The Pennsylvania State University

The Graduate School

Department of Agricultural and Biological Engineering

EFFECTS OF NATURAL AND ANTHROPOGENIC DRIVERS ON EMERGING

ORGANIC CONTAMINANTS IN WASTEWATER AND DRINKING WATER

SYSTEMS: OCCURRENCE AND REMOVAL

A Dissertation in

BioRenewable Systems

Faith Awino Kibuye

Submitted in Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

August 2019

The dissertation of Faith Awino Kibuye was reviewed and approved* by the following:

Heather E. Gall Assistant Professor of Agricultural and Biological Engineering Dissertation Advisor, Chair of Committee

Herschel A. Elliott Professor of Agricultural and Biological Engineering

John E. Watson Professor of Soil Science and Soil Physics

Rachel A. Brennan Associate Professor of Civil and Environmental Engineering

Paul H. Heinemann Professor of Agricultural and Biological Engineering Head of Department of Agricultural and Biological Engineering

*Signatures are on file in the Graduate School

iii

ABSTRACT

Indirect potable water reuse is considered an effective management practice for augmenting declining water resources. However, the occurrence of emerging organic contaminants (EOCs) such as prescription and non-prescription drugs, health and beauty products, among others in wastewater effluent challenges reuse practices due to the underlying risk to aquatic ecosystem health and potential human health impacts. It is important therefore to understand the fate, transport and risks of EOCs in surface and groundwater drinking water sources that are impacted by various indirect potable reuse practices. The objectives of this study were to evaluate the occurrence and removal of selected EOCs during wastewater treatment, assess their fate, transport and risks in impacted surface and groundwater sources, and evaluate their removal during conventional drinking water treatment.

While land application of wastewater effluent is beneficial for recharging groundwater aquifers and avoiding direct pollutant discharges to surface waters, it provides a pathway of EOCs that persist in wastewater effluent to underlying aquifers. The extent to which effluent irrigation activities at the Penn State Living Filter has impacted groundwater was investigated. Commonly used pharmaceuticals (acetaminophen, ampicillin, caffeine, naproxen, ofloxacin, sulfamethoxazole, and trimethoprim) were targeted for analysis in wastewater influent, effluent and monitoring wells at the spray irrigation site. In wastewater influent, acetaminophen and trimethoprim were the most frequently detected (93%) above the limit of quantification (LOQ), while in the effluent, caffeine and trimethoprim were detected most frequently (70%). Acetaminophen and

iv caffeine were generally well removed (>88%) during wastewater treatment while other compounds including antibiotics and the anti-inflammatory drug naproxen were removed to a lesser extent and in some cases, were present at higher concentrations in effluent samples. The removal efficiencies of the pharmaceuticals varied seasonally with the least removals recorded in colder months. The impact of long-term wastewater irrigation on groundwater was observed through the presence of studied compounds at levels above the

LOQ. Detection frequencies were however lower in groundwater samples compared to the effluent, with sulfamethoxazole (40%) and caffeine (32%) as the most frequently detected compounds. Similarly, average concentrations of pharmaceuticals in groundwater were nearly two orders of magnitude lower than concentrations in the effluent. Effluent irrigation performs an ecosystem service by mitigating an ecosystem risk to aquatic organisms as wastewater effluent posing medium to high risk to aquatic systems is not discharged directly in streams but allowed to infiltrate through the soil and decrease in concentration before recharging groundwater. Furthermore, human health risk assessments indicate that concentrations of studied EOCs in groundwater, which is used as a drinking water source, appear to pose minimal risk.

Domestic drinking water wells are common in the U.S. and serve about 1 million homes and farms in the commonwealth of Pennsylvania. These private wells are often located in areas served by onsite wastewater treatment systems such as septic systems where treated domestic wastewater effluent is discharged in subsurface leach fields for further treatment before recharging groundwater. The occurrence, range of concentrations, and potential human health risks of seven pharmaceutical compounds (acetaminophen,

v ampicillin, caffeine, naproxen, ofloxacin, sulfamethoxazole, and trimethoprim) in 26 private wells located in central PA were evaluated. Ofloxacin (100%) and sulfamethoxazole (58%) were the most frequently detected compounds while naproxen was not detected in any sample and other pharmaceuticals were present in <50% of samples. Fate and transport modelling in the vadose zone suggest that detection frequencies and concentrations in groundwater are influenced by physicochemical properties of pharmaceutical compounds including sorption potential and biodegradation rates in soil.

Average concentrations were typically < 20 µg/L in private wells however, groundwater concentrations were higher when compared to nearby surface water samples. Additionally, human health risk calculations were conducted and revealed that none of the concentrations observed in the groundwater samples posed significant human health risk.

Factors influencing the fate and transport of 20 EOCs in surface water sources impacted by both point source inputs from wastewater discharges and non-point inputs were investigated during a 2-yr. monitoring study. Six drinking water sources consisting of 3 riverine and reservoir types located in the Susquehanna River Basin were used as study sites. Higher detection frequencies and concentrations of targeted EOCs were observed in riverine sources in comparison to reservoir sources where longer residence times can promote natural attenuation processes. Detection frequencies of EOCs were linked to dominant land use types in associated watersheds with lower frequencies of detections observed in highly forested watersheds potentially due to reduced pollutant sources.

Seasonal influences in concentrations of EOCs were observed as higher concentrations were recorded in the fall, winter and spring seasons since these colder seasons are

vi associated with lower biodegradation rates in surface water. Statistically significant

(p<0.05) seasonal variations were observed for acetaminophen, caffeine, naproxen, and sulfamethoxazole. EOCs depicted diverse hydrologic contaminant transport responses.

EOCs of wastewater origin such as metformin and sulfamethoxazole exhibited dilution responses whereby aqueous concentrations reduced during high streamflow conditions.

Thiamethoxam, a neonicotinoid insecticide depicted concentration accretion patterns with lower in stream concentrations during low streamflow conditions and higher levels following high precipitation events which results in surface runoff that transports agricultural EOCs from diffuse sources. Risk calculations indicate that measured EOCs posed medium to high risk to aquatic organisms, however human health risks through consumption of fish was low.

The occurrence of pharmaceuticals in various surface water sources raises concern over their removal through conventional drinking water treatment processes. A 1-yr. monitoring study was conducted in two conventional drinking water treatment plants

(DWTPs) to evaluate seasonality in the occurrence of seven pharmaceutical compounds

(acetaminophen, ampicillin, caffeine, naproxen, ofloxacin, sulfamethoxazole, and trimethoprim) in source waters, intermediate treatment steps and finished treated water distributed to consumers. At least one pharmaceutical compound was quantified in 88% of samples collected from untreated source waters, 80% of intermediate treatment step samples, and 64% of finished drinking water samples. Average concentrations in source water samples ranged between 0.12-14.66 µg/L to 0.05 µg/L-7.87 µg/L in finished drinking water. In general, concentrations of pharmaceuticals in finished water samples were lower

vii than source water samples indicating that most of the pharmaceutical compounds are degraded during water treatment, though occurrence in finished water imply incomplete removal during treatment. Concentrations in source and treated drinking water varied seasonally with the highest concentrations recorded during fall, winter and spring seasons while lower concentrations occurred in the summer months. Environmental risk assessment reveal that the studied pharmaceuticals pose medium to high risk to the aquatic organisms. Although occurrence of pharmaceuticals in finished drinking water indicate their ability to persist during water treatment, concentrations measured in finished water pose minimal human health risks.

viii

TABLE OF CONTENTS List of Figures ...... xi List of Tables ...... xiv Acknowledgments...... xvii

Chapter 1: Introduction ...... 1

Background and Justification ...... 1 Dissertation Overview: Statement of Goals and Objectives ...... 4 References ...... 6

Chapter 2: Literature Review ...... 8

Background ...... 8 Sources and pathways of EOCs into aquatic systems ...... 9 EOCs of Interest and their Physicochemical Properties ...... 12 Attenuation of EOCs in Wastewater Treatment Systems ...... 20 Occurrence of EOCs in Groundwater ...... 22 Factors Influencing Fate and Transport of EOCs in Groundwater...... 25 Occurrence of EOCs in Surface Water ...... 28 Factors Influencing Fate and Transport of EOCs in Surface Water ...... 29 Attenuation of EOCs During Drinking Water Treatment ...... 31 Ecological and Human Health Impacts of EOCs ...... 34 Ecological and Human Health Risk Assessments ...... 36 Literature Gaps in the study of EOCs in Natural and Engineered Systems...... 39 Summary Statement and Potential Contributions ...... 41 References ...... 41

Chapter 3: Fate of Pharmaceuticals in a Spray-Irrigation System: From Wastewater to Groundwater ...... 52

Abstract...... 53 Introduction ...... 54 Materials and Methods ...... 57 Study Site Description ...... 57 Sample Collection ...... 61 Wastewater Samples ...... 61 Groundwater Samples ...... 62 Sample Handling and Targeted Analysis of PPCPs ...... 62 Risk Calculations ...... 65 Ecological Risk Calculations ...... 65 Human Health Risk Calculations ...... 66 Results and Discussion ...... 67

Summary of Occurrence in WWTP Influent and Effluent ...... 67 Seasonal Variations in WWTP Influent and Effluent ...... 70 Wastewater Removal Efficiencies ...... 74 Loads of PPCPs Spray-Irrigated at the Living Filter ...... 76 Occurrence and Concentrations in Groundwater ...... 77 Risk Assessments ...... 85 Conclusion ...... 88 References ...... 90

Chapter 4: Occurrence, Concentrations, and Risks of Pharmaceutical Compounds in Private wells in Central Pennsylvania ...... 97

Abstract...... 98 Introduction ...... 99 Materials and Methods ...... 102 Sample Collection ...... 104 Modeling Approach ...... 106 Risk Calculations ...... 110 Results ...... 111 Predicted contaminant travel time and delivery ratios to groundwater ...... 115 Risk Calculations ...... 116 Discussion ...... 117 Human health risk assessment...... 123 Conclusions ...... 124 References ...... 125

Chapter 5: Seasonal variations of emerging organic contaminants (EOCs) in drinking water sources in the Susquehanna River Basin...... 131

Abstract...... 132 Methodology ...... 136 Site Description ...... 136 Sample Processing and Targeted Analysis of EOCs ...... 139 Statistical Analysis ...... 141 Risk calculations ...... 143 Results and Discussion ...... 147 Spatial Variation and Influence of Source Water Type ...... 149 Seasonal Variations ...... 151 Concentration-Discharge (C-Q) Relationships ...... 154 Relationships with Water Quality Indicators ...... 159 Ecological and Human Health Risk Assessment ...... 161 Conclusion ...... 166 References ...... 168

Chapter 6: Occurrence and Seasonal Variations of Pharmaceutical Compounds in Source and Treated Drinking Water ...... 173

Abstract...... 173 Introduction ...... 174 Site Descriptions ...... 177 Sample Collection ...... 179 Targeted Pharmaceutical Analysis ...... 179 Risk calculations ...... 181 Results and Discussion ...... 183 Overview of Occurrence in Drinking Water Sources ...... 183 Occurrence during Drinking Water Treatment ...... 186 Occurrence in Finished Water ...... 195 Seasonal Variations ...... 198 Risk Assessment ...... 204 Conclusion ...... 207 References ...... 209

Chapter 7: Conclusion ...... 215

Summary of Key Findings and Implications ...... 215 Recommendations for Future Research ...... 222 References ...... 224

Appendix A: Supplementary Information- Fate of Pharmaceuticals in a Spray-Irrigation System: From Wastewater to Groundwater Sample Preparation and Analysis ...... 225

Appendix B: Master Well Owner Network (MWON) Volunteer Communication Templates ...... 233

LIST OF FIGURES

Figure 2-1: Sources, pathways, and receptors of PPCPs and other EOCs. Adapted from Stuart et al. (2012)...... 11

Figure 3-1: Schematic of Penn State’s wastewater treatment plant unit operations. Stars indicate where samples were collected for pharmaceutical analysis...... 58

Figure 3-2: Map of the Penn State Living Filter site showing the two irrigation sites (State Gamelands and Astronomy sites), the WWTP, effluent pumping station, monitoring wells and the general groundwater flow direction...... 61

Figure 3-3: Monthly variations in the concentration of PPCPs in WWTP influents and effluents, along with average air temperature. Bars from October 2016 – February 2017 represent mean concentrations and error bars represent maximum and minimum concentrations from the weekly samples collected in Phase I sampling period (October – February)...... 72

Figure 3-4: Seasonal variations in the mean concentration of PPCPs in WWTP influent and effluent. Error bars represent standard deviations in concentrations...... 73

Figure 3-5: Monthly variations in overall removal efficiencies of PPCPs during the sampling period. Negative values indicate that the concentrations in the effluent were higher than in the influent...... 75

Figure 3-6: Daily PPCP loads (g d-1) and corresponding flow rates (millions of liters per day, MLD) to the Living Filter to be spray-irrigated during the study period...... 77

Figure 3-7: Seasonal variations in the mean concentration of PPCPs in WWTP effluent and groundwater. Error bars represent maximum and minimum concentrations...... 85

Figure 4-1. Map of private groundwater well sampling locations and surface water sampling location in the West Branch of the Susquehanna River Basin...... 103

Figure 4-2: Front view of (a) a typical septic seepage bed per regulatory standards in Pennsylvania and (b) contaminant transport model schematic ...... 108

Figure 4-3. Frequencies of detection and maximum concentrations of each compound of interest in groundwater well samples ...... 113

Figure 5-1: Land use maps of the study sites in the Susquehanna River Basin. Sites A and C have two sub-watersheds serving as surface water inputs to the reservoirs...... 137

xii

Figure 5-2: Detection frequencies and mean concentrations of EOC measured >LOD in Phase I (n=78) and Phase II (n=161) sampling periods. Error bars represent standard deviations and stars indicate compounds detected >LOQ only once during Phase II sampling period...... 148

Figure 5-3: Total EOC detection in samples collected in Phase I by percentage forested and developed land use in respective study site watersheds...... 149

Figure 5-4: Detection frequencies >LOD and average concentrations in samples collected in Phase I by source water types (riverine and reservoirs)...... 151

Figure 5-4: Seasonal variations in mean concentrations in samples collected in Phase I by source water types (riverine and reservoirs). Concentrations LOD are reported as LOQ/2...... 152

Figure 5-6: Seasonal variations in detection frequencies >LOD and mean concentrations in samples collected in Phase II. Standard deviations are shown as error bars for compounds detected >LOD in more than one sample...... 154

Figure 5-7: Concentration-discharge (C-Q) relationships for EOCs detected >LOQ plotted with flow duration curve during Phase II sampling period. CVC and CVQ represent the coefficients of variation for EOC concentration and discharge, respectively...... 155

Figure 5-8. Fluxes of dissolved EOCs with discharge in January through March sampling period...... 157

Figure 5-9. Fluxes of dissolved EOCs with discharge in April and May sampling period ...... 158

Figure 5-10. Fluxes of dissolved EOCs with discharge in June sampling period ...... 159

Figure 5-11: % Risk contributions of EOCs to overall risk quotients for fish, Daphnia and algae...... 165

Figure 6-1: Schematic diagram showing drinking water treatment at DWTP A and B. Post filtration and clear well samples are referred to as ‘combined filter effluent (CFE)’ and ‘clear well effluent (CWEFF)’, respectively, while samples collected at the final treatment steps were called ‘finished water’...... 178

Figure 6-2: Concentrations of pharmaceutical compounds >LOD in samples collected at each treatment step in DWTP B. Each open circle dot represents one sample >LOD. Darker circles represent overlapping concentration from multiple samples. Ampicillin was not detected in any sample and is not represented in this figure...... 188

xiii

Figure 6-3: Concentrations of pharmaceutical compounds >LOD in samples collected in source water and after each treatment step and finished drinking water in DWTP A. Each open circle dot represents one sample >LOD. Darker circles represent overlapping concentration from multiple samples. CFE-combined filter effluent; CWEFF-clear well effluent...... 191

Figure 6-4: Concentrations of pharmaceutical compounds >LOD in source and finished water in DWTP A and B...... 197

Figure 6-5: Seasonal variations in pharmaceutical concentrations in source water, intermediate treatment steps and treated finished water in DWTP A. CFE-combined filter effluent; CWEFF-clear well effluent...... 200

Figure 6-6: Seasonal variations in pharmaceutical concentrations in source water, intermediate treatment steps and treated finished water in DWTP B. CFE-combined filter effluent...... 201

Figure 6-7: Risk Quotients for fish, Daphnia, and algae from exposure levels corresponding to average pharmaceutical concentrations in source water...... 205

Figure 6-8: Human Health Risk Quotients for adult population (age>18) from exposure levels corresponding to average pharmaceutical concentrations in finished water...... 206

xiv

LIST OF TABLES

Table 2-1: Physical and chemical properties of non-prescription pharmaceuticals, a personal care product and a metabolite ...... 15

Table 2-2: Physical and chemical properties of human prescription pharmaceutical compounds ...... 17

Table 2-3: Physical and chemical properties of human and veterinary antibiotics ...... 19

Table 2-4: Physical and chemical properties of veterinary antibiotics ...... 20

Table 3-1: Physicochemical properties of selected pharmaceuticals...... 63

Table 3-2: Summary statistics for influent and effluent samples from Penn State's wastewater treatment plant...... 69

Table 3-3: Summary statistics for seasonal variations of wastewater influent and effluent ...... 73

Table 3-4: Summary statistics for loads of PPCPs spray-irrigated at Living Filter ...... 76

Table 3-5: Water quality indicators measured during groundwater sampling ...... 78

Table 3-6: Summary statistics for groundwater samples ...... 81

Table 3-7: Summary statistics variations of PPCPs in groundwater based on land use .. 83

Table 3-8: Summary statistics for seasonal variations of PPCPs in groundwater ...... 84

Table 3-9: EC50 (µg/L) and corresponding PNECs (Shown in parentheses) and calculated risk quotients in WWTP effluent ...... 87

Table 3-10: Parameters used in risk quotient calculations and risk quotient values for groundwater samples ...... 88

Table 4-1: Physicochemical properties of selected pharmaceutical compounds...... 104

Table 4-2: Model parameters used for the contaminant transport calculations between septic tank leach field and groundwater well...... 107

Table 4-3. Silt loam soil properties used in model calculations...... 109

Table 4-4: Summary of pharmaceutical concentrations (µg/L) in groundwater and surface water samples...... 112

Table 4-5: Household source water characteristics and measured concentrations in groundwater and field blank samples ...... 114

Table 4-6. Calculated mean travel and delivery ratio (M/M0) of pharmaceuticals to groundwater ...... 116

Table 4-7: Acceptable daily intakes (ADIs), estimated drinking water equivalent levels (DWELs) and corresponding risk quotients for selected pharmaceuticals of interest. ... 116

Table 5-1: Watershed characteristics of the selected study sites...... 136

Table 5-2: Physicochemical characteristics of selected compounds ...... 141

Table 5-3: EC50 and LC50 (mg/L) used to calculate PNEC (by dividing EC50 by an assessment factor of 1000), for fish, algae and Daphnia ...... 146

Table 5-4: Summary statistics for seasonal variations in riverine and reservoir sources during Phase I (row 1: mean ± standard deviation, row 2: detection frequency)...... 152

Table 5-5: Summary statistics for seasonal variations at Site E during Phase II sampling period...... 153

Table 5-6: Calculated PNEC and seasonal risk quotients (RQ) for fish, algae and Daphnia during phase I sampling period ...... 162

Table 5-7: Calculated PNEC and seasonal risk quotients for fish, algae and Daphnia during phase II sampling period ...... 163

Table 5-8: Predicted concentrations in fish, predicted human dose and human health risk quotients ...... 166

Table 6-1: Physicochemical properties of selected pharmaceuticals...... 181

Table 6-2: EC50 and LC50 (mg/L) used to calculate PNEC (by dividing EC50 by an assessment factor of 1000), for fish, algae and Daphnia; and Acceptable daily intakes (ADI) used to calculate Drinking water equivalent levels (DWELs) for human health risk assessment...... 182

Table 6-3: Concentrations of pharmaceuticals (µg/L) and detection frequencies in samples collected from DWTP A ...... 184

xvi

Table 6-4: Concentrations of pharmaceuticals (µg/L) and detection frequencies in samples collected from DWTP B ...... 185

xvii

ACKNOWLEDGMENTS

My utmost gratitude to God Almighty for His blessings and grace.

I am deeply grateful to my advisor Dr. Heather Gall for her invaluable guidance every step of the way, unwavering support, kindness and generosity. Your passion and dedication as an educator and researcher are an inspiration to me. My heartfelt appreciations go to my dissertation committee, Drs. Rachel Brennan, Herschel Elliott, and

Jack Watson for their helpful insight and contributions towards my research objectives. Dr.

Watson, thank you for all the hours you spent teaching me soil physics and modelling solute transport. Dr. Elliott, thank you for teaching me about the chemistry of water treatment and for your words of wisdom and encouragement. I also received great mentorship and support from Dr. Tamie Veith of the USDA-ARS Pasture Systems and

Watershed Management Research Unit. Thank you so much for your willingness to guide and help towards accomplishing set goals, and for going out of your way. Thank you all so much for guiding me through this journey, asking me questions that honed my critical thinking, and for being approachable and allowing me to learn from you.

This work would not have been possible without the analytical expertise of Dr. Kyle

Elkin of the USDA-ARS Pasture Systems and Watershed Management Research Unit.

Thank you for all your hard work and diligence towards analytical method development and sample analysis. To Bryan Swistock, thank you so much for all your help with field work at the drinking water treatment plants. I am very thankful for all you help with objective 2 (Chapter 4) which would have been challenging without your extension

xviii expertise. For her help with initial site visits and field days, I am also thankful to Amy

Galford.

Thank you, Jeremy Harper, for helping me deploy dataloggers at my sampling sites during frozen conditions. Also, for their help towards field work, I thank Dan Leavy, Zach

Kliueber, Juan Li Zhu (Visiting Scholar), Tulio de Souza, Melanie Norwin, Hongzeng Zhu,

Megan Miller, Shannon Jacobs, and Laura Saleh; former and current undergraduate students who were very enthusiastic to learn about my research project and provide helping hands during long field days. Thank you, Odette Mina, for your support with field work and for accompanying me to field sites. I would like to express my gratitude to Brittany

Ayers, an honors student who collected part of the data presented in Objective 1 (Chapter

3). To the current members of the phenomenal Gall-TV lab group: Katie Hayden, Talia

Leventhal, Lidiia Iavorivska, and Joe Chandler and Marlene Carla Ndoun, thank you all for being a joy to work with and for always offering a helping hand.

My sincere thanks to individuals from the Penn State Office of Physical Plant.

Thank you, John Gaudlip, David Swisher, Joseph Swanderski, for your support and help with coordinating sampling events at the Wastewater treatment plant and the Living Filter.

Thank you, Robert Franks, for all your help with wastewater samples, and Anthony

Caldana and Nick Allison for the groundwater sampling at the Living Filter. I learned a lot working with you.

