cultivation and molecular methods for Enterococci spp. at
Lake Michigan beaches (r = 0.65 and r = 0.68, respectively) with CE measurements consistently
higher than CFU measurements (5-10 times higher as reported by Byappanahalli et al. 2010 and
16 times higher as reported by Haugland et al. 2005). Although relationships between qPCR and
culture based methods are suggested, each study reports that associations varied with respect to
environment, method, or pollution type. For example, Converse et al. (2012) found positive
correlations between culture and molecular based approaches for Enterococci spp., but note
association strength depends on whether the waterbody is dominated by point (r = 0.38 to 0.83)
or non-point sources (r = 0.19 to 0.34). Several studies and reviews report inconsistent
relationships between bacterial levels measured using non-cultivation techniques (i.e. qPCR) and
fecal indicator bacteria measured via cultivation methods (Santo Domingo et al. 2007; Haack et
al. 2009; Stapleton et al. 2009). To date, a single, universally accepted relationship between
cultivation and molecular based methods has not been adopted.

In addition to the relationship between indicators and methods, several studies focused on
relationships between pathogens and fecal indicator organisms with conflicting results.

181

Schriewer et al. (2010) found qPCR detection of Bacteroidales had direct and significant
correlations with Cryptosporidium spp. occurrences in estuary and river environments (r = 0.21,
p = 0.013). However, multiple studies report that indicator organisms rarely correlate with
pathogen detection. For instance, Harwood et al. (2005) and Wilkes et al. (2009) report routine
detection of fecal indicators (Escherichia coli (E. coli), Clostridium perfringens, enterococci,
total and fecal coliforms, and F-specific coliphage) in absence of pathogens (Campylobacter
spp., Salmonella spp., Giardia, Cryptosporidium, Listeria monocytogenes, and E. coli O157:H7)
using cultivation methods. Hellein et al. (2011) measured enterococci using two United State
Environmental Protection Agency (USEPA) approved methods (cultivation and molecular) to
conclude both techniques produced poor correlations with molecular based Campylobacter spp.
presence.

Finally, uncertainty surrounds microbial monitoring and human health implications. E. coli and
enterococci, the most common microorganisms for determining recreational water quality safety,
were directly linked to adverse health outcomes via epidemiological studies at multiple beaches
throughout the US (Cabelli et al. 1982; Cabelli 1983). A meta-analysis of 27 studies concluded
that E. coli and enterococci consistently represented gastrointestinal illness in fresh and marine
waters, respectively (Wade et al. 2003). Using qPCR and cultivation techniques to measure
Enterococcus, Wade et al. (2008) noted significant associations between gastrointestinal illness
in swimmers and bacterial concentrations in both marine and freshwater recreational beaches;
specifically a single log10 increase of daily averages for Enterococcus or Bacteroidales (16S
rRNA sequence from Dick and Fields 2004) (calibrated cell equivalents, CCE) doubled the risk
of gastrointestinal illness in swimmers (Wade et al. 2010). One study investigating method
182

specific health implications claimed molecular and cultivation methods produced similar
correlations between water exposure and gastrointestinal illnesses (Colford et al. 2012). Others
support these findings and suggest that molecular and culture methods represent similar
implications for beach actions (i.e. beach closure or advisory numbers) (Shibata et al. 2010;
Lavender and Kinzelman 2009). However, Colford et al. (2007) have also reported no existing
associations between illnesses and enterococci via cultivation or molecular methods during a
cohort study at a single marine beach. Cumulatively, implications for water quality management
following modification of microbial monitoring programs remain unclear, regardless of detection
organism or method. The contrast in bacteria and health relationships for studies focused on
single or multiple sites illustrates the need for fine tuning criteria at more localized scales.

The USEPA suggests new water quality criteria that include the use of qPCR for marine and
freshwater beach monitoring (USEPA 2011). Following epidemiological studies at nine beaches
from the Great Lakes and marine waters influenced by sewage treatment discharge, the USEPA
made public a new molecular method and criteria for beach closures/advisories targeting
Enterococci spp. with a geometric mean of 475 CCE 100 ml
1000 CCE 100 ml

-1

-1

(USEPA 2011) which represented the 75

and a statistical threshold level of

th

percentile of the microbial data

distributions found during the epidemiological studies. However, this value for the molecular
methods (versus cultivation criterion that is already used in the states) may not be appropriate at
sites with diffuse sources of pollution and watersheds with heterogeneous landscape patterns.
This manuscript aims to 1) evaluate molecular and cultivation based regulatory criteria in a
flashy watershed receiving primarily non-point source pollution, 2) investigate potential sub-

183

criteria for molecular markers not included in current regulations, and 3) analyze a long term
monitoring data set for implications of water quality interpretation following a method shift.

4.2. Materials and methods
4.2.1. Sampling location and collection
The Mitchell Creek (Michigan, USA) drains a watershed composed of urban (23.4%),
agriculture (37.7%), forest/open (24.7%), wetlands (14.0%), and water (0.1%) (Figure 4.1.).
Surface water grab samples (n = 111) were collected from four Mitchell Creek sites and one
Grand Traverse Bay beach site between June and September 2010 using sterile one liter Nalgene
bottles. All samples were placed in coolers on ice, transported to the laboratory, and analyzed
within four hours. Samples were collected under wet and dry conditions based on a
predetermined threshold of 5.1 mm cumulative rainfall in the 48 hours prior to sample collection.
These threshold levels were associated with rapid changes in surface water velocity and
discharge.

184

N
W

E
S

Figure 4.1. Mitchell Creek watershed and sampling locations within Michigan and the Great Lakes.
185

4.2.2. Enumeration of bacteria using cultivation techniques
Water subsamples (100 ml) were analyzed for E. coli and enterococci using chromogenic
substrate methods Quanti-Tray 2000 Colilert® and Enterolert®, (IDEXX Laboratories, Inc.,
Westbrook, ME) respectively. E. coli and enterococci trays were incubated for 24 hours at 35 °C
and 41 °C, respectively. Wells that fluoresced yellow were counted as E. coli positive and wells
that fluoresced blue were count as enterococci positive in their respective trays. E. coli and
enterococci concentrations were calculated from IDEXX Quanti-Tray®/2000 MPN table and
-1

reported as Most Probable Number (MPN) 100 ml . Stock cultures of E. coli (ATCC 15597)
and Enterococcus faecium (ATCC 35667) were used as positive controls.

4.2.3. Enumeration of bacteria using molecular techniques
A total of 900 ml per sample was filtered through a 47 mm 0.45 µm pores size nitrocellulose
membrane filter and then immersed into 25 ml of sterile phosphate buffer solution in a 50 ml
centrifuge tube. The solution and filter was vortexed at high speed (3200 RPM) for 2 minutes,
followed by filter removal, and the suspension was centrifuged at 4500 x g for 20 minutes. Using
sterile pipettes, 23 ml of supernatant was decanted and the remaining volume re-suspended to
form a 2 ml concentrate. A volume of 200 μl of the concentrated suspension was then used for
DNA extraction by QIAamp Mini DNA kit according to the manufacturer’s instructions (Qiagen,
Valencia, CA). The total volume of 200 μl of the concentrated samples were extracted for DNA
and stored in -20 °C until analyzed with qPCR. A negative control was included during filtration
and extraction, consisting of molecular grade water in lieu of sample product.

186

Standards for qPCR were prepared by extracting DNA from bacterial strains Escherichia coli
ATCC 15597 and Enterococcus faecalis ATCC 19433. Bacteroides thetaiotaomicron genomic
DNA was purchased from ATCC (29148D-5). The E. coli uidA, Enterococci spp. 23S rRNA,
and B. thetaiotaomicron a-mannanase genes were amplified separately using published primer
sets (Srinivasan et al. 2011; Frahm and Obst 2003; Yampara-Iquise et al. 2008). Polymerase
chain reaction (PCR) was performed in a 25 µl total reaction mix which contained 15 µl Hotstart
DNA Polymerase Mastermix, 0.4 mM of each primer, 2 µl of the template DNA, and molecular
grade water (QIAGEN, Valencia, CA, USA) to a final volume of 25 µl. The amplified PCR
products for all three target genes were cloned into TOPO PCR 2.1 and transformed with the
TOPO10 competent cells (Invitrogen Inc., Carlsbad, CA, USA) according to the protocol
provided by the manufacturer. Plasmids were extracted with QIAprep Spin MiniPrep kit
(QIAGEN, Valencia, CA, USA) and were sequenced at the Research Technology Support
Facility (RTSF) at Michigan State University to confirm the insertion of the target inside the
vector. The plasmids were quantified using Nano-Drop spectrophotometer (NanoDrop
Technologies Inc., Wilmington, DE, USA) and then serially diluted ten-fold to construct qPCR
0

6

standard curves. Triplicates of plasmid concentrations ranging from 10 to 10 copies per 5 µl
were used for the standard curve.

E. coli qPCR assay targeting the uidA gene (Srinivasan et al. 2011) had a detection limit of 100
copies per 5 µl. The E. coli qPCR reaction mix consisted of 10 µl of Taqman Light Cycler
Mastermix (Roche, Indianapolis, IN), 0.2 µM each of forward and reverse primers (Eurofins
MWG Operon, USA), 0.1 µM of probe (Eurofins MWG Operon, USA), 5 µl of template, and
nuclease free water to a final volume of 20 µl. The assay was carried out in LightCycler 2.0 ®
187

(Roche, Indianapolis, IN) through the following temperature profiles: initial denaturation for
10min at 95 °C followed by 45 cycles of denaturation for 6s at 95 °C; annealing for 8s at 58 °C
and extension at 72 °C for 8s.

Enterococcus qPCR assay targeting the 23S rRNA gene (Frahm and Obst 2003) had a detection
limit of 10 copies per 5 µl. The enterococcus qPCR reaction mix consisted of 10 µl of Taqman
Light Cycler Mastermix (Roche, Indianapolis, IN), 0.2 µM each of forward and reverse primers
(Eurofins MWG Operon, USA), 0.1 µM of the probe (Eurofins MWG Operon, USA), and 5 µl
of template and nuclease free water to a final volume of 20 µl. The assay was carried out in
LightCycler 2.0 ® (Roche, Indianapolis, IN) through the following temperature profiles: initial
denaturation at 10min at 95 °C followed by 45 cycles of denaturation for 15s at 95 °C; annealing
for 30s at 60 °C and extension at 72 °C for 15s.

The α-1-6, mannanase (BT 3501) targeting B. thetaiotaomicron qPCR assay (Yampara-Iquise et
al. 2008) had a detection limit of 10 copies per 5 µl. The reaction mix for qPCR B.
thetaiotaomicron consisted of 10 µl of Taqman Light Cycler Mastermix (Roche, Indianapolis,
IN), 0.2 µM each of forward and reverse primers (Eurofins MWG Operon, USA), 0.1 µM of the
probe (Eurofins MWG Operon, USA), and 5 µl of template and nuclease free water to a final
volume of 20 µl. The assay was carried out in LightCycler 2.0 ® (Roche, Indianapolis, IN)
through the following temperature profiles: initial denaturation at 10min at 95°C followed by 50
cycles of denaturation for 15s at 94 °C; annealing for 60s at 60 °C and extension at 72 °C for 5s.
Triplicate analysis was done for all dilutions, positive controls, and negative controls for all three
markers.

188

The copies of the corresponding genes were converted to cell equivalents (CE). In the cases of E.
coli and B. thetaiotaomicron, only one copy of the target gene is present in a cell, thus one copy
number corresponds to one cell. Viau and Peccia (2009) suggest four copies of 23S rRNA
present per enterococci cell, therefore DNA copies-to-cell conversions of enterococci qPCR
targets were based on a 4:1 ratio. All final concentrations for qPCR analyses were reported as
-1

qPCR cell equivalents (CE) 100 ml .

To examine DNA extract for inhibitory substances, five replicates from each sample DNA
extract initially negative for E. coli, Enterococci spp. or B. thetaiotaomicron DNA were pooled.
All samples were then diluted 10 and 100 times. Molecular grade water (control), undiluted, 10and 100- dilutions of DNA were spiked with known amounts of E. coli, enterococcus, or B.
thetaiotaomicron DNA and analyzed by real time PCR. The threshold cycle values of these
spiked DNA samples were compared to those of the DNA samples from distilled water spiked
with the same concentration of the target DNA of the respective assay.

4.2.4. Statistical analysis
Cultivation (E. coli and enterococci) and molecular (E. coli, Enterococci spp., and B.
thetaiotaomicron) results were compared by space, time, and detection technique to identify
differences and associations using independent t-tests, Kruskal-Wallis one-way ANOVA, and
Spearman rank correlation coefficient tests. Classification And Regression Tree (CART) analysis
was also used to compare E. coli (molecular), Enterococci spp. (molecular), B. thetaiotaomicron
(molecular), E. coli (cultivation) and enterococci (cultivation) results following Martin et al.
189

(2011) and Wilkes et al. (2011). CART is a trial and error method that attempts to split
dependent variables into homogeneous categories based on independent variables that influence
the dependent variable (target organism). All CART analyses were performed using R software
system (R foundation for Statistical Computing). CART has been previously used to investigate
pathogenic bacteria and parasite relationships with environmental and land use factors (Wilkes et
al. 2011) and to predict the occurrence of fecal indicator bacteria with respect to physiochemical
variables (Bae et al. 2010).

CART models start out with a parent or root node which contains all available data. CART then
looks at all independent variables (splitting variables) and selects the single variable that
produces the two most different groups of dependent variables based on predefined splitting
criterion and regression analysis. In this study, splitting criteria were developed using recursive
partitioning algorithm and a 10-fold cross validation. A 10-fold cross validation breaks all data
into 10 subsets and calculates the split based on nine of the ten subsets. Each time a group is split
per above criteria the binary splits are called child nodes. This method is used for each group
until a stopping rule is reached. For this project, the stopping criterion was set at a minimum of 5
observations per subgroup (Martin et al. 2011). A terminal node is defined as a child node which
has met the defined stopping rule.

Fully grown trees often require pruning to ensure significant variable associations are not missed
as a result of following the splitting and stopping criteria (Lemon et al. 2003). Pruning is the
process of growing trees until they reach stopping criteria and then removing less statistically
significant results from the analysis. Trees were pruned according to the one-standard error rule

190

(Breiman et al. 1984; Venables and Ripley 1999; De’ath and Fabricous 2000). This rule
minimizes the cross-validated error of the model which has been shown to produce optimal sized
trees and produce more stable tree sizes across replications compared to the 0-SE pruning
method and (Breiman et al. 1984; Questier et al. 2005).

Competitor and surrogate variables were identified for each node by investigating detailed
CART outputs. Competitor splits are those variables that have similar complexity parameter
values compared to the primary split. A complexity parameter compares the complexity (number
of terminal nodes) to the cross-validated error for each group. For this project the complexity
parameter was set at 0.05. Surrogate splits are alternative variables that split the subgroup into
very similar groups. An example of a CART output is presented below in Figure 4.2.

191

Figure 4.2. Example of a Classification And Regression Tree (CART) output. Root nodes
contain all available data and are split into binary groups using recursive partitioning algorithm
and 10-fold cross validation with a complexity parameter value of 0.05. Primary splitting
variables and values are described for each child node. Terminal nodes (bottom of the tree)
include the mean concentration and number of target organism cases in each node. Each node
was derived based on mean value of each response variable, group size, and defining variables.

Regulatory criteria outcomes were compared between samples based on USEPA suggested water
-1

-1

quality criteria for E. coli (235 MPN 100 ml ), enterococci (61 MPN 100 ml ) and Enterococci
-1

spp. (1000 cell equivalents 100 ml ). Regulatory actions were defined as any act (i.e. closure,
advisory, or warning) that would result from water quality exceeding the suggested criteria.
Ratios were developed by diving cultivation results by qPCR results and then compared.
Significant differences between regulatory outcomes were identified using independent t-tests
and Chi-square power tests. Binomial regression tests were used to calculate the probability of
increased bacteria incidences when one regulatory criterion was exceeded.

192

-1

E. coli daily geometric mean concentrations (MPN 100 ml ) measured between 2001 and 2011
(n = 189) at Traverse City State Park beach (TCSP) were collected from Michigan Beach Guard
database (http://www.deq.state.mi.us/beach/). Independent t-tests were used to compare E. coli
cultivation results from the long term dataset and this project dataset at TCSP to identify if water
quality measured during this project accurately represented historical water quality.

Sample concentrations below method detection were assigned a value equal to half of the
detection limit with respect to each method. The geometric means from all molecular assay
replicates were calculated and used for statistical analysis. Kruskal-Wallis one-way ANOVA,
independent t-tests, Chi-square power tests, and correlation coefficients were calculated using
IMB SPSS Statistics (v.19.0) or SigmaPlot (v.11.0). Significance threshold was set at an (α) of
0.05.

4.3. Results
4.3.1. Comparing culture versus molecular results
In total, 111 samples were analyzed for E. coli (cultivation and qPCR), enterococci (cultivation
and qPCR), and B. thetaiotaomicron (qPCR). Results for each assay are summarized in Table
4.1. The highest single sample and geometric mean concentrations for both cultivation assays (E.
coli and enterococci) were reported at MC5, the site with the highest percent agricultural land
use (41%) in the upstream catchment. For all molecular assays, the highest geometric mean and
single sample concentrations were detected at MC2 (570 m upstream from Creek outlet) and
TCSP, respectively. At the TCSP beach, cultivation based E. coli and enterococci means were
193

statistically lower than cultivation E. coli and enterococci averaged across all creek sites (p <
0.05) and molecular based Enterococci spp. means were statistically lower than molecular based
E. coli and B. thetaiotaomicron (p < 0.05). Amongst all data, all three molecular assay means
were statistically different from both cultivation means (p < 0.001). However wet weather
sample means of all assays were not statistically different from dry weather samples (p ≥ 0.134).

Table 4.1. Log10 transformed bacterial concentrations from a Great Lakes water system and the
respective number of regulatory exceedances.
Assay

Method

Sample
number

Range

Geometric
Mean

Regulatory
1
exceedance (n)

E. coli

Cultivation
-1
(MPN 100 ml )

111
(23)

0.30-3.38
(0.30-2.89)

2.20
(1.22)

65
(2)

Enterococci spp.

Cultivation
-1
(MPN 100 ml )

111
(23)

0.78-3.47
(0.78-2.74)

2.14
(1.22)

89
(3)

E. coli

qPCR
-1
(CE 100 ml )

111
(23)

3.02-5.72
(3.26-5.72)

2

4.23
(4.27)

NA

Enterococci spp.

qPCR
-1
(CE 100 ml )

111
(23)

1.53-4.47
(1.82-4.47)

3

2.86
(2.90)

66
(12)

B. thetaiotaomicron

qPCR
-1
(CE 100 ml )

111
(23)

1.43-6.84
(1.92-6.84)

3

3.77
(3.85)

NA

1

4

1

Values in parenthesis represent measurements from TCSP beach only. Regulator exceedances
based on USEPA suggested criteria for freshwater recreational waters resulted in advisories at
2
-1 3
TCSP but not in Creek sites. Method detection limit of 3.9 log10 copies 100 ml . Method
-1

detection limit of 2.9 log10 copies 100 ml . Non-detections were reported as half the method
4

detection limit. NA represents the absence of established health based criteria.

194

Cultivatable E. coli and enterococci concentrations followed close spatial and temporal patterns
(Figure 4.3.) which were not statistically different over time (except on August 8, 2010) or space
throughout Mitchell Creek (except at TCSP beach) (p ≥ 0.066). Similarly, the molecular assays
B. thetaiotaomicron, E. coli, and Enterococci spp. followed spatial and temporal trends with each
other (Figure 4.3.) and were not statistically different over space (p ≥ 0.276) or time (p < 0.001).
Spearman rank correlations of assay concentrations (Table S.4.1.) indicate strong, positive
relationships between (1) cultivation E. coli and cultivation enterococci (r = 0.877, p < 0.001)
and (2) all molecular assays (r = 1.000, p < 0.001).

During dry weather, moderate correlations were identified between (1) molecular Enterococci
spp. and cultivation E. coli (r < 0.69, p < 0.05) and (2) molecular Enterococci spp. and molecular
E. coli (r < -0.87, p < 0.05). During wet weather, moderate correlations were found between
molecular B. thetaiotaomicron and molecular E. coli (r > 0.87, p < 0.05), but cultivation
enterococci were inversely related to B. thetaiotaomicron, molecular E. coli, and molecular
Enterococci spp. (p < 0.05). Overall, E. coli (molecular), Enterococci spp. (molecular), and B.
thetaiotaomicron (molecular) means were not significantly different between beach and creek
sites (p ≥ 0.746), but E. coli (cultivation) and enterococci (cultivation) were significantly
different between creek sites and the beach site (p < 0.001).

195

(Log10 CE 100 ml-1)

Concentration

7
6
5
4
3

(Log10 MPN 100 ml-1)

Concentration

4
3
2
1

65

70

75

80

85

90

95 100 105 110

8/7/2010

8/7/2010

8/7/2010

8/7/2010

8/7/2010

8/8/2010

8/10/2010

8/17/2010

60

8/14/2010

55

8/7/2010

50

8/7/2010

45

8/7/2010

40

8/7/2010

35

8/7/2010

30

8/7/2010

25

7/18/2010

20

7/15/2010

15

7/12/2010

10

7/1/2010

5

6/28/2010

6/24/2010

0

8/7/2010

0

Sample
(Collection date)

Figure 4.3. A spatial and temporal depiction of molecular assays (TOP FIGURE) B. thetaiotaomicron (CIRCLE), E. coli (SQUARE),
and Enterococci spp. (TRIANGLE) and cultivation assays (BOTTOM FIGURE) E. coli (SQUARE) and enterococci (TRIANGLE)
concentrations. Lines are presented to discern between organisms (B. thetaiotaomicron-DASHED, enterococci-SOLID, E. coliDOTTED). Each collection date (n = 11) depicts results from four creek sites (MC2-RED, MC3-GREEN, MC5-BLUE, MC6-PINK)
and one beach (TCSP-BLACK). On August 7, 2010 consecutive hourly samples (n = 12) were collected from each site for a total of
60 samples. In total 111 samples were collected during this study.
196

Molecular and cultivation indicator ratios were highly variable amongst all data. The two
cultivation tests used for regulatory purposes were compared to the qPCR signal for Enterococci
spp. (which as previously mentioned could also be used for regulatory purposes). For the
complete data set, cultivation E. coli per molecular Enterococci spp. ratios averaged 0.84 and
cultivation enterococci per molecular Enterococci spp. ratios were slightly lower and averaged
0.81 (p < 0.001). Ratios were not different between wet (n = 75; E. coli MPN: Enterococci spp.
qPCR average ratio = 0.84; enterococci MPN: Enterococci spp. qPCR average ratio = 0.83) and
dry (n = 36; E. coli MPN: Enterococci spp. qPCR average ratio = 0.84; enterococci MPN:
Enterococci spp. qPCR average ratio = 0.80) weather samples (p ≥ 0.424).

The ratios for TCSP beach and creek sites specifically were significantly higher in the creek (E.
coli MPN: Enterococci spp. qPCR average ratio = 0.93; enterococci MPN: Enterococci spp.
qPCR average ratio = 0.90) than at the beach (E. coli MPN: Enterococci spp. qPCR average ratio
= 0.51; enterococci MPN: Enterococci spp. qPCR average ratio = 0.47; p < 0.001).

4.3.2. Indicators and criteria
Regulatory outcomes based on USEPA single sample maximum criteria were applied to all
-1

samples (n = 111) and identified 65 samples above the E. coli criterion (235 MPN 100 ml ), 89
-1

samples exceeded the enterococci criterion (61 MPN 100 ml ), and 66 samples exceeded the
-1

Enterococci spp. criterion (1000 CCE 100 ml ). Regulatory outcomes based on Enterococci
spp. qPCR results agreed with E. coli and enterococci cultivation regulatory outcomes in 50.5%
and 55.9% of samples, respectively. In comparison, 78.4% of samples were in regulatory

197

agreement using E. coli and enterococci cultivation based methods. Thirty-four percent of
samples exceeded all three criteria while 8.1% of samples were below all three criteria.
Comparisons of regulatory outcomes between each criterion are described in Figure S.4.1.

Based on binomial regression analysis, it was determined that cultivation based criteria (i.e. E.
coli and enterococci) exceedances were not predictive of increased molecular incidence of B.
thetaiotaomicron, E. coli, and Enterococci spp. (p ≥ 0.357). Similarly, molecular Enterococci
spp. criteria exceedances were not predictive of increased cultivation E. coli or enterococci levels
(p ≥ 0.162). Conversely, comparisons of criteria to organisms detected using the same method
(i.e. cultivation verse cultivation OR molecular verse molecular) demonstrated highly significant
2

relationships (p < 0.001; Chi ≥ 82.1; df = 1).

4.3.3. Implications for use historical data sets
Analysis aimed at identifying implications for comparing water quality datasets following a
method shift was performed on TCSP beach sample datasets. Cultivation E. coli results from
long term beach monitoring data (2001-2011, n = 189) were compared to cultivation based E.
coli results measured during this project (2010, n = 23) at the same location, as shown in Figure
-1

4.4. At TCSP beach, the long term E. coli mean (1.29 log10 MPN 100 ml ) was not statistically
-1

different (p = 0.116) from this project E. coli mean (1.12 log10 MPN 100 ml ). In comparison,
-1

molecular E. coli results measured during this project (4.26 log10 CE 100 ml ) were statistically
different than cultivation E. coli results from the long term data set (p < 0.001).

198

35
Long term E. coli (cultivation)
30

Project E. coli (cultivation)
Project E. coli (molecular)

Frequency (%)

25
20
15
10
5
0
0

1

2

3

4

5

6

Log10 E. coli (MPN or CCE 100 ml-1)
Figure 4.4. A frequency distribution of cultivation E. coli measurements from a long term beach
monitoring database (2001-2011, n = 189), cultivation E. coli measurements during project
(2010, n = 23), and molecular E. coli measurements during project (2010, n = 23) at TCSP beach
during 2 time periods. Cultivation E. coli concentrations between time periods were within the
normal expected distribution (p = 0.116) while molecular based E. coli concentrations from the
project were outside expected normal distributions of long term cultivation E. coli (p < 0.001).

Classification and Regression Tree (CART) analysis was used to further investigate relationships
between detection methods. CART results showed cultivation results mostly explained other
cultivation concentrations and molecular results mainly explained other molecular results (Figure
4.5.). Molecular results did not include any significant surrogate or primary splits associated with
cultivation variables. Similarly, cultivation results did not include any significant primary or

199

surrogate splits associated with molecular variables. To test the predictive ability of cultivation
based organisms for molecular organism concentrations, CART models targeting molecular
bacteria in the root node were developed using only cultivation based organisms as independent
variables. Likewise, to test the predictive ability of molecular based organisms for cultivated
bacteria concentrations, CART models targeting cultivated organisms in the root node were
developed using only molecular based organisms as independent variables. These reduced
models identified that cultivation results predicted no more than 6% of molecular concentrations
and molecular variables predicted less than 12% of cultivation variables (Table S.4.2.).
Cumulatively, these models suggest qPCR assays poorly predict cultivation results and vice
versa, likely being driven by the beach analyses.

200

Figure 4.5. Classification And Regression Tree (CART) analysis of (A) E. coli MPN, (B) enterococci MPN, (C) E. coli CE, (D)
Enterococci spp. CE, and (E) B. theta CE. Binary splitting of variables identified best categories according to splitting criteria. The
target organism is bolded in the top rectangle. Independent splitting variables and splitting value are presented for each branch of the
tree. Target organism means and target organism cases (in parenthesis) are described for each terminal node (bottom rectangle).

201

Figure 4.5. (cont’d)

4.4. Discussion
Using the USEPA suggested criteria for cultivation E. coli, cultivation enterococci, and
molecular based Enterococci spp., water quality in the study area was unsafe for total body
contact during the project timeframe. Molecular E. coli measurements at the beach ( X = 4.27
A

-1

E

A

-1

log10 copies 100 ml ) and across all creek sites ( X = 4.23 log10 copies 100 ml ) were greater
A

E

A

than levels reported by Lee et al. (2012) at four other Great Lakes beaches (~3.0 log10 copies
-1

100 ml ). Concentrations of the B. thetaiotaomicron marker averaged 3.77 log10 copies 100 ml
1

-

across all sites, within reported ranges of a human impacted urban creek in the US (Yampara-

Iquise et al. 2008) and similar to levels reported in tertiary treated sewage effluent (Srinivasan et
al. 2011). Together, these results represent elevated health risks in the study area stemming from
the presence of human fecal contamination.

202

Although direct correlations between E. coli MPN and Enterococci spp. CE concentrations were
identified, CART analysis demonstrated weak predictive power between assays. Specifically, no
molecular variables predicted cultivated variables and no cultivation variables predicted
molecular variables. Interestingly, Enterococci spp. (CE) concentrations were more closely
related to cultivation E. coli than cultivation enterococci amongst all sites. Biochemical based
cultivation methods targeting enterococci used during the current study favor the growth of E.
faecium and E. faecalis (USEPA 2002B), with higher concentrations of these species found in
human feces (Scott et al. 2002). In comparison, the molecular Enterococci spp. target is not
specific to humans as others have reported finding it in chickens at levels of > 4.46 log10 copies
-1

g

6

-1

wet weight (Wise and Siragusa 2007) in cattle at 10 copies g
6

2011), and in gulls at levels up to 10 CE g

-1

dry weight(Rogers et al.

wet weight. E. coli is found in virtually all warm

blooded animals and gulls (Winfield and Groisman 2003). Perhaps the strong relationships
between molecular Enterococci spp. and cultivation E. coli (i.e. two general targets) are due to
the non-point sources which dominate this watershed.

B. thetaiotaomicron α-1-6 mannanase is a human sewage specific marker (Yampara-Iquise et al.
2008) with a reported human pollution specificity of 97% (Aslan and Rose, 2012). B.
thetaiotaomicron results indicate a significant presence of human fecal contamination.
Interestingly, no point source sewage discharges are located directly in the creek watershed but
there are approximately 1600 on-site septic systems in the watershed (Luscz and Hyndman, in
prep.). These results suggest human fecal contamination is entering the Creek from faulty
infrastructure such as septic systems, sewer pipes, or illicit stormwater connections. However, B.

203

thetaiotaomicron increases were not associated with cultivation E. coli and cultivation
enterococci when they exceeded their respective criterion; suggesting the B. thetaiotaomicron
marker was not a suitable sub-criterion for cultivation based E. coli or enterococci and their
associated health risk implications. However, the B. thetaiotaomicron measured during this
project are indicative of increased human health risk in both the Mitchell Creek and at TCSP
beach as the geometric means were higher than those reported at three marine beaches ( X = 2.95,
A

E

A

-1

2.99, and 3.11 log10 CE 100 ml ) where a direct association between Bacteriodales daily
averages and gastrointestinal illnesses was demonstrated (Wade et al. 2010).

The findings of this project do not support the use of a single universal factor which describes
the mathematical relationship between cultivation and molecular results and illustrates the
difficulty of defining a generalized relationship factor on a national or global scale, as supported
by Converse et al. (2012). Whitman et al. (2010) demonstrated high uncertainty and poor
correlations between molecular and cultivation methods when microbial concentrations were
close to 1.0 log10 CFU 100 ml

-1

and associations between methods were influenced by local

factors. However, in the current study, such trends were not identified as overall cultivation E.
coli and cultivation enterococci concentrations were generally greater than 2.1 log10 MPN 100
-1

ml . At TCSP beach, E. coli and enterococci means were generally low (1.2 log10 MPN 100 ml

-

1

) and the ratio variability was high. Overall, qPCR to cultivation ratios at creek sites were closer

to one and nearly two-times higher than those found at TCSP beach. This indicates qPCR and
cultivation concentrations were more similar in creek water than beach water. These results
support those of Byappanahalli et al. (2010) who reported more agreement in qPCR to
204

cultivation ratios at a river site than to a nearby Great Lakes beach using similar microbial
methods.

It was initially theorized that the different rates of DNA persistence and viable organism survival
significantly influenced molecular and cultivation method relationships, especially when
comparing beach to creek results. Walters et al. (2009) reported enterococci CE took eight times
longer than cultivated enterococci to decline by the same order of magnitude (T90 = 8.28 days
and 1.04 days, respectively) in a microcosm experiment. Cultivated E. coli was shown to decay
slower/faster than molecular E. coli in water (T90 = 21.7 hours and T99 = 5.65 days,
respectively) (Jin et al. 2004; Liang et al. 2012). Parallel research in the study area (Chapter 3)
measured river discharge rates which were used to estimate the average transport time from the
most upstream sampling point to the creek mouth (3.2 linear km) at 6 hours. Additional analysis
from the same project indicated microbial contamination entering the Grand Traverse Bay from
the Mitchell Creek can impair the TCSP designated swim area (500 linear m) in less than 1 hour.
Comparing such temporal transport estimates to previously reported decay rates of the assayed
markers, indicates there was not sufficient time for bacterial or DNA degradation to significantly
influence results. However, the conditions required for the rapid transport of pollution from the
creek to the TCSP beach (wind and high creek discharge) were not always present during sample
collection, leaving open the theory that transport DNA persistence and viable organism survival
rates may influence molecular and cultivation relationships.