I am very grateful for the collaborative support from the Pennsylvania American

Water Company without which Objective 3 and 4 would not have been possible. Thank you, John Yamona, for your support. I am also extremely grateful for Nicholas Kapelan,

xix

Emery Yurko, Jeremy Resseguie, Scott Sharp, John Prawzdik, Mike Barger, Ryan

Troutman of the Pennsylvania American Water Company for their help in coordinating field days and granting access to sampling sites. With their help with water treatment plant sampling, I am very grateful for Dave Richie, Scott Roads, Laura Walter, Richard Bitting, and Terry Patrick.

This study was funded by the Pennsylvania Sea Grant and the Pennsylvania State

University Office of Physical Plant (OPP). Other funds came from the Pennsylvania State

University College of Agriculture 2017 Graduate Student Competitive Grants Program. I am very grateful for my fellowship support from the Pennsylvania State University

Department of Agricultural and Biological Engineering.

For being the initial support towards the gift of education, I am forever indebted to

STARS Children Africa. Thank you, Zawadi Africa Education Fund, for opening my first opportunity for higher education. For the support, mentorship and for introducing me to scientific research, I appreciate you Dr. Samuel Darko of Benedict College. And finally, thank you to my family: my mother, Mary and my siblings, Angela, Billy, Irene, and Joshua for their sacrifice, support, patience, endless love and prayers. Thank you, Emmanuel

Mwenje, for the encouragement and moral support.

For the women very close to my heart, who may not have had the opportunities I had, but whose sacrifices and encouragement are the reasons I have made it thus far.

For my mother, Mary Kibuye and my sister, Angela Adhiambo.

And for my son, Manuel.

Chapter 1

Introduction

Background and Justification

Conserving the integrity of global water resources is one of the most vital environmental issues of the 21st century. Currently, water quality in many groundwater and surface water sources is threatened by increased anthropogenic pollution from industrial, agricultural and domestic wastewater thereby increasing the demand for freshwater in adequate quantity and quality. The occurrence of emerging organic contaminants (EOCs) in aquatic systems due to wastewater discharges is one of the existing water quality issues.

Research has shown that EOCs, including but not limited to human and veterinary antibiotics, human anti-inflammatory drugs, stimulants, ingredients in health and beauty products, and agricultural pesticides, once introduced into aquatic systems can persist in the environment resulting in deleterious ecological effects. Some EOCs are endocrine disrupting compounds (EDCs) and can impact reproductive systems in aquatic organisms

(Leet et al., 2011). Other studies have documented toxicity of various EOCs to aquatic organisms (La Farré et al., 2008) and raised concerns regarding the links between the occurrences of antibiotics in the environment and the spread of antibiotic resistance (Scott et al., 2016). Since these aquatic systems are also used as sources for potable water production, there are inherent concerns around the potential impacts of EOCs consumed by humans inadvertently through drinking water.

Wastewater reuse is becoming a common practice worldwide, particularly in regions impacted by water scarcity driven by drought and/or increasing human demand.

Wastewater reuse through irrigation of secondary treated wastewater on cultivated and forested land provides a practical means for sustaining agricultural production while managing limited water resources as irrigated wastewater recharges underlying aquifers.

In domestic set ups, domestic wastewater is reused following septic system treatment whereby septic effluent is discharged in underground leach fields for further treatment before recharging groundwater. In both municipal wastewater treatment plants (WWTPs) and domestic septic effluent reuse systems, underlying soils are employed as biogeochemical filters to process wastewater inputs. Since both WWTPs and septic systems incompletely degrade EOCs such as human pharmaceuticals and personal care products (PPCPs) (Carrara et al., 2008; Hedgespeth et al., 2012), it is therefore essential to evaluate water quality impacts with regards to the occurrence of PPCPs as result of such wastewater reuse practices especially where impacted aquifers are also used as drinking water sources. By assessing long term groundwater impact from these wastewater reuse systems, the embedded questions on understanding the capacity of soils to process and assimilate pollutants of wastewater origin (Keestra et al., 2016) are also addressed.

About 50% of surface water used as drinking water sources in the U.S. are impacted with de facto water reuse, the incidental presence of wastewater effluent in a downstream drinking water source (Rice & Westerhoff, 2014). Consequently, studies have established the ubiquitous occurrence of PPCPs among other EOCs in various surface water sources

(Focazio et al., 2008; Kolpin et al., 2002). Surface water sources are also vulnerable to

3 other diffuse sources of EOCs such as combined sewer overflows (CSOs) and surface runoff from urban and agricultural areas that can result in high loads of EOCs during periods of high precipitation (Benotti & Brownawell, 2007). Though spatio-tempral variations in the occurrence of EOCs in surface water sources have been studied (Kolpin et al., 2002; Reif et al., 2012), limited attention has been given to the temporal variations of EOCs in surface water sources as a function of hydroclimatic factors. Understanding the hydroclimatic influences on EOCs is essential to establish transport characteristics during varying streamflow conditions. Characterizing the fluxes of EOCs in surface water sources used for potable water production is additionally necessary to establish loads of EOCs in drinking water treatment plant (DWTP) influents and understand how source water concentrations influence removal in DWTPs. Furthermore, most studies on emerging contaminants have focused on the fate of various EOCs in WWTPs and impacted surface water sources whereas the temporal variations in the occurrence and removal of EOCs in

DWTPs has received limited investigative attention. Though most EOCs are not regulated water pollutants, there are concerns of probable human health impacts necessitating the need to understand removal of EOCs during drinking water treatment and factors influencing levels in treated drinking water.

Dissertation Overview: Statement of Goals and Objectives

The goal of this study was to evaluate the occurrence, fate and transport of selected emerging organic contaminants in surface and groundwater sources and to understand their removal during wastewater treatment and drinking water treatment systems. Chapter 2 provides a review of the literature relevant to the study. The specific objectives and dissertation organization are discussed in the following narrative.

Objective 1 (Chapter 3 a ): To understand the fate of seven pharmaceutical compounds including non-steroidal anti-inflammatory drugs (acetaminophen and naproxen), a stimulant (caffeine), and human antibiotics (ampicillin, ofloxacin, sulfamethoxazole, and trimethoprim) in a coupled human-natural system: a wastewater treatment plant and impacted groundwater wells at a forested and cropped spray-irrigation site. As a site with a multi-decadal history of frequent wastewater inputs, results of this study provide insight into the long-term sustainability of spray-irrigation activities and potential impacts to underlying groundwater aquifers utilized as a drinking water source.

Measured effluent concentrations are used to conduct ecological risk assessments to evaluate the ecosystem service provided by the Living Filter study site by spray irrigating effluent as opposed to directly discharging to surface water sources. Potential human health risks through drinking water are also assessed using values measured from impacted groundwater that is used as a drinking water source.

aReprinted from Science of the Total Environment, 654, F.A. Kibuye, H.E. Gall, K.R. Elkin, B. Ayers, T.L. Veith, M. Miller, S. Jacob, K.R. Hayden, J.E. Watson, H.A. Elliott, Fate of pharmaceuticals in a spray-irrigation system: From wastewater to groundwater, 197-208, (2019). DOI: 10.1016/j.scitotenv.2018.10.442

Objective 2 (Chapter 4a): To evaluate the occurrence of selected pharmaceutical compounds including non-steroidal anti-inflammatory drugs (acetaminophen and naproxen), a stimulant (caffeine), and human antibiotics (ampicillin, ofloxacin, sulfamethoxazole, and trimethoprim) in private domestic groundwater sources in central

Pennsylvania, U.S. Vadose zone modelling are conducted to understand how a compound’s physicochemical characteristics influence fate and transport and subsequent groundwater impact. Differences in occurrence between groundwater and surface water sources during the same sampling period are also assessed. Measured groundwater concentrations are used to perform a human health risk assessment assuming groundwater was consumed without further treatment.

Objective 3 (Chapter 5): To characterize hydroclimatic and seasonal influences on the fluxes of selected EOCs in surface water drinking water sources. The study also evaluates the spatial and temporal variations of selected EOCs in the Susquehanna River

Basin and the variations in concentrations due to source water types and surrounding land use. Targeted analysis was conducted for 20 EOCs including human prescription drugs, non-steroidal anti-inflammatory drugs, a stimulant and its metabolite, a common ingredient in health and beauty products, veterinary antibiotics and an agricultural pesticide.

Measured concentrations are used to conduct ecological and human health risk assessments.

aReprinted from the Journal of Environmental Quality, F.A. Kibuye, H.E. Gall, K.R. Elkin, B. Swistock, J.E. Watson, T.L. Veith, H. A. Elliott, Occurrence, concentrations, and risks of pharmaceutical compounds in private wells in Central Pennsylvania. 2019. DOI: 10.2134/jeq2018.08.0301

Objective 4 (Chapter 6): To assess seasonal variations in concentrations of selected pharmaceuticals including non-steroidal anti-inflammatory drugs (acetaminophen and naproxen), a stimulant (caffeine), and human antibiotics (ampicillin, ofloxacin, sulfamethoxazole, and trimethoprim) in drinking water sources, intermediate water treatment steps and finished drinking water distributed to consumers. Two conventional drinking water treatment plants employing different treatment schemes were used as study sites. Measured concentrations in source water and finished drinking water are used to conduct ecological and human health risk assessments respectively.

Chapter 7 provides a summary of key findings, conclusions and implications pertaining to each research objective. Directions for future research are also outlined.

References

Benotti, M. J., & Brownawell, B. J. (2007). Distributions of pharmaceuticals in an urban estuary during both dry-and wet-weather conditions. Environmental Science & Technology, 41(16), 5795-5802.

Carrara, C., Ptacek, C. J., Robertson, W. D., Blowes, D. W., Moncur, M. C., Sverko, E. D., & Backus, S. (2008). Fate of pharmaceutical and trace organic compounds in three septic system plumes, Ontario, Canada. Environmental Science & Technology, 42(8), 2805-2811.

Focazio, M. J., Kolpin, D. W., Barnes, K. K., Furlong, E. T., Meyer, M. T., Zaugg, S. D., Barber, L.B., & Thurman, M. E. (2008). A national reconnaissance for pharmaceuticals and other organic wastewater contaminants in the United States—II) Untreated drinking water sources. Science of the Total Environment, 402(2-3), 201-216.

Hedgespeth, M. L., Sapozhnikova, Y., Pennington, P., Clum, A., Fairey, A., & Wirth, E. (2012). Pharmaceuticals and personal care products (PPCPs) in treated wastewater discharges into Charleston Harbor, South Carolina. Science of the Total Environment, 437, 1–9.

Keesstra, S. D., Bouma, J., Wallinga, J., Tittonell, P., Smith, P., Cerdà, A., Montanarella, L., Quinton, J.N., Pachepsky, Y., van der Putten, W.H., & Bardgett, R. D. (2016). The significance of soils and soil science towards realization of the United Nations Sustainable Development Goals. Soil, 2(2), 111-128.

Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D., Barber, L.B. & Buxton, H.T. (2002). Pharmaceuticals, Hormones, and Other Organic Wastewater Contaminants in U. S. Streams, 1999 - 2000: A National Reconnaissance. Environmental Science & Technology, 36(6), 1202–1211.

La Farré, M., Pérez, S., Kantiani, L., & Barceló, D. (2008). Fate and toxicity of emerging pollutants, their metabolites and transformation products in the aquatic environment. TrAC Trends in Analytical Chemistry, 27(11), 991-1007.

Leet, J. K., Gall, H. E., & Sepúlveda, M. S. (2011). A review of studies on androgen and estrogen exposure in fish early life stages: effects on gene and hormonal control of sexual differentiation. Journal of Applied Toxicology, 31, 379-398.

Reif, A. G., Crawford, J. K., Loper, C. A., Proctor, A., Manning, R., & Titler, R. (2012). Occurrence of pharmaceuticals, hormones, and organic wastewater compounds in Pennsylvania waters, 2006-09. US Department of the Interior, US Geological Survey.

Rice, J., & Westerhoff, P. (2014). Spatial and temporal variation in de facto wastewater reuse in drinking water systems across the USA. Environmental Science & Technology, 49(2), 982- 989.

Scott, G. I., Porter, D. E., Norman, R. S., Scott, C. H., Uyaguari-Diaz, M. I., Maruya, K. A., Weisberg, S.B., Fulton, M.H., Wirth, E.F., Moore, J., & Pennington, P. L. (2016). Antibiotics as CECs: an overview of the hazards posed by antibiotics and antibiotic resistance. Frontiers in Marine Science, 3, 24.

Chapter 2

Literature Review

Background

Emerging organic contaminants (EOCs) is a collective term referring to organic contaminants that have been quantified at trace concentrations (ng/L-µg/L) in the environment over the past two decades. EOCs are numerous and diverse hence, they are lumped into classes describing their purpose, use or characteristic for instance: pharmaceutical compounds including antimicrobials, analgesics, human and veterinary antibiotics; ingredients in personal care products; natural and synthetic hormones; agricultural pesticides; among many others. EOCs are universally used in industrial, medicinal, agricultural and domestic settings with the general goal of attaining high quality living standards through improving personal hygiene and health, enhancing animal growth and health, and increasing agricultural yields. When EOCs are introduced in the aquatic environment, they are however deemed as pollutants of concern due to their known or suspected endocrine disrupting characteristics (Leet et al., 2011; Veldhoen et al., 2006), and their potential adverse effects to environmental and human health (Boxall et al., 2012;

Fent et al., 2006).

Most EOCs are not new chemicals; they have likely been present in the environment since each chemical began to be produced, possibly for as long as several decades or longer. Despite their long-term use and likely occurrence in environmental

9 systems, they are classified as ‘emerging’ contaminants. This is because most EOCs have just recently been detected and quantified in various environmental systems owing to the advancements in analytical methods and instrumentation. Additionally, the term

“emerging” is used because in comparison to other pollutants such as heavy metals, little is known about their environmental fate, transport, and impacts to characterize their risks.

Therefore, EOCs lack regulatory standards in wastewater and drinking water treatment systems and there is no requirement for their removal during treatment. Some EOCs such as pharmaceuticals, hormones and pesticides are however included in the United States

Environmental Protection Agency’s (U.S. EPA) Contaminant Candidate List (CCL) 3 and

4 as contaminants expected to occur in public water systems that may need regulation under

Safe Drinking Water Act (SDWA) (https://www.epa.gov/ccl).

Sources and pathways of EOCs into aquatic systems

The occurrence of EOCs in aquatic environments follows a “source–pathway– receptor” model as illustrated in Figure 2-1 (Stuart et al., 2012) whereby receptors can include groundwater and surface waters sources (e.g. lakes, rivers, estuarine waters, and creeks). Based on the sources highlighted in Figure 2-1, the occurrence of EOCs in receiving aquatic systems is subject to influences from surrounding land use and the distribution of sources of both point and non-point origin.

Point source pollution originates from distinct sources whose “inputs into aquatic systems can often be defined in a spatially explicit manner” (Stuart et al., 2012). Discharges from wastewater treatment plants (WWTPs) are considered the most important point

10 source of EOCs into the environment. Incompletely metabolized pharmaceuticals in the body are eliminated via excretion (Aksu & Tunç, 2005; Jjemba, 2006), while topically applied drugs and personal care products are not exposed to metabolic degradation and are washed off during routine bathing and washing (Ternes et al., 2004), in addition to any pharmaceuticals disposed through flushing down the drain (Glassmeyer et al., 2009).

Together all these pharmaceuticals and personal care products (PPCPs) constitute domestic waste stream discharged to WWTPs. Due to inadequate degradation and removal during wastewater treatment (Ternes et al., 2004; Verlicchi et al., 2012), WWTP effluent contain a mixture of dissolved EOCs and their metabolites that are typically discharged as a single point source into receiving aquatic systems.

Combined sewer systems, comprising combined collection system of both domestic sewage and storm water runoff, are significant sources of EOCs into surface water (Phillips et al., 2012) especially when flows exceed the capacity of WWTPs during heavy precipitation events. Thus high loads of EOCs are reported in aquatic systems following combined sewer overflow (CSO) discharges (Benotti & Brownawell, 2007; Phillips et al.,

2012).

Onsite wastewater treatment systems such as septic systems insufficiently remove

PPCPs (Conn, 2006). Effluent from these systems are commonly discharged into underground leach fields for further treatment as wastewater plume percolates through the soil and eventually recharge groundwater (Conn, 2006; Yang et al., 2016). Therefore,

PPCPs not fully degraded can impact underlying groundwater. Due to groundwater

11 influence on surface water systems, aquifers impacted by septic plumes can act as a source of pharmaceuticals to nearby surface water sources (Standley et al., 2008).

1Figure 2-1: Sources, pathways, and receptors of PPCPs and other EOCs. Adapted from Stuart et al. (2012).

Non-point source pollution originates from “poorly defined, diffuse sources that typically occur over broad geographical scales” (Stuart et al., 2012). EOCs from such sources are often mobilized through surface runoff during periods of high precipitation.

About 70% of manufactured antibiotics are for animal use (Reif et al., 2012). In addition

12 to these pharmaceuticals and synthetic growth hormones administered to promote animal health and growth, animals also naturally excrete hormones. Similarly, biosolids, organic matter recycled from sewage, contain EOCs that sorb to solids during wastewater treatment. Therefore, agricultural fields amended with animal manure and biosolids can act as sources of pharmaceuticals and hormones into surface water via surface runoff or can impact groundwater through leaching processes (Kümmerer, 2003; Sabourin et al.,

2009). Wastewater reuse through irrigation of secondary WWTP effluent onto cultivated and forested fields introduces EOCs present in effluent into groundwater aquifers (Ayers et al., 2017; McEachran et al., 2017) and can impact surface water sources via surface runoff from irrigated sites (Pedersen et al., 2005). About 8.1 mg/kg of active pharmaceuticals and personal care products (PPCPs) from municipal solid wastes degrade or dissolve into landfill leachate (Musson & Townsend, 2009) which can be directly washed into surface waters during surface runoff events or leach into groundwater.

EOCs of Interest and their Physicochemical Properties

This study will focus on human antibiotics including ampicillin, ofloxacin, sulfamethoxazole, and trimethoprim; non-steroidal anti-inflammatory drugs: acetaminophen and naproxen; antidiuretic drug metformin; a stimulant, caffeine; and an antimicrobial, triclosan. Together they will be referred to as pharmaceuticals and personal care products (PPCPs). These compounds were selected by the funding agency the

Pennsylvania Sea Grant. They exhibit a wide range of physicochemical properties and are therefore representative of a broader array of compounds. Additionally, acetaminophen,

13 naproxen, sulfamethoxazole, and trimethoprim are ranked among compounds to be prioritized in the monitoring of emerging contaminants in aquatic systems and drinking water based on their occurrence in the environment, ecological effects and potential human health impacts (Kumar & Xagoraraki 2010).

Aside from the compounds requested by the funding agency, eight compounds consisting of human and veterinary antibiotics including: chlortetracycline HCl, tetracycline HCl, oxytetracycline HCl, erythromycin, sulfadiazine, sulfadimethoxine, sulfamethazine and tylosin were targeted for analysis. Additionally, three more compounds consisting of an antihistamine drug called cimetidine, a caffeine metabolite known as theobromine, and a neonicotinoid insecticide thiamethoxam were also targeted for analysis.

These compounds were selected based on standard availability and analytical capabilities, frequency of use, and a hypothesized occurrence in the studied surface water sites.

Acetaminophen and naproxen are non-prescription non-steroidal anti- inflammatory drugs. Acetaminophen (N-(4-hydroxyphenyl) acetamide) commonly known as paracetamol is the most extensively used non-prescription analgesic and antipyretic drug in the U.S. (Bedner & MacCrehan, 2006). It is also applied as an intermediate for other pharmaceuticals, for azo dyes and photographic chemicals production (Tabassum et al.,

2011). When consumed, 2-3% is excreted in the parent compound form (Monteiro &

Boxall, 2010). Naproxen ((2S)-2-(6-methoxynaphthalen-2-yl) propanoic acid) is an anti- inflammatory/analgesic drug that is found in over the counter medications, but sometimes higher doses are prescribed. Less than 1% of naproxen is excreted unchanged (Monteiro &

Boxall, 2010).

Stimulants consisted of caffeine and its metabolite, theobromine. Caffeine (3,7- trimethylpurine-2,6-dione) is an alkaloid occurring in more than 60 plant species and is a stimulant constituent of various beverages (tea, coffee, caffeinated soft drinks and energy drinks); numerous food products (chocolate, dairy desserts, and pastries), and is used in enhancing the effects of some analgesics. The average consumption of caffeine per person in the U.S. is 210 mg/day (Buerge et al., 2003). Caffeine is metabolized in humans into three major metabolites including theobromine, paraxanthine and theophylline (Cornish &

Christman, 1957). Theobromine (3, 7-Dimethylxanthine), is also an alkaloid present in cacao beans.

Triclosan (5-chloro-2-(2,4-dichlorophenoxy) phenol) is an antimicrobial agent present in several personal hygiene products widely used in dental products such as toothpastes. Though triclosan has since been banned by the Food and Drug Administration

(FDA) in the U.S. due to a widespread risk to antimicrobial resistance (McNamara & Levy,

2016), triclosan was previously used in a wide range of over the counter topical products including soaps, lotions, and deodorants with an estimated worldwide annual production of 1500 tons (Bester, 2005).

Physical and chemical characteristics of the described EOCs including non- prescription pharmaceutical compounds, the stimulant caffeine and its metabolite, and triclosan, an antimicrobial ingredient in personal care products are summarized in Table 2-

1Table 2-1: Physical and chemical properties of non-prescription pharmaceuticals, a personal care product and a metaboliteab

Antibiotics used primarily by human applications consisted of ampicillin, ofloxacin, sulfamethoxazole, and trimethoprim. Ampicillin ((2S,5R,6R)-6-[[(2R)-2- amino-2-phenylacetyl]amino]-3,3-dimethyl-7-oxo-4-thia-1-azabicyclo[3.2.0]heptane-2- carboxylic acid) is a penicillin beta lactam used to prevent and treat bacterial infections from gram-positive and gram-negative bacteria in both animals and humans (Serrano,

2005). It was once the most widely used antibiotic in the U.S. (Gunther et al., 1993) because of its low toxicity and a wide range of antimicrobial properties (Pang et al., 2016), however the development of bacterial resistance is a threat to its application. Following human

aSource: http://www.chemspider.com/ b Source: https://pubchem.ncbi.nlm.nih.gov/

16 consumption, about 30-60% of ampicillin is excreted unchanged (Monteiro & Boxall,

2010).

Ofloxacin (5-[(3,4,5-trimethoxyphenyl) methyl] pyrimidine-2,4-diamine) is a fluoroquinolone antibiotic used to treat urinary and respiratory tract infections and is active against both gram-positive and gram-negative bacteria in humans with some veterinary application (Yassine et al., 2017). About 70-98% of ofloxacin is excreted by humans following consumption (Monk & Campoli-Richards 1987).