A comparison of the current project dataset with respect to regulatory criteria showed cultivation
of enterococci would have resulted in the greatest number of regulatory based actions (i.e.
205

closure, advisory, or warning), suggesting the enterococci cultivation method may offer the most
protection of public health but also potentially the result of false-positive as suggested by
Kinzelman et al. (2003). These results support findings by Kinzelman et al. (2003) who
compared cultivation E. coli and enterococci threshold levels per USEPA suggestions and found
monitoring for enterococci would have resulted in an additional 56 water quality advisories
compared to E. coli. In the current study, three-quarters of samples tested for cultivation E. coli
and enterococci were in agreement with criteria based regulatory outcomes. However, 55.1% and
45.6% of Enterococci spp. CE criteria exceedances agreed with cultivation based criteria
exceedances in the creek and at the beach, respectively. Although the total numbers of regulatory
outcomes during the project were similar between Enterococci spp. (qPCR) and E. coli (MPN),
monitoring for Enterococci spp. (qPCR) alone would have resulted in 65% fewer regulatory
actions (i.e. water quality above criteria) in the creek and 91% more regulatory actions at the
beach compared to monitoring only for cultivated E. coli.

This project focused on a small watershed dominated by non-point source pollutants, much
different than those used for criteria development (USEPA 2011). Nearly all the beaches selected
for epidemiological studies during USEPA criteria development were impacted by sewage and
the non-point source impacted beach was found to have poor illness to indicator correlations
(USEPA 2010). The current study identified direct associations between molecular Enterococci
spp. and cultivation E. coli assays, supporting previous research by Lavender and Kinzelman
(2009), Whitman et al. (2010), and Converse et al. (2012). Furthermore, the total number of
individual regulatory exceedances was similar for cultivation E. coli and molecular Enterococci
spp. but the pairwise agreement between was much lower when beaches and creek sites were

206

separated, implying these two criterions and methods do not provide similar levels of human
health protection across all water types. It is recommended that further studies focus on
molecular methods which develop a regulatory target as well as persistence studies that address
the transport and fate of molecular markers from rivers to beaches.

207

APPENDIX

208

Table S.4.1. Spearman’s rank correlation matrix among microorganism detection methods.
E. coli
-1
(MPN 100 ml )
E. coli
-1
(MPN 100 ml )
Enterococcus
-1
(MPN 100 ml )
E. coli
-1
(CE 100 ml )
Enterococcus spp.
-1
(CE 100 ml )
B. thetaiotaomicron
-1
(CE 100 ml )

Enterococcus
-1
(MPN 100 ml )

E. coli
-1
(CE 100 ml )

Enterococcus
-1
(CE 100 ml )

r coefficient
p Value
Sample n
r coefficient
p Value
Sample n

0.877
< 0.001
111

r coefficient

-0.100

-0.187

p Value

0.298

0.049

111

111

r coefficient

-0.099

-0.188

1.000

p Value

0.299

0.048

< 0.001

111

111

111

r coefficient

-0.101

-0.189

1.000

1.000

p Value

0.290

0.047

< 0.001

< 0.001

111

111

111

111

Sample n

Sample n

Sample n

209

B. thetaiotaomicron
-1
(CE 100 ml )

Figure S.4.1. Analysis of regulatory based outcomes occurring during entire project according to USEPA suggested criteria. Values
presented represent number of cases (total n = 111).

210

Table S.4.2. CART results describing the ability of different methods to predict microorganism concentrations. Model results include
model target assay and interactions between assays measured using the different method (i.e. molecular (target) associations with
cultivation only and vice versa). Refer to methods and Figure 4.2. for interpretation of CART analysis.
Model target assay

Root node

Complexity
2
parameter (R )

Child node

Complexity
2
parameter (R )

Total predictive
2
Value (R )

B. thetaiotaomicron
-1
(CE 100 ml )

Enterococcus
-1
(MPN 100 ml )

0.06

-

-

0.06

E. coli
-1
(CE 100 ml )

Enterococcus
-1
(MPN 100 ml )

0.06

-

-

0.06

Enterococcus spp.
-1
(CE 100 ml )

Enterococcus
-1
(MPN 100 ml )

0.06

-

-

0.06

E. coli
-1
(MPN 100 ml )

B. thetaiotaomicron
-1
(CE 100 ml )

0.05

B. thetaiotaomicron
-1
(CE 100 ml )

0.05

0.10

Enterococcus
-1
(MPN 100 ml )

B. thetaiotaomicron
-1
(CE 100 ml )

0.07

B. thetaiotaomicron
-1
(CE 100 ml )

0.05

0.12

211

REFERENCES

212

REFERENCES

Aslan, A. and Rose, J. (2012). Evaluation of the host specificity of Bacteroides thetaiotaomicron
alpha mannanese gene as a sewage pollution marker. Letters in Applied Microbiology, submitted
for publication
Bae, H. K., Olson, B., Hsu, K.-L., and Sorooshian, S. (2010). Classification And Regression Tree
(CART) analysis for indicator bacterial concentration prediction for a Californian coastal area.
Water Science and Technology, 61, 545-553.
Breiman, L., Friedman, J., Olshen, R., and Stone, C. (1984). Classification and Regression Trees.
Chapman and Hall, New York.
Byappanahalli, M., Whitman, R., Shively, D., and Nevers, M. (2010). Linking non-culturable
(qPCR) and culturable enterococci densities with hydrometeorological conditions. The Science of
the Total Environment, 408, 3096-3101.
Cabelli, V., Dufour, A., McCabe, L., and Levin, M. (1982). Swimming-associated gastroenteritis
and water quality. American Journal of Epidemiology, 115, 606-616.
Cabelli, V. J. Health effects for marine recreation waters; USEPA 600/1-80-031; Health Effects
Research Laboratory: Research Triangle Park, NC, 1983.
Colford, J., Wade, T., Schiff, K., Wright, C., Griffith, J., Sandhu, S., Burns, S., Sobsey, M.,
Lovelace, G., and Weisberg, S. (2007). Water quality indicators and the risk of illness at beaches
with nonpoint sources of fecal contamination. Epidemiology, 18, 27-35.
Colford, J., Schiff, K., Griffith, J., Yau, V., Arnold, B., Wright, C., Gruber, J., Wade, T., Burns,
S., Hayes, J., McGee, C., Gold, M., Cao, Y., Noble, R., Haugland, R., and Weisberg, S. (2012).
Using rapid indicators for enterococcus to assess the risk of illness after exposure to urban runoff
contaminated marine water. Water Research, 46, 2176-2186.
Converse, R., Griffith, J., Noble, R., Haugland, R., Schiff, K., and Weisberg, S. (2012).
Correlation between quantitative PCR and culture-based methods for measuring Enterococcus
spp. over various temporal scales at three California marine beaches. Applied and Environmental
Microbiology, 78, 1237-1242.
De’ath, G., and Fabricius, K. (2000). Classification and Regression Trees: A powerful yet simple
technique for ecological data analysis. Ecology, 81, 3178-3192.

213

Dick, L.K. and Field, K.G. (2004). Rapid estimation of numbers of fecal Bacteroidetes by use of
a quantitative PCR assay for 16S rRNA genes. Applied and Environmental Microbiology, 70,
5695-5697.
European Commission (EC). (2006). Directive 2006/7/EC of the European Parliament and of the
Council of 15 February 2006 concerning the management of bathing water quality and repealing
Directive 76/160/EEC.
Field, K.G. and Samadpour, M. (2007). Fecal source tracking, the indicator paradigm, and
managing water quality. Water Research, 41, 3517-3538.
Frahm, E., and Obst, U. (2003). Application of the fluorogenic probe technique (TaqMan PCR)
to the detection of Enterococcus spp. and Escherichia coli in water samples. Journal of
Microbiological Methods, 52, 123-131.
Girones, R., Ferrús, M.A., Alonso, J.L., Rodriguez-Manzano, J., Calgua, B., Corrêa, A.D.A.,
Hundesa, A., Carratala, A., and Bofill-Mas, S. (2010). Molecular detection of pathogens in
water--the pros and cons of molecular techniques. Water Research, 44, 4325-4339.
Haack, S.K., Duris, J.W., Fogarty, L.R., Kolpin, D.W., Focazio, M.J., Furlong, E.T., and Meyer,
M.T. (2009). Comparing wastewater chemicals, indicator bacteria concentrations, and bacterial
pathogen genes as fecal pollution indicators. Journal of Environmental Quality, 38, 248-258.
Harwood, V.J., Levine, A.D., Scott, T.M., Chivukula V., Lukasik, J., Farrah, S.R., Rose, J.B.
(2005). Validity of the indicator organism paradigm for pathogen reduction in reclaimed water
and public health protection. Applied and Environmental Microbiology, 71, 3163-3170.
Haugland, R., Siefring, S., Wymer, L., Brenner, K., and Dufour, A. (2005). Comparison of
Enterococcus measurements in freshwater at two recreational beaches by quantitative
polymerase chain reaction and membrane filter culture analysis. Water Research, 39, 559-68.
Hellein, K.N., Battie, C., Tauchman, E., Lund, D., Oyarzabal, O.A., and Lepo, J.E. (2011).
Culture-based indicators of fecal contamination and molecular microbial indicators rarely
correlate with Campylobacter spp. in recreational waters. Journal of Water Health, 9, 695-707.
Jin, G., Jeng, H.W., Bradford, H., and Englande, A.J. (2004). Comparison of E. coli, enterococci,
and fecal coliform as indicators for brackish water quality assessment. Water Environment
Research, 76, 245-255.
Kinzelman, J., Ng, C., Jackson, E., Gradus, S., and Bagley, R. (2003). Enterococci as indicators
of Lake Michigan recreational water quality: Comparison of two methodologies and their
impacts on public health regulatory events. Applied and Environmental Microbiology, 69, 92-96.
Lavender, J., and Kinzelman, J. (2009). A cross comparison of qPCR to agar-based or defined
substrate test methods for the determination of Escherichia coli and enterococci in municipal
water quality monitoring programs. Water Research, 43, 4967-4979.
214

Lee, C., Agidi, S., Marion, J. W., and Lee, J. (2012) Arcobacter in Lake Erie beach waters: An
emerging gastrointestinal pathogen linked with human-associated fecal contamination. Applied
and Environmental Microbiology, 78, 5511-5519.
Lemon, S.C., Roy, J., Clark, M.A., Friedmann, P.D., and Rakowski, W. (2003). Classification
and Regression Tree Analysis in public health: methodological review and comparison with
logistic regression. The Society of Behavioral Medicine, 26, 172-181.
Liang, Z., He, Z., Zhou, X., Powell, C.A., Yang, Y., Roberts, M. G., and Stoffella, P.J. (2012).
High diversity and differential persistence of fecal Bacteroidales population spiked into
freshwater microcosm. Water Research, 46, 247-257.
Luscz, E. and Hyndman, D. (in preparation). Modeling Nutrient Loading to Watersheds in the
Great Lakes Basin: A Detailed Source Model at the Regional Scale.
Martin, S., Soranno, P., Bremigan, M., and Cheruvelil, K. (2011). Comparing hydrogeomorphic
approaches to lake classification. Environmental Management, 48, 957-974.
Questier, F., Put, R., Coomans, D., Walczak, B., and Heyden, Y.V. (2005). The use of CART
and multivariate regression trees for supervised and unsupervised feature selection.
Chemometrics and Intelligent Laboratory Systems, 76, 45-54.
Rabinovici, S., Bernknopf, R.L., and Wein, A.M. (2004). Economic and health risk trade-offs of
swim closures at a Lake Michigan beach. Environmental Science and Technology, 38, 27372745.
Santo Domingo, J.W., Bambic, D.G., Edge, T.A., and Wuertz, S. (2007). Quo vadis source
tracking? Towards a strategic framework for environmental monitoring of fecal pollution. Water
Research, 41, 3539-3552.
Schriewer, A., Miller, W., Byrne, B., Miller, M., Oates, S., Conrad, P., Hardin, D., Yang, H.,
Chouicha, N., Melli, A., Jessup, D., Dominik, C., and Wuertz, S. (2010). Presence of
Bacteroidales as a predictor of pathogens in surface waters of the central California coast.
Applied and Environmental Microbiology, 76, 5802-5814.
Scott, T. M., Rose, J. B., Jenkins, T. M., Farrah, S. R., and Lukasik, J. (2002). Microbial Source
Tracking: Current methodology and future directions. Applied and Environmental Microbiology,
68, 5796-5803.
Siefring, S.C., Varma, M., Atikovic, E., Wymer, L., and Haugland, R.A. (2008). Improved realtime PCR assays for the detection of fecal indicator bacteria in surface waters with different
instrument and reagent systems. Journal of Water and Health, 6, 225-237.
Shibata, T., Solo-Gabriele, H., Sinigalliano, C., Gidley, M., Plano, L., Fleisher, J., Wang, J.,
Elmir, S., He, G., Wright, M., Abdelzaher, A., Ortega, C., Wanless, D., Garza, A., Kish, J., Scott,
215

T., Hollenbeck, J., Backer, L., and Fleming, L. (2010). Evaluation of conventional and
alternative monitoring methods for a recreational marine beach with nonpoint source of fecal
contamination. Environmental Science and Technology, 44, 8175-8181.
Srinivasan, S., Aslan, A., Xagoraraki, I., Alocilja, E., and Rose, J. (2011). Escherichia coli,
enterococci, and Bacteroides thetaiotaomicron qPCR signals through wastewater and septage
treatment. Water Research, 45, 2561-2572.
Stapleton, C.M., Kay, D., Wyer, M.D., Davies, C., Watkins, J., Kay, C., Mcdonald, A., Porter, J.,
and Gawler, A. (2009). Evaluating the operational utility of a Bacteroidales quantitative PCRbased MST approach in determining the source of faecal indicator organisms at a UK bathing
water. Water Research, 43, 4888-4899.
United States Environmental Protection Agency (USEPA). (2002A). Method 1603: Escherichia
coli (E. coli) in Water by membrane filtration using modified membrane-thermotolerant
Escherichia coli agar (Modified mTEC). EPA 821-R-02-023.
United States Environmental Protection Agency (USEPA), O. of W. (2002B). Method 1600:
Enterococci in water by membrane filtration using membrane-Enterococcus indoxyl-B-DGlucoside agar (mEI). (EPA 821-R-02-022).
United States Environmental Protection Agency (USEPA). (2010). Report on 2009 National
Epidemiologic and Environmental Assessment of Recreational Water Epidemiology Studies
(NEEAR 2010 - Surfside & Boquerón). EPA-600-R-10-168. Available at:
http://www.epa.gov/neear/files/Report2009v5_508comp.pdf
United States Environmental Protection Agency (USEPA). (2011). Recreational Water Quality
Criteria Draft Report No: 820-D-11-002.
Venables, W. and Ripley, B. (1999). Modern applied statistics with S-PLUS, 3rd edn. Springer,
New York.
Viau, E.J. and Peccia, J. (2009). Evaluation of the enterococci indicator in biosolids using
culture-based and quantitative PCR assays. Water Research, 43, 4878-4887.
Wade, T.J., Calderon, R.L., Sams, E., Beach, M., Brenner, K.P., Williams, A.H., and Dufour, A.
(2006). Rapidly measured indicators of recreational water quality are predictive of swimmingassociated gastrointestinal illness. Environmental Health Perspectives, 114, 24-28.
Wade, T., Calderon, R., Brenner, K., Sams, E., Beach, M., Haugland, R., Wymer, L., and
Dufour, A. (2008). High sensitivity of children to swimming-associated gastrointestinal illness:
Results using a rapid assay of recreational water quality. Epidemiology, 19, 375-383.
Wade, T.J., Pai, N., Eisenberg, J.N.S., and Colford, J.M. (2003). Do U.S. Environmental
Protection Agency water quality guidelines for recreational waters prevent gastrointestinal

216

illness? A systematic review and meta-analysis. Environmental Health Perspectives, 111, 11021109.
Wade, T., Sams, E., Brenner, K., Haugland, R., Chern, E., Beach, M., Wymer, L., Rankin, C.,
Love, D., Li, Q., Noble, R., and Dufour, A. (2010). Rapidly measured indicators of recreational
water quality and swimming-associated illness at marine beaches: A prospective cohort study.
Environmental Health, 9, 66.
Walters, S.P., Yamahara, K.M., and Boehm, A.B. (2009). Persistence of nucleic acid markers of
health-relevant organisms in seawater microcosms: Implications for their use in assessing risk in
recreational waters. Water Research, 43, 4929-4939.
Whitman, R., Ge, Z., Nevers, M., Boehm, A., Eunice, C., Haugland, R., Lukasik, A., Molina, M.,
Przybyla-Kelly, K., Shively, D., White, E., Zepp, R., and Byappanahalli, M. (2010). Relationship
and variation of qPCR and culturable enterococci estimates in ambient surface waters are
predictable. Environmental Science and Technology, 44, 5049-5054.
Wilkes, G., Edge, T., Gannon, V., Jokinen, C., Lyautey, E., Medeiros, D., Neumann, N.,
Ruecker, N., Topp, E., and Lapen, D. (2009). Seasonal relationships among indicator bacteria,
pathogenic bacteria, Cryptosporidium oocysts, Giardia cysts, and hydrological indices for
surface waters within an agricultural landscape. Water Research, 43, 2209-2223.
Wilkes, G., Edge, T., Gannon, V., Jokinen, C., Lyautey, E., Neumann, N., Ruecker, N., Scott, A.,
Sunohara, M., Topp, E., and Lapen, D. (2011). Associations among pathogenic bacteria,
parasites, and environmental and land use factors in multiple mixed-use watersheds. Water
Research, 45, 5807-5825.
Winfield, M.D. and Groisman, E.A. (2003). Role of nonhost environments in the lifestyles of
Salmonella and Escherichia coli. Applied and Environmental Microbiology, 69, 3687-3694.
Wise, M.G. and Siragusa, G.R. (2007). Quantitative analysis of the intestinal bacterial
community in one- to three-week-old commercially reared broiler chickens fed conventional or
antibiotic-free vegetable-based diets. Journal of Applied Microbiology, 102, 1138-49.
World Health Organization (WHO). (2003). Guidelines for safe recreational water environments.
Volume 1, Coastal and fresh waters. 253 pp.
Yampara-Iquise, H., Zheng, G., Jones, J., and Carson, C. (2008). Use of a Bacteroides
thetaiotaomicron-specific alpha-1-6, mannanase quantitative PCR to detect human faecal
pollution in water. Journal of Applied Microbiology, 105, 1686-1693.

217

CHAPTER 5.
MICROBIAL RESPONSES TO LAND, PHYSICAL, CHEMICAL, ENVIRONMENTAL,
AND HYDROLOGICAL FACTORS

218

5.1. Introduction
Natural landscapes are defined using geological properties including topography and soil
composition (Saatchi et al. 2009). However, human impacts modify natural land use/cover
resulting in permanent changes of hydrogeological cycles (Breuer et al. 2009; Vörösmarty and
Sahagian 2000). Specifically, decreased precipitation infiltration (i.e. increased surface runoff)
was associated with deforestation (Germer et al. 2009) and increased impervious surface area
(Arnold and Gibbons 1996). Overland flows concentrate pollutants and rapidly transport them
down gradient where they eventually enter larger systems and become magnified in aquatic
environments, impacting water quality (Falkenmark 2011; Evers et al. 2011). The greatest
potential for pollution transport across the landscape is the physical movement of water itself
(Falkenmark 2011). A number of models have been developed to calculate overland and surface
water flows (Katz et al. 1995; Ray et al. 2010) and nutrient/chemical transport (Cha et al. 2010),
but few studies have focused on microbial movement from land to water, particularly nontraditional coliform bacteria.

Larger waterbodies receive multiple inputs (rivers, point sources, and non-point sources) are
often at highest risk of long term and chronic water degradation. Microbial loading and
deposition posed the greatest risks near recreational areas, drinking water intakes, and
fishing/shellfish harvesting zones where human exposure was highest (Kistemann et al. 2002;
Wong et al. 2009; Almeida and Soares 2012). These highly visible areas receive more attention
than the actual source(s) since identifying the origin of all pollution in complex watersheds
would require extensive time and intellectual investments. For instance, Soranno et al. (2011)
suggested addressing water quality concerns in diverse watersheds would require comprehensive

219

investigations at multiple temporal and spatial scales during various environmental and
hydrological conditions, coupled with predictive models. Grayson et al. (1997) suggested another
technique referred to as a ‘snapshot’ approach. This method captures water quality
characteristics at a single point in time or condition across broad spatial areas, while providing
information regularly missed during routine monitoring. Compared to long term comprehensive
investigations, the snapshot approach reduces the number of samples, cost, and personnel
resources required to address pollution sources.

Escherichia coli (E. coli) are commonly used to describe relative risk during routine monitoring
in lieu of pathogen detection. However, E. coli is not source specific and tracing pollution in
water bodies to a specific land use has proven difficult. These types of studies have rarely
produced definitive conclusions because bacteria respond rapidity to flows, are not specific to
one source, and have finite lives but can regrow under certain environmental conditions (Vega et
al. 1998; Alm et al. 2006). Using molecular approaches (DNA detection via qPCR), specific
source targets can be isolated in complex systems and have recently been used to investigate land
use and water quality impairments (Peed et al. 2011). Furtula et al. (2012) demonstrated
ruminant, pig, and dog fecal contamination in an agriculturally dominated river (Canada) using
multiple Bacteroides markers. More specifically, the molecular based marker Bacteroides
thetaiotaomicron α-1-6 mannanase (B. thetaiotaomicron) gene was shown to be highly specific
to human fecal material (Yampara-Iquise et al. 2008; Srinivasan et al. 2011), but no studies have
used this particular marker to link water quality to specific land use patterns.

220

Reference conditions are defined as the condition that exists under minimal disturbance and
representative of a group with similar physical, chemical, and biological characteristics
(Reynoldson et al. 1997). Reference conditions have been used to assess aquatic resources and
habitat expectations by measuring the presence of macroinvertebrates, fish, and diatoms
(Reynoldson et al. 1997; Davies and Jackson 2006; Carlisle et al. 2008). However, microbial
reference conditions have not been adequately explored or defined. Tiefenthaler et al. (2009)
suggested microbial reference conditions based on 15 low-impaired California streams with
-1

geometric mean concentrations of cultivatable E. coli (1.0 log10 MPN 100 ml ) and enterococci
-1

(1.2 log10 MPN 100 ml ) below State water quality thresholds and were considered a low
potential health risk. In the Great Lakes, no E. coli reference conditions have been described,
however a health threshold has been suggested by the USEPA at a geometric mean of 2.10 log10
-1

E. coli MPN 100 ml for recreating in freshwater. Ideally, in truly pristine water there would be
no B. thetaiotaomicron as it is a highly human specific molecular marker (Yampara-Iquise et al.
2008), but the marker can be detected in sewage effluent following complete treatment and
disinfection (Srinivasan et al. 2011). Microbial reference conditions could help fine-tune general
water quality criteria for specific watersheds, water quality changes over time associated with
degradation, and assessing management actions and goals.

In response to increased water degradation from human stressors and the lack of microbial
reference conditions in the Great Lakes basin, this paper aims to 1) Examine the occurrence of
Escherichia coli (E. coli) and a human specific source marker (B. thetaiotaomicron) in river
systems under baseflow conditions; 2) identify specific land uses that modify reference levels of
221

fecal contamination in rivers; and 3) determine key chemical, physical, environmental, and
hydrological variables driving water quality of rivers draining to the Great Lakes. To address
these objectives, land use, hydrological, geological, physical, chemical, biological, and microbial
variables measured from spatially independent rivers under baseflow conditions were coupled
with Classification And Regression Tree analysis. This statistical approach was investigated to
better understand relationships in a variety of watersheds with the hope to eventually support
water and landscape decision making and reduce human health risk from pathogen exposure.

5.2. Materials and methods
5.2.1. Study area
This study involved Lower Peninsula (Michigan, USA) river watersheds draining to the Great
Lakes. Watersheds were selected using the following criteria: 1) 30 large watersheds were
deemed essential as they represented 80% of land cover of the Lower Peninsula; and 2) of the
remaining smaller watersheds, 70 were randomly selected around the state. Smaller watersheds
were further filtered in the field according to timing and personnel logistics. In total, 64 river
systems were selected and represented 84% of Michigan’s Lower Peninsula drainage area
(Figure 5.1.). All sampling locations were located at bridge crossings and selected on the criteria
that each was reasonably accessible, had adequate flow which could be measured using an
Acoustic Doppler Current Profiler, river water dominated discharge, and the maximum amount
of upstream land use was captured while meeting the above criteria.

222

N
E

W
S

Figure 5.1. Sampled river systems and catchment areas in Michigan (USA) and NLCD 2006
land use in Michigan.

5.2.2. Water sample collection
Grab samples were collected from each river between October 1 and October 13, 2010 during
baseflow hydrologic stage conditions. Baseflow was determined using gage data and reasonable
assumptions gained from team member’s prior field experiences in Michigan and the Great
Lakes. Water temperature (°C), specific conductance (microsiemens per centimeter), and
-1

dissolved oxygen (mg l ) were measured on site using YSI 600R Sonde (YSI Incorporated,

223

United States). Field samples were placed on ice in coolers and transported to Michigan State
University for analysis at respective laboratories. The same methods were used for samples
collected during spring thaw (March 4-23, 2011) and after summer rainfall (June 1-28, 2011).

5.2.3. Water analysis
Each sample was assayed for water chemistry as summarized in Table 5.1. Microbial analyses
were performed within 24 hours of collection, whereas DNA analysis was performed at a later
date on frozen sample concentrate (described below). All samples were tested for E. coli using
IDEXX Colilert® Quanti-Tray 2000®. Following incubation at 35°C (± 0.5 °C) for 24 hour (± 2
hour), yellow and fluorescent wells were reported positive for E. coli, compared to a most
-1

probable number (MPN) table, and reported as MPN 100 ml . Escherichia coli C-3000 (ATCC
15597) were used as positive controls for verification of media integrity. Sterile water was used
for negative controls to verify method integrity. E. coli measurements below method detection
-1

capabilities (1.0 MPN 100 ml ) were assigned a value equal to the detection limit.

Samples were analyzed for the human specific marker B. thetaiotaomicron α-1-6 mannanase
(5’CATCGTTCGTCAGCAGTAACA3’; 5’CCAAGAAAAAGGGACAGTGG3’) according to
Yampara-Iquise et al. (2008). Analysis was performed by filtering 900 ml of water through a
0.45 µm hydrophilic mixed cellulose esters filter. Each filter was placed into a 50 ml centrifuge
tube containing 20 ml of sterile Phosphate Buffered Water, vortexed, and centrifuged (30
minutes; 4000 x g; 21 °C). Eighteen ml were decanted from the tube and the remaining eluent
and pellet were stored at -80 °C. DNA was extracted from 200 µl of the thawed pellet via
QIAamp® DNA mini kit protocol. Quantitative polymerase chain reaction (qPCR) for B.
224

thetaiotaomicron was performed following Yampara-Iquise et al. (2008) with a probe
modification (Srinivasan et al. 2011) using a Roche Light-Cycler® 2.0 Instrument (Roche
Applied Sciences). Each B. thetaiotaomicron assay was carried out with 10 µl of LightCycler
480 Probe Mastermix (Roche Applied Sciences), 0.4 µl forward and reverse primers, 0.2 µl
probe #62 (6FAM-ACCTGCTG-NFQ; Roche Applied Sciences Universal Probe Library), 1.0 µl
Bovine Serum Albumin, 3.0 µl nuclease free water, and 5.0 µl of extracted DNA and processed
in triplicates. The qPCR analyses included a 15 minute, 95 °C pre-incubation cycle, followed by
50 amplification cycles, and a 0.5 minute 40 °C cooling cycle. A diluted plasmid standard was
included during each qPCR run as a positive control and molecular grade water was used in
place of DNA template for negative controls. B. thetaiotaomicron gene copies were converted to
-1

cell equivalents (CE) and reported as qPCR CE 100 ml . For B. thetaiotaomicron, one copy of
the target gene is present in each cell, thus one copy number corresponds to one cell.

E. coli and B. thetaiotaomicron results were reported as concentrations instead of loads to be
consistent with USEPA’s total maximum daily load (TMDL) recommendations. The USEPA
-1

suggests a recreational water quality threshold for E. coli of 2.37 log10 MPN 100 ml , above
which full body submersion is not recommended.

225

Table 5.1. Summary of chemical and nutrient methods with respective references.
Assay

Units

Ammonia

µg l

Calcium

mg l

Chlorine (Cl-)

mg l

Magnesium

mg l

Nitrate/nitrite

µg l

Pheophytin corrected
chlorophyll a

µg l

-1

Phenate method

-1

Flame atomic absorption
spectrophotometry
Dionex membrane-suppression
ion chromatography
Flame atomic absorption
spectrophotometry

-1
-1

-1

-1

pH
Potassium

Total chlorophyll a

µg l

Hamilton et al. 2009

Standard Methods 4500P.E.*

-1

Dionex membrane- suppression
ion chromatography
Second derivative spectroscopy
following persulfate digestion
Ascorbic acid method following
persulfate digestion
Second derivative spectroscopy
following persulfate digestion
Ascorbic acid method following
persulfate digestion
Fluorometry following ethanol
extraction

µg l

µg l

Standard Methods
10200.H*

Ascorbic acid method

Sulfate (SO4)

Total phosphorus

Standard Methods 4500NO3-E*

-1

µg l

µg l

Wetzel and Likens 2000

Hamilton et al. 2009

Soluble reactive
phosphorus

Total nitrogen

Wetzel and Likens 2000;
Hamilton et al. 2009

Flame atomic absorption
spectrophotometry (0.5% HNO3
preservative)

mg l

µg l

Fluorometry with
pheophytincorrection following
ethanol extraction
Hydrolab multisonde

Wetzel and Likens 2000

Hamilton et al. 2009

mg l

µg l

Cadmium reduction

Reference
Standard Methods 4500NH3-G*

Flame atomic absorption
spectrophotometry (0.5% HNO3
preservative)

Sodium

Total dissolved
nitrogen
Total dissolved
phosphorus

Method description

-1

-1

-1
-1
-1
-1
-1

*APHA (1999)

226

Hamilton et al. 2009
Crumpton et al. 1992
Standard Methods 4500P.E and 4500-N.C*
Crumpton et al. 1992
Standard Methods 4500P.E and 4500-N.C*
Standard Methods
10200.H*

5.2.4. Hydrometry
Hourly precipitation data were extracted from Next Generation Radar (NEXRAD) through the
National Climate Data Center (http://www.ncdc.noaa.gov/nexradinv/). The radar stations were
located in Grand Rapids, Gaylord, and Detroit (Michigan) and had a base reflectivity of 0.50
2

degree with an elevation range of 124 nautical miles and 16 km cells. Hourly precipitation
averages across each watershed were used to calculate total rainfall with weighted averages
applied to watersheds partially contained in NEXRAD cells. Precipitation was categorized into
cumulative hourly totals (mm) prior to sample collection and reported as mm per time prior to
sample collection.

Real-time river velocity was measured at each site during sample collection using an Acoustic
Doppler Current Profiler (ADCP), USGS stream gauges, or current-meter via wading following
USGS protocol (Jarrett 1991). River discharge was calculated from flow velocities and reported
3 -1

as m s .

5.2.5. Land use
Land use, watershed delineation, and septic system estimates were defined using ESRA ArcMap
GIS software. The spatial analyst watershed tool was used to develop surface watersheds for
each sampling point at 1/3 Arc-Second resolution contour lines on a GCS North American
coordinate system. Two watersheds were defined for each river and referred to as 1) full
watersheds which included the entire upstream drainage area and 2) reduced watersheds which
included drainage boundaries upstream of the sampling site to the nearest lakes, reservoirs, and
ponds, referred to as ‘lakes’ from here on. A 60 m riparian buffer (referred to as ‘buffer’ from
227

here on) was applied to each watershed. A digital map of land cover from Landsat imagery at 30
meter resolution and the National Land Cover Database (2006) was used to define land use of
each watershed and buffer. Land use was categorized using NLCD classification system with 16
categories and further refined to seven categories using Anderson Land Cover Classification
System Level 1 (Anderson et al. 1976). Table 5.2. describes the Anderson classifications and
equivalent NLCD categories.

Table 5.2. Anderson level 1 land use classifications and descriptions.
Classification Description

Examples

Urban

Intensive use with
structures covering
the majority of land

Cities, shopping,
industrial, and
commercial
centers

Agricultural

Land used for food
production

Pasture and hay (81)
Cultivated crops (82)

Open

Predominant
natural vegetation
is grass or shrubs
Closed canopy at
least 10% from
timber quality trees
Area predominantly
cover by water
throughout year
Land with water
table near land
surface for
significant portion
of year
Land that has less
than one-third
vegetative cover.

Pasture, row crop,
orchards, confined
feeding operations
Herbaceous, shrub,
brush
Deciduous,
coniferous, and
mixed forested
Streams, lakes,
bays, and
reservoirs
Marshes, swamps,
perched bogs

Deciduous forest (41)
Evergreen forest (42)
Mixed forest (43)
Water (11)

Beaches, exposed
rock, gravel pits

Barren (31)

Forest

Water

Wetland

Barren

228

Associated NLCD classifications
(Code)
Developed open space (21)
Developed low intensity (22)
Developed Medium intensity (23)
Developed high intensity (24)

Shrub and scrub (52)
Grassland and herbaceous (71)

Woody wetland (90)
Emergent herbaceous wetland (95)

Using GIS programs, household locations from the 2010 US Census data were compared to
current municipal sewage treatment infrastructure locations to produce a list and location of
households that likely utilize on-site septic systems to treat wastewater (Luscz and Hyndman, in
2

prep.). Estimated septic system numbers (per watershed) and densities (per km ) in catchment
and 60m buffered areas were calculated for each of the 64 river systems.