Trimethoprim (5-[(3,4,5-trimethoxyphenyl) methyl] pyrimidine-2,4-diamine) and sulfamethoxazole (4-amino-N-(5-methyl-1,2-oxazol-3-yl) benzenesulfonamide) are bacteriostatic antibiotics commonly used to treat urinary tract infections. Trimethoprim enhances the effect of sulfonamides and is therefore frequently employed as a potentiator for sulfamethoxazole (Bushby & Hitchings, 1968) in a combination called Bactrim. About

80% and 30% of trimethoprim and sulfamethoxazole, respectively, are excreted upon administration (Verlicchi et al., 2010). The resistance of trimethoprim and many sulfonamides is however spreading rapidly (Sköld, 2001).

Metformin (3-(diaminomethylidene)-1,1-dimethylguanidine) is an oral antidiuretic drug used to treat type II diabetes and is a potential anticancer agent, as it decreases the risk of various cancers (Kasznicki et al., 2014; Leone et al., 2014). Its wide usability makes it one of the most prescribed pharmaceuticals in the world though, metformin is not metabolized by the human body and is 100% excreted (Oosterhuis et al., 2013).

Cimetidine (1-cyano-2-methyl-3-[2-[(5-methyl-1H-imidazol-4-yl) methylsulfanyl] ethyl] guanidine) is a commonly used prescription histamine H2 receptor antagonist that is

17 prescribed to reduce the secretion of gastric acid in patients with ulcers. When consumed, about 70% of cimetidine is excreted unchanged in the urine (Brimblecombe et al., 1975).

Table 2-2 below summarizes the physical and chemical properties of these prescription drugs.

2Table 2-2: Physical and chemical properties of human prescription pharmaceutical compoundsab

Tetracycline HCl ((4S,4aS,5aS,6S,12aR)-4-(dimethylamino)-1,6,10,11,12a- pentahydroxy-6-methyl-3,12-dioxo-4,4a,5,5a-tetrahydrotetracene-2- carboxamide;hydrochloride), oxytetracycline HCl ((4S,4aR,5S,5aR,6S,12aR)-4-

(dimethylamino)-1,5,6,10,11,12a-hexahydroxy-6-methyl-3,12-dioxo-4,4a,5,5a-

a Source: http://www.chemspider.com/ b Source: https://pubchem.ncbi.nlm.nih.gov/

18 tetrahydrotetracene-2-carboxamide), erythromycin

((3R,4S,5S,6R,7R,9R,11R,12R,13S,14R)-6-[(2S,3R,4S,6R)-4-(dimethylamino)-3- hydroxy-6-methyloxan-2-yl]oxy-14-ethyl-7,12,13-trihydroxy-4-[(2R,4R,5S,6S)-5- hydroxy-4-methoxy-4,6-dimethyloxan-2-yl]oxy-3,5,7,9,11,13-hexamethyl- oxacyclotetradecane-2,10-dione), and sulfadiazine (4-amino-N-pyrimidin-2- ylbenzenesulfonamide) are antibiotics of both human and veterinary application.

Tetracycline HCl is used to treat a broad array of bacterial infections as well as upper respiratory infections while oxytetracycline HCl is a broad-spectrum antibiotic used to cure respiratory tract, urinary tract as well as ear, eye and skin infections. About 80-90% of

Tetracycline (HCL) is excreted following human consumption (Monteiro & Boxall, 2010).

Erythromycin is used for a respiratory tract and skin infections and is also used to cure some sexually transmitted infections. When consumed about 12-15% of erythromycin is excreted by humans (Monteiro & Boxall, 2010). Sulfadiazine is a sulfonamide antibiotic that is used for a broad range of infections including respiratory tract and ear infections.

Their physical and chemical properties of these antibiotics are summarized in Table 2-3 below.

a 3Table 2-3: Physical and chemical properties of human and veterinary antibiotics

Sulfamethazine (4-amino-N-(4,6-dimethylpyrimidin-2-yl)benzenesulfonamide), sulfadimethoxine (4-amino-N-(2,6-dimethoxypyrimidin-4-yl)benzenesulfonamide), chlortetracycline HCl ((4S,4aS,6S,12aR)-7-chloro-4-(dimethylamino)-1,6,10,11,12a- pentahydroxy-6-methyl-3,12-dioxo-4,4a,5,5a-tetrahydrotetracene-2- carboxamide;hydrochloride), and tylosin (2-[(4R,5S,6S,7R,9R,11E,13E,15R,16R)-6-

[(2R,3R,4R,5S,6R)-5-[(2S,4R,5S,6S)-4,5-dihydroxy-4,6-dimethyloxan-2-yl]oxy-4-

(dimethylamino)-3-hydroxy-6-methyloxan-2-yl]oxy-16-ethyl-4-hydroxy-15-

[[(2R,3R,4R,5R,6R)-5-hydroxy-3,4-dimethoxy-6-methyloxan-2-yl]oxymethyl]-5,9,13-

a Source: http://www.chemspider.com/ b Source: https://pubchem.ncbi.nlm.nih.gov/

20 trimethyl-2,10-dioxo-1-oxacyclohexadeca-11,13-dien-7-yl]acetaldehyde) are antibiotics used mostly for animals (Kemper 2008) and properties are summarized in Table 2-4.

a 4Table 2-4: Physical and chemical properties of veterinary antibiotics

Attenuation of EOCs in Wastewater Treatment Systems

The potential removal mechanisms of EOCs in WWTPs include sorption to solids, biological and chemical transformation, and volatilization. Thus, EOC removal is controlled by individual physicochemical properties including solubility, biodegradability as indicated with biodegradation constants (Kbiol), hydrophobicity an indication of a compound’s solid-water distribution coefficients (Kd), and volatility all of which vary

a Source: http://www.chemspider.com/ b Source: https://pubchem.ncbi.nlm.nih.gov/

21 vastly between compounds. During biological treatment, attenuation routes can include biodegradation or sorption to solids. Thus, EOCs with high Kbiol and low Kd values are well removed through biological transformation; EOCs with low Kbiol and high Kd values are efficiently removed through sorption to solids; and compounds with low Kbiol and Kd values are neither transformed nor sorbed (Suárez et al., 2008). Highly hydrophilic compounds and those resistant to microbial degradation have minimal removal during activated sludge treatments in WWTPs (Nakada et al., 2006), and can thus be removed through chemical oxidation processes depending on their reactivities (Verlicchi et al. 2010). Highly volatile compounds are primarily removed in aerated treatment steps (Suárez et al., 2008) and can also be removed through sorption or microbial degradation as a function of their physicochemical characteristics.

Removal efficiencies in WWTPs are also influenced by factors including activated sludge characteristics such as age and sludge residence times (Ternes et al., 2004), and seasonality (Hedgespeth et al., 2012). Compounds detected in WWTP influents are dependent on usage by the population served (Vatovec et al., 2016) and seasonal trends in influent concentrations correspond to higher usage of antibiotics and anti-inflammatory drugs in colder months (Golovko et al., 2014; Hedgespeth et al., 2012) and elevated concentrations of seasonal drugs such as anti-histamine medications during the onset of spring allergy season (Vatovec et al., 2016). Removal efficiencies in WWTPs are low during colder seasons (fall, winter and spring) due to lower biodegradation rates as a result of lower temperatures (Golovko et al., 2014; Hedgespeth et al., 2012;Vieno et al., 2005).

Onsite wastewater treatment systems involve pumping of domestic wastewater to a septic tank for organic matter digestion followed by effluent dispersal in underground leach fields for further natural treatment. These are common wastewater disposal mechanisms in rural and urbanizing communities (Yang et al., 2016). Septic tank effluents generally have low water quality as indicated by high alkalinity, total suspended sediments, and carbonaceous biochemical oxygen demand (Garcia et al., 2013). EOCs in domestic effluent can be removed through degradation processes in septic tanks and both sorption to the soil matrix and microbial degradation in leach fields. Wastewater plumes from this dispersal infiltrate to groundwater (Carrara et al., 2008; Conn et al., 2006). In a septic system study, compounds such as acetaminophen, caffeine, codeine, carbamazepine, cotinine, sulfamethoxazole, and trimethoprim were detected in a school septic tank effluent at concentrations >100 µg/L and carbamazepine, sulfamethoxazole, and nicotine were detected after percolation through a 2 m sand vadose zone (Godfrey et al., 2007). Elevated concentrations of PPCPs were also reported in septic tank effluents and receiving groundwater (Carrara et al., 2008). Comparable to WWTPs, PPCPs in septic systems are unique to consumer usage and a compound’s removal is dependent on its (previously discussed) properties and is influenced by hydraulic loading rates (Conn, 2006).

Occurrence of EOCs in Groundwater

Generally, occurrence of EOCs in groundwater is less studied compared to surface water systems. Groundwater quantity and quality is nonetheless dynamic and is often

23 influenced by both long- and short-term climatic conditions, water withdrawal patterns, as well as surrounding land use (García-Galán et al., 2010). Furthermore, groundwater is an important water source that ought to be conserved because it contributes to flow in several surface water sources.

Barnes et al. (2008) conducted the first nationwide study consisting of 47 groundwater sites across 18 states in the U.S. representing a variety of land uses, climatic conditions, and hydrogeology. At least one compound was detected in 81% of the groundwater sites sampled with maximum concentrations <5 µg/L. Detected PPCPs included sulfamethoxazole, triclosan, caffeine, and acetaminophen that were present in

23%, 15%, 13%, and 6% of studied sites respectively, but trimethoprim was not detected in any sample. A second nationwide study on both surface and groundwater used for drinking water production by Focazio et al. (2008) elucidated that groundwater sources were less polluted with EOCs in comparison to surface water sources as majority of EOCs were detected in surface water sources.

When 1231 public drinking water wells located in heavily urbanized and agricultural regions in California were studied, carbamazepine, sulfamethoxazole, acetaminophen, caffeine, codeine, and trimethoprim were the most frequently detected compounds with maximum concentrations <2 µg/L (Fram & Belitz, 2011). Pharmaceutical detections were significantly positively correlated with the percentage of developed/urban land use around wells (Fram & Belitz, 2011). In a statewide study in Pennsylvania (Reif et al., 2012), agricultural land use was found to have low impact on groundwater quality.

From 2006-2009, quarterly groundwater samples were collected from groundwater wells

24 serving livestock and only a nicotine metabolite, cotinine and the antibiotics sulfamethoxazole from human use and veterinary antibiotic tylosin were detected of the analyzed 44 compounds.

Domestic impact on groundwater quality has been studied at onsite wastewater systems. Because septic systems are reported to inadequately remove most PPCPs

(Carrarra et al., 2008; Del Rosario et al., 2014; Godfrey et al., 2007; Yang et al., 2016), a variety of compounds are detected in groundwater sources impacted by onsite wastewater treatment. 20 domestic drinking water wells located on properties served exclusively by septic systems in Cape Cod, Massachusetts were sampled and analyzed for 117 organic wastewater compounds. At least one pharmaceutical compound was detected in 65% of the wells with sulfamethoxazole and carbamazepine detected most frequently, though concentrations of most compounds were <100 ng/L (Schaider et al., 2016). In Nebraska private wells, Verstraeten et al. (2005) reported the most detections for caffeine (47%), acetaminophen (26%), trimethoprim (8%), and sulfamethoxazole (8%) with maximum concentrations of 0.12 µg/L, 0.015 µg/L, 0.58 µg/L, and 0.15 µg/L respectively. Overall, the occurrence of EOCs in private wells is less studied. Because the Environmental

Protection Agency’s (EPA) Safe Drinking Water Act does not regulate private well sources, the responsibility of ensuring safe drinking water falls to the private well owners though few homeowners regularly sample their wells to determine whether their water is meeting the EPA’s primary drinking water standards (Focazio et al., 2006). Since septic systems are considered as hot-spots of EOCs to groundwater (Yang et al, 2016) in both properly functioning and malfunction septic systems (James et al., 2016), more studies are

25 needed to understand the efficiency of removal of EOCs by septic systems and their impacts on groundwater.

In a groundwater aquifer impacted by WWTP effluent irrigation for 20 yrs., compounds such as caffeine, sulfamethoxazole, cotinine, and carbamazepine were detected in water samples at lower concentrations than the irrigated effluent (McEachran et al.,

2017). Ayers et al. (2017) also demonstrated the ability of PPCPs to persist in the vadose zone and impact underlying groundwater following a 40-yr irrigation of secondary effluent in agricultural and forested land. The most detected compounds in the wells were caffeine and ofloxacin. Concentrations were lower in groundwater than in the irrigated effluent, implying that the soil matrix is an effective biogeochemical filter for PPCPs. Though soil is demonstrated as an efficient biogeochemical filter for EOCs, long term performance of soils at wastewater reuse sites should be investigated to ensure the soil’s filtering capacity is not exceeded. Comparing EOCs in groundwater impacted by both long-term irrigation and septic leachate is also necessary to gage the effectiveness of both wastewater disposal mechanisms.

Factors Influencing Fate and Transport of EOCs in Groundwater

The frequency of occurrence of EOCs in groundwater can be influenced by physical factors such as surrounding land use, well location in relation to pollutant sources, and well depth. In a study by Fram & Belitz, (2011) a significant positive correlation of pharmaceutical detections with the percentage of developed/urban land use around wells

26 was observed. A higher frequency of detection of PPCPs was recorded in shallower wells than in deeper ones (Barnes et al.,2008) and those that were <14m deep (Verstraeten et al.,

2005). In private wells impacted with septic waste plumes, higher concentrations and detection frequencies were recorded in wells situated down gradient of the leach beds

(Phillips et al., 2015) and wells closer (< 30 m) to septic systems (Verstraeten et al., 2005).

Groundwater pollution extent can also be influenced by groundwater residence times, pollutant loading rates, and vadose zone characteristics such as organic matter content, clay content, and redox conditions (Lapworth et al., 2012).

Compound characteristics such as hydrophobicity and biodegradability also influence fate and transport of EOCs and their overall impact on groundwater.

Hydrophobicity of a compound influences its ability to sorb into organic matter and clay material in soil. Hydrophilic EOCs with low log KOW values have higher solubility thus have the tendency to contaminate groundwater more than hydrophobic compounds that are less mobile and interact more with soil materials (Del Rosario et al., 2014). For example caffeine is highly hydrophilic (log KOW = -0.07) and is detected at high frequencies in groundwater (Barnes et al., 2008; Fram & Belitz, 2011; James et al., 2016; Verstraeten et al., 2005). EOCs with the lowest mean log KOW values dominate groundwater samples

(McEachran et al., 2017) indicating that such compounds contaminate groundwater more than their hydrophobic counterparts. The transport of ionic compounds are more complex since their fate is dependent on pH conditions (Lapworth et al., 2012). In some cases, highly hydrophobic compounds are detected in groundwater because some compounds are charged at certain soil pH values and therefore can be mobile in soil due to charge repulsion

27 with the soil matrix. For example, ibuprofen (Log KOW = 3.97) is anionic and was more mobile in negatively charged soils resulting in its occurrence in groundwater (Del Rosario et al., 2014).

As EOCs travel through the soil they are subjected to microbial degradation.

Degradation rates vary depending on the degradation half-lives of EOCs in soil

(Grossberger et al., 2014). Highly mobile EOCs with long half-lives can persist in soil and contaminate groundwater easily. However, compounds with short half-lives such as caffeine may not be detected in groundwater despite a high mobility through soil due to a fast degradation rate (Phillips et al., 2015). High loading rates can however offset fast biodegradation rates resulting in detection of easily degradable EOCs in groundwater

(McEachran et al., 2017). Redox conditions of the soil and changes in the wastewater plumes can also influence biodegradation rates (Carrara et al., 2008). Some compounds are highly degraded in oxic conditions than in anoxic conditions. Biotransformation of caffeine in sediments was significantly higher in oxic conditions than under anoxic conditions

(Bradley et al., 2007). Several estrogenic compounds including 17β-estradiol, estrone, nonylphenol, and diethoxycarboxylates, and caffeine were monitored in septic systems and down gradient groundwater. Findings indicated that these compounds are favorably removed in oxic conditions since the highest concentrations were recorded in the anoxic/suboxic portions of the wastewater plume while lowest concentrations were in the more oxic shallow wells (Swartz et al., 2006). Similar findings were recorded by Carrara et al. (2008) where naproxen and ibuprofen was highly transported in anoxic conditions, but degraded in oxic conditions.

Occurrence of EOCs in Surface Water

In 2002, the first large-scale investigation of EOCs in U.S. streams measured various classes of EOCs in 139 streams selected due to their proximity to potential sources

(Kolpin et al., 2002). Collected samples were tested for a diverse list of EOCs including pharmaceuticals, hormones, plasticizers, poly aromatic hydrocarbons, pesticides and other wastewater organic compounds, with 80% of sampled sites testing positive for at least one

EOC. Caffeine and triclosan were among the most frequently detected compounds.

Trimethoprim, acetaminophen, and sulfamethoxazole were also detected but in less than

25% of the sites. Non-prescription drugs were present at higher concentrations of up to 10

µg/L in comparison to prescription drugs that were < 2 µg/L.

A second nationwide study investigated the occurrences of various EOCs in both groundwater and surface water drinking water sources (Focazio et al., 2008). There was a higher detection frequency in the surface water sources than groundwater sources. Top five highest detections in surface water sites were steroids, non-prescription drugs, fragrances, antibiotics, and pesticides. In a study comparing surface water types, Wang et al. (2011) indicated that concentrations of detected compounds were higher in river water sources as opposed to lakes and reservoirs. This can be attributed longer residence times in reservoirs and lakes that permit for more environmental attenuation processes such as sorption to sediments and biodegradation (Glassmeyer et al., 2017).

A statewide study in Pennsylvania (PA) collected data on the occurrence of a wide range of EOCs in PA waters. Samples were collected from a range of study sites that included streams used as drinking water sources. A wide range of EOCs was detected in

29 selected 27 streams, but pharmaceuticals were the most frequently quantified class. The most frequently detected pharmaceuticals were caffeine (75%), sulfamethoxazole (40%), acetaminophen (25%), carbamazepine (20%), trimethoprim (8%), and ofloxacin (2%).

Factors Influencing Fate and Transport of EOCs in Surface Water

The occurrence of EOCs in surface water sources is highly influenced by surrounding land use. Reif et al. (2012) reported fewer detection of EOCs in highly forested watersheds and detections increased with decreasing forested land use within a watershed.

Human PPCPs are dominant in surface waters impacted with WWTP effluent especially at sites downstream of discharge locations, while agricultural EOCs such as pesticides and veterinary antibiotics are more associated with high agricultural land uses within a watershed (Fairbairn et al., 2016).

In stream concentrations are further observed to exhibit seasonal variations explained by regional seasonal trends that reflect seasonal use of PPCPs (Yu et al., 2013) and removal extent in WWTPs (Hedgespeth et al., 2012) along with seasonal agricultural application of pesticides, manure, and biosolids (Bernot et al., 2013). Higher concentrations of PPCPs in surface water bodies are observed in colder months (Mu et al.,

2017; Vieno et al., 2005; Wang et al., 2010) as opposed to warmer months, while agricultural pollutants are dominant in warmer seasons of the year following increased application and surface runoff events (Bernot et al., 2013; Gómez et al., 2012).

Observed seasonal trends can further be due to seasonally influenced attenuation processes in surface water. EOCs in surface water are subject to environmental attenuation through photodegradation processes. As much as photodegradation processes are influenced by a compound’s susceptibility to photolysis, seasonal factors such as sunlight irradiation, lengths of day, cloud cover, and surface water turbidity influence photodegradation rates. Lower concentrations of EOCs in surface water systems in summer months (Azzouz & Ballesteros, 2013; Padhye et al., 2014; Vieno et al., 2005) are attributed to increased photolysis. Summer months are also associated with high temperatures which can increase other attenuation processes such as biodegradation (Kunkel & Radke, 2008), resulting in lower aqueous concentrations. Dissolved concentrations of EOCs can also be reduced when hydrophobic compounds sorb to suspended sediments in water, a process that is also influenced by aquatic temperatures (ten Hulscher, 1996).

Seasonal variations in surface water can further be attributed seasonally variant hydroclimatic patterns. Streamflow conditions can influence concentrations of EOCs in surface water. Dilution processes occurring during periods of prolonged rainfall result in high stream discharge and water levels that can result in lower concentrations of EOCs

(Reif et al., 2012; Vieno et al., 2005; Wang et al., 2010). Precipitation events can however offset dilution patterns when pollutants are washed off from various point and non-point sources to increase loads of EOCs in surface water (Benotti & Brownawell 2007; Bernot et al., 2013; Gómez et al., 2012). Depending on upstream distribution of potential sources and the intensity of precipitation events, EOCs are bound to depict different mobilization characteristics in surface water. Pollutant mobilization as a function of streamflow

31 conditions and the subsequent impact on water quality is widely studied for conventional pollutants such as suspended sediments, nutrients, biological oxygen demand (BOD), chemical oxygen demand (COD), and microbiological impairment (e.g. cryptosporidium, and Escherichia coli) (Ashley et al., 2003; Gasperi et al., 2010; Gromaire et al., 2001;

Madoux-Humery et al., 2013). However, few studies exist on wet weather transport characteristics of EOCs in surface water. This is important for surface water sources that are used for potable water production since drinking water intakes are vulnerable to fluctuations in source water quality and can inform plant operational practices to control concentrations of EOCs in finished drinking water.

Attenuation of EOCs During Drinking Water Treatment

Conventional water treatment process consists of coagulation, flocculation, sedimentation, filtration, and disinfection. Some plants also employ advanced treatments such as ozone oxidation and membrane filtration processes such as reverse osmosis, ultrafiltration, nanofiltration, and microfiltration. EOCs can be removed through various mechanisms including settling during sedimentation processes; sorption to flocs, filter media, and activated carbons; and chemical oxidation processes.

During coagulation and flocculation treatment, dissolved hydrophobic EOCs can sorb to suspended sediments and natural organic matter (Adams et al., 2002), and ionic compounds can have electrostatic interactions with floc particles (Vieno et al., 2007).

Therefore, EOCs can be simultaneously removed with suspended sediments and colloids during settling processes in sedimentation tanks. Similarly, EOCs sorbed to floc particles

32 can be removed concurrently with floc during filtration processes. Removal during filtration is higher for hydrophobic compounds as opposed to highly polar compounds

(Kim et al., 2007).

Advanced treatment processes including membrane filtration processes such as ultrafiltration, microfiltration, reverse osmosis, and nanofiltration are also used in some treatment plants. These treatment techniques remove pollutants through steric hindrances, electrostatic interactions, hydrophobic interactions, and membrane adsorption and are influenced by EOC physicochemical properties including molecular size and width, charge, polarity, diffusivity, and hydrophobicity. Membrane properties such as permeability, porosity, surface charge, and operating conditions such as flow rate can also influence contaminant removal (Bellona et al., 2004; Kimura et al., 2003; Snyder et al.,

2003).

Sorption entails the transfer of EOCs from aqueous phase to the solid media, thus sorption is controlled by adsorbent-adsorbate interactions such as hydrogen and covalent bonding; dipole–dipole interactions; electrostatic and van der Waals forces; and hydrophobic interactions (Aksu & Tunç, 2005; Huerta-Fontela et al., 2011; Snyder et al.,

2003). Sorption is the main removal mechanism in granular activated carbon (GAC) filters and powdered activated carbon (PAC) treatment and is controlled by hydrophobic interactions between the sorbent and solute, thus a compound’s KOW can be used a predictor of its removal in activated carbon systems. Higher sorption rates are predicted for compounds with high KOW (Westerhoff et al., 2005). Sorption extent is also influenced by

33 activated carbon intrinsic properties such as surface area, pore distribution, surface functional groups, and adsorbent concentration (Huerta-fontela et al., 2011).

Oxidation processes including permanganate oxidation, chlorination, and chloroamination are important in removal of hydrophilic EOCs that fail to sorb to suspended sediments and filter media (de Jesus Gaffney et al., 2015). Chlorine mostly reacts through electrophilic substitution and addition and can also selectively react with electron rich bonds in organic compounds (Aga, 2007). The efficacy of chlorination on

EOCs is influenced by free available chlorine species (HOCl/OCl-) at various pHs, the speciation characteristics of EOCs (neutral, cationic, or anionic) (Qiang et al., 2006), as well as functional groups existing on their structures (Rivera-Utrilla et al., 2013).

In comparison to WWTPs, limited studies have focused on the fate of EOCs in drinking water treatment plants (DWTPs) with the first nationwide reconnaissance study in U.S. DWTPs conducted recently by Glassmeyer et al. (2017). The efficiencies of water treatment unit processes on the removal of various EOCs are commonly evaluated in simulated and pilot scale studies (Broséus et al., 2009; Li et al., 2017; Ma et al., 2017;

Rozas et al., 2017; Westerhoff et al., 2005). However, these studies might not capture the effect of environmental factors such as the occurrence of complex mixtures of EOCs in source water as well as concentration fluxes in source water due to seasonal and hydroclimatic influences. In a full scale DWTP evaluation of removal of EOCs, Padhye et al.

(2014) pointed out that most compounds studied in a full scale DWTP did not follow their reactivity trends reported in bench scale studies possibly due to more complex conditions in DWTPs that are generally not accounted for experimentally.

Removal efficiencies of EOCs in DWTPs can vary seasonally due to temperature influences on chemical oxidation rates and photolytic degradation in outdoor coagulation, flocculation and sedimentation basins. Though seasonal variations in drinking water treatment for EOCs is understudied, Azzouz & Ballesteros, (2013) indicate that the removal extent during potassium permanganate (KMnO4) oxidation and alum aided coagulation treatment step in a full scale DWTP were three times greater in the hottest period than in the coldest period. Similarly, more compounds were detected after chlorine oxidation in the fall and winter periods in comparison to summer months likely due to higher reactivities in the summer. Though bench scale studies suggest that the removal efficiency of EOCs during water treatment is independent initial concentrations

(Westerhoff et al., 2005), observed seasonal variations during drinking water treatment may reflect seasonal variations in source waters that are used for potable water production.

However, studies on how removal extent of EOCs in full scale DWTPs vary as function of source water concentrations are lacking.

Ecological and Human Health Impacts of EOCs

The occurrence of EOCs in aquatic systems raise concerns about potential ecological and human health impacts. Pharmaceutical compounds remain active even at trace concentrations and may evoke various responses (Jjemba, 2006). Inherent or acquired resistances in microbial and bacterial species developed through the occurrence of pharmaceuticals in the environment is a public health problem (Boxall et al., 2012) because it can lead to the development of treatment-resistant illnesses. Furthermore, the presence

35 of a mixture of pharmaceuticals in one sample at a point in time, fundamentally suggests potential interactive effects from such mixtures that may be synergistic in nature. Some

EOCs are endocrine disrupting compounds (EDCs) and can change hormonal signaling in organisms, causing direct effects by inhibiting growth and maturation of tissues in mammalian development or indirect effects, such as impacting the reproductive cycle or resulting in the development secondary sex characteristics (Mantovani et al., 1999). Other potential effects of EDCs include interfering with the nervous systems, metabolism, and even carcinogenic impacts (Colborn et al., 1993).

The occurrence of EOCs in the environment can contribute to loss of biodiversity.

Oaks et al. (2004) linked veterinary diclofenac to the decline of oriental white-backed vultures in Pakistan and India. Other ecotoxicological effects of EOCs that could lead to a decline in organismal population include: decline in the hatching of eggs of birds, fishes and turtles; interference with the reproductive system of fishes, reptiles, birds, and mammals; and alterations in the immunologic system of marine mammals (Esplugas et al.,

2007). Triclosan and triclocarban, common compounds in personal care products, are capable of hindering algal growth (Liu & Wong, 2013). Furthermore, triclosan is an EDC that affect thyroid hormone mediated gene expression in the North American anuran, Rana catesbeiana tadpoles (Veldhoen et al., 2006).

Humans are primarily exposed to EOCs discharged in aquatic systems through drinking water, though exposure can also occur from other environmental sources such as agricultural products as well as from fish consumption. The World Health Organization

(WHO) (2012) has concluded adverse human health effects are unlikely following chronic

36 exposure through drinking water. However, some studies have indicated potential side effects such as development of allergic reactions to beta-lactams, such as penicillin G and methicillin (Kümmerer, 2009). Furthermore, Pomati et al. (2006) observed growth inhibition effects on human embryonic cells exposed to a mixture of pharmaceuticals at trace, environmentally relevant, doses.

The impacts of low concentration PPCPs in aquatic systems are largely documented for non-target species through risk assessments. However, human health related impacts resulting from inadvertent consumption of low concentrations PPCPs in drinking water are not well established. Preliminary risk assessment studies suggest no adverse effect on human health, but the mode of action of mixtures in the human body is not yet understood.

Ecological and Human Health Risk Assessments

Most EOCs are not regulated in drinking water, surface water, or wastewater, but to work towards the regulation process, systematic risk assessment is needed for both non- target species and human health. Environmental risk posed by EOCs in aquatic systems is estimated using representative taxa such as algae, Daphnia and fish. A hazard quotient

(HQ), which is the ratio of measured environmental concentrations and no observed effect concentration (NOEC) or the predicted no effect concentrations (PNEC) of EOCs, is used as a measure of risk. In the absence of NOEC values, EC50 or LC50 representing the concentration at which 50% of population exhibits a response and the concentration which is lethal to 50% of the population, respectively, are used. HQ values ≥1 indicate potential environmental risk while values < 1 represents minimal risk (Chen et al., 2016; Ginebreda

37 et al., 2010). To calculate HQ, PNECs are first estimated by dividing the EC50 values by an assessment factor of 1000, which is an uncertainty factor used due to extrapolation of data from other species in the environment, after which HQ values are calculated by dividing measured concentrations in water by the PNECs. This risk assessment technique is however limited as the cumulative effects of mixtures of EOCs on the aquatic organisms are not evaluated. Additionally, the risk evaluation does not assess associated chronic exposure effects.

The most common exposure route for humans from untreated surface water is through the consumption of fish. The concentrations or dose of exposure of fish or other organisms from EOCs in surface water can be estimated as follows (Muñoz et al., 2010):

퐶푓푖푠ℎ = C푊푎푡푒푟 × 퐵퐶퐹푓푖푠ℎ × 퐵푀퐹푓푖푠ℎ (Equation 2-1)

-1 -1 where Cfish is the estimated concentration in fish (µg kg wt), BCF (L kgwt ) is the bioconcentration factora in fish and BMF (dimensionless) is the biomagnificationb factor in fish. A predicted human exposure dose via consumption of fish can therefore be estimated by:

퐶 × 퐼푛푡푎푘푒 퐷표푠푒 (µ푔 푘푔 −1푑−1) = 푓푖푠ℎ 푓푖푠ℎ (Equation 2-2) 푏푤 BW where Cfish is the predicted concentration in fish, intake is the human fish consumption rate and BW is the body weight.

a The ratio of the concentration of a compound in organismal tissues to that in the environment (i.e. surface water) b Tendency of a pollutant to concentrate/increase from one trophic level to the next.

Other human exposure route is through drinking water. The average daily potential dose received by a person through drinking water can be calculated as follows:

퐶 ×퐷푊퐼 퐷표푠푒 (µ푔. 푘푔 −1. 푑−1) = 푒푓푓 (Equation 2-3) 푏푤 퐵푊 where Ceff is the is the concentration of EOC in DWTP effluent water sample, DWI is the drinking water intake (L/d) where age specific values can be applied (de Jesus Gaffney et al., 2015). Accuracy in risk assessment can however be improved by the calculation of risk quotients that are calculated by dividing the concentrations measured in effluent sample by a corresponding drinking water equivalent level (DWEL). Calculation of DWEL incorporates the acceptable daily intake (ADI) of a compound, which represents the level of daily intake that has no adverse health effects in the population (Schwab et al., 2005), consumers body weight, DWI which represents the ingestion rate, and the frequency of exposure (FOE) typically 350 /365 d (de Jesus Gaffney et al., 2015).

퐴퐷퐼 ×퐵푊 퐷푊퐸퐿 (µ푔. 퐿−1푑−1) = (Equation 2-4) 퐷푊퐼×퐹푂퐸

DWELs can subsequently be divided by the concentrations in final DWTP effluent to result in a risk quotient to characterize associated human health risks.

In the absence of ADI data on the effects of PPCPs on human health, lowest therapeutic doses (de Jesus Gaffney et al., 2015) or extrapolated No observable adverse effect levels (NOAEL) from animal data are used. When NOAEL data is not available the lowest observable adverse effect level are used (Blanset et al., 2007). Due to the extrapolation from animal data, uncertainty factors (UF) are used during the calculation of

ADI.

푁푂퐴퐸퐿 (µ푔.푘푔−1.푑) 퐴퐷퐼 (µ푔. 푘푔−1. 푑−1) = (Equation 2-5) 푈퐹1 ×푈퐹2 ×푈퐹3 ×푈퐹4

Literature Gaps in the study of EOCs in Natural and Engineered Systems

The ubiquitous occurrence of EOCs in the environment is of concern considering that they occur as complex mixtures, which can result in synergistic effects on non-target species. One of the most effective ways to manage and reduce contamination of environmental systems is effective contaminant source control. While most studies have focused on understanding occurrence and fate of EOCs in natural and engineered systems, limited scientific investigation has been given to establishing best management practices for reducing the presence of EOCs in the environment. Furthermore, natural and anthropogenic factors influencing spatio-temporal variability of EOCs in contaminant sources and impacted environmental systems remain understudied (Petrie et al., 2015). For instance, few studies focus on the EOC mobilization and transport dynamics in aquatic systems as a function of hydroclimatic influences on streamflow conditions (Benotti &

Brownawell, 2007; Launay et al., 2013; Pailler et al., 2009; Zhang et al., 2016). Studies that do seek to examine EOC variability in aquatic systems over seasons or hydrologic conditions generally conduct minimal sampling campaigns that are likely not fully representative of the range of seasonal and hydrologic variability (Cantwell, 2018; Loraine

& Pettigrove, 2006; Mijangos et al., 2018; Moreno-González et al., 2013).

The sources of EOCs in the environment have been identified to be domestic, agricultural and industrial wastewaters (Kolpin et al., 2002). Therefore, many studies have

40 performed monitoring campaigns for EOCs in various wastewater treatment facilities. In contrast, limited scientific attention has been given to their fate during drinking water treatment. The efficiencies of water treatment unit processes on the removal of various

EOCs have been evaluated in lab and pilot scale studies (Broséus et al., 2009; Li et al.,

2017; Ma et al., 2017; Rozas et al., 2017; Westerhoff et al., 2005). However, fewer studies have holistically assessed the occurrence and removal of EOCs in full scale DWTPs and factors influencing removal efficiencies (Azzouz & Ballesteros, 2013; Glassmeyer et al.,

2017; Padhye et al., 2014; Stackelberg et al., 2007).

In agricultural wastewater reuse sites and domestic onlot wastewater treatment systems, soils are used to filter wastewater inputs that eventually replenish groundwater sources. It is predicted that soils will be required to recycle larger wastewater inputs due to increasing wastewater volumes (Keetsra et al., 2016). Therefore, more studies are needed to understand the capacity of soils to assimilate wastewater inputs containing a wide array of EOCs and other pollutants. Understanding the factors influencing soil’s biogeochemical filtering potential is thus essential for the protection of groundwater sources. Additionally, design recommendations for onlot wastewater systems are necessary to manage groundwater quality impact from persistent EOCs.

The toxicological significance of chronic versus acute low-level dose exposure to non-target organisms in the environment is not well established as the toxicology of most

EOCs is not well understood. Literature gaps further exist on the toxicological impacts of complex mixtures EOCs that non-target organisms are exposed to. These limitations also apply to potential human health impacts that are largely unknown. Furthermore, current

41 risk assessment techniques may underestimate risk since approaches are limited by the non- holistic approach that fails to incorporate the full exposure spectrum such as the frequency and duration of exposure to EOCs, exposure history, and interactions between mixtures of

EOCs that an organism is exposed to i.e. if the risk from the exposure is cumulative or synergistic (Daughton, 2004; Ginebreda et al., 2010; Schwab et al., 2005).

Summary Statement and Potential Contributions

Aquatic systems used as drinking water sources are impacted by either planned or unplanned indirect potable reuse of municipal wastewater. Since most EOCs are incompletely degraded during wastewater treatment, various classes of EOCs have been detected in aquatic systems and negatively impact ecological health and pose potential human health impacts. This dissertation investigates and contributes to the understanding of the occurrence, fate, transport and risk levels of a wide array of EOCs in (i) wastewater and groundwater impacted by beneficial reuse through effluent irrigation, (ii) private groundwater in households relying on septic systems for treatment of domestic wastewater,

(iii) surface water sources used for potable water production that are impacted various point and non-point wastewater inputs, and (iv) untreated and treated drinking water.

References

Adams, C., Wang, Y., Loftin, K., & Meyer, M. (2002). Removal of antibiotics from surface and distilled water in conventional water treatment processes. Journal of Environmental Engineering, 128(3), 253-260.

Aga, D. S. (Ed.). (2007). Fate of pharmaceuticals in the environment and in water treatment systems. CRC Press (217-290)

Aksu, Z., & Tunç, Ö. (2005). Application of biosorption for penicillin G removal: comparison with activated carbon. Process Biochemistry, 40(2), 831-847.

Ashley, R., Crabtree, B., Fraser, A., & Hvitved-Jacobsen, T. (2003). European Research into Sewer Sediments and Associated Pollutants and Processes. Journal of Hydraulic Engineering, 129(4), 267–275.

Ayers, B., Elkin, K., Kibuye, F., & Gall, H. E. (2017). Pharmaceuticals at Penn State’s Living Filter: From Wastewater to Groundwater. American Society of Agricultural and Biological Engineering (ASABE), Paper No. 1700255

Azzouz, A., & Ballesteros, E. (2013). Influence of seasonal climate differences on the pharmaceutical, hormone and personal care product removal efficiency of a drinking water treatment plant. Chemosphere, 93(9), 2046-2054.

Barnes, K. K., Kolpin, D. W., Furlong, E. T., Zaugg, S. D., Meyer, M. T., & Barber, L. B. (2008). A national reconnaissance of pharmaceuticals and other organic wastewater contaminants in the United States—I) Groundwater. Science of the Total Environment, 402(2-3), 192- 200.

Bedner, M., & MacCrehan, W. A. (2006). Transformation of acetaminophen by chlorination produces the toxicants 1, 4-benzoquinone and N-acetyl-p-benzoquinone imine. Environmental Science & Technology, 40(2), 516-522.

Bellona, C., Drewes, J. E., Xu, P., & Amy, G. (2004). Factors affecting the rejection of organic solutes during NF/RO treatment—a literature review. Water Research, 38(12), 2795-2809.

Benotti, M. J., & Brownawell, B. J. (2007). Distributions of pharmaceuticals in an urban estuary during both dry-and wet-weather conditions. Environmental science & technology, 41(16), 5795-5802.

Bernot, M. J., Smith, L., & Frey, J. (2013). Human and veterinary pharmaceutical abundance and transport in a rural central Indiana stream influenced by confined animal feeding operations (CAFOs). Science of the Total Environment, 445, 219-230.

Bester, K. (2005). Fate of triclosan and triclosan-methyl in sewage treatment plants and surface waters. Archives of Environmental Contamination and Toxicology, 49(1), 9-17.

Blanset, D. L., Zhang, J., & Robson, M. G. (2007). Probabilistic estimates of lifetime daily doses from consumption of drinking water containing trace levels of N, N-diethyl-meta- toluamide (DEET), triclosan, or acetaminophen and the associated risk to human health. Human and Ecological Risk Assessment, 13(3), 615-631.

Boxall, A.B., Rudd, M.A., Brooks, B.W., Caldwell, D.J., Choi, K., Hickmann, S., Innes, E., Ostapyk, K., Staveley, J.P., Verslycke, T. & Ankley, G.T., (2012). Pharmaceuticals and

personal care products in the environment: what are the big questions? Environmental Health Perspectives, 120(9), 1221.

Bradley, P. M., Barber, L. B., Kolpin, D. W., McMahon, P. B., & Chapelle, F. H. (2007). Biotransformation of caffeine, cotinine, and nicotine in stream sediments: implications for use as wastewater indicators. Environmental Toxicology and Chemistry, 26(6), 1116-1121.

Brimblecombe, R. W., Duncan, W. A. M., Durant, G. J., Emmett, J. C., Ganellin, C. R., & Parsons, M. E. (1975). Cimetidine—a non-thiourea H2-receptor antagonist. Journal of International Medical Research, 3(2), 86-92.

Broséus, R., Vincent, S., Aboulfadl, K., Daneshvar, a., Sauvé, S., Barbeau, B., & Prévost, M. (2009). Ozone oxidation of pharmaceuticals, endocrine disruptors and pesticides during drinking water treatment. Water Research, 43(18), 4707–4717.

Buerge, I. J., Poiger, T., Müller, M. D., & Buser, H. R. (2003). Caffeine, an anthropogenic marker for wastewater contamination of surface waters. Environmental Science & Technology, 37(4), 691-700.

Bushby, S. R. M., & Hitchings, G. H. (1968). Trimethoprim, a sulphonamide potentiator. British Journal of Pharmacology and Chemotherapy, 33(1), 72-90.

Cantwell, M.G., Katz, D.R., Sullivan J.C., Shapley, D., Lipsomb, J., Epstein, J., Juhl, A.R., Knudson, C., & O’Mullan, G.D. (2018). Spatial patterns of pharmaceuticals and wastewater tracers in the Hudson River Estuary. Water Research, 137, 335-343.

Chen, G., Liu, X., Tartakevosky, D., & Li, M. (2016). Risk assessment of three fluoroquinolone antibiotics in the groundwater recharge system. Ecotoxicology and Environmental Safety, 133, 18-24.

Colborn, T., Vom Saal, F. S., & Soto, A. M. (1993). Developmental effects of endocrine-disrupting chemicals in wildlife and humans. Environmental Health Perspectives, 101(5), 378.

Cornish, H. H., & Christman, A. A. (1957). A study of the metabolism of theobromine, theophylline, and caffeine in man. Journal of Biological Chemistry, 228(1), 315-323.

Daughton, C. G. (2004). Non-regulated water contaminants: emerging research. Environmental Impact Assessment Review, 24(7-8), 711-732. de Jesus Gaffney, V., Almeida, C. M., Rodrigues, A., Ferreira, E., Benoliel, M. J., & Cardoso, V. V. (2015). Occurrence of pharmaceuticals in a water supply system and related human health risk assessment. Water Research, 72, 199-208.

Del Rosario, K. L., Mitra, S., Humphrey Jr, C. P., & O'Driscoll, M. A. (2014). Detection of pharmaceuticals and other personal care products in groundwater beneath and adjacent to onsite wastewater treatment systems in a coastal plain shallow aquifer. Science of the Total Environment, 487, 216-223.

Esplugas, S., Bila, D. M., Krause, L. G. T., & Dezotti, M. (2007). Ozonation and advanced oxidation technologies to remove endocrine disrupting chemicals (EDCs) and pharmaceuticals and personal care products (PPCPs) in water effluents. Journal of Hazardous Materials, 149(3), 631–642.

Fairbairn, D. J., Karpuzcu, M. E., Arnold, W. A., Barber, B. L., Kaufenberg, E. F., Koskinen, W. C., Novak, P.J., Rice, P.J., & Swackhamer, D. L. (2016). Sources and transport of contaminants of emerging concern: a two-year study of occurrence and spatiotemporal variation in a mixed land use watershed. Science of the Total Environment, 551, 605-613.

Fent, K., Weston, A. A., & Caminada, D. (2006). Ecotoxicology of human pharmaceuticals. Aquatic toxicology, 76(2), 122-159.

Focazio, M. J., Tipton, D., Dunkle Shapiro, S., & Geiger, L. H. (2006). The Chemical Quality of Self‐Supplied Domestic Well Water in the United States. Groundwater Monitoring & Remediation, 26(3), 92-104.

Focazio, M.J., Kolpin, D.W., Barnes, K.K., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Barber, L.B. & Thurman, M.E., 2008. A national reconnaissance for pharmaceuticals and other organic wastewater contaminants in the United States—II) Untreated drinking water sources. Science of the Total Environment, 402(2-3), 201-216

Fram, M. S., & Belitz, K. (2011). Occurrence and concentrations of pharmaceutical compounds in groundwater used for public drinking-water supply in California. Science of the Total Environment, 409(18), 3409–3417.

Garcia, S. N., Clubbs, R. L., Stanley, J. K., Scheffe, B., Yelderman Jr, J. C., & Brooks, B. W. (2013). Comparative analysis of effluent water quality from a municipal treatment plant and two on-site wastewater treatment systems. Chemosphere, 92(1), 38-44.

García-Galán, M. J., Garrido, T., Fraile, J., Ginebreda, A., Díaz-Cruz, M. S., & Barceló, D. (2010). Simultaneous occurrence of nitrates and sulfonamide antibiotics in two ground water bodies of Catalonia (Spain). Journal of Hydrology, 383(1-2), 93-101.

Gasperi, J., Gromaire, M. C., Kafi, M., Moilleron, R., & Chebbo, G. (2010). Contributions of wastewater, runoff and sewer deposit erosion to wet weather pollutant loads in combined

sewer systems. Water Research, 44(20), 5875–5886.

Ginebreda, A., Muñoz, I., de Alda, M. L., Brix, R., López-Doval, J., & Barceló, D. (2010). Environmental risk assessment of pharmaceuticals in rivers: relationships between hazard indexes and aquatic macroinvertebrate diversity indexes in the Llobregat River (NE Spain). Environment International, 36(2), 153-162.

Glassmeyer, S. T., Hinchey, E. K., Boehme, S. E., Daughton, C. G., Ruhoy, I. S., Conerly, O., Daniels, R.L., Lauer, L., McCarthy, M., Nettesheim, T.G., & Sykes, K. (2009). Disposal practices for unwanted residential medications in the United States. Environment International, 35(3), 566-572.

Glassmeyer, S.T., Furlong, E.T., Kolpin, D.W., Batt, A.L., Benson, R., Boone, J.S., Conerly, O., Donohue, M.J., King, D.N., Kostich, M.S. & Mash, H. E. (2017). Nationwide reconnaissance of contaminants of emerging concern in source and treated drinking waters of the United States. Science of the Total Environment, 581, 909-922.