The National Pollutant Discharge Elimination System (NPDES) from the Environmental
Protection Agency's Enforcement and Compliance History was used to estimate ammonia and
-1

total phosphorus loads (kg year ) from point source effluents discharging upstream of the
sampling point for each river (Luscz and Hyndman, in prep.).

5.2.6. Statistical analysis
A constant value of one was added to E. coli and B. thetaiotaomicron concentrations prior to log
transformation and statistical analysis. Soil hydraulic conductivity underwent natural log
transformations prior to statistical analyses. Spearman Correlation tests were used to examine
relationships amongst physical, chemical, weather, river discharge, land use, estimated pollution
discharges and microbial measurements. Significance threshold was set at (α) 0.05. Descriptive
statistics were performed using IBM SPSS Statistics software (Version 19.0).

Classification And Regression Tree (CART) analysis was used to compare E. coli and B.
thetaiotaomicron results to chemical, hydrological, physical, environmental, and land use
variables. Three E. coli and B. thetaiotaomicron model scenarios were created: 1) only land use

229

variables; 2) only nutrient, chemical, hydrological, precipitation, and physical variables; and 3)
all variables combined.

CART attempts to split dependent variables into homogeneous categories based upon the
influence of independent variables on the dependent variable (target organism). CART was
previously used to investigate pathogenic bacteria and parasite relationships with environmental
and land use factors (Wilkes et al. 2011), to classify lakes based on chemistry and clarity (Martin
et al. 2011), and to predict the occurrence of fecal indicator bacteria with respect to
physiochemical variables (Bae et al. 2010).

CART models started out with a parent or root node containing all available data. Then all
independent variables were examined and the variable that produced the two most different
2

groups of dependent variables, using regression analysis based on R and pre-defined splitting
criteria were selected (splitting variables). In this study, splitting criteria were developed using
recursive partitioning algorithm and a 10-fold cross validation. A 10-fold cross validation broke
all data into 10 subsets and calculated the split based on nine of the ten subsets. Each time a
group split (per above criteria) the binary splits were called child nodes. This method was used
for each group until reaching a stopping rule set at a minimum of five observations per subgroup
(Martin et al. 2011). A terminal node was defined as a child node which met the defined stopping
rule.

Fully grown trees often required pruning to ensure significant variable associations were not
missed as a result of following the splitting and stopping criteria (Lemon et al. 2003). Pruning is
230

the process of producing the trees until they reach the stopping criterion and then removing the
less statistically significant results from the analysis. Pruning followed the 1-standard error rule
(Breiman et al. 1984; Venables and Ripley 1999; De’ath and Fabricous 2000) which minimized
the cross-validated error of the model. This approach was shown to produce optimal sized and
more stable tree sizes across replications compared to the 0-standard error pruning method
(Breiman et al. 1984; Questier et al. 2005).

Competitor and surrogate variables were identified for each node by investigating detailed
CART outputs. Competitor splits are those variables with similar complexity parameter values
compared to the primary split. A complexity parameter compares the number of terminal nodes
(complexity) to the cross-validated error for each group. For this project the complexity
parameter was set at 0.05. Surrogate splits were alternative variables that split the subgroup into
very similar groups. All CART analyses were performed using R software system (R foundation
for Statistical Computing).

An example of a CART output is presented below (Figure 5.2.). At the top of the tree, a parent or
root node is presented with the primary splitting variables and values described for each child
node. At the bottom of the tree, terminal nodes include the mean concentration and number of
target organism cases in each node.

231

Figure 5.2. Classification And Regression Tree analysis output example. Root nodes contain all
available data and are split into binary groups using recursive partitioning algorithm and 10-fold
cross validation with a complexity parameter value of 0.05. Primary splitting variables and
values are described for each child node. Terminal nodes (bottom of the tree) include the mean
concentration and number of target organism cases in each node. Each node was derived based
on mean value of each response variable, group size, and defining variables.

5.3. Results
Sixty-four river catchments were sampled during baseflow, spring thaw, and summer rain
conditions. E. coli and B. thetaiotaomicron results for all sites under each of the three conditions
are presented in Table S.5.1. However, this manuscript and all presented calculations and results,
address the baseflow conditions only. Baseflow conditions offer an opportunity to define
reference conditions which provide a measuring point for temporal changes and management
goals. Future work will compare microbial analysis across baseflow, spring thaw, and summer
rainfall events.

232

5.3.1. Microbial water quality
The first goal of this project was to examine E. coli CFU and B. thetaiotaomicron cell equivalent
concentrations in rivers under baseflow conditions from 64 rivers systems in the Lower
Peninsula of Michigan (USA). E. coli ranged between 0.20 and 3.0 log10 MPN 100ml

-1

with a

-1

geometric mean of 1.4 log10 MPN 100ml . E. coli levels were below the detection limit (< 1
-1

MPN 100 ml ) at four rivers. B. thetaiotaomicron concentrations ranged between 4.2 and 5.9
log10 CE 100 ml

-1

-1

with a geometric mean of 5.1 log10 CE 100 ml . Interestingly, B.

thetaiotaomicron was present in all samples even in the absence of E. coli. Figure 5.3. illustrates
the ranges of E. coli and B. thetaiotaomicron measured in each river system.

233

A.

B.
E. coli
Log concentrations

B. thetaiotaomicron
Log concentrations

N
E

W
S
Kilometers
-1

-1

Figure 5.3. (A.) E. coli (log10 MPN 100 ml ) and (B.) B. thetaiotaomicron (log10 CE 100 ml ) concentrations measured at 64 river
catchments under baseflow conditions. E. coli and B. thetaiotaomicron categories were evenly split across the concentration range.
-1
Areas in black were not sampled. The USEPA health exposure criterion for E. coli is 2.37 log10 MPN 100 ml , shown as the two
highest categories in the E. coli figure and was detected at nine rivers. No single river sample had measurable concentrations of both
microorganisms in the highest concentration categories.
234

Nine rivers exceeded USEPA’s suggested criterion for safe contact (2.37 log10 E. coli MPN
-1

100ml ). At these nine rivers, E. coli ranged from 2.4 to 3.0 log10 MPN 100ml
geometric mean of 2.3 log10 MPN 100 ml
log10 CE 100 ml

-1

-1

-1

with a

and B. thetaiotaomicron ranged from 4.6 to 5.6
-1

with a geometric mean of 5.0 log10 CE 100 ml . In the rivers meeting

USEPA criterion (n = 55), E. coli and B. thetaiotaomicron geometric means were 1.3 log10 MPN
100 ml
100 ml

-1

-1

-1

and 5.2 log10 CE 100 ml , respectively. E. coli ranged from 0.30 to 2.3 log10 MPN
-1

and B. thetaiotaomicron ranged from 4.2 to 5.9 log10 CE 100 ml . A comparison of

microorganism geometric means at sites above and below criterion indicated E. coli were
statistically different (p < 0.001) while B. thetaiotaomicron were not different (p = 0.433)
between the two groups. Correlation analysis between E. coli and B. thetaiotaomicron at sites
below criteria were statistically related (r = 0.308, p = 0.022) while E. coli and B.
thetaiotaomicron at sites above E. coli criteria were not statistically related (r = 0.159, p =
0.683).

5.3.2. Land use
Land use classifications for each river system at the full watershed, reduced watershed, and
reduced watershed riparian buffer are summarized in Table 5.3 and detailed in Table S.5.2
according to Anderson Land Use Classification systems level one. The land use composition
over the entire project area is also illustrated in Figure 5.1. Overall, full watershed sizes ranged
2

2

2

from 2.9 km to 12853 km (X = 1377 km ). Urban development averaged 16.7% and 21.3% in
the full and reduced watersheds, respectively. In the reduced watersheds, urban coverage
235

exceeded 90% at four sites while 34 sites had less than 10% urban coverage. Agriculture in the
full and reduced watersheds averaged 27.9% and 27.2%, respectively. Forest, water, and wetland
cover in full watersheds averaged 31.3%, 2.67%, and 14.0%, respectively. Forest, water, and
wetland cover in reduced watersheds averaged 29.0%, 1.61%, and 13.9%, respectively. In the
reduced watershed and buffered areas, significant associations were identified between E. coli
and percent agriculture (r > 0.345, p < 0.005) and water (r > -0.311, p < 0.01) coverage. B.
thetaiotaomicron was also associated with agriculture cover at the reduced watershed and
buffered area (r > 0.250, p < 0.05).

Impervious surface coverage averaged 5.5% in the buffers and 7.5% in the watersheds with a
low of 0.0% and a high of 55.9%. Impervious surface coverage in the buffer and watershed were
correlated to septic density at the same spatial scale (r ≥ 0.370, p < 0.001). In the reduced
watersheds, septic system numbers in the reduced watersheds ranged between 0 and 63624
systems per watershed (X = 6063 systems per watershed). Similar septic system densities were
-2

-2

observed in the buffer (X = 17 systems km ), reduced watershed (X = 19 systems km ), and
-2

full watersheds (X = 16 systems km ). B. thetaiotaomicron was statistically related to the
number of septic systems in the watershed (r = 0.634, p < 0.001). The average soil hydraulic
conductivity in the buffer and watershed was 1.78 m day
-1

conductivity ranged from 0.40 to 4.7 m day

-1

-1

and 2.21 m day , respectively. Soil
-1

in the buffer and 0.52 to 4.7 m day

in the

watershed. No E. coli or B. thetaiotaomicron correlations were identified with impervious
2

surface coverage, septic system density per km , soil hydraulic conductivity, or estimated total
point source loadings of total nitrogen and phosphorus.
236

Of the 64 sampled rivers, six were located in Areas of Concern (AOC) as identified by the
Michigan Department of Environmental Quality (http://www.michigan.gov/deq/0,4561,7-1353313_3677_15430-240913--,00.html). The E. coli geometric mean for these six sites (1.9 log10
-1

MPN 100 ml ) was not statistically different than the overall E. coli geometric mean for the
entire project (p = 0.345). However, the B. thetaiotaomicron geometric mean (5.6 log10 CE 100
-1

ml ) was statistically higher than non-AOC sites (p = 0.002). No significant land use
correlations existed with either microorganism at these six sites (p < 0.05), indicating another
land use characteristic was associated with increases of this human fecal marker.

237

Table 5.3. Land use summary for full watersheds, reduced watersheds, and reduced watersheds
riparian buffers (60 m).
Scale-Parameter

Minimum

Mean

Maximum

Standard deviation

2.88

1377

12854

2431

0.00

16

114

19.5

Impervious surface (km )

0.41

5.13

56.9

9.80

Urban (%)

3.16

16.7

99.7

0.21

Agriculture (%)

0.00

28.0

74.2

0.22

Open (%)

0.00

6.97

20.1

0.05

Forest (%)

0.19

31.4

70.7

0.18

Water (%)

0.00

2.68

23.7

0.04

Wetland (%)

0.07

14.0

48.3

0.10

Barren (%)

0.00

0.31

2.45

0.00

0.15

366

4065

630

0.00

19.0

102

18.1

Impervious surface (km )

0.40

7.50

55.9

13.6

Urban (%)

3.10

21.3

99.7

26.2

Agriculture (%)

0.00

27.2

77.4

24.0

Open (%)

0.00

6.16

18.8

5.27

Forest (%)

0.00

29.0

71.2

19.5

Water (%)

0.00

1.61

15.4

3.32

Wetland (%)

0.00

13.9

47.9

12.1

Barren (%)

0.00

0.77

31.1

3.87

Watershed

A

2

Area (km )
-2

Septic density (Number km )
2

Watershed

B

2

Area (km )
-2

Septic density (Number km )
2

A

B

Entire upstream drainage area including lakes; Watersheds were defined as the total
upstream area to the nearest lake draining to each rivers respective sampling point

238

Table 5.3. (cont’d)
Scale-Parameter
60 m riparian buffer

Minimum

Mean

Maximum

Standard deviation

B

2

0.06

46.0

497

78.3

0.00

17.0

161

29.0

Impervious surface (km )

0.00

5.50

42.7

9.64

Urban (%)

0.00

18.9

98.3

23.0

Agriculture (%)

0.00

21.4

72.1

21.7

Open (%)

0.00

3.64

19.4

3.80

Forest (%)

0.00

22.1

62.6

14.9

Water (%)

0.00

6.09

63.2

12.0

Wetland (%)

0.00

27.3

76.3

17.9

Barren (%)

0.00

0.59

24.9

3.12

Area (km )
-2

Septic density (Number km )
2

5.3.3. Physical, chemical, environmental, and hydrology
Variables thought to be influencing water quality in Great Lakes tributaries were measured
according to Table 5.1. Dissolved oxygen ranged from 5.9 to 13.3 mg l
potassium ranged from 0.43 to 9.8 mg l
-1

-1

-1

-1

(X = 9.8 mg l ),

-1

(X = 2.2 mg l ), and total phosphorus ranged from 7.7

-1

to 396 mg l (X = 37.8 mg l ). Water temperatures ranged between 7.0°C and 17.5°C (X = 13.1
°C) and were directly correlated to urban (r = 0.466) and water (r = 0.328) coverage while
inversely correlated to open (r = -0.580), forest (r = -0.429), and wetland (r = -0.440) coverage at
the watershed scale (p < 0.008), demonstrating the urban heat effect.

239

Analysis of the data was undertaken during base flow when precipitation was not significantly
influencing water quality or quantity. Six hour cumulative precipitation amounts ranged from 0.0
3 -1

3 -1

to 9.2 mm and averaged 0.14 mm. River discharge (X = 6.7 m s ) ranged from 0.01 to 57 m s
3 -1

-2

and discharge per area ranged from 0.0 to 84 m s km . Discharge for each river system is
provided in Table S.5.1. Descriptive statistics for all physical, chemical, and hydrological
variables are provided in Table S.5.3. A complete chemical, nutrient, environmental, and
hydrological analysis is provided in Martin et al. (in preparation for publication).

5.3.5. CART analysis of microbial water quality
CART analysis was used to determine associations between water quality variables and
microorganisms for full and reduced watersheds. Reduced watershed analysis was performed
only on river systems where sampling points were not located at the lake outlet (n = 53). Eleven
2

sites were located at lake outlets resulting in defined watershed which averaged 108 km ,
2

significantly smaller than the overall watershed size (X = 366 km ). In comparison to the other
53 systems, these eleven watersheds had 9X greater water coverage (6% watershed and 23%
buffer) and 5X less agriculture coverage (6% watershed and 4% buffer). These eleven sites were
removed from the reduced watershed and buffer CART analysis as it was assumed retention time
in the lentic water systems reduced microbe inputs derived from land use activity. CART
analyses for each model scenario are summarized in Table 5.4. and presented in Figure S.5.1.

240

Table 5.4. CART analyses for E. coli and B. thetaiotaomicron as dependent variables and land use, nutrient, chemical, hydrological,
and environmental as independent variables.
Watershed Model Scenario
scale

Target organism
(Total CP)

Primary split

Full
(n=64)

E. coli
(47.4%)

Total phosphorus

All data

Number of target
organisms

Geometric mean
(log10 CE or MPN)

A

24

0.98

A

40

2.02

D

19

1.73

D

45

5.32

C

43

1.91

C

Total phosphorus

Split value

21

1.06

D

19

1.73

D

45

5.32

A

24

0.98

A

40

2.02

B

36

5.35

B

28

4.89

< 19.0
> 19.0

B. thetaiotaomicron Septic system #
(36.5%)
Septic system #
Land use data

< 1622

E. coli
(28.8%)

< 42.5

Forest
Forest

> 1622

> 42.5

B. thetaiotaomicron Septic system #
(51.8%)
Septic system #
Nutrient, chemical,
precipitation,
and physical

< 1622

E. coli
(63.8%)

< 19.0

Total phosphorus
Total phosphorus

B. thetaiotaomicron Dissolved oxygen
(25.8%)
Dissolved oxygen
A

-1 B

-1 C

D

> 1622

> 19.0

< 10.0
> 10.0

μg l ; mg l ; percent land cover; total number of estimated septic systems in the watershed. *Reduced watersheds excluded
upstream lakes and reservoirs and included 53 river systems as 11 sampling points were located at the lake outlet, resulting in
substantially smaller watersheds and minimizing the influence of land use characteristics on water.