Godfrey, E., Woessner, W. W., & Benotti, M. J. (2007). Pharmaceuticals in on‐site sewage effluent and ground water, western Montana. Groundwater, 45(3), 263-271.

Golovko, O., Kumar, V., Fedorova, G., Randak, T., & Grabic, R. (2014). Seasonal changes in antibiotics, antidepressants/psychiatric drugs, antihistamines and lipid regulators in a wastewater treatment plant. Chemosphere, 111, 418-426.

Gromaire, M. C., Garnaud, S., Saad, M., & Chebbo, G. (2001). Contribution of different sources to the pollution of wet weather flows in combined sewers. Water research, 35(2), 521-533.

Grossberger, A., Hadar, Y., Borch, T., & Chefetz, B. (2014). Biodegradability of pharmaceutical compounds in agricultural soils irrigated with treated wastewater. Environmental Pollution, 185, 168-177.

Gunther, M. R., Mao, J. I. N. G. H. E., & Cohen, M. S. (1993). Oxidant-scavenging activities of ampicillin and sulbactam and their effects on neutrophil functions. Antimicrobial Agents and Chemotherapy, 37(5), 950-956.

Huerta-Fontela, M., Galceran, M. T., & Ventura, F. (2011). Occurrence and removal of pharmaceuticals and hormones through drinking water treatment. Water Research, 45(3), 1432-1442.

James, C. A., Miller-Schulze, J. P., Ultican, S., Gipe, A. D., & Baker, J. E. (2016). Evaluating Contaminants of Emerging Concern as tracers of wastewater from septic systems. Water Research, 101, 241–251.

Jjemba, P. K. (2006). Excretion and ecotoxicity of pharmaceutical and personal care products in the environment. Ecotoxicology and Environmental Safety, 63(1), 113-130.

Kasznicki, J., Sliwinska, A., & Drzewoski, J. (2014). Metformin in cancer prevention and therapy. Annals of Translational Medicine, 2(6).

Keesstra, S. D., Bouma, J., Wallinga, J., Tittonell, P., Smith, P., Cerdà, A., ... & Bardgett, R. D. (2016). The significance of soils and soil science towards realization of the United Nations Sustainable Development Goals. Soil.

Kemper, N. (2008). Veterinary antibiotics in the aquatic and terrestrial environment. Ecological indicators, 8(1), 1-13.

Kim, S. D., Cho, J., Kim, I. S., Vanderford, B. J., & Snyder, S. A. (2007). Occurrence and removal of pharmaceuticals and endocrine disruptors in South Korean surface, drinking, and waste waters. Water Research, 41(5), 1013-1021.

Kimura, K., Amy, G., Drewes, J., & Watanabe, Y. (2003). Adsorption of hydrophobic compounds onto NF/RO membranes: an artifact leading to overestimation of rejection. Journal of Membrane Science, 221(1-2), 89-101.

Kolpin, D. W., Furlong, E. T., Meyer, M. T., Thurman, E. M., Zaugg, S. D., Barber, L. B., & Buxton, H. T. (2002). Pharmaceuticals, hormones, and other organic wastewater contaminants in US streams, 1999− 2000: A national reconnaissance. Environmental Science & Technology, 36(6), 1202-1211.

Kumar, A., & Xagoraraki, I. (2010). Pharmaceuticals, personal care products and endocrine- disrupting chemicals in US surface and finished drinking waters: a proposed ranking system. Science of the Total Environment, 408(23), 5972-5989.

Kümmerer, K. (2003). Significance of antibiotics in the environment. Journal of Antimicrobial Chemotherapy, 52(1), 5-7.

Kümmerer, K. (2009). Antibiotics in the aquatic environment–a review–part I. Chemosphere, 75(4), 417-434. Kunkel, U., & Radke, M. (2008). Biodegradation of acidic pharmaceuticals in bed sediments: insight from a laboratory experiment. Environmental Science & Technology, 42(19), 7273- 7279.

Lapworth, D. J., Baran, N., Stuart, M. E., & Ward, R. S. (2012). Emerging organic contaminants in groundwater: a review of sources, fate and occurrence. Environmental Pollution, 163, 287-303

Launay, M., Dittmer, U., Steinmetz, H., & Kuch, B. (2013). Temporal variations of organic

pollutants levels during storm events in an urban receiving water body. Novatech 2013, 1– 10.

Leone, A., Di Gennaro, E., Bruzzese, F., Avallone, A., & Budillon, A. (2014). New perspective for an old antidiabetic drug: metformin as anticancer agent. In Advances in Nutrition and Cancer (pp. 355-376). Springer, Berlin, Heidelberg.

Li, C., Dong, F., Crittenden, J. C., Luo, F., Chen, X., & Zhao, T. (2017). Kinetics and mechanism of 17 b -estradiol chlorination in a pilot-scale water distribution systems. Chemosphere, 178, 73–79.

Liu, J. L., & Wong, M. H. (2013). Pharmaceuticals and personal care products (PPCPs): a review on environmental contamination in China. Environment International, 59, 208-224.

Loraine, G. A., & Pettigrove, M. E. (2006). Seasonal variations in concentrations of pharmaceuticals and personal care products in drinking water and reclaimed wastewater in southern California. Environmental Science & Technology, 40(3), 687-695.

Ma, L., Li, J., & Xu, L. (2017). Aqueous chlorination of fenamic acids: Kinetic study, transformation products identification and toxicity prediction. Chemosphere, 175, 114- 122.

Madoux-Humery, A. S., Dorner, S., Sauvé, S., Aboulfadl, K., Galarneau, M., Servais, P., & Prévost, M. (2013). Temporal variability of combined sewer overflow contaminants: evaluation of wastewater micropollutants as tracers of fecal contamination. Water Research, 47(13), 4370-4382.

Mantovani, A., Stazi, A. V., Macri, C., Maranghi, F., & Ricciardi, C. (1999). Problems in testing and risk assessment of endocrine disrupting chemicals with regard to developmental toxicology. Chemosphere, 39(8), 1293-1300.

McEachran, A. D., Shea, D., & Nichols, E. G. (2017). Pharmaceuticals in a temperate forest-water reuse system. Science of the Total Environment, 581, 705-714.

McNamara, P. J., & Levy, S. B. (2016). Triclosan: an instructive tale. Antimicrobial agents and chemotherapy, AAC-02105.

Mijangos, L., Ziarrusta, H., Ros, O., Kortazar, L., Fernández, L.A., Olivares, M., Zuloaga, O., Prieto, A., & Etxebarria, N. (2018). Occurrence of emerging pollutants in estuaries of the Basuqe Country: Analysis of sources and distribution, and assessment of the environmental risk. Water Research, 147, 152-163.

Moreno-González, R., Campillo, J. A., García, V., & León, V. M. (2013). Seasonal input of regulated and emerging organic pollutants through surface watercourses to a

Mediterranean coastal lagoon. Chemosphere, 92(3), 247-257.

Monk, J.P. & Campoli-Richards, D.M. Drugs (1987) 33: 346. https://doi.org/10.2165/00003495- 198733040-00003

Monteiro S.C., Boxall A.B. (2010) Occurrence and Fate of Human Pharmaceuticals in the Environment. In: Reviews of Environmental Contamination and Toxicology. Reviews of Environmental Contamination and Toxicology, vol 202. Springer, New York, NY

Mu, R., Shi, H., Adams, C., Eichholz, T., & Ma, Y. (2017). Detection, occurrence, and removal of selected pharmaceuticals in Missouri source and finished drinking waters. Urban Water Journal, 14(7), 704-712.

Muñoz, I., Martínez Bueno, M. J., Agüera, A., & Fernández-Alba, A. R. (2010). Environmental and human health risk assessment of organic micro-pollutants occurring in a Spanish marine fish farm. Environmental Pollution, 158(5), 1809–1816.

Musson, S. E., & Townsend, T. G. (2009). Pharmaceutical compound content of municipal solid waste. Journal of Hazardous Materials, 162(2-3), 730-735.

Nakada, N., Tanishima, T., Shinohara, H., Kiri, K., & Takada, H. (2006). Pharmaceutical chemicals and endocrine disrupters in municipal wastewater in Tokyo and their removal during activated sludge treatment. Water research, 40(17), 3297-3303.

Oaks, J. L., Gilbert, M., Virani, M. Z., Watson, R. T., Meteyer, C. U., Rideout, B. A., Shivaprasad, H.L., Ahmed, S., Chaudhry, M.J.I., Arshad, M., & Mahmood, S. (2004). Diclofenac residues as the cause of vulture population decline in Pakistan. Nature, 427(6975), 630.

Oosterhuis, M., Sacher, F., & ter Laak, T. L. (2013). Prediction of concentration levels of metformin and other high consumption pharmaceuticals in wastewater and regional surface water based on sales data. Science of The Total Environment, 442, 380-388.

Padhye, L. P., Yao, H., Kung'u, F. T., & Huang, C. H. (2014). Year-long evaluation on the occurrence and fate of pharmaceuticals, personal care products, and endocrine disrupting chemicals in an urban drinking water treatment plant. Water Research, 51, 266-276.

Pailler, J.Y., Guignard, C., Meyer, B., Iffly, J. F., Pfister, L., Hoffman, L., & Krein, A. (2009). Behaviour and fluxes of dissolved antibiotics, analgesics, and hormones during flood events in a small heterogenous catchment in the Grand Duchy of Luxembourg. Water, Air and Soil Pollution, 203, 79-98.

Pang, Y., Huang, J., Xi, J., Hu, H., & Zhu, Y. (2016). Effect of ultraviolet irradiation and chlorination on ampicillin-resistant Escherichia coli and its ampicillin resistance gene. Frontiers of Environmental Science & Engineering, 10(3), 522-530.

Pedersen, J. A., Soliman, M., & Suffet, I. H. (2005). Human pharmaceuticals, hormones, and personal care product ingredients in runoff from agricultural fields irrigated with treated wastewater. Journal of Agricultural and Food chemistry, 53(5), 1625-1632.

Petrie, B., Barden, R., & Kasprzyk-Hordern, B. (2015). A review on emerging contaminants in wastewaters and the environment: current knowledge, understudied areas and recommendations for future monitoring. Water research, 72, 3-27.

Phillips, P. J., Chalmers, A. T., Gray, J. L., Kolpin, D. W., Foreman, W. T., & Wall, G. R. (2012). Combined sewer overflows: an environmental source of hormones and wastewater micropollutants. Environmental Science & Technology, 46(10), 5336-5343.

Phillips, P. J., Schubert, C., Argue, D., Fisher, I., Furlong, E. T., Foreman, W., Gray, J., & Chalmers, A. (2015). Concentrations of hormones, pharmaceuticals and other micropollutants in groundwater affected by septic systems in New England and New York. Science of the Total Environment, 512, 43-54.

Pomati, F., Castiglioni, S., Zuccato, E., Fanelli, R., Vigetti, D., Rossetti, C., & Calamari, D. (2006). Effects of a complex mixture of therapeutic drugs at environmental levels on human embryonic cells. Environmental Science & Technology, 40(7), 2442-2447.

Qiang, Z., Macauley, J. J., Mormile, M. R., Surampalli, R., & Adams, C. D. (2006). Treatment of antibiotics and antibiotic resistant bacteria in swine wastewater with free chlorine. Journal of Agricultural and Food Chemistry, 54(21), 8144-8154.

Rivera-Utrilla, J., Sánchez-Polo, M., Ferro-García, M. Á., Prados-Joya, G., & Ocampo-Pérez, R. (2013). Pharmaceuticals as emerging contaminants and their removal from water. A review. Chemosphere, 93(7), 1268-1287.

Rozas, O., Baeza, C., Núñez, K., Rossner, A., Urrutia, R., & Mansilla, H. D. (2017). Organic micropollutants ( OMPs ) oxidation by ozone : Effect of activated carbon on toxicity abatement. Science of the Total Environment, 590–591, 430–439.

Sabourin, L., Beck, A., Duenk, P. W., Kleywegt, S., Lapen, D. R., Li, H., Metcalfe, C.D., Payne, M. Topp, E. (2009). Runoff of pharmaceuticals and personal care products following application of dewatered municipal biosolids to an agricultural field. Science of the Total Environment, 407(16), 4596–4604.

Schaider, L. A., Ackerman, J. M., & Rudel, R. A. (2016). Septic systems as sources of organic wastewater compounds in domestic drinking water wells in a shallow sand and gravel aquifer. Science of the Total Environment, 547, 470–481.

Schwab, B. W., Hayes, E. P., Fiori, J. M., Mastrocco, F. J., Roden, N. M., Cragin, D., Meyerhoff, R.D., Vincent, J.D., & Anderson, P. D. (2005). Human pharmaceuticals in US surface waters: a human health risk assessment. Regulatory Toxicology and Pharmacology, 42(3), 296-312.

Serrano, P. H. (2005). Responsible use of antibiotics in aquaculture (No. 469). Food & Agriculture Org.

Sköld, O. (2001). Resistance to trimethoprim and sulfonamides. Veterinary research, 32(3-4), 261- 273.

Snyder, S. A., Westerhoff, P., Yoon, Y., & Sedlak, D. L. (2003). Pharmaceuticals, personal care products, and endocrine disruptors in water: implications for the water industry. Environmental Engineering Science, 20(5), 449-469.

Stackelberg, P. E., Gibs, J., Furlong, E. T., Meyer, M. T., Zaugg, S. D., & Lippincott, R. L. (2007). Efficiency of conventional drinking-water-treatment processes in removal of pharmaceuticals and other organic compounds. Science of The Total Environment, 377(2– 3), 255–272.

Standley, L. J., Rudel, R. A., Swartz, C. H., Attfield, K. R., Christian, J., Erickson, M., & Brody, J. G. (2008). Wastewater‐contaminated groundwater as a source of endogenous hormones and pharmaceuticals to surface water ecosystems. Environmental Toxicology and Chemistry, 27(12), 2457-2468.

Stuart, M., Lapworth, D., Crane, E., & Hart, A. (2012). Review of risk from potential emerging contaminants in UK groundwater. Science of the Total Environment, 416, 1-21.

Suárez, S., Carballa, M., Omil, F., & Lema, J. M. (2008). How are pharmaceutical and personal care products (PPCPs) removed from urban wastewaters? Reviews in Environmental Science and Bio/Technology, 7(2), 125-138.

Swartz, C. H., Reddy, S., Benotti, M. J., Yin, H., Barber, L. B., Brownawell, B. J., & Rudel, R. A. (2006). Steroid estrogens, nonylphenol ethoxylate metabolites, and other wastewater contaminants in groundwater affected by a residential septic system on Cape Cod, MA. Environmental Science & Technology, 40(16), 4894-4902.

Tabassum, S., Sabir, S., Sulaiman, O., Rafatullah, M., Khan, I., & Hashim, R. (2011). Oxidative degradation of acetaminophen by permanganate in neutral medium-A kinetic and mechanistic pathway. Journal of Dispersion Science and Technology, 32(2), 217-223. ten Hulscher, D. E. (1996). Effect of temperature on sorption coefficients and sorption kinetics of organic micropollutants-a review. Chemosphere, 32, 609-626.

Ternes, T. a., Joss, A., & Siegrist, H. (2004). Peer Reviewed: Scrutinizing Pharmaceuticals and Personal Care Products in Wastewater Treatment. Environmental Science & Technology, 38, 392A–399A.

Vatovec, C., Phillips, P., Van Wagoner, E., Scott, T. M., & Furlong, E. (2016). Investigating dynamic sources of pharmaceuticals: Demographic and seasonal use are more important than down-the-drain disposal in wastewater effluent in a University City setting. Science of the Total Environment, 572, 906-914.

Veldhoen, N., Skirrow, R. C., Osachoff, H., Wigmore, H., Clapson, D. J., Gunderson, M. P., Van Aggelen, G., & Helbing, C. C. (2006). The bactericidal agent triclosan modulates thyroid hormone-associated gene expression and disrupts postembryonic anuran development. Aquatic Toxicology, 80(3), 217-227.

Verlicchi, P., Galletti, A., Petrovic, M., & Barceló, D. (2010). Hospital effluents as a source of emerging pollutants: an overview of micropollutants and sustainable treatment options. Journal of Hydrology, 389(3-4), 416-428.

Verstraeten, I. M., Fetterman, G. S., Meyer, M. T., Bullen, T., & Sebree, S. K. (2005). Use of tracers and isotopes to evaluate vulnerability of water in domestic wells to septic waste. Groundwater Monitoring & Remediation, 25(2), 107-117.

Vieno, N. M., Härkki, H., Tuhkanen, T., & Kronberg, L. (2007). Occurrence of pharmaceuticals in river water and their elimination in a pilot-scale drinking water treatment plant. Environmental Science & Technology, 41(14), 5077-5084.

Vieno, N. M., Tuhkanen, T., & Kronberg, L. (2005). Seasonal variation in the occurrence of pharmaceuticals in effluents from a sewage treatment plant and in the recipient water. Environmental Science & Technology, 39(21), 8220-8226.

Wang, C., Shi, H., Adams, C. D., Gamagedara, S., Stayton, I., Timmons, T., & Ma, Y. (2011). Investigation of pharmaceuticals in Missouri natural and drinking water using high performance liquid chromatography-tandem mass spectrometry. Water Research, 45(4), 1818-1828.

Westerhoff, P., Yoon, Y., Snyder, S., & Wert, E. (2005). Fate of endocrine-disruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environmental Science & Technology, 39(17), 6649-6663.

World Health Organization, 2012. Pharmaceuticals in Drinking-water. WHO Press, Geneva.

Yang, Y. Y., Toor, G. S., Wilson, P. C., & Williams, C. F. (2016). Septic systems as hot-spots of pollutants in the environment: Fate and mass balance of micropollutants in septic drainfields. Science of the Total Environment, 566, 1535-1544.

Yassine, M. H., Rifai, A., Hoteit, M., & Mazellier, P. (2017). Study of the degradation process of ofloxacin with free chlorine by using ESI-LCMSMS: Kinetic study, by-products formation pathways and fragmentation mechanisms. Chemosphere, 189, 46-54.

Yu, Y., Wu, L., & Chang, A. C. (2013). Seasonal variation of endocrine disrupting compounds, pharmaceuticals and personal care products in wastewater treatment plants. Science of the Total Environment, 442, 310-316.

Zhang, S., Pang, S., Wang, P., Wang, C., Han, N., Liu, B., Han, B., Li, Y., & Anim-Larbi, K. (2016). Antibiotic concentration and antibiotic-resistant bacteria in two shallow urban lakes after stormwater event. Environmental Science and Pollution Research, 23(10), 9984-9992.

Chapter 3

Fate of Pharmaceuticals in a Spray-Irrigation System: From Wastewater to Groundwater

Reprinted from Science of the Total Environment, 654, F.A. Kibuye, H.E. Gall, K.R. Elkin, B. Ayers, T.L. Veith, M. Miller, S. Jacob, K.R. Hayden, J.E. Watson, H.A. Elliott, Fate of pharmaceuticals in a spray-irrigation system: From wastewater to groundwater, 197-208, (2018), with permission from Elservier. DOI: 10.1016/j.scitotenv.2018.10.442

Graphical Abstract

Highlights.

• 7 pharmaceuticals were tracked through a wastewater treatment plant and reuse site. • Removal efficiencies through the treatment plant varied by season and by compound. • Pharmaceuticals posed a medium to high risk to aquatic ecosystem health. • Groundwater concentrations were ~ 2 orders of magnitude lower than those in effluent. • Pharmaceutical concentrations in groundwater posed minimal human health risk.

Abstract

Land application of wastewater effluent is beneficial for recharging groundwater aquifers and avoiding direct pollutant discharges to surface waters. However, the fate of non- regulated organic wastewater pollutants, such as pharmaceuticals and personal care products (PPCPs), in such wastewater reuse systems is understudied. Here, a 14-month study (October 2016 through December 2017) was conducted to evaluate the fate and potential risks of seven commonly used PPCPs in a local wastewater treatment plant

(WWTP) and samples were collected from 13 groundwater monitoring wells at a spray- irrigation site where effluent has been spray-irrigated since the early 1980s.

Acetaminophen and trimethoprim were the most frequently detected (93%) PPCPs in

WWTP influent, while in the effluent, caffeine and trimethoprim were detected most frequently (70%). Wastewater treatment generally reduced concentrations of acetaminophen and caffeine by >88%; however, some compounds had low removal or were present at higher concentrations in the effluent compared with influent (e.g. naproxen, sulfamethoxazole, trimethoprim and ofloxacin). Seasonal trends were observed, with higher PPCP concentrations in the WWTP influent and effluent in the winter. Risk calculations conducted on the wastewater effluent suggest that the risk posed by PPCPs that persisted in the effluent are medium to high to aquatic organisms. Detection frequencies of PPCPs were lower in groundwater samples compared to the effluent, with sulfamethoxazole (40%) and caffeine (32%) as the most frequently detected compounds.

Similarly, average concentrations of PPCPs in groundwater were nearly two orders of magnitude lower than concentrations in the effluent. Minimal seasonal influence was

54 observed for groundwater samples. Human health risk assessments indicate that concentrations in groundwater, which is used as a drinking water source, appear to pose minimal risk

Keywords. emerging contaminants, groundwater quality, wastewater irrigation, wastewater treatment, water reuse

Introduction

Land-application of wastewater effluent on cropped and forested fields has been embraced since the 1970s (Tzanakakis et al., 2014). Because effluent is rich in inorganic nitrogen, land-application on cropped and forested land provides a nutrient source, reducing the need for synthetic fertilization while simultaneously providing water for agricultural use (Angelakis et al., 1999; Candela et al., 2007; Haruvy, 1997). Furthermore, land application of wastewater treatment plant (WWTP) effluent reduces the cost of wastewater treatment by leveraging the ability of the soil matrix to act as a biogeochemical filter for wastewater before recharging underlying aquifers (Aiello et al., 2007; Chen et al.,

2006; De Vries, 1972; Ferguson, 1983; Lubello et al., 2004; Vazquez-Montiel et al., 1996).

However, land application of wastewater effluent can inadvertently introduce a range of contaminants that persist through wastewater treatment processes into the environment.

The extent to which soils in these water reuse systems can act as effective filters for various wastewater pollutants has been investigated previously. Candela et al. (2007) conducted a two-year study on a golf course located in a mild Mediterranean climate that received >2100 mm of both wastewater irrigation and precipitation. They reported

55 increased soil salinity in the top 50 cm and overall groundwater hardness. However, no pathogen risk as a result of wastewater irrigation was observed in the aquifer. Heavy metals have also been reported to increase in groundwater following wastewater irrigation but have remained below levels of concern for human health (Ramirez-Fuentes et al., 2002).

Nitrate concentrations in groundwater underlying wastewater irrigation sites have been demonstrated to vary with soil types and depth to underlying aquifer (Tang et al., 2004), duration of irrigation (Ramirez-Fuentes et al., 2002) and synthetic fertilizer application

(Candela et al., 2007).