241

Table 5.4. (cont’d)
Watershed Model Scenario
scale
Reduced* All data
(n=53)

Target organism
(Total CP)
E. coli
(36.4%)

Primary split
Potassium
Potassium

Number of target
organisms

Geometric mean
(log10 CE or MPN)

B

15

1.29

B

38

2.04

D

25

4.87

D

28

5.50

C

32

2.04

C

21

1.50

D

25

4.87

D

28

5.50

B

15

1.29

B

38

2.04

B

28

5.45

B

25

4.93

Split value
< 0.91
> 0.91

B. thetaiotaomicron Septic system #
(50.2%)
Septic system #
Land use data

< 1912

E. coli
(22%)

< 1.28

Mixed forest
Mixed forest

> 1912

> 1.28

B. thetaiotaomicron Septic system #
(50.2%)
Septic system #
Nutrient, chemical,
precipitation,
and physical

< 1912

E. coli
(58.7%)

< 0.91

Potassium
Potassium

B. thetaiotaomicron Dissolved oxygen
(33.5%)
Dissolved oxygen

242

> 1912

> 0.91
< 10.0
> 10.0

CART models developed using only land variables indicated low forest coverage (< 42.5% in the
watershed) had the strongest association with the highest E. coli levels (X > 1.91 log10 MPN 100
-1

ml ). Septic system numbers had the strongest association with elevated B. thetaiotaomicron
concentrations. The number of septic tanks required to impair water quality varied depending on
whether upstream lakes were included (i.e. full watershed; > 1621 septic systems per watershed)
or excluded (i.e. reduced watershed; > 1912 septic systems per watershed). The number of septic
systems located within the 60 m riparian buffer was a competitor variable for B.
thetaiotaomicron in the full watershed model (improvement difference between primary and
competitor variable = 0.79%), but no competitor variables were identified for E. coli.

CART models developed with only nutrient, chemical, and hydrological independent variables
-1

identified the highest E. coli concentrations in the full watershed (X = 2.02 log10 MPN 100 ml ;
2

-1

R total = 0.64) were associated with total phosphorus levels above 19.0 μg l , while in the
reduced watersheds the highest E. coli concentrations were associated with potassium levels
-1

-1

2

above 0.91 mg l (X = 2.31 log10 MPN 100 ml ; R total = 0.545). Dissolved oxygen below 10
mg l

-1

-1

explained the highest levels of B. thetaiotaomicron (X > 5.35 log10 CE 100 ml ) in the

full and reduced watersheds. No strong competitor variables were identified for E. coli or B.
thetaiotaomicron.

Finally, models were developed that incorporated all independent variables. Interestingly, E. coli
outputs for these models were nearly identical to models developed with only nutrient, chemical,
243

and hydrological variables. However, B. thetaiotaomicron outputs developed with all variables
were nearly identical to the models developed with only land use variables and were heavily
influenced by the number of septic systems located in the watershed. Septic system numbers in
2

the full watershed were directly correlated to B. thetaiotaomicron concentrations (R = 0.338).
No competitor variables were identified for E. coli in this model, but as seen in the B.
thetaiotaomicron land use model, the number of septic systems located within the 60 m riparian
buffer was a competitor variable.

5.4. Discussion
E. coli is commonly used as an indicator of fecal contamination and pathogens in freshwater
rivers and lakes. As shown in this study, E. coli can be found in a variety of stream systems
under baseflow conditions. In general, observed E. coli densities were below USEPA water
quality criteria. The E. coli levels in this study were within previously observed and reported
ranges in Great Lakes tributary rivers (Byappanahalli et al. 2003; Byappanahalli et al. 2006;
Nevers et al. 2007). The overall B. thetaiotaomicron geometric mean was a single log higher
than secondary treated sewage effluent, while the highest concentrations in the current study
were 1.5 logs higher than biologically treated septage effluent (Srinivasan et al. 2011).
Collectively, the E. coli and B. thetaiotaomicron results suggest human fecal contamination was
impairing river water quality under baseflow conditions.

A comprehensive review by Wade et al. (2003) found E. coli levels in freshwater below 2.23
log10 MPN 100 ml

-1

were associated with low relative risks for swimmers compared to non244

swimmers. Since the E. coli geometric mean concentration observed in this study (1.4 log10
-1

MPN 100ml ) was below the safety level reported by Wade et al. (2003), it is suggested that a
reference condition for E. coli be established at 1.4 log10 MPN 100ml

-1

for Michigan rivers.

Wade et al. (2006) reported positive associations between occurrence of illness and molecular
detection of Bacteroides at one Great Lakes beach with a geometric mean concentration of 3.08
-1

log10 CE 100 ml , but the authors note the associations were statistically weak (p < 0.1).
Additionally, Yampara-Iquise et al. (2008) reported B. thetaiotaomicron levels ranged from 5.8
to 9.8 log10 copies 100 ml

-1

in multiple urban, agricultural, and small town creek systems

representing various levels of human impact. In the current study, the mean B. thetaiotaomicron
-1

concentration (5.1 log10 CE 100 ml ) was 1.6 times higher than levels measured by Wade et al.
(2006) but below the range reported by Yampara-Iquise et al. (2008). Therefore, establishing a B.
thetaiotaomicron reference condition for Michigan Rivers will require additional sample
collection and analysis. While the concept of a reference condition lies in the notion of minimal
impact (Reynoldson et al. 1997), this study examined a variety of river types including highly
urbanized systems, as it is widely recognized that few streams are truly unimpaired in the Great
Lakes. It is also understood that E. coli levels in watersheds will likely differ from the reference
condition depending on temporal changes and geographic, natural, and anthropogenic
characteristics. However, establishing such levels in the Great Lakes is important to define
acceptable disturbance levels, support management decisions, and define long term water quality
changes.

245

On average, E. coli concentrations were 0.5 logs higher in rivers with less than 1.27% forest
coverage in the 60 m riparian buffer suggests that the presence of a small forest cover will
improve microbial water quality, perhaps by slowing the transport of bacteria through soil or
overland. However, the overall watershed forest coverage also had a significant (indirect)
influence on E. coli concentrations; suggesting that forests are bacteria sinks or more likely, the
land use replacing forest are a source of bacteria. Overall, land use had relatively little influence
on baseflow E. coli concentration as shown between the three CART scenario outputs. This was
not surprising since precipitation, the primary driver of microbes from land to water, was
purposefully excluded from sampling events.

It was thought that specific land use characteristics would better explain microorganism
occurrence in water compared to overall land use. However, the number of septic systems per
watershed was the only land use characteristic associated with B. thetaiotaomicron using CART.
The direct and significant correlation identified between total number of septic systems in the
full watershed and B. thetaiotaomicron concentrations illustrates a significant problem for
Michigan with an estimated 1.4 million on-site septic systems (MDEQ 2009). Similar amounts
of B. thetaiotaomicron variation were explained regardless of lake presence in the upstream
watershed (i.e. full verse reduced watersheds). However, the number of septic systems associated
with elevated B. thetaiotaomicron concentrations increased when upstream lakes were removed
from model development; indicating lakes are a B. thetaiotaomicron sink under baseflow
conditions. Although B. thetaiotaomicron was primarily explained by the number of septic
systems per watershed, dissolved oxygen was also identified as an explanatory variable,
indicating the presence of other potential sources of organic material leading to lower dissolved

246

oxygen (i.e. sewage discharge or combined sewer overflows) as reported by Hvitved-Jacobson
(1982) and Gammons et al. (2011). It is suggested that future analysis include incremental spatial
assessment of B. thetaiotaomicron upstream and downstream from waste water treatment plant
discharges.

Five of the 64 sampled river systems were placed on the AOC list due in part to beach closures,
but only one river was explicitly characterized as having chronic bacterial contamination (Rouge
th

River). At this site, E. coli was above the 95 percentile for microbial distributions of the project
-1

dataset (2.68 log10 MPN 100 ml ). In comparison to the overall geometric means, this
watershed had 3X more septic systems (n = 20175), 4X more urban coverage (85%), 5X more
impervious surface coverage (35%), and 4X more total nitrogen and total phosphorus loads
-1

associated with point source discharges (715149 and 128982 kg year , respectively); indicating
extreme watershed specific influences were masked during CART modeling. Further
investigations should focus on smaller spatial scales, specifically on the effects of combined
sewer overflows (although likely not an issue during baseflow conditions), waste water treatment
infrastructure, local low impact development and fertilizer use policies in relation to water
quality.

CART analysis identified higher levels of potassium and lower discharge rates resulted in the
highest measurements of E. coli. Inverse relations between discharge and E. coli (r = -0.517, p <
3 -

0.001) suggest that E. coli could accumulate in streams when discharge rates are below 0.66 m s
1

, above which E. coli would be transported downstream. Potassium has been linked to water
247

softener regeneration waste, domestic sewage, forest clear cutting, and leaching of bio-solid and
manure application in agriculture fields (Lynch and Corbett 1990; Wang et al. 1999; Thomas
2000; Chambers et al. 2002). Specifically, potassium levels reported in the literature averaged 25
mg l

-1

in secondary sewage effluent (Emongor and Ramolemana 2004), 38.5 mg l

tank effluent (Brandes 1977), 6.78 mg l

-1

-1

in septic

in a river with a watershed comprised predominantly

by agriculturale (Neal et al. 2000), and 16 mg l

-1

in a sewage impacted river (Gunkel et al.

2007). In the current study, elevated E. coli levels were associated with potassium levels above
-1

-1

0.91 mg l , but the overall potassium average was 2.2 mg l , much lower than previous reports.
Observed potassium levels were more similar to those presented by Katz et al. (2011) and Wolf
et al. (2004), indicative of contaminated groundwater. Thus, potassium levels above 0.91 mg l

-1

may be a suitable indicator of elevated E. coli levels resulting from contaminated groundwater
entering rivers during baseflow conditions.

Using a snapshot approach, this study found multiple Great Lake tributary rivers contained
human fecal contamination under baseflow conditions. Results suggest a regional E. coli
reference condition could be established below the current USEPA freshwater recreational
criterion. Furthermore, the impact septic systems have on surface water quality was shown,
highlighting the need for increased monitoring of on-site wastewater treatment systems.
Michigan does not have a statewide sanitary code which has allowed septic systems to go
unchecked for decades. If these systems are not addressed at a state level, continued chronic
water quality impairments are expected.

248

Acknowledgements
I would like to acknowledge special assistance from Sherry Martin, Anthony Kendall, Steve
Hamilton, David Hyndman, Emily Luscz, Bobby Chrisman, Rebecca Ives, Sarah AcMoody, and
Seth Hunt that provided vital support during this project. Partial funding for this project came
from NOAA GLERL grant titled “Land Use Change and Agricultural Lands Indicators and
Tipping Points.”

249

APPENDIX

250

Table S.5.1. E. coli and B. theta levels measured in 64 Michigan rivers under baseflow, spring thaw, and summer rain conditions.

Site
River system
ID
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
23
24

St. Joseph
Paw Paw River
Kalamazoo
Grand
Muskegon
White River
Pere Marquette
Big Sable River
Little Manistee
Manistee
Bear Creek
Betsie
Platte
Boardman
Elk-Torch
Cheboygan
Black
Thunder Bay
Au Sable
Au Gres
Rifle
Black River
Pine River

E. coli
-1
(MPN 100 ml )
Spring Summer
Baseflow
thaw
rain
1.0
1.4
1.8
1.8
1.7
1.1
1.6
1.6
1.5
1.5
1.3
1.1
1.8
1.8
1.6
2.1
1.6
2.0
2.2
2.0
2.0
0.7
0.3
0.3
1.4
1.4
1.7
0.2
0.2
1.5
2.1
1.3
2.2
2.0
0.9
1.9
0.2
0.2
1.0
1.1
0.2
1.6
0.2
0.2
0.7
0.2
0.2
1.0
1.2
0.3
1.4
0.5
1.0
1.0
1.2
0.5
1.2
2.1
1.1
1.9
1.6
1.6
2.3
1.1
2.8
2.6
2.3
1.7

B. thetaiotaomicron
-1
(CE 100 ml )
Spring Summer
Baseflow
thaw
rain
5.3
4.4
3.1
5.7
4.2
3.6
5.9
4.0
3.3
5.5
4.1
3.5
5.4
3.9
3.7
5.8
4.2
3.6
5.8
4.1
3.3
5.3
3.7
3.5
5.4
3.8
3.2
4.7
3.6
3.4
5.9
3.6
3.0
5.8
4.0
3.1
5.2
3.7
3.3
5.3
3.7
3.4
4.5
3.3
2.6
4.9
3.2
3.3
4.8
3.6
3.3
4.9
4.0
3.3
5.6
3.9
3.5
5.3
2.6
2.6
5.5
4.1
3.8
5.4
4.1
5.4
3.5
2.6
251

Discharge
3 -1
(m s )
Baseflow
57.28
8.01
0.00
50.69
37.77
7.70
12.20
2.87
4.49
42.76
3.01
22.83
4.82
6.68
12.82
26.05
11.58
12.59
32.46
1.07
5.04
0.66
0.02

Table S.5.1. (cont’d)
E. coli
-1
Site
(MPN 100 ml )
River system
ID
Spring Summer
Baseflow
thaw
rain
25 Belle River
2.0
2.5
1.7
26 Clinton River
2.0
1.8
1.9
27 River Rouge
2.7
2.9
2.8
28 Huron
1.9
1.3
1.9
29 Raisin
1.4
1.9
2.3
31 South Branch Black River
2.3
2.5
1.6
32 North Branch Black River
2.2
2.2
1.7
33 Macatawa River
1.3
2.8
2.1
34 Pine Creek
2.1
2.6
2.3
35 Pigeon River
2.7
2.9
2.3
36 Rush Creek
2.3
2.1
2.5
37 Buck Creek
2.2
2.2
2.3
39 Sand Creek
2.3
2.9
1.9
40 Bass River
2.1
2.4
2.3
41 Little Pigeon Creek
2.5
2.2
2.3
43 Black Creek
3.0
2.0
2.1
48 Silver Creek
1.7
1.8
1.8
51 Flower Creek
2.5
2.2
2.7
52 Stony Lake Outlet
0.5
1.3
1.1
55 Swan Creek
2.5
2.7
2.7
56 Lincoln River
2.2
1.5
2.5
57 Crystal River
0.7
0.6
1.2
59 Belangers Creek
1.3
1.7
1.8

B. thetaiotaomicron
-1
(CE 100 ml )
Spring Summer
Baseflow
thaw
rain
5.6
3.4
3.3
5.5
4.4
3.1
5.5
4.5
3.5
5.8
4.7
3.5
5.4
3.9
2.6
5.9
3.5
3.2
5.6
4.1
3.2
5.6
4.1
2.6
4.9
3.6
3.3
5.0
4.0
3.1
5.4
4.0
2.6
4.6
3.7
3.6
4.9
4.1
2.6
5.1
4.2
2.9
4.9
4.0
3.5
4.8
3.9
3.9
4.4
3.5
2.6
4.6
3.7
2.6
5.2
4.3
3.7
4.9
3.8
3.4
5.0
3.8
2.6
4.6
3.8
2.6
4.7
3.3
2.6
252

Discharge
3 -1
(m s )
Baseflow
1.36
0.00
0.32
7.70
4.73
1.10
1.49
0.16
0.27
0.08
0.12
0.64
0.31
0.20
0.03
0.66
0.37
0.34
1.26
0.34
0.66
1.57
0.08

Table S.5.1. (cont’d)

Site
River system
ID
60
62
63
64
65
66
67
69
70
71
73
91
94
97
101
102
103
104

Mitchell Creek
Jordan River
Monroe Creek
Boyne River
Bear River
Carp River
Ocqueoc River
Trout River
Little Trout River
Long Lake Creek
Tawas River
Harrington Drain
Marsh Creek
Sandy Creek
Cass River
Flint River
Shiawassee River
Tittabawassee River

E. coli
-1
(MPN 100 ml )
Spring Summer
Baseflow
thaw
rain
2.2
0.9
2.6
0.9
0.8
1.6
1.2
1.2
2.3
1.2
0.5
1.5
1.5
0.9
1.9
1.0
0.8
1.8
0.9
0.7
1.7
1.3
0.9
1.6
2.4
1.0
2.1
1.8
0.8
1.1
1.2
0.8
2.0
2.0
1.9
3.0
2.4
1.5
2.5
2.2
1.7
2.4
1.2
2.7
2.5
1.9
2.7
2.1
2.0
2.5
2.1
1.9
1.0
2.5

B. thetaiotaomicron
-1
(CE 100 ml )
Spring Summer
Baseflow
thaw
rain
4.8
4.0
3.1
4.4
2.6
2.6
4.5
2.6
2.6
5.4
2.6
2.6
4.8
3.2
3.5
5.0
3.9
2.9
4.7
4.1
2.6
4.8
3.7
2.9
4.9
4.0
3.3
4.2
3.5
3.1
4.5
4.1
2.6
4.5
4.3
3.4
5.3
4.1
2.6
4.9
3.8
2.6
5.4
3.9
2.6
5.7
5.2
2.6
4.7
4.4
2.6
5.6
4.4
2.6

253

Discharge
3 -1
(m s )
Baseflow
0.29
4.11
0.08
1.75
1.58
1.76
2.39
0.41
0.06
0.03
1.59
0.01
0.13
0.02
1.95
6.31
4.36
17.47

Table S.5.2. Land use composition of defined river catchments using Anderson Land Use Classification systems.
Site ID