As wastewater reuse becomes increasingly common worldwide, the presence of pharmaceuticals and personal care products (PPCPs) known to persist in treated wastewater effluent has raised concerns about their potential impacts to ecosystem and human health.

Because PPCPs lack water quality standards, any removal through WWTPs is coincidental rather than deliberate. Many pharmaceuticals persist through wastewater treatment processes in an active form, even if they are partially degraded or biotransformed (Jim et al., 2006; Kosma et al., 2014). Although little data regarding the removal of PPCPs in wastewater reuse systems exist, data collected within a 7.5 m2 garden irrigated by WWTP effluent in Jerez de la Frontera, Spain reported accumulation of PPCPs in the soil profile up to a depth of 150 cm (Biel-Maeso et al., 2018). Consistently, data collected from a municipal wastewater reuse system in a forested setting in North Carolina suggest the system was able to effectively attenuate PPCPs present in the wastewater effluent, with concentrations in the groundwater several orders of magnitude smaller than those in the wastewater effluent and soil (McEachran et al., 2017).

Despite previous studies that have found soil to be an effective biogeochemical filter for PPCPs, the long-term success of land application systems remains uncertain (Akber et al., 2008), particularly for sites with a multi-decadal history of frequent wastewater inputs.

Factors such as the presence of preferential flow paths (Assadian et al., 2005; Gall et al.,

2016) and clogged soil pores (De Vries, 1972) can impact soil-water interactions, hence minimizing pollutant removal. Eventually, under frequent, long-term irrigation, the soil's capacity to act as an effective biogeochemical filter for pollutants may be reduced, leading to increased pollutant levels in groundwater over time (Ramirez-Fuentes et al., 2002). It is thus essential to frequently monitor groundwater impacted by effluent irrigation to establish overall long-term performance of soil as a biogeochemical filter.

Depending on respective physicochemical properties, PPCPs that enter the soil travel through soil horizons at different rates that determine the extent of their impact on groundwater. Factors such as total wastewater loading rates, groundwater residence times, and physical and chemical characteristics, including soil organic matter and clay content, can influence transport of PPCPs through soil and groundwater (Lapworth et al., 2012).

Generally, PPCP occurrence in groundwater is studied less often than occurrence in surface water, with Barnes et al. (2008) providing the first nationwide reconnaissance study in the

United States on the presence of PPCPs in groundwater.

The goal of this study, therefore, was to understand the fate of seven pharmaceuticals

(acetaminophen, ampicillin, caffeine, naproxen, ofloxacin, sulfamethoxazole and trimethoprim) through a WWTP and their subsequent transport to groundwater wells at a spray-irrigation site. The targeted pharmaceutical compounds encompass a wide range of

57 physicochemical properties and are expected to provide representative results for a broader range of PPCPs of interest. The Pennsylvania State University (Penn State) has been spray- irrigating all of its treated wastewater at a 245-ha spray-irrigation field known as the Living

Filter in Central Pennsylvania since the early 1980s, and Penn State uses the aquifer to which the wastewater is irrigated as its drinking water supply. A 14-month study at Penn

State's WWTP and spray-irrigation field to sample wastewater influent, effluent and 13 groundwater monitoring wells at the Living Filter was conducted with the following research objectives: (i) understand the performance of the WWTP for removing pharmaceuticals; (ii) assess the ability of the soil profile at the spray-irrigation field to provide further treatment of the pharmaceuticals that persist in the effluent; and (iii) conduct a preliminary risk assessment to understand potential ecological and human health impacts of the Living Filter groundwater. The results of this research can provide insight into the long-term sustainability of spray-irrigation activities and potential impacts to groundwater aquifers that are recharged by effluent and also serve as drinking water sources.

Materials and Methods

Study Site Description

Wastewater Treatment Plant

The WWTP at Penn State was first installed in 1913 and has expanded over time to accommodate increasing flow rates as the student population and campus size have grown.

It currently has a design capacity of 18.2 million L per day (MLD), with nearly all (97%)

58 of the influent flow from Penn State and a small percentage (~3%) from the surrounding municipality. While most WWTPs have relatively consistent average flow rates throughout the year, Penn State's WWTP experiences significant variability in flow rates, ranging from

5.7–7.6 MLD during fall and spring semesters, and as low as 1.9 MLD when students are off campus during holiday breaks.

The unit operations in the treatment plant consist of initial screening followed by aerated primary settling. The flow then splits into three different treatment steps, with 50% directed through an activated sludge tank, 25% directed to a trickling filter, and the remaining 25% directed through an anoxic tank to promote denitrification (Figure 3-1).

Flow from the trickling filters and anoxic tank are then combined in a secondary aeration tank and eventually combined with activated sludge treatment at the chlorine disinfection contact chamber. The treated effluent is then stored for no longer than 6 h before it is pumped to the spray-irrigation fields at the Living Filter.

2Figure 3-1: Schematic of Penn State’s wastewater treatment plant unit operations. Stars indicate where samples were collected for pharmaceutical analysis.

Wastewater Spray-Irrigation Site

The Living Filter is a 245-ha site to which wastewater effluent is pumped approximately 4 km for spray-irrigation (Figure 3-2). Operation of the site is permitted by the Pennsylvania Department of Environmental Protection (PA DEP), with a maximum irrigation rate of approximately 5 cm ha−1 wk.−1. However, typically only 45–60% of this amount is applied. The site has 177 irrigation laterals (above-ground pipes, as shown in

Figure 3- 2) with >3000 spray heads. Wastewater effluent is sprayed 24 h a day, 7 d a week, on a rotating schedule such that any field is irrigated for 12 h and rested for at least 6.5 d to follow permit requirements.

Hydrogeological characteristics of the site were fully characterized at the beginning of effluent irrigation and are described in detail by Parizek et al. (1967). The study site is underlain by limestone and dolomite bedrock cut by thrust faults that facilitate extensive weathering. Overall, the depth to bedrock at the site varies from <10 to 50 m (O'Driscoll and Parizek, 2003). The soils have been mapped as primarily Hagerstown silty clay loam, a fine, mixed, semiactive, mesic Typic Hapludalf and Hublersburg silt loam, a clayey, illitic, mesic Typic Hapludult.

The regional water table occurs at depths of 30 – 100 m below the surface. Aquifer transmissivity is estimated to be about 1240 m2 d-1 to 1860 m2 d-1. The relatively high transmissivity values and flat water table indicate the potential of free groundwater movement from irrigation sites and other groundwater recharge areas (Parizek et al., 1967).

Groundwater elevation maps are presented in Appendix A Figures A-1 and A-2. The general direction of groundwater flow is shown in Figure 3-2.

The Living Filter’s land use is approximately half forested (State Gamelands site) and half agricultural (Astronomy site), as shown in Figure 3-2. Cropped lands are commonly rotated with primarily wheat (Triticum aestivum), corn silage (Zea mays L.), soybeans (Glycine max), rye (Secale cereale), and sorghum–sudangrass (Sorghum bicolor x S. bicolor var. Sudanese), while the forested areas are comprised of white oak (Quercus alba) and other mixed hard woods (Andrews et al., 2016). During the study period, 7 wells

(W7, G12, G10, W6, W5, W1, and W2) in the forested and 6 wells (P5, P4, F3, P2, P1, and

P3) in the agricultural sites, for a total of 13 wells, were monitored for PPCPs. These wells are also sampled monthly to monitor nitrate as per permit requirements. Most of the wells are directly impacted with effluent irrigation while 5 wells (W2, P4, F3, P1 and P2) are in locations that do not receive direct spray irrigation but are influenced by groundwater movement from irrigated sites (Figure 3-2).

3Figure 3-2: Map of the Penn State Living Filter site showing the two irrigation sites (State Gamelands and Astronomy sites), the WWTP, effluent pumping station, monitoring wells and the general groundwater flow direction.

Sample Collection Wastewater Samples

Samples were collected from the WWTP influent and effluent (Figure 3-1) in two phases: in Phase I, 24-h composite samples were collected weekly from October 2016 through February 2017, while in Phase II, 24-h composite samples were collected monthly from April to December 2017. The influent and effluent concentrations were used to calculate removal efficiencies for the WWTP.

Groundwater Samples

Monthly groundwater samples were collected from October 2016 through October

2017 from each of the 13 monitoring wells at the Living Filter. Samples for PPCP analysis were collected at the same time as the water samples that are analyzed for nitrate under PA

DEP permit requirements. During each sampling event, depth to water was determined in each well prior to purging. The wells were then purged for at least 1 borehole volume.

Water samples were collected when indicator parameters (temperature, pH, conductivity, and dissolved oxygen) measured using calibrated handheld meters were stabilized. All samples were collected in trace-cleaned 250-mL amber glass bottles with polytetrafluoroethylene (PTFE)-lined caps.

Sample Handling and Targeted Analysis of PPCPs

Seven pharmaceutical compounds were targeted for quantitative analysis. These compounds include four antibiotics (ampicillin, sulfamethoxazole, ofloxacin, and trimethoprim), two non-steroidal anti-inflammatory medications (acetaminophen and naproxen), and a stimulant (caffeine). These seven PPCPs exhibit a wide range of physicochemical properties (Table 3-1) and are therefore representative of a broader array of PPCPs.

PPCPs were analyzed using a high-resolution accurate mass (HRAM) Q Exactive

Orbitrap mass spectrometer (ThermoFisher Scientific, Bremen, Germany), interfaced to the chromatography system through a heated electrospray injection (HESI) source. A 500-

µL volume of sample was concentrated to a 20-µLvolume using an inline concentrator

63 column Hypersil Gold aQ 20x2.1 mm 12-µm (ThermoFisher, Sunnyvale, CA). The sample was then injected onto a 100 x 2.1mm 3-µm Hypersil Gold aQ analytical column and eluted using the gradient reported in the Appendix A (Table A-1). Following each injection, the concentrator column, injection port and injection valve were washed with 10 volumes of

100% methanol followed by 5 volumes of methanol buffer (Eluent B) and allowed to equilibrate with aqueous mobile phase (Eluent A) for 5 column volumes prior to the loading of the next sample.

5Table 3-1: Physicochemical properties of selected pharmaceuticals. Pharmaceutical aaChemical aMolar Mass % Half- a a a Compound Formula (g/mol) pKa log KOW log KOC Excreted life in soil Antibiotics bb a Ampicillin C16H19N3O4S 349.40 2.50; 7.30 1.35 2.00* 30-60 1d* cc g Sulfamethoxazole C10H11N3O3S 253.28 1.60; 5.70 0.89 1.86 30 39 d dd h Ofloxacin C18H20FN3O4 361.37 5.97; 9.28 -0.39 4.64 70-98 4.10 y c g Trimethoprim C14H18N4O3 290.32 7.12 0.91 1.88 80 62 d Anti-inflammatory Drugs ee a Acetaminophen C8H9NO2 151.17 9.38 0.46 1.32 5 13-40 d ff b g Naproxen C14H14O4 230.26 4.15 3.18 2.52 < 1 2 d Stimulant h ii j Caffeine C8H10N4O2 194.19 10.40 -0.07 2.87 -- 35 h

The limit of detection (LOD) for all PPCPs was 0.01 g/L (signal:noise ratio ≥ 3) and the limit of quantification (LOQ) was 0.1 g/L (signal:noise ratio ≥ 10) except for ofloxacin with an LOD and LOQ of 0.3 g/L and 3g/L, respectively. In all cases, the

a https://pubchem.ncbi.nlm.nih.gov b Monteiro & Boxall, (2010) c Verlicchi et al., (2010) d Monk & Campoli-Richards, (1987) e Parfit, (1999) fOosterhuis et al., (2013) g Wu et al., (2012) h Walters et al., (2010) i Martínez-Hernández et al., (2016) Ka = Acid dissociation constant; KOW = octanol-water partition coefficient; KOC=organic carbon partition coefficient; *values for amoxicillin used due to lack of data for ampicillin

64 reporting limits of the analytes measured between the LOD and LOQ were determined as one half of the LOQ as stated in EPA Method 301 (USEPA, 2018). For calculations related to the removal efficiencies and performance of the WWTP, concentrations below the LOD were taken as LOD/sqrt(2) (Antweiler & Taylor, 2008; Aruga, 1997; Gall et al., 2014; Mina et al., 2016). More details on the high performance liquid chromatography (HPLC) and mass spectrometer are summarized in Appendix A.

All samples were collected in 250-mL amber glass bottles with polytetrafluoroethylene (PTFE)-lined caps following EPA Method 1694 (USEPA, 2007).

Nine field blank samples were collected and handled using similar protocols as PPCP samples. Caffeine, naproxen trimethoprim and ofloxacin were detected in some field blanks samples and average values were used to censor PPCP samples. To preserve sample integrity and to minimize degradation, all samples were preserved on ice during transportation to the United States Department of Agriculture-Agricultural Research

Service (USDA-ARS) laboratory in University Park, PA and stored at 4°C before processing within 48 h of collection.

Instrument calibration curves for the target compounds were calculated using seven standard solutions over a range of 0.1–500 μg/L. The correlation coefficients (R2) of the calibration curves were ≥0.98. Each batch of samples was analyzed in sequence with a matrix blank, calibration standards, inter-sample blanks and an isotope labeled spike

(caffeine-d3) bracketing every 20 unknowns. Percent recoveries were determined from 12 samples for each PPCP that spanned the calibration range of 0.1–500 μg/L. The percent recoveries ranged between 75 and 95% for all target compounds. The highest recoveries

65 were observed for acetaminophen (90–95%) and caffeine (>95%), followed by trimethoprim (>85%) and sulfamethoxazole (80–85%). The percent recoveries for naproxen and ofloxacin were 75–80% and >75%, respectively.

Risk Calculations

Ecological Risk Calculations

Spray-irrigation activities at the Living Filter began as an alternative to discharging directly into a nearby surface water body (Spring Creek) due to water quality concerns related to the stream's designation as a cold-water trout fishery. To assess the ecosystem service function of the Living Filter in protecting Spring Creek from PPCPs present in the wastewater effluent, an ecological risk assessment was performed in which it was assumed that the levels of PPCPs in the WWTP effluent were impacting a surface water containing representative taxa such as fish, algae and Daphnia. A risk quotient (RQ) was calculated by dividing average pharmaceutical concentrations in effluent during the study period by the corresponding predicted no effect concentrations (PNEC). PNECs were calculated by dividing EC50 (median effective concentration for test species) obtained from literature for each PPCP by an arbitrary uncertainty factor of 1000 (Sanderson et al., 2003). RQ values >1 suggest a potential ecological risk, while values <1 suggest minimal risk.

Human Health Risk Calculations

Groundwater at the Living Filter has the potential of impacting groundwater sources used for drinking water production for Penn State. In this case, advanced drinking water treatment (i.e., nanofiltration and activated carbon) occurs before on-campus distribution; however, in many areas groundwater is frequently used for drinking with little or no treatment. Therefore, a human health risk assessment was performed assuming human exposure levels at average concentrations of PPCPs measured at the Living Filter wells. Risk estimations were calculated for adult populations with a 50th percentile body weight (BW) of 60 kg and daily drinking water intake (DWI) of 2.04 L d−1 for a frequency of exposure (FOE) of 350 d yr−1 (350 d/365 d = 0.96) (de Jesus Gaffney et al., 2015).

Average concentrations measured in groundwater were used as exposure levels while acceptable daily intakes (ADIs), obtained from literature for respective PPCPs, were used as exposure endpoints since ADIs represent daily exposure levels without adverse effects on a population (Schwab et al., 2005). Drinking water equivalent levels (DWELs) were then estimated as shown below (Equation 3-1).

퐴퐷퐼 ×퐵푊 퐷푊퐸퐿 (µ푔 퐿−1푑−1) = (Equation 3-1) 퐷푊퐼×퐹푂퐸

A drinking water risk quotient, which is calculated as the ratio of groundwater concentration to the estimated DWEL, was then used to characterize risk. A quotient greater than unity suggests possible risk from utilizing the groundwater as a drinking water supply.

Results and Discussion

Summary of Occurrence in WWTP Influent and Effluent

Acetaminophen and trimethoprim were the most frequently detected compounds in the WWTP, each present in 93% of influent samples (Table 3-2), while caffeine and sulfamethoxazole were present in 80% and 70% of influent samples, respectively.

Ampicillin, naproxen and ofloxacin were above the LOQ in fewer than half of the influent samples. Effluent samples typically had lower percentages of samples above the LOQ compared to influent samples, which suggested degradation or transformation during wastewater treatment. Trimethoprim and caffeine remained present above the LOQ in the effluent most frequently (70% of effluent samples), with sulfamethoxazole and acetaminophen present above the LOQ in 63% and 57% of effluent samples respectively, while naproxen, ampicillin and ofloxacin remained above the LOQ in less than half of the effluent samples (Table 3-2). Similar detection frequencies in influent and effluent were observed for other wastewater studies, such that acetaminophen, caffeine, sulfamethoxazole and trimethoprim were among the most frequently detected PPCPs in

WWTPs (Hedgespeth et al., 2012; Kosma et al., 2014; Sui et al., 2011).

Influent concentrations of PPCPs spanned several orders of magnitude, with concentrations as low as the LOQ (0.1 g/L) to as high as >4700 g/L for caffeine. Average concentrations in wastewater influent were highest for acetaminophen, caffeine, and naproxen, while the antibiotics sulfamethoxazole, ofloxacin, and trimethoprim had the lowest average concentrations (Table 3-2). Acetaminophen and caffeine were present in

68 influent at average concentrations of 654.6 g/L and 841.32 g/L, respectively, up to eight- fold higher than WWTP values previously reported by Hedgespeth et al. (2012) of 70 –

150 g/L and 40 – 100 g/L for caffeine and acetaminophen, respectively. Kasprzyk-

Hordern et al. (2009) measured acetaminophen in WWTP influents at concentrations ranging between 68 – 483 g/L. In WWTPs in Riverside County, California, acetaminophen and naproxen had the highest mean concentrations with influent concentrations as high as 218 g/L and 210 g/L, respectively (Yu et al., 2013), however, maximum concentrations for naproxen reported in that study were one order of magnitude lower than maximum values in the current study (Table 3-2). Observed concentrations for antibiotics sulfamethoxazole, trimethoprim and ofloxacin in urban and hospital wastewaters (Kasprzyk-Hordern et al., 2010; Kosma et al., 2014; Verlicchi et al., 2010) were generally similar to the current data set, though maximum concentrations in the current dataset were roughly one order of magnitude higher.

Observed variations in influent concentrations can be attributed to varying regional pharmaceutical usage. During the study period, up to about 2000 defined daily doses

(DDD) for naproxen and 400 DDD for trimethoprim and sulfamethoxazole were dispensed at the University Health Services and neighboring pharmacies in a single day, not including prescriptions collected from other pharmacies and over the counter purchases of naproxen and acetaminophen. Ofloxacin was also largely dispensed, but in the form of ear and eye drops. High concentrations observed in influent may thus be attributed to high pharmaceutical usage in a short period of time.

6Table 3-2: Summary statistics for influent and effluent samples from Penn State's wastewater treatment plant. Mean Median Minimum Maximum Standard % n (µg/L) (µg/L) (µg/L) (µg/L) Deviation > LOQ (µg/L) Influent Acetaminophen 654.60 198.66 5.58 3,274.56 1,005.15 93% Ampicillin 283.08 1.02 0.05 3,920.43 1,006.63 43% Caffeine 841.32 68.50 0.05 4,705.52 1,668.04 80% Naproxen 452.17 28.70 0.05 2,063.84 732.05 30% Ofloxacin 100.62 1.50 1.50 2,413.07 454.74 20% Sulfamethoxazole 25.03 0.89 0.05 129.29 43.47 70% Trimethoprim 9.80 2.19 0.16 209.85 39.23 93% Effluent Acetaminophen 5.83 0.20 0.05 112.78 23.38 57% Ampicillin 25.98 0.14 0.05 195.83 58.07 23% Caffeine 23.23 0.40 0.05 258.68 59.76 70% Naproxen 3,765.07 33.68 0.05 34,718.99 9,925.96 43% Ofloxacin 22.15 1.50 1.50 413.75 81.57 17% Sulfamethoxazole 68.30 2.30 0.05 317.47 113.86 63% Trimethoprim 2.03 0.74 0.05 15.14 3.10 70% Total n=30 (Phase 1 n at each sampling point = 20 and Phase II n at each sampling point = 10)

Effluent concentrations generally spanned fewer orders of magnitude than influent concentrations, with maximum concentrations for each PPCP of interest ~ 400 g/L or less, except for naproxen (>30,000 g/L). Mean concentrations in wastewater effluent were also reduced by one to two orders of magnitude relative to mean influent concentrations, such that all mean effluent concentrations, apart from naproxen, were <70 g/L compared to mean influent concentrations <850 g/L. Average concentrations measured in the current study were at least one order of magnitude higher than values reported in literature with concentrations for acetaminophen and caffeine (Hedgespeth et al., 2012; Sui et al., 2011); and sulfamethoxazole and trimethoprim (Golovko et al., 2014; Sui et al., 2011) typically

<1 g/L in effluent. While similar average concentrations for acetaminophen ranging between 0.8 – 13.2 g/L in effluent have been reported (Kasprzyk-Hordern et al., 2009), naproxen concentrations in the current study were up to three orders of magnitude higher

70 than values reported in literature (Kasprzyk-Hordern et al., 2009; Yu et al., 2013). For antibiotics ofloxacin and trimethoprim, average values observed in the current study were within ranges reported in literature (Kasprzyk-Hordern et al., 2009; Radjenović et al.,

2009), however maximum values were around one order of magnitude higher and mean values observed for sulfamethoxazole were at least one order of magnitude higher than those reported by other researchers (Kasprzyk-Hordern et al., 2009; Kosma et al., 2014;

Radjenović et al., 2009). Variation in effluent concentrations can be associated with different WWTP characteristics that can influence PPCP removal during wastewater treatment and the high influent loads observed in the study. WWTP operational factors such as hydraulic residence times, flow rates, sludge age and wastewater characteristics are factors that vary between WWTPs and can influence removal extent reported between studies.

Seasonal Variations in WWTP Influent and Effluent

PPCPs exhibited diverse monthly variations in the WWTP throughout the 14- month sampling period. In general, higher concentrations in both influent and effluent were observed in the colder months (winter and spring) (Figure 3-3, Table 3-3). The anti- inflammatory drug, acetaminophen, and caffeine were frequently detected throughout the sampling period, but were detected at the highest concentrations in the winter, with average concentrations of more than 1500 g/L. While caffeine was not detected in the spring, acetaminophen was detected throughout the sampling period. Hedgespeth et al. (2012) reported comparable findings, with higher concentrations of caffeine and acetaminophen

71 in winter associated with higher consumption in winter and a continued ingestion in other seasons of the year. Similarly, mean concentrations observed in the current study for the antibiotics ampicillin (560.4 g/L) and sulfamethoxazole (60.2 g/L) were highest during winter, whereas ofloxacin (843.2 g/L) and trimethoprim (72.01 g/L) were highest in the spring. The highest average concentration for the analgesic naproxen (>2000 g/L) was also observed in the spring season. Other studies have also reported higher influent concentrations of various pharmaceuticals in the colder months of the year due to increased consumption rates (Golovko et al., 2014; Hedgespeth et al., 2012; Yu et al., 2013).

Overall, PPCPs detected in influent were also present in effluent samples (Figure

3-3). Similar trends were observed in the effluent samples, with average concentrations for sulfamethoxazole (177.9 µg/L), ampicillin (32.7 µg/L) and caffeine (71.17 µg/L) highest in the winter samples, and ofloxacin (166.3 µg/L) and trimethoprim (6.8 µg/L) highest in the spring (Figure 3-4; Table 3-3). Contrary to expectation, concentrations of naproxen

(18,373 µg/L) and acetaminophen (37.6 µg/L) were highest in the summer. Effluent concentrations were expected to be high in the colder seasons and lower in warmer seasons since removal efficiencies in WWTPs are low in colder seasons due to lower biodegradation rates as a result of low temperatures (Vieno et al., 2005) in addition to higher concentrations in influent. Overall, the seasonal trends observed in the current study are consistent with those reported by other wastewater studies (Hedgespeth et al., 2012;

Sui et al., 2011).

4Figure 3-3: Monthly variations in the concentration of PPCPs in WWTP influents and effluents, along with average air temperature. Bars from October 2016 – February 2017 represent mean concentrations and error bars represent maximum and minimum concentrations from the weekly samples collected in Phase I sampling period (October – February).

7Table 3-3: Summary statistics for seasonal variations of wastewater influent and effluent Mean Concentrations (µg/L) Fall Winter Spring Summer Influent Acetaminophen 173.38±138.19 1760.80±1158.13 10.87±5.67 65.02±71.61 Ampicillin 0.08±0.04 560.42±1481.62 46.51±14.59 61.20±12.44 Caffeine 50.68±32.88 2639.23±2144.24

5Figure 3-4: Seasonal variations in the mean concentration of PPCPs in WWTP influent and effluent. Error bars represent standard deviations in concentrations.

Wastewater Removal Efficiencies

The removal efficiencies for each of the PPCPs of interest were calculated on a monthly basis to better understand performance over time (Figure 3-5). Overall, the removal efficiencies varied by compound, with the highest average removal efficiencies

(>80%) observed for acetaminophen and caffeine, while the lowest removal efficiencies

(<50%) were observed for naproxen and sulfamethoxazole. Low removal efficiencies occurred in the winter for caffeine, naproxen, ofloxacin, sulfamethoxazole, and trimethoprim and can likely be attributed to lower biodegradation rates due to reduced wastewater temperature (Vieno et al., 2005). For acetaminophen and ampicillin, however, the lowest removal efficiencies occurred in the spring and summer. In several instances, the concentrations of PPCPs in WWTP effluent were higher than influent concentrations and were reported here as negative removal efficiencies (Figure 3-5). Similar seasonal variations have been reported by other studies (Golovko et al., 2014; Hedgespeth et al.,

2012; Yu et al., 2013).

Overall, the removal efficiency results are consistent with trends observed in the literature. Caffeine and acetaminophen generally have high removal efficiencies (>95%;

Hedgespeth et al., 2012; Kasprzyk-Hordern et al., 2009; Kosma et al., 2014), while instances of negative removal have also been observed in other studies (Carballa et al.,

2004; Göbel et al., 2007; Zorita et al., 2009). Sulfamethoxazole is generally found to have higher effluent concentrations due to the transformation of metabolites back to the parent compound (Göbel et al., 2007). Moderately hydrophobic compounds such as naproxen

(Table 3-1) may sorb to solids leading to an underestimation of influent loads since only

75 aqueous phase concentrations were monitored. Furthermore, effluent concentrations may be higher due to desorption from particles (Zorita et al., 2009). Low to moderate removal efficiencies observed in the current study are consistent with findings by others for naproxen (Carballa et al., 2004; Kasprzyk-Hordern et al., 2009; Nakada et al., 2006) and ofloxacin (Zorita et al., 2009). Low average removals were exhibited by ampicillin (56%) and trimethoprim (48%), however removal rates for both fluctuated from <10% to >95%.

Similar trimethoprim removal efficiencies have been reported in other studies (Golovko et al., 2014; Kosma et al., 2014; Verlicchi et al., 2012).

6Figure 3-5: Monthly variations in overall removal efficiencies of PPCPs during the sampling period. Negative values indicate that the concentrations in the effluent were higher than in the influent.

Loads of PPCPs Spray-Irrigated at the Living Filter

Mass fluxes of PPCPs leaving the WWTP were calculated to estimate the daily quantities of PPCPs spray-irrigated at the Living Filter. Fluxes were calculated using average flow during each 24 h sampling period and corresponding measured concentration in WWTP effluent. The highest loads were observed for naproxen ranging between 27 –

211 kg d-1 (Table 3-4), while average loads for other PPCPs were as low as 17.6 g d-1and as high as 373 g d-1. The trends in the loads reflected both seasonal trends in the wastewater flow as well as trends in the effluent concentrations (Figure 3-6).

8 Table 3-4: Summary statistics for loads of PPCPs spray-irrigated at Living Filter Mean Median Minimum Maximum SD (g/d) (g/d) (g/d) (g/d) (g/d) Acetaminophen 76.4 0.9 0 834.1 239 Ampicillin 255.6 1.1 0.2 1504.7 501.6 Caffeine 81.9 0.9 0.2 1060.2 273.2 Naproxen 27,218 226.8 0.2 210,783 64696 Ofloxacin 304.5 11.1 5.7 3775.8 965 Sulfamethoxazole 373.2 5.6 0.4 2225.6 779.8 Trimethoprim 17.6 4.6 0.2 138.2 35.6

For naproxen, loads were highest during the summer months despite lower flow rates due to highly elevated concentrations in the effluent while for ofloxacin, loads were highest in the spring due to high effluent concentrations (Table 3-4) and the highest observed flow rates during the study period. Caffeine and sulfamethoxazole loads were highest in the winter months due to elevated concentrations in the winter that were accompanied by high flow rates. Loads were higher than those in literature (Radjenović et al., 2009; Zorita et al., 2009) due to higher effluent concentrations in the current dataset.

-1 7Figure 3-6: Daily PPCP loads (g d ) and corresponding flow rates (millions of liters per day, MLD) to the Living Filter to be spray-irrigated during the study period.

Occurrence and Concentrations in Groundwater

Groundwater samples were collected from 13 monitoring wells at the Penn State

Living Filter for a one-year period (October 2016 – October 2017) to examine the effects of effluent irrigation on groundwater. During the sampling period, the depth to the well water levels at the 13 wells ranged from approximately 20 to 92 m below ground level.

Water quality indicators such as water temperature, dissolved oxygen, pH, and specific conductance were also measured and ranged as follows: 7.3 – 18.2 ºC, 0.6 – 11.8 mg/L,

6.6 – 9.5, and 0.15 – 0.93 mS/cm respectively and are provided in Table 3-5.

9Table 3-5: Water quality indicators measured during groundwater sampling Mean Median Minimum Maximum SD pH 7.51 7.59 6.57 9.50 0.44 Dissolved oxygen (mg/L) 7.85 8.60 0.60 11.80 2.77 Conductivity (mS/s) 0.58 0.65 0.15 0.96 0.21 Temperature °C 11.76 11.80 7.30 18.20 2.09 Depth to water (m) 55.07 55.78 19.84 92.51 23.97

During the sampling period, concentrations of each of the targeted PPCPs were detected above the LOQ in at least one groundwater sample. The most frequently detected compounds in the groundwater samples were sulfamethoxazole and caffeine, which were present above the LOQ in 40% and 32% of the groundwater samples, respectively (Table

3-6). Naproxen was quantified in 19% of the groundwater samples, while acetaminophen, ampicillin, trimethoprim, and ofloxacin were above the LOQ in fewer than 15% of the groundwater samples. Caffeine and sulfamethoxazole are also among the most frequently detected PPCPs in other groundwater studies (Ayers et al., 2017; Barnes et al., 2008; Fram

& Belitz, 2011; McEachran et al., 2017; Vulliet & Cren-Olivé, 2011) and similar low detection frequencies for trimethoprim, ofloxacin and naproxen have been observed by

Vulliet & Cren-Olivé (2011).The detection of PPCPs in groundwater is a clear indication of wastewater contribution to groundwater at the site over the 40-year irrigation period.

However, the groundwater samples typically had lower percentages of samples above the

LOQ compared to the WWTP effluent and were generally one to two orders of magnitude lower in concentration compared to the WWTP effluent, suggesting at least some removal of the PPCPs via degradation (Carrara et al., 2008) or sorption to the soil (Lapworth et al.,

2012) that occurs as the compounds travel through the soil (Tables 3-2 and 3-6).

Furthermore, uptake by plants may be another removal pathway of PPCPs in reclaimed water (Franklin et al., 2016).

Some PPCPs are hydrophilic and/or have long biodegradation half-life in soil and thus may have lower attenuation rates in soil, making them more likely to reach groundwater (Del Rosario et al., 2014; McEachran et al., 2017). The most frequently detected compounds in the groundwater samples were sulfamethoxazole, a moderately persistent antibiotic with a soil half-life of 39 d (Wu et al., 2012) and caffeine, a highly hydrophilic compound that likely moves readily with groundwater. Additionally, the extent of groundwater pollution by PPCPs in a wastewater reuse site can be controlled by other factors such as total wastewater loading rates, soil chemical conditions, organic matter content, clay content and groundwater residence time (Lapworth et al., 2012). Processes such as desorption of previously sorbed PPCPs back into the aqueous phase and ionic interaction between PPCPs and soil can also increase the likelihood of leaching to groundwater (Lapworth et al., 2012). Previous unsaturated zone studies at the Living Filter have reported the accumulation of sulfamethoxazole, ofloxacin and trimethoprim in irrigated soils (Franklin et al., 2018) and estrogens in both irrigated and non-irrigated soils

(Woodward et al., 2014). Besides sorption to soil materials and degradation processes, concentrations in groundwater may also be reduced through dilution processes from precipitation and potential surface water influx from neighboring surface water sources

(Figure 3-2).

Groundwater concentration ranges measured in the current study were consistent to those in other groundwater studies. Median concentrations for acetaminophen, caffeine,

80 sulfamethoxazole and trimethoprim were similar to those reported by Fram and Belitz

(2011) from groundwater used for public drinking water in California, however maximum concentrations in the present study were one order of magnitude higher. Average concentration for naproxen (37.7 μg/L) and ofloxacin (13.53 μg/L) were higher than values reported in literature, with concentrations for naproxen (McEachran et al., 2017; Vulliet and Cren-Olivé, 2011) and ofloxacin (Teijon et al., 2010) typically <1 μg/L in literature.

Higher concentrations in the present study can be attributed to the potential high loading rates of the studied pharmaceuticals if previous loading through irrigation activities at the site were similar to those reported here (Figure 3-5). Other site-specific factors may also play an important role on the attenuation of PPCPs through the soil matrix. The presence of preferential flow paths can reduce soil-water interactions by increasing non-uniform macropore flow of irrigated water, thus minimizing the ability of soil to act as a reactive filter (e.g., Gall et al., 2016). The frequency of occurrence of preferential flow paths at the

Living Filter site has been reported to be 47% and 45% at the cropped and forested irrigated sites, respectively (Hopkins et al., 2016).

10Table 3-6: Summary statistics for groundwater samples Mean Median Minimum Maximum Standard % n (µg/L) (µg/L) (µg/L) (µg/L) Deviation > LOQ (µg/L) Irrigated Groundwater Wells Acetaminophen 0.14 0.05 0.05 0.93 0.22 9% Ampicillin 0.71 0.65 0.20 1.22 0.35 11% Caffeine 2.44 0.15 0.05 14.15 4.66 32% Naproxen 41.96 38.84 6.95 98.39 26.61 18% Ofloxacin 3.16 3.21 1.50 7.10 2.02 5% Sulfamethoxazole 2.52 0.13 0.05 27.41 6.20 44% Trimethoprim 0.27 0.05 0.05 1.51 0.42 7% Non-irrigated Groundwater Wells Acetaminophen 0.84 0.05 0.05 15.58 3.47 7% Ampicillin 0.97 0.48 0.36 3.69 1.21 13% Caffeine 3.01 0.18 0.05 14.05 4.94 33% Naproxen 31.90 18.76 3.22 97.95 34.07 20% Ofloxacin 31.67 5.12 1.50 114.94 55.62 4% Sulfamethoxazole 1.49 0.12 0.05 9.54 3.06 35% Trimethoprim 1.49 0.15 0.05 6.95 3.05 7% All Groundwater Wells Acetaminophen 0.42 0.05 0.05 15.58 2.19 9% Ampicillin 0.82 0.65 0.20 3.69 0.82 11% Caffeine 2.65 0.16 0.05 14.15 4.73 32% Naproxen 37.70 27.01 3.22 98.39 29.77 19% Ofloxacin 13.53 3.21 1.50 114.94 33.73 4% Sulfamethoxazole 2.13 0.12 0.05 27.41 5.23 40% Trimethoprim 0.61 0.12 0.05 6.95 1.62 7% n Irrigated groundwater wells = 83; n Non-irrigated groundwater wells = 54; n All groundwater wells = 137

Comparison of Irrigated and Non-Irrigated Wells

Minimal variations in concentration were observed in wells located in irrigated and non-irrigated areas of the Living Filter. Average concentrations in both irrigated and non- irrigated wells spanned the same orders of magnitude, with concentrations typically

<4 μg/L, except for ofloxacin, which was detected at concentrations an order of magnitude higher at the non-irrigated wells. Overall, similarities in concentrations at the sites may be expected due to the influence of groundwater flow at the site leading to flow from irrigated

82 areas impacting wells in non-irrigated areas and the multi-decadal history of effluent irrigation activities. Interestingly, the highest concentrations observed across the site were generally in well P2, which does not receive direct irrigation from laterals at the site. This can be linked with potential influence from other wells (Figure 3-2). Higher nitrate levels, typically ranging between 5 and 8 mg/L-NO3 have been observed in this well in contrast to lower levels found in other monitoring wells at the site (unpublished data collected for site permitting requirements). Given that the direction of groundwater flow is generally towards P2, it is likely that groundwater impacted by irrigation activities upgradient is influencing the quality of water observed at P2. Similarly, maximum concentrations in the irrigated sites were mostly from wells W5 and G10 (Table A-3), and these wells have also been observed to have elevated NO3 in comparison to other wells at the site.

High nitrate levels in wells with wastewater effluent as the dominant sources of nitrate have been correlated with high concentrations and detection frequencies of PPCPs in private wells and has thus been used to indicate wastewater impact on groundwater

(Schaider et al., 2016). Differences in the prevalence of PPCPs in the wells may also be influenced by land use, with wells W1, W2, W5, W6, W7, G10 and G12 in forested parts of the Living Filter and wells P1, P2, P3, P4, P5 and F3 in agricultural parts of the site

(Table 3-7). Based on the groundwater flow direction, wells in the agricultural land use area of the Living Filter are influenced by flow from the forested land use, which may lead to an increase in PPCPs in the agricultural areas. Lower concentrations in forested sites may also be due to higher soil organic carbon in forested land use systems. Land use effects on soil organic carbon have been shown to be more important than climatic factors, such

83 as precipitation, whereby forested land use systems have been reported to result in up to

57% soil organic carbon in comparison to other land uses (Jobbágy and Jackson, 2000). In a forested wastewater reuse system, sorption of PPCPs to soil organic carbon was demonstrated as an important attenuation route of PPCPs (McEachran et al., 2017).

11Table 3-7: Summary statistics variations of PPCPs in groundwater based on land use

Mean Median Minimum Maximum Standard % n (µg/L) (µg/L) (µg/L) (µg/L) Deviation > LOQ (µg/L) Forested Site: Wells W1, W2, W5, W6, W7, G10, G12 Acetaminophen 0.15 0.05 0.05 8.00 0.23 11% Ampicillin 0.73 0.71 0.20 9.00 0.35 12% Caffeine 2.55 0.15 0.05 27.00 4.88 37% Naproxen 40.27 29.09 4.12 98.39 31.94 18% Ofloxacin 3.86 3.25 1.50 8.73 2.72 7% Sulfamethoxazole 3.68 0.15 0.05 37.00 7.27 32% Trimethoprim 0.17 0.05 0.05 6.00 0.20 7% Cropped Sites: Wells P1, P2, P3, P4, P5, F3 Acetaminophen 0.71 0.05 0.05 15.58 3.17 6% Ampicillin 0.95 0.48 0.36 32.73 1.22 11% Caffeine 2.83 0.18 0.05 97.95 4.58 27% Naproxen 35.13 24.93 3.22 114.94 28.51 20% Ofloxacin 39.31 1.50 1.50 114.94 65.50 2% Sulfamethoxazole 0.98 0.12 0.05 9.54 2.42 48% Trimethoprim 1.49 0.17 0.05 6.95 2.73 8% LOQ=Limit of quantification (0.1 µg/L) for all compounds except ofloxacin (3 µg/L); n Forested site=73; n Cropped site=66; LOD=Limit of detection (0.01 µg/L) for all compounds except ofloxacin (0.3 µg/L); Concentrations are reported as 0.05 µg/L or 1.5 µg/L (ofloxacin) when concentrations are above the LOD but less than the LOQ.

Seasonal Variations in Groundwater Concentrations

Minimal seasonal variations were observed in groundwater; however, PPCP concentrations detected in the groundwater samples were higher during the first few months of the sampling period (fall 2016 and winter 2017) (Figure 3-7 and Table 3-8).

Since groundwater concentrations are controlled by factors such as groundwater residence

84 time, dilution, and vadose zone transport processes, negligible seasonal variations in groundwater is expected (Teijon et al., 2010). Biodegradation rates of PPCPs in soil are likely decreased during lower temperatures. In a wastewater irrigated site, Biel-Maeso et al. (2018) reported higher concentrations of PPCPs in the soil profile during colder seasons attributed to increased persistence because of reduced microbial activity. Reduced biodegradation rates in combination with other factors such as the presence of preferential flow paths in a wastewater reuse site can influence concentrations in groundwater during seasons of high precipitation and during the freeze-thaw cycles in the winter by either increasing levels in groundwater or having a diluting effect in groundwater concentrations.

12Table 3-8: Summary statistics for seasonal variations of PPCPs in groundwater Fall 2016 Winter 2017 Spring 2017 Summer 2017 Fall 2017 Acetaminophen 0.56±2.66 0.12±0.21

8Figure 3-7: Seasonal variations in the mean concentration of PPCPs in WWTP effluent and groundwater. Error bars represent maximum and minimum concentrations.

Risk Assessments

Ecological Risk Assessment

The ecological risk calculations revealed that the mean concentrations in the

WWTP effluent would pose medium to high risk to aquatic organisms (Table 3-9) if they were discharged to a surface water system. Naproxen, sulfamethoxazole and ofloxacin posed high risk, with RQ values >1. Naproxen posed high risk for all three taxa and sulfamethoxazole posed high risk to Daphnia and algae. Ofloxacin had high RQs above 1 for algae and fish while caffeine posed high risk to Daphnia. Verlicchi et al. (2012) reported high risk (RQ > 1) from sulfamethoxazole and ofloxacin, while medium risk was

86 observed from naproxen (0.1 < RQ < 1). Kosma et al. (2014) also reported medium to high risk from sulfamethoxazole and trimethoprim in effluent and low to medium risk for naproxen. Higher RQ values for naproxen in the current study can be explained by highly elevated concentrations of naproxen in effluent. These findings indicate that if effluent is to be discharged in a surface water system, medium to high acute risk to aquatic life is expected. However, since the wastewater is spray-irrigated rather than released directly to surface water, the Living Filter functions to mitigate this risk. The estimations of RQ are made each compound at a time without incorporating the potential effects from occurring in mixtures.

Human Health Risk Assessment

Potential human health risk from exposure through groundwater as a drinking water source was estimated using average concentrations in groundwater. Use of average rather than maximum concentrations is expected to provide conservative risk estimates. It was assumed that mean values best represent the average concentrations an adult would consume from this drinking water source. Risk quotient calculated for human health were found to be <1 (Table 3-10) for an intake rate of 2 L d−1, indicating minimal risk.

13Table 3-9: EC50 (µg/L) and corresponding PNECs (Shown in parentheses) and calculated risk quotients in WWTP effluent EC50 and PNEC (µg/L) Risk Quotients

Daphnia Daphnia Compound magnia Algae Fish magnia Algae Fish Acetaminophen a9.E+03 (9.2) a1.E+05 (134) a4.E+05 (378) 0.63 0.04 0.02 Ampicillin -- b>1.E+06 -- -- 0.03 -- Caffeine c2 E+04 (16) -- -- 1.45 -- -- d2 E+04* Naproxen d2 E+04*(22) d3 E+04* (34) 251 171.14 110.74 (15) e3 E+04 Ofloxacin f5 E+03 (4.74) f2 E+04(16) 0.70 4.67 1.38 (31.75) e3 E+04 Sulfamethoxazole d5 E+04* (51) d9 E+05* (890) 2.71 1.34 0.08 (25.2) g1 E+05 Trimethoprim a2.E+04 (16) d8 E+05*(795) 0.02 0.13 0.0026 (120.7)

Similar risk calculation for treated drinking water by de Jesus Gaffney et al. (2015) found RQ values to be less than unity; however, RQ values were generally higher in the present study as risk was assessed for untreated groundwater. In a forest wastewater reuse system, McEachran et al. (2017) also found groundwater concentrations to pose low human health risks. Actual risks are expected to be lower, as Penn State further treats the groundwater at a drinking water treatment plant, which further reduces concentrations of

PPCPs. However, there are uncertainties with risk calculations since they do not assess the potential impact of chronic consumption of water with PPCPs and interactions due to occurrence of mixtures of PPCPs in drinking water.

a Grung et al., (2008) b Eguchi et al., (2004) c Lilius et al., (1995) d Sanderson et al., (2003) e Isidori et al., (2005) f Ferrari et al., (2004) g Kim et al., (2007) -- no EC50 values and calculated risk quotients; * Values estimated using ECOSAR

14Table 3-10: Parameters used in risk quotient calculations and risk quotient values for groundwater samples Mean Concentrations ADI DWEL Risk (µg/L) (µg kg -1d-1) (µg L-1 d-1) Quotient Acetaminophen 0.4 a340 10428.6 4.03E-05 Ampicillin 0.8 b0.5 15.3 5.35E-02 Caffeine 1.6 c30 920.2 2.88E-03 Naproxen 43.6 d5.7 174.8 2.16E-01 Ofloxacin 3.6 d7.1 217.8 6.21E-02 Sulfamethoxazole 2.2 a130 3987.4 5.34E-04 Trimethoprim 0.2 a4.2 128.8 4.74E-03

Conclusion

To assess the performance of a coupled human-natural system (engineered wastewater treatment plant followed by natural treatment at a spray-irrigation site) for removal of PPCPs, water samples were collected during a 14-month study period from the

WWTP influent, effluent and 13 groundwater monitoring wells at the spray-irrigation site.

High removal efficiencies were observed for acetaminophen (>88%) during wastewater treatment. Other compounds, especially antibiotics and the anti-inflammatory drug naproxen, were removed to a lesser extent with instances of negative removal potentially due to metabolite reversal to parent compound or desorption from organic matter.

Concentrations of PPCPs were found to be high during winter and spring season corresponding to both high consumer use during these seasons and reduced wastewater removal efficiencies because of low temperatures.

a Schwab et al., (2005) b Vragović et al., (2012) c McEachran et al., (2017) d Prosser & Sibley, (2015)

At the forested and agricultural wastewater spray-irrigation site that has been in operation for over 40 years, the impact of long-term wastewater irrigation on groundwater was observed by the presence of studied PPCPs at levels above method detection and quantification limits. Though groundwater concentrations were significantly lower than those measured in wastewater effluent, a combination of long-term irrigation and groundwater flow influences are observed at the Living Filter as non-irrigated wells impacted by flow from irrigated sites depicted similar and, in some cases, higher levels of

PPCPs than irrigated wells. Furthermore, considering the persistent nature of most PPCPs and prolonged wastewater irrigation at the site, there is an underlying risk of continued persistence of PPCPs in groundwater, creating the need for improvements in management practices at the site.

The Living Filter performs an ecosystem service by mitigating an ecosystem risk to aquatic organisms as WWTP effluent found to pose medium to high risk to aquatic systems is not discharged in streams but rather, allowed to infiltrate through the soil and decrease in concentration before recharging groundwater. Though groundwater can act as a source of PPCPs to surface water systems, concentrations of PPCPs are often reduced and/or diluted to lessen ecological health risks. Thus, in terms of limiting PPCPs in the environment, land application of WWTP effluent offers a more effective management practice than direct discharge of effluent to surface water. Detection frequencies of PPCPs above the limit of quantification (LOQ = 0.1 μg/L and 3 μg/L) were lower in groundwater ranging from 4 to 40% in comparison to WWTP effluent (23%–70%). Average concentrations of PPCPs detected in groundwater were also typically two orders of

90 magnitude lower (0.4–37.7 μg/L) than those measured in WWTP effluents (2.0–

3765 μg/L). Additionally, the concentrations measured in the groundwater were found to pose minimal risk to human health through use of the groundwater as a drinking water source. These results suggest that the ecosystem services provided by a wastewater reuse system is a feasible option to reduce the risk posed by PPCPs in wastewater effluent.

References

Aiello, R., Cirelli, G. L., & Consoli, S. (2007). Effects of reclaimed wastewater irrigation on soil and tomato fruits: a case study in Sicily (Italy). Agricultural Water Management, 93(1-2), 65-72.

Akber, A., Mukhopadhyay, A., Al-Senafy, M., Al-Haddad, A., Al-Awadi, E., & Al-Qallaf, H. (2008). Feasibility of long-term irrigation as a treatment method for municipal wastewater using natural soil in Kuwait. Agricultural Water Management, 95(3), 233-242.

Andrews, D. M., Robb, T., Elliott, H., & Watson, J. E. (2016). Impact of long-term wastewater irrigation on the physicochemical properties of humid region soils: “The Living Filter” site case study. Agricultural Water Management, 178, 239-247.

Angelakis, A. N., Marecos Do Monte, M. H. F., Bontoux, L., & Asano, T. (1999). The status of wastewater reuse practice in the Mediterranean basin: Need for guidelines. Water Research, 33(10), 2201–2217.

Antweiler, R. C., & Taylor, H. E. (2008). Evaluation of statistical treatments of left-censored environmental data using coincident uncensored data sets: I. Summary statistics. Environmental Science & Technology, 42(10), 3732-3738.

Aruga, R. (1997). Treatment of responses below the detection limit: some current techniques compared by factor analysis on environmental data. Analytica Chimica Acta, 354(1-3), 255-262.

Assadian, N. W., Di Giovanni, G. D., Enciso, J., Iglesias, J., & Lindemann, W. (2005). The transport of waterborne solutes and bacteriophage in soil subirrigated with a wastewater blend. Agriculture, Ecosystems and Environment, 111(1-4), 279–291.

Biel-Maeso, M., Corada-Fernández, C., & Lara-Martin, P. A. (2018). Monitoring the occurrence of pharmaceuticals in soils irrigated with reclaimed wastewater. Environmental Pollution, 235, 312-321.

Candela, L., Fabregat, S., Josa, A., Suriol, J., Vigués, N., & Mas, J. (2007). Assessment of soil and groundwater impacts by treated urban wastewater reuse. A case study: Application in a golf course (Girona, Spain). Science of the Total Environment, 374(1), 26-35. Carrara, C., Ptacek, C. J., Robertson, W. D., Blowes, D. W., Moncur, M. C., Sverko, E. D., & Backus, S. (2008). Fate of pharmaceutical and trace organic compounds in three septic system plumes, Ontario, Canada. Environmental Science & Technology, 42(8), 2805-2811.

Carballa, M., Omil, F., Lema, J. M., Llompart, M., Garcı́a-Jares, C., Rodrı́guez, I., ... & Ternes, T. (2004). Behavior of pharmaceuticals, cosmetics and hormones in a sewage treatment plant. Water Research, 38(12), 2918-2926.

Chen, J., Tang, C., & Yu, J. (2006). Use of 18O, 2H and 15N to identify nitrate contamination of groundwater in a wastewater irrigated field near the city of Shijiazhuang, China. Journal of Hydrology, 326(1-4), 367-378. de Jesus Gaffney, V., Almeida, C. M., Rodrigues, A., Ferreira, E., Benoliel, M. J., & Cardoso, V. V. (2015). Occurrence of pharmaceuticals in a water supply system and related human health risk assessment. Water Research, 72, 199-208.

De Vries, J. (1972). Soil filtration of wastewater effluent and the mechanism of pore clogging. Journal (Water Pollution Control Federation), 565-573.

Eguchi, K., Nagase, H., Ozawa, M., Miyamoto, K., & Yoshimura, H. (2004). Evaluation of antimicrobial agents for veterinary use in the ecotoxicity test using microalgae, Chemosphere, 57(11), 1733-1738.

Ferguson, B. K. (1983). Environmental planning for wastewater land application: Lessons from Penn State's “Living Filter”. Landscape Planning, 10(3), 205-218.

Ferrari, B., Mons, R., Vollat, B., Fraysse, B., Paxēaus, N., Giudice, R.L., Pollio, A. & Garric, J., (2004). Environmental risk assessment of six human pharmaceuticals: are the current environmental risk assessment procedures sufficient for the protection of the aquatic environment? Environmental Toxicology and Chemistry, 23(5), 1344-1354.

Franklin, A. M., Williams, C. F., & Watson, J. E. (2018). Assessment of Soil to Mitigate Antibiotics in the Environment Due to Release of Wastewater Treatment Plant Effluent. Journal of Environmental Quality.

Franklin, A.M., C.F. Williams, D.M. Andrews, E.W. Woodward, and J.E. Watson. 2016. Uptake of three antibiotics and an antiepileptic drug by wheat crops spray irrigated with wastewater treatment plant effluent. Journal of Environmental Quality. 45:546–554.

Gall, H. E., Basu, N. B., Mashtare, M. L., Rao, P. S. C., & Lee, L. S. (2016) Assessing the impacts of anthropogenic and hydro-climatic drivers on estrogen legacies and trajectories. Advances in Water Resources, 87, 19-28.

Gall, H. E., Sassman, S. A., Jenkinson, B., Lee, L. S., & Jafvert, C. T. (2014). Hormone loads exported by a tile‐drained agroecosystem receiving animal wastes. Hydrological Processes, 28(3), 1318-1328.

Göbel, A., McArdell, C. S., Joss, A., Siegrist, H., & Giger, W. (2007). Fate of sulfonamides, macrolides, and trimethoprim in different wastewater treatment technologies. Science of the Total Environment, 372(2-3), 361-371. Golovko, O., Kumar, V., Fedorova, G., Randak, T., & Grabic, R. (2014). Seasonal changes in antibiotics, antidepressants/psychiatric drugs, antihistamines and lipid regulators in a wastewater treatment plant. Chemosphere, 111, 418-426.

Grung, M., Källqvist, T., Sakshaug, S., Skurtveit, S., & Thomas, K. V. (2008). Environmental assessment of Norwegian priority pharmaceuticals based on the EMEA guideline. Ecotoxicology and Environmental Safety, 71(2), 328-340.

Haruvy, N. (1997). Agricultural reuse of wastewater: nation-wide cost-benefit analysis. Agriculture, Ecosystems & Environment, 66(2), 113-119.

Hopkins, I., Gall, H., & Lin, H. (2016). Natural and anthropogenic controls on the frequency of preferential flow occurrence in a wastewater spray irrigation field. Agricultural Water Management, 178, 248-257.

Isidori, M., Lavorgna, M., Nardelli, A., Pascarella, L., & Parrella, A. (2005). Toxic and genotoxic evaluation of six antibiotics on non-target organisms. Science of the Total Environment, 346(1-3), 87-98.

Jim, T. Y., Bouwer, E. J., & Coelhan, M. (2006). Occurrence and biodegradability studies of selected pharmaceuticals and personal care products in sewage effluent. Agricultural Water Management, 86(1-2), 72-80.

Jobbágy, E. G., & Jackson, R. B. (2000). The vertical distribution of soil organic carbon and its relation to climate and vegetation. Ecological Applications, 10(2), 423-436.

Kasprzyk-Hordern, B., Dinsdale, R. M., & Guwy, A. J. (2009). The removal of pharmaceuticals, personal care products, endocrine disruptors and illicit drugs during wastewater treatment and its impact on the quality of receiving waters. Water Research, 43(2), 363-380.

Kim, Y., Choi, K., Jung, J., Park, S., Kim, P. G., & Park, J. (2007). Aquatic toxicity of acetaminophen, carbamazepine, cimetidine, diltiazem and six major sulfonamides, and their potential ecological risks in Korea. Environment International, 33(3), 370-375.

Kosma, C. I., Lambropoulou, D. A., & Albanis, T. A. (2014). Investigation of PPCPs in wastewater treatment plants in Greece: occurrence, removal and environmental risk assessment. Science of the Total Environment, 466, 421-438. Lapworth, D. J., Baran, N., Stuart, M. E., & Ward, R. S. (2012). Emerging organic contaminants in groundwater: a review of sources, fate and occurrence. Environmental Pollution, 163, 287-303.

Lilius, H., Hästbacka, T., & Isomaa, B. (1995). A comparison of the toxicity of 30 reference chemicals to Daphnia magna and Daphnia pulex. Environmental Toxicology and Chemistry: An International Journal, 14(12), 2085-2088.

Lubello, C., Gori, R., Nicese, F. P., & Ferrini, F. (2004). Municipal-treated wastewater reuse for plant nurseries irrigation. Water Research, 38(12), 2939-2947.

Martínez-Hernández, V., Meffe, R., López, S. H., & de Bustamante, I. (2016). The role of sorption and biodegradation in the removal of acetaminophen, carbamazepine, caffeine, naproxen and sulfamethoxazole during soil contact: a kinetics study. Science of The Total Environment, 559, 232-241.

McEachran, A. D., Shea, D., & Nichols, E. G. (2017). Pharmaceuticals in a temperate forest-water reuse system. Science of the Total Environment, 581, 705-714.

Mina, O., Gall, H. E., Saporito, L. S., & Kleinman, P. J. (2016). Estrogen transport in surface runoff from agricultural fields treated with two application methods of dairy manure. Journal of Environmental Quality, 45(6), 2007-2015.

Monk, J.P. & Campoli-Richards, D.M. Drugs (1987) 33: 346. https://doi.org/10.2165/00003495- 198733040-00003 Monteiro S.C., Boxall A.B. (2010) Occurrence and Fate of Human Pharmaceuticals in the Environment. In: Reviews of Environmental Contamination and Toxicology. Reviews of Environmental Contamination and Toxicology, vol 202. Springer, New York, NY.

Nakada, N., Tanishima, T., Shinohara, H., Kiri, K., & Takada, H. (2006). Pharmaceutical chemicals

and endocrine disrupters in municipal wastewater in Tokyo and their removal during activated sludge treatment. Water Research, 40(17), 3297-3303.

O’Driscoll, M. A., & Parizek, R. R. (2003). The hydrologic catchment area of a chain of karst wetlands in central Pennsylvania, USA. Wetlands, 23(1), 171-179.

Parfit, K. (1999). Martindale: The complete drug reference. The Pharmaceutical Press, London, 778, 854-857. Parizek, R.R., L.T. Kardos, W.E. Sopper, E.A. Myers, D.E. Davis, M.A. Farrell, N.B. Nesbitt. 1967. The Pennsylvania State University studies No. 23 B Wastewater Renovation and Conservation. The Pennsylvania State University, University Park, PA, USA.

Prosser, R. S., & Sibley, P. K. (2015). Human health risk assessment of pharmaceuticals and personal care products in plant tissue due to biosolids and manure amendments, and wastewater irrigation. Environment International, 75, 223-233.

Radjenović, J., Petrović, M., & Barceló, D. (2009). Fate and distribution of pharmaceuticals in wastewater and sewage sludge of the conventional activated sludge (CAS) and advanced membrane bioreactor (MBR) treatment. Water Research, 43(3), 831-841.

Ramirez-Fuentes, E., Lucho-Constantino, C., Escamilla-Silva, E., & Dendooven, L. (2002). Characteristics, and carbon and nitrogen dynamics in soil irrigated with wastewater for different lengths of time. Bioresource Technology, 85(2), 179-187.

Schwab, B.W., Hayes, E.P., Fiori, J.M., Mastrocco, F.J., Roden, N.M., Cragin, D., Meyerhoff, R.D., Vincent, J.D. and Anderson, P.D., (2005). Human pharmaceuticals in US surface waters: a human health risk assessment. Regulatory Toxicology and Pharmacology, 42(3), 296-312.

Sui, Q., Huang, J., Deng, S., Chen, W., & Yu, G. (2011). Seasonal variation in the occurrence and removal of pharmaceuticals and personal care products in different biological wastewater treatment processes. Environmental Science & Technology, 45(8), 3341-3348.

Tang, C., Chen, J., Shindo, S., Sakura, Y., Zhang, W., & Shen, Y. (2004). Assessment of groundwater contamination by nitrates associated with wastewater irrigation: a case study in Shijiazhuang region, China. Hydrological Processes, 18(12), 2303-2312.

Teijon, G., Candela, L., Tamoh, K., Molina-Díaz, A., & Fernández-Alba, A. R. (2010). Occurrence of emerging contaminants, priority substances (2008/105/CE) and heavy metals in treated wastewater and groundwater at Depurbaix facility (Barcelona, Spain). Science of the Total Environment, 408(17), 3584-3595.

Tzanakakis, V., Koo-Oshima, S., Haddad, M., Apostolidis, N., & Angelakis, A. (2014). The history of land application and hydroponic systems for wastewater treatment and reuse. Evolution of Sanitation and Wastewater Technologies through the Centuries; IWA Publishing: London, UK, 457.

US EPA (U.S. Environmental Protection Agency). 2007. Method 1694: Pharmaceuticals and personal care products in water, soil, sediment and biosolids by HPLC/MS/MS. US EPA Office of Water and Office of Science and Technology. EPA-821-R-08-002. Available at: https://www.epa.gov/sites/production/files/201510/documents/method_1694_2007.pdf

US EPA (U.S. Environmental Protection Agency). 2018. Method 301: Field validation of pollutant measurement methods from various waste media. Available at: https://www.epa.gov/sites/production/files/2018-03/documents/method_301_3-26- 2018_1.pdf

Vazquez-Montiel, O., Horan, N. J., & Mara, D. D. (1996). Management of domestic wastewater for reuse in irrigation. Water Science and Technology, 33(10-11), 355-362.

Verlicchi, P., Al Aukidy, M., & Zambello, E. (2012). Occurrence of pharmaceutical compounds in urban wastewater: removal, mass load and environmental risk after a secondary treatment—a review. Science of the Total Environment, 429, 123-155.

Vragović, N., Bažulić, D., & Zdolec, N. (2012). Dietary exposure assessment of ß-lactam antibiotic residues in milk on Croatian market. Croatian Journal of Food Science and Technology, 4(1), 81-84.

Vulliet, E., & Cren-Olivé, C. (2011). Screening of pharmaceuticals and hormones at the regional scale, in surface and groundwaters intended to human consumption. Environmental Pollution, 159(10), 2929-2934.

Walters, E., McClellan, K., & Halden, R. U. (2010). Occurrence and loss over three years of 72 pharmaceuticals and personal care products from biosolids–soil mixtures in outdoor mesocosms. Water Research, 44(20), 6011-6020.

Woodward, E. E., Andrews, D. M., Williams, C. F., & Watson, J. E. (2014). Vadose zone transport of natural and synthetic estrogen hormones at penn state’s “living filter” wastewater irrigation site. Journal of Environmental Quality, 43(6), 1933-1941.

Wu, Y., Williams, M., Smith, L., Chen, D., & Kookana, R. (2012). Dissipation of sulfamethoxazole and trimethoprim antibiotics from manure-amended soils. Journal of Environmental Science and Health, Part B, 47(4), 240-249.

Zorita, S., Mårtensson, L., & Mathiasson, L. (2009). Occurrence and removal of pharmaceuticals in a municipal sewage treatment system in the south of Sweden. Science of the Total Environment, 407(8), 2760-2770.

Chapter 4

Occurrence, Concentrations, and Risks of Pharmaceutical Compounds in Private wells in Central Pennsylvania

Reprinted from the Journal of Environmental Quality, F.A. Kibuye, H.E. Gall, K.R. Elkin, B. Swistock, J.E. Watson, T.L. Veith, H. A. Elliott, Occurrence, concentrations, and risks of pharmaceutical compounds in private wells in Central Pennsylvania. 2019, with permission from Journal of Environmental Quality. DOI: 10.2134/jeq2018.08.0301

Graphical Abstract:

Highlights.

• 26 private wells in Pennsylvania were sampled for 7 pharmaceutical compounds. • At least 1 compound was detected per groundwater sample at ng - g/L levels. • On average, concentrations were higher in private wells than nearby surface water. • Concentrations were generally highest for the most frequently detected compounds. • Concentrations were low enough that no adverse human health effects are expected.

Abstract

Over the counter and prescription medications are routinely present at detectable levels in surface and groundwater bodies. The presence of these emerging contaminants has raised both environmental and public health concerns, particularly when the water is used for drinking either directly or with additional treatment. However, the frequency of occurrence, range of concentrations, and potential human health risks are not well understood, especially for groundwater supplies. Private wells are often not tested for contaminants regulated by drinking water standards and are even less frequently tested for emerging contaminants. By partnering with the Pennsylvania Master Well Owner

Network, water samples were collected from 26 households with private wells in the West

Branch of the Susquehanna River Basin in central Pennsylvania in winter 2017. All samples were analyzed for six pharmaceuticals: acetaminophen, ampicillin, naproxen, ofloxacin, sulfamethoxazole, and trimethoprim, and one over-the-counter stimulant: caffeine. At least one compound was detected at each site. Ofloxacin and naproxen were the most and least frequently detected compounds, respectively. Concentrations from the groundwater wells were higher than those of nearby surface water samples. However, risk calculations revealed that none of the concentrations measured in groundwater samples posed significant human health risk. A simple, physicochemical-based modeling approach was employed to predict pharmaceutical transport from septic absorption field to groundwater and further elucidate variations in detection frequencies. Findings indicate that although septic tanks may act as contaminant sources for groundwater wells, the

99 human health impacts from trace-level pharmaceuticals that may be present are likely minimal.

Keywords: Emerging contaminants, groundwater quality, private wells, septic tanks, pharmaceuticals

Introduction

Groundwater is an important supply of drinking water globally. It is estimated that half of the population accesses potable water from groundwater aquifers (Smith et al.,

2016). In the United States, approximately 13 million households utilize private wells as a drinking water source (USEPA, 2008). Furthermore, homeowners with private wells commonly have septic tanks on their property to treat their wastewater (Schaider et al.,

2014). About 25% of domestic wastewater in the United States is treated with septic systems, while some rural and forested areas in the Northeastern United States have up to

85% of residents relying on septic systems (Swartz et al., 2006). When these systems are functioning properly, treated wastewater effluent is dispersed into subsurface absorption fields for further treatment before recharging underlying groundwater.

Although septic tanks and absorption fields are ideally installed downgradient of domestic supply wells, domestic wastewater can still degrade well water quality, especially if a septic tank is not maintained, was improperly installed, or has passed the designed lifespan. Per regulations, a horizontal distance of at least 30 m is required between a septic absorption field and a domestic supply well (US Department of Housing and Human

Development, 2012). In Pennsylvania, a minimum vertical distance of 1.2 m is required

100 between the bottom of an absorption unit and a downward limiting zone of bedrock or a seasonal high water table (Pennsylvania Code, 1997).

In addition to commonly regulated contaminant issues in private wells, such as fecal coliform, E. coli, and nitrate (Swistock et al., 2013), emerging contaminants, including pharmaceuticals, ingredients in personal care products, and other organic wastewater compounds, pose potential threats to groundwater quality. Most emerging contaminants are known to incompletely degrade in wastewater treatment system including public wastewater treatment plants (Verlicchi et al., 2012) and septic systems (Godfrey et al.,

2007). Consequently, compounds that persist or are incompletely degraded may travel with wastewater plumes and contaminate groundwater, thus making septic systems important pollutant sources to surrounding domestic groundwater sources (Yang et al., 2016; Katz et al., 2010; Carrara et al., 2008; Swartz et al., 2006). Additionally, impacted groundwater can act as a source of pharmaceuticals and other classes of emerging contaminants to nearby surface water sources (Standley et al., 2008).

Nearly 10-20% of septic systems nationally are functioning poorly (USEPA, 2008), thereby increasing the risk of groundwater contamination both by pollutants regulated by the Environmental Protection Agency’s (EPA) drinking water standards as well as unregulated emerging contaminants, such as pharmaceuticals. In a statewide study of 701 private wells in Pennsylvania, Swistock et al. (2013) found that 41% of private water wells failed at least one drinking water standard. Drinking water standards for nitrate (10 mg-

N/L), pH (6.5-8.5), arsenic (0.01 mg/L), and lead (0.015 mg/L) were exceeded in up to

20% of sampled private wells. Because the EPA’s Safe Drinking Water Act does not

101 regulate private well sources, the responsibility of ensuring safe drinking water falls to the private well owner. Few homeowners regularly sample their wells to determine if the

EPA’s primary drinking water standards are met (Focazio et al., 2006).

Occurrences of pharmaceuticals in private well sources can indicate septic system failure in treating these compounds, as they are good markers of human wastewater impacts in domestic groundwater (James et al., 2016; Barnes et al., 2008) and surface water sources

(Focazio et al., 2008; Kolpin et al., 2002). Furthermore, it is important to measure the levels of pharmaceuticals in drinking water sources to better understand exposure levels and associated human health risks. Conventional drinking water treatment reduces levels of pharmaceuticals in drinking water sources, albeit with differing removal efficiencies depending on treatment technology and the specific pharmaceutical compound

(Glassmeyer et al., 2017; USEPA, 2010). However, communities