Site Description

Agriculture Rangeland Forest
(%)
(%)
(%)

Water Wetland Barren
(%)
(%)
(%)

St. Joseph

Area Urban
2
(km ) (%)
11061 14.3

1

59.5

1.1

10.4

2.4

12.2

0.2

2

Paw Paw River

1027

11.5

47.5

2.6

21.1

1.4

15.6

0.3

3

Kalamazoo

5002

14.1

47.8

1.8

21.5

2.1

12.4

0.4

4

Grand

12854 12.7

55.3

1.0

16.6

1.5

12.7

0.3

5

Muskegon

6418

7.6

19.6

9.6

40.5

3.9

18.6

0.1

6

White River

1049

5.2

20.3

9.7

49.7

0.7

14.3

0.1

7

Pere Marquette

1790

5.0

9.3

8.3

61.6

1.2

14.5

0.1

8

Big Sable River

476

5.3

11.5

8.0

52.4

5.0

17.0

0.8

9

Little Manistee

526

4.7

3.9

12.5

68.6

0.7

9.6

0.1

10

Manistee

3559

5.7

9.6

15.9

56.4

1.4

10.9

0.1

11

Bear Creek

350

6.3

13.8

20.1

37.6

2.3

19.7

0.1

12

Betsie

618

8.1

7.6

13.1

46.2

9.9

15.0

0.1

13

Platte

471

6.6

9.9

13.4

56.2

7.5

6.1

0.2

14

Boardman

716

10.8

10.4

18.8

46.7

2.1

10.9

0.2

15

Elk-Torch

1308

7.6

14.4

13.6

45.4

11.3

7.4

0.2

16

Cheboygan

2317

6.4

8.2

11.6

51.0

8.1

14.5

0.2

17

Black

1509

5.5

4.4

12.1

47.1

3.9

27.0

0.1

18

Thunder Bay

2241

6.3

11.0

8.8

40.2

2.7

31.0

0.1

19

Au Sable

5287

8.4

3.2

14.5

58.9

2.0

12.8

0.1

20

Au Gres

987

6.4

23.3

7.6

37.7

2.2

22.0

0.7

21

Rifle

858

9.3

16.5

8.9

44.2

1.6

19.4

0.1

23

Black River

1250

6.2

74.2

1.2

10.6

0.1

7.5

0.1

254

Table S.5.2. (cont’d)
Site ID

Site Description

Urban
(%)

Agriculture Rangeland Forest
(%)
(%)
(%)

Water Wetland Barren
(%)
(%)
(%)

Pine River

Area
2
(km )
440

24

9.0

46.5

3.2

33.3

0.3

7.5

0.1

25

Belle River

512

9.5

59.7

1.7

19.0

0.3

9.7

0.2

26

Clinton River

1880

51.5

20.2

1.3

14.9

2.8

8.6

0.7

27

River Rouge

1033

82.9

5.4

0.5

7.2

0.7

2.9

0.3

28

Huron

2298

32.5

24.5

1.2

21.8

4.2

15.1

0.6

29

Raisin

2683

10.8

67.4

0.8

11.1

1.4

8.3

0.2

31

South Branch Black River

313

9.1

45.8

4.4

22.8

1.2

16.5

0.2

32

North Branch Black River

398

7.0

43.6

5.6

24.8

1.7

17.1

0.2

33

Macatawa River

292

23.5

67.8

0.7

4.0

0.2

3.1

0.9

34

Pine Creek

48

48.4

30.9

1.1

12.1

0.3

6.1

1.1

35

Pigeon River

102

11.0

66.0

2.0

15.3

0.1

5.1

0.5

36

Rush Creek

152

56.4

31.5

0.4

7.6

1.1

2.3

0.6

37

Buck Creek

3

91.3

0.0

0.6

1.4

0.9

5.8

0.0

39

Sand Creek

142

19.1

60.8

0.8

11.3

0.2

7.6

0.3

40

Bass River

127

11.1

63.6

2.1

16.0

0.2

6.6

0.5

41

Little Pigeon Creek

14

18.9

16.4

6.2

41.9

0.0

16.3

0.3

43

Black Creek

136

14.9

34.8

5.3

29.9

4.8

10.1

0.2

48

Silver Creek

41

11.7

0.6

15.2

63.7

4.2

4.4

0.2

51

Flower Creek

79

10.2

45.6

10.7

27.7

0.6

3.4

1.8

52

Stony Lake Outlet

160

10.1

37.7

11.7

35.1

1.0

4.1

0.3

55

Swan Creek

54

5.5

57.9

8.2

15.5

1.3

11.5

0.0

255

Table S.5.2. (cont’d)
Site ID

Site Description

Urban
(%)

Agriculture Rangeland Forest
(%)
(%)
(%)

Water Wetland Barren
(%)
(%)
(%)

Lincoln River

Area
2
(km )
215

56

5.6

33.2

11.8

30.6

2.1

16.5

0.2

57

Crystal River

110

4.7

3.4

8.7

53.7

23.7

3.3

2.4

59

Belangers Creek

25

6.7

38.4

12.7

30.8

1.5

9.9

0.0

60

Mitchell Creek

38

28.3

22.8

16.3

19.4

0.2

13.0

0.1

62

Jordan River

174

3.2

7.8

6.5

70.7

0.0

11.8

0.1

63

Monroe Creek

27

4.2

22.3

8.8

44.5

2.2

18.1

0.0

64

Boyne River

199

8.3

16.1

10.8

54.5

0.6

9.4

0.2

65

Bear River

293

6.4

13.3

7.0

48.5

6.6

18.1

0.2

66

Carp River

119

6.2

8.6

7.7

22.0

7.0

48.3

0.1

67

Ocqueoc River

369

4.7

6.5

11.5

43.4

2.2

31.4

0.3

69

Trout River

82

4.6

13.5

9.5

28.8

0.1

43.1

0.3

70

Little Trout River

28

5.4

27.8

7.5

14.3

0.1

44.6

0.3

71

Long Lake Creek

162

5.7

11.7

7.1

20.7

15.7

39.1

0.1

73

Tawas River

403

8.4

7.1

6.9

51.6

2.0

24.0

0.0

91

Harrington Drain

53

99.7

0.0

0.0

0.2

0.0

0.1

0.0

94

Marsh Creek

78

72.0

4.7

1.7

15.4

0.0

5.9

0.2

97

Sandy Creek

82

26.2

58.7

1.3

10.6

0.0

2.7

0.4

101

Cass River

2174

6.9

57.4

2.2

19.7

0.2

13.5

0.1

102

Flint River

3206

21.0

40.6

2.0

24.1

1.6

10.4

0.3

103

Shiawassee River

1517

15.7

52.5

0.7

17.0

2.2

11.4

0.4

104

Tittabawassee River

6211

8.6

32.8

7.3

30.6

1.5

19.1

0.2

256

Table S.5.3. Descriptive statistics of physical-chemical, and hydrological variables measured during baseflow conditions at 64 rivers.
th

Parameter

UNIT

Ammonia

µg l

Calcium

mg l

Chlorine (Cl-)

mg l

Dissolved oxygen

mg l

Dissolved organic carbon

mg l

Magnesium

mg l

Nitrate/nitrite

µg l

Pheophytin corrected
chlorophyll a

µg l

Count Minimum

Mean

Maximum Standard deviation

th

5
Percentile

95
Percentile

-1

63

0.00

23.6

280.0

45.57

0.00

98.50

-1

63

30.04

62.4

160.6

21.56

33.82

98.22

-1

63

3.44

42.3

291.8

54.43

5.85

174.79

-1

64

5.90

9.82

13.3

1.66

7.15

12.21

-1

63

1.60

6.14

26.8

4.23

2.12

15.55

-1

63

7.03

18.4

34.2

6.27

10.28

29.05

-1

64

0.00

858.3

5638.9

1310.26

0.00

4095.58

-1

59

0.03

0.82

4.42

1.04

0.07

3.42

63

7.90

8.20

8.38

0.11

7.99

8.35

-1

63

0.43

2.20

9.79

1.90

0.45

6.02

-1

63

3.03

27.0

199.3

36.86

3.40

113.00

-1

64

0.523

2.21

4.66

1.14

0.620

4.20

-1

63

257.00

527.0

1589.0

264.16

265.20

1039.80

-1

64

0.86

23.3

266.0

45.03

2.10

87.01

-1

63

2.39

32.1

169.8

30.47

5.55

89.58

pH
Potassium

mg l

Sodium

mg l

Soil hydraulic
conductivity

m day

Specific conductance

μS cm

Soluble reactive P

µg l

Sulfate

µg l

2

*Precipitation measured at hourly averages from 16 km NEXRAD cells and reported in cumulative mm per time

257

Table S.5.3. (cont’d)
th

Parameter

UNIT

Total dissolved N

µg l

Total dissolved P

µg l

Total N

µg l

Total P

µg l

Total chlorophyll a

µg l

Count Minimum

th

Mean

Maximum

Standard deviation

5
Percentile

95
Percentile

-1

64

0.00

1423.3

6033.7

1346.46

337.55

5414.14

-1

64

3.11

25.2

292.3

38.57

3.94

58.04

-1

64

81.82

1082.1

5583.1

1129.31

110.80

3610.59

-1

64

7.70

37.8

395.5

52.44

8.91

102.54

-1

59

0.07

1.57

7.76

1.92

0.20

7.41

64
64
64
64
64
64
64
64
64

0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00

0.14
1.89
3.42
4.42
6.04
7.66
8.26
9.00
11.62

9.20
77.89
78.61
78.61
78.61
80.07
80.46
87.30
92.55

1.15
10.20
11.64
11.86
11.85
13.66
13.45
14.06
16.46

0.00
0.00
0.00
0.00
0.00
0.00
0.029
0.029

7.91
30.8
31.0
31.0
34.2
34.2
34.2
57.2

Precipitation*
6 hour
12 hour
18 hour
24 hour
2 day
3 day
4 day
6 day
8 day

mm
mm
mm
mm
mm
mm
mm
mm
mm
3 -1

Discharge

m s

63

0.00

6.74

57.28

12.48

0.017

43.4

Water temperature

°C

64

6.98

13.09

17.50

2.56

8.2

16.6

258

(A.1.)

(A.2.)

(A.1.)

(A.2.)

(A.3.)

(A.3.)

Figure S.5.1. CART tree outputs for E. coli and B. thetaiotaomicron at the (A) full and (B) reduced watersheds developed with (1) all
data, (2) land use variables only, and (3) nutrient, chemical, and environmental variables only.
259

Figure S.5.1. (cont’d)
(B.1.)

(B.2.)

(B.1.)

(B.2.)

(B.3.)
(B.3.)

260

REFERENCES

261

REFERENCES

Alm, E.W., Burke, J., and Hagan, E. (2006). Persistence and potential growth of the fecal
indicator bacteria, Escherichia coli, in shoreline sand at Lake Huron. Journal of Great Lakes
Research, 32, 401-405.
Almeida, C. and Soares, F. (2012). Microbiological monitoring of bivalves from the Ria
Formosa Lagoon (south coast of Portugal): A 20 years of sanitary survey. Marine Pollution
Bulletin, 64, 252-262
American Public Health Association (APHA). (1998). Standard methods for the examination of
water and wastewater (American Public Health Association, Washington, D.C), 20th ed.
Anderson, J.R., Hardy, E.E., Roach, J.T., and Witmer, R.E. (1976). A land use and land cover
classification system for use with remote sensor data (USGS Publication No. 964). Washington,
DC: US Government Printing Office.
Arnold, C.L. and Gibbons, C.J. (1996). Impervious surface coverage: The emergence of a key
environmental indicator. Journal of American Planning Association, 62, 243-258.
Bae, H.K., Olson, B., Hsu, K.L., and Sorooshian, S. (2010). Classification and Regression Tree
(CART) analysis for indicator bacterial concentration prediction for a Californian coastal area.
Water Science and Technology, 61, 545-553.
Brandes, M. (1977). Effective by phosphorus removal by adding alum to septic tank. Water
Pollution Control Federation, 49, 2285-2296.
Breiman L., Friedman J.H., Olshen R.A., and Stone C.J. (1984). Classification and regression
trees. Chapman and Hall, New York.
Breuer, L., Huisman, J.A., Willems, P., Bormann, H., Bronstert, A., Croke, B.F.W., Frede, H.G., Gräff, T., Hubrechts, L., Jakeman, A.J., Kite, G., Lanini, J., Leavesley, G., Lettenmaier, D.P.,
Lindström, G., Seibert, J., Sivapalan, M., and Viney, N.R. (2009). Assessing the impact of land
use change on hydrology by ensemble modeling (LUCHEM). I: Model intercomparison with
current land use. Advances in Water Resources, 32, 129-146.
Byappanahalli, M.N., Fowler, M., Shively, D., and Whitman, R.L. (2003). Ubiquity and
persistence of Escherichia coli in a Midwestern coastal stream. Applied and Environmental
Microbiology, 69, 4549-4555.

262

Byappanahalli, M.N., Whitman, R.L., Shively, D.A, Sadowsky, M.J., and Ishii, S. (2006).
Population structure, persistence, and seasonality of autochthonous Escherichia coli in
temperate, coastal forest soil from a Great Lakes watershed. Environmental Microbiology, 8,
504-513.
Carlisle, D.M., Hawkins, C.P., Meador, M.R., Potapova, M., and Falcone, J. (2008). Biological
assessments of Appalachian streams based on predictive models for fish, macroinvertebrate, and
diatom assemblages. Journal of the North American Benthological Society, 27, 16-37.
Cha, Y., Stow, C.A., Reckhow, K.H., DeMarchi, C., and Johengen, T.H. (2010). Phosphorus
load estimation in the Saginaw River, MI using a Bayesian hierarchical/multilevel model. Water
Research, 44, 3270-3282.
Chambers, B., Royle, S., Hadden, S., and Maslen, S. (2002). The use of biosolids and other
organic substances in the creation of soil-forming materials. Water and Environment Journal, 16,
34-39.
Crumpton, W.G., Isenhart, T.M., and Mitchell. P.D. (1992). Nitrate and organic N analyses with
second derivative spectroscopy. Limnology and Oceanography, 37, 907-913.
Davies, S.P. and Jackson, S.K. (2006). The biological condition gradient: A descriptive model
for interpreting change in aquatic ecosystems. Ecological Applications, 16, 1251-1266.
De’ath, G. and Fabricius, K. (2000). Classification And Regression Trees: A powerful yet simple
technique for ecological data analysis. Ecology, 81, 3178-3192.
Emongor, V.E. and Ramolemana, G.M. (2004). Treated sewage effluent (water) potential to be
used for horticultural production in Botswana. Physics and Chemistry of the Earth, 29, 11011108.
Evers, D.C., Wiener, J.G., Basu, N., Bodaly, R.A., Morrison, H.A., and Williams, K.A. (2011).
Mercury in the Great Lakes region: Bioaccumulation, spatiotemporal patterns, ecological risks,
and policy. Ecotoxicology, 20, 1487-1499.
Falkenmark, M. (2011). Water-A Reflection of Land Use: Understanding of Water Pathways and
Quality Genesis. International Journal of Water Resources Development, 27, 13-32.
Furtula, V., Osachoff, H., Derksen, G., Juahir, H., Colodey, A., and Chambers, P. (2012).
Inorganic nitrogen, sterols and bacterial source tracking as tools to characterize water quality and
possible contamination sources in surface water. Water Research, 46, 1079-1092.
Gammons, C.H., Babcock, J.N., Parker, S.R., and Poulson, S.R. (2010). Diel cycling and stable
isotopes of dissolved oxygen, dissolved inorganic carbon, and nitrogenous species in a stream
receiving treated municipal sewage. Chemical Geology, 283, 44-55.
263

Germer, S., Neill, C., Vetter, T., Chaves, J., Krusche, A.V., and Elsenbeer, H. (2009).
Implications of long-term land-use change for the hydrology and solute budgets of small
catchments in Amazonia. Journal of Hydrology, 364, 349-363.
Grayson, R.B., Gippel, C.J., Finlayson, B.L., and Hart, B.T. (1997). Catchment-wide impacts on
water quality: The use of “snapshot” sampling during stable flow. Journal of Hydrology, 199,
121-134.
Gunkel, G., Kosmol, J., Sobral, M., Rohn, H., Montenegro, S., and Aureliano, J. (2007). Sugar
cane industry as a source of water pollution - Case study on the situation in Ipojuca River,
Pernambuco, Brazil. Water, Air, and Soil Pollution, 180, 261-269.
Hamilton, S.K., Bruesewitz, D.A., Horst, G.P., Weed, D.B., and Sarnelle, O. (2009). Biogenic
calcite–phosphorus precipitation as a negative feedback to lake eutrophication. Canadian
Journal of Fisheries and Aquatic Sciences, 66, 343-350.
Hvitved-Jacobsen, T. (1982). The impact of combined sewer overflows on the dissolved oxygen
concentration of a river. Water Research, 16, 1099-1105.
Jarrett, R. (1991). Wading measurements of vertical velocity profiles. Geomorphology, 4, 243247.
Katz, B.G., Eberts, S.M., and Kauffman, L.J. (2011). Using Cl/Br ratios and other indicators to
assess potential impacts on groundwater quality from septic systems: A review and examples
from principal aquifers in the United States. Journal of Hydrology, 397, 151-166.
Katz, D.M., Watts, F.J., and Burroughs, E.R. (1995). Effects of surface roughness and rainfall
impact on overland flow. Journal of Hydraulic Engineering, 121, 546-553.
Kistemann, T., Claßen, T., Koch, C., Dangendorf, F., Fischeder, J., Gebel, J., Vacata, V., and
Exner, M. (2002). Microbial load of drinking water reservoir tributaries during extreme rainfall
and runoff. Applied and Environmental Microbiology, 68, 2188-2197.
Lemon, S.C., Roy, J., Clark, M.A., Friedmann, P.D., and Rakowski, W. (2003). Classification
and Regression Tree Analysis in public health: Methodological review and comparison with
logistic regression. The Society of Behavioral Medicine, 26, 172-181.
Luscz, E. and Hyndman, D. (in preparation). Modeling Nutrient Loading to Watersheds in the
Great Lakes Basin: A Detailed Source Model at the Regional Scale.
Lynch, J.A. and Corbett, E.S. (1990). Evaluation of best management practices for controlling
nonpoint pollution from silvicultural operations. Water Resources Bulletin, 26, 41-52.
Martin, S. and Hyndman, D. (In prep.). Chemical and nutrient responses to land, physical, ,
environmental, and hydrological factors in Michigan, USA.
264

Martin, S., Soranno, P., Bremigan, M.T., and Cheruvelil, K.S. (2011). Comparing
hydrogeomorphic approaches to lake classification. Environmental Management, 48, 957-974.
Michigan Department of Environmental Quality (MDEQ). (2009). Michigan’s nonpoint source
program plan.
Neal, C., Williams, R., Neal, M., Bhardwaj, L., Wickham, H., Harrow, M., and Hill, L. (2000).
The water quality of the River Thames at a rural site downstream of Oxford. The Science of the
Total Environment, 251/252, 441-457.
Nevers, M.B., Whitman, R.L., Frick, W.E., and Ge, Z. (2007). Interaction and influence of two
creeks on Escherichia coli concentrations of nearby beaches: Exploration of predictability and
mechanisms. Journal of Environmental Quality, 36, 1338-1345.
Patz, J.A., Vavrus, S.J., Uejio, C.K., and McLellan, S.L. (2008). Climate change and waterborne
disease risk in the Great Lakes region of the U.S. American journal of Preventive Medicine, 35,
451-458.
Peed, L.A., Nietch, C.T., Kelty, C.A., Meckes, M., Mooney, T., Sivaganesan, M., and Shanks, O.
(2011). Combining land use information and small stream sampling with PCR-based methods for
better characterization of diffuse sources of human fecal pollution. Environmental Science and
Technology, 45, 5652-5659.
Questier, F., Put, R., Coomans, D., Walczak, B., and Heyden, Y.V. (2005). The use of CART
and multivariate regression trees for supervised and unsupervised feature selection.
Chemometrics and Intelligent Laboratory Systems, 76, 45-54.
Ray, D.K., Duckles, J.M., and Pijanowski, B.C. (2010). The impact of future land use scenarios
on runoff volumes in the Muskegon River Watershed. Environmental Management, 46, 351-366.
Reynoldson, T.B., Norris, R.H., Resh, V.H., Day, K.E., and Rosenberg, D.M. (1997) The
reference condition : a comparison of multimetric and multivariate approaches to assess waterquality impairment using benthic macroinvertebrates. Journal of the North American
Benthological Society, 16, 833-852.
Saatchi, S., Malhi, Y., Zutta, B., Buermann, W., Anderson, L.O., Araujo, A. M., Phillips, O.L.,
Peacock, J., Ter Steege, H., Lopez Gonzalez, G., Baker, T., Arroyo, L., Almeida, S., Higuchi, N.,
Killeen, T., Monteagudo, A., Neill, D., Pitman, N., Prieto, A., Salomão, R., Silva, N.,
Vásquez Martínez, R., Laurance, W., and Ramírez, H.A. (2009). Mapping landscape scale
variations of forest structure, biomass, and productivity in Amazonia. Biogeosciences
Discussions, 6, 5461-5505
Soranno, P.A., Wagner, T., Martin, S.L., McLean, C., Novitski, L.N., Provence, C.D., and
Rober, A.R. (2011). Quantifying regional reference conditions for freshwater ecosystem
265

management: A comparison of approaches and future research needs. Lake and Reservoir
Management, 27, 138-148.
Srinivasan, S., Aslan, A., Xagoraraki, I., Alocilja, E., and Rose, J.B. (2011). Escherichia coli,
enterococci, and Bacteroides thetaiotaomicron qPCR signals through wastewater and septage
treatment. Water Research, 45, 2561-2572.
Stone, D.A., Weaver, A.J., and Zwiers, F.W. (2000). Trends in Canadian Precipitation Intensity.
Atmosphere and Ocean, 38, 321-347.
Thomas, M.A. (2000). The effect of residential development on ground-water quality near
Detroit, Michigan. Journal of the American Water Resources Association, 36, 1023-1038.
Tiefenthaler, L.L., Stein, E.D., and Lyon, G.S. (2009). Fecal indicator bacteria (FIB) levels
during dry weather from Southern California reference streams. Environmental Monitoring and
Assessment, 155, 477-92.
Vega, M., Pardo. R., Barrado, E., and Debân, L. (1998). Assessment of seasonal and polluting
effects on the quality of river water by exploratory data analysis. Water Research, 32, 35813592.
Venables, W.N. and Ripley, B.D. (1999). Modern applied statistics with S-PLUS, 3rd edn.
Springer, New York
Vörösmarty, C. and Dork, S. (2000). Anthropogenic Disturbance of the Terrestrial Water Cycle.
BioScience, 50, 753-765.
Wade, T.J., Calderon, R.L., Sams, E., Beach, M., Brenner, K.P., Williams, A.H., and Dufour, A.
(2006). Rapidly measured indicators of recreational water quality are predictive of swimmingassociated gastrointestinal illness. Environmental Health Perspectives, 114, 24-28.
Wade, T.J., Pai, N., Eisenberg, J.N.S., and Colford, J.M. (2003). Do U.S. Environmental
Protection Agency Water Quality guidelines for recreational waters prevent gastrointestinal
illness? A systematic review and meta-analysis. Environmental Health Perspectives, 111, 11021109.
Wang, S.J., Fox, D.G., Cherney, D.J., Klausner, S.D., and Bouldin, D.R. (1999). Impact of dairy
farming on well water nitrate level and soil content of phosphorus and potassium. Journal of
Dairy Science, 82, 2164-2169.
Wetzel, R.G. and Likens, G.E. (2000). Limnological analyses, 3rd ed. Springer-Verlag.
Wilkes, G., Edge, T., Gannon, V., Jokinen, C., Lyautey, E., Neumann, N., Ruecker, N., Scott, A.,
Sunohara, M., Topp, E., and Lapen, D. (2011). Associations among Pathogenic Bacteria,
266

Parasites, and Environmental and Land use Factors in Multiple Mixed-Use Watersheds. Water
Research, 45, 5807-5825.
Wolf, L., Held, I., Eiswirth, M., and Hötzl, H. (2004). Impact of leaky sewers on groundwater
quality. Acta Hydrochimica et Hydrobiologica, 32, 361-373.
Wong, M., Kumar, L., Jenkins, T.M., Xagoraraki, I., Phanikumar, M.S., and Rose, J.B. (2009).
Evaluation of public health risks at recreational beaches in Lake Michigan via detection of
enteric viruses and a human-specific bacteriological marker. Water Research, 43, 1137-1149.
Yampara-Iquise, H., Zheng, G., Jones, J.E., and Carson, C.A. (2008). Use of a Bacteroides
thetaiotaomicron-specific alpha-1-6, mannanase quantitative PCR to detect human faecal
pollution in water. Journal of Applied Microbiology, 105, 1686-1693.

267

CHAPTER 6.
CONCLUSION

268

6.1. Goals of the Research and Results of the Multi-scale Studies.
Overall, this project aimed to measure and assess fecal indicator bacteria in water and describe
their response to land and climate variability. Focused investigations occurred at various spatial
and temporal scales across multiple watersheds in order to describe causes, sources, and
processes associated with the fate of fecal pollution in water.

Investigations in the Saginaw Bay, a shallow wind driven system that receives agriculture, urban,
and forested runoff, focused on a narrow spatial scale directly at the beach. The tools used in this
study included E. coli, enterococci, C. perfringens, coliphage CN-13 and F+amp, enterococcus
surface protein (esp), and Bacteroides human and bovine markers. Results identified sediment
and stranded algae mats were acting as reservoirs of E. coli, enterococci, C. perfringens, and
coliphage in the nearshore area. These reservoirs had the greatest influence on microbial water
quality when precipitation, wind, and wave action resuspended sediments and detached bacteria
into the water column. Elevated concentrations of E. coli, enterococci, and C. perfringens were
observed in these reservoirs (compared to the surrounding water column) in absence of recent
contamiantion, potentially as a result of regrowth and persistence. Positive detections of the
enterococci surface protein gene marker confirmed that some human fecal contamination was
impacting Saginaw Bay water quality. It was clear from this work that in addition to the
nearshore reservoirs, upstream activities required further investigations to accuaretly describe
contamination sources. This required going from the beach environment to the watershed level
and adding more sensitive genetric markers and land use investigations.

269

The sequential investigation increased the spatial and temporal assessment and focused on a
single watershed (Mitchell Creek) draining to the Grand Traverse Bay. This highly modified and
flashy watershed discharges to the East Grand Traverse Bay near the Traverse City State Park
beach. Using the fecal indicator E.coli, microbial water quality at Traverse City State Park beach
routinely met water quality safety thresholds but the nearby Mitchell Creek was deemed unsafe
for swimming at all sampling locations. Enterococci, C. perfringens, coliphage CN-13, and
Bacteroides theataiotaomicron were added to the toolbox to further assess water quality.

Major findings identified that the Mitchell Creek was heavily impacted by human population and
human fecal contamination. This contamination was primarily precipitation driven and had the
ability to influence Traverse City State Park beach water quality. Results confirmed
environmental conditions (precipitation, air and water temperature, river discharge, and solar
radiation) had significant effects on microbial water quality. Furthermore, land use coverage
(urban, agriculture, and wetlands) and wastewater treatment discharge (as a proxy for human
population) were identified as potential sources of fecal contamination in this area. As there are
no sewage discharges located directly in this watershed, non-point sources (e.g. septic tanks,
leaky sewers) were considered the likely source of the human sewage marker. Although it was
proposed that the microbes detected in water originated from land, a single land use type was not
identified as a primary source of bacteria using E. coli, enterococci, C. perfringens, or coliphage.
At the conclusion of this project, it was not clear whether the results observed in this system
could be directly translated to larger-scale, basin wide assessment using the same tools; leading
to the design and implementation of a broader spatial survey.

270

A survey of Michigan Lower Peninsula rivers draining to the Great Lakes was conducted under
baseflow hydrologic conditions. The study design included analysis of nutrients, ions, isotopes,
physical parameters, land use characteristics, and microbes (E. coli and Bacteroides
thetaiotaomicron) from 64 rivers. This approach was used as a way to improve scientific
understanding of water pollution processes at the state level. At this scale, land use
characteristics (primarily the number of septic systems in the watershed) more accurately
predicted Bacteroides thetaiotaomicron concentrations than land use type. Total phosphorus and
potassium best predicted E. coli at the entire watershed and entire watershed excluding upstream
lakes, respectively. Specifically, phosphorus was retained in lakes while potassium, presumably
from agriculture and septic system leachate, was transported through groundwater into rivers.
Similarly, B. thetaiotaomicron was retained in lake systems as a result of increased water
retention time, increased residential density, and DNA degradation. B. thetaiotaomicron detected
in rivers likely entered via groundwater from nearby septic systems and sewage discharge.
Strong correlations between B. thetaiotaomicron and the total number of septic systems
illustrated the significant influence of septic systems on microbial water quality. Baseflow
analysis represented a microbial reference condition for Michigan rivers which can be used to
assess future changes to water quality stemming from climate change or anthropogenic
influences.

Overall, the Sagianw Bay project idenfied the processes of microbial contamination at the beach
but failed to identify the processes by which water initially became impacted. In the Mitchell
Creek, sources and transport mechanisms led to impacted beaches but specific land use types
were not implicated using general bacteria. The statewide survey identified correlations between
271

land use characterizations (i.e. septic system numbers per watershed) and microbial water
quality, indicating land use types were too general to draw significant links with fecal indicator
bacteria.

6.2. Implications for managing and improving water quality in the Great Lakes
The Clean Water Act set the lofty goal of making all waters safe for fishing and swimming.
While progress has been made in many systems, the overall goal has not been reached. In order
to eliminate water pollution, significant and continuous efforts, enforcement, and funding must
occur using a top down approach. While it is important to protect and improve the water quality
of large waterbodies and systems, this dissertation illustrated how the health of small streams can
have drastic implications for surrounding communities and waterbodies. By allocating greater
attention to small systems, improvements in Great Lake water quality can be achieved and
costly, large scale projects can be avoided. Only through continuous monitoring, application,
enforcement, and perseverance of the Clean Water Act can the health of the Great Lakes be
guaranteed for future generations.

A series of regulatory based outcome analysis in the Grand Traverse Bay area using multiple
indicator organisms determined that monitoring for enterococci using cultivation methods would
result in the greatest number of regulatory actions according to the USEPA suggested criteria. It
is not assumed that the greater number of closures represented a greater protection for bathers
compared to E. coli outcomes. Analysis identified a significant disconnect between molecular
and cultivation based results in different water system types (i.e. creek or beach). Specifically in
beach water, analysis of criteria for E. coli cultivation, enterococci cultivation, and Enterococcus
272

spp. molecular methods suggest that molecular based Enterococcus spp. would have resulted in
the most regulatory actions. Unlike cultivation based enterococci, this molecular approach may
provide a greater level of protection as results can be produced more rapidly, allowing for water
quality managers to make regulatory decisions on the same day.

Climate investigations found that watershed and site specific precipitation thresholds exist for
microbial water quality. Projected temperature increaes in the Great Lakes basin will lead to
more frequent algae blooms as surface water flow descreases and nutrient loading increases. As
shown in this dissertation, algae mat occurrence can impair nearshore water and represents a
potential threat to human health. Additionally, more frequent and intense precipitation events are
projected, which will lead to shallow groundwater contamination and require a shift in the design
and implementation of on-site wastewater treatment systems. Climate change, population
growth, and land use development in the Great Lakes will drive the availability and quality of
water, compounding water infrastructure stress. Such forecasts indicate the need to anticipate
long term water use shifts and prepare for their associated implications on infrastructure.
Congress should mandate integrated assessments for watersheds most at risk and fund more
effective stormwater management based on current drinking and wastewater infrastructure
capacities.

6.3. Recommentations
Based on the cummulatve results from these studies, the following management actions are
recommended:

273

1. Stranded algae mats should be removed from the nearshore area early in the morning
under low wind speed and wave height and when 48 hour anticendent precipitation is
below 6.4 mm. Such actions would allow for solar inactivation of bacteria, reduce risk to
bathers entering water later in the day, and improve beach water quality. Significant
financial and managerial attention must focus on combined sewer overflow and on-site
wastewater treatment systems in order to reduce the discharge, direct or diffuse, of
human fecal material to the Saginaw Bay.

2. Michigan must adopt a statewide sanitary code to level the playing field between all
counties, eventually leading to improved design, maintanence, and governance of on-site
wastewater treatment systems. Creation of a statwide sanitary code would bridge the
knowledge gap between those who know about potential environmental hazards and
those who need to know in order to provide effective changes in policy and technology
that would improve water quality throughout Michigan.

3. Molecular and cultivation based monitoring methods in the nearshore areas must continue
to be evaluated both at large spatial scales and at site-specific beaches. Weak
relationships between molecular and cultivation results confirm criteria are not
interchangeable between all water systems and provide different levels of protection.
Further exploration of the suggested recreational water quality monitoring marker
Enterococcus spp. is reguired at beaches impacted by non-point sources of pollution.

274

4. Michigan must implement a surveylence system for on-site wastewater treatment
systems. This system will allow prioritization of remediation efforts and promote more
effective treatment systems on a large scale. A new framework should include detailed
history of each system across the state inclusive of molecular source tracking tools but
should not be contstrained by political boundaries. Ideally this system would facilitate
local level assessments, actions, and adaptations through land and water management
plans at a watershed scale.

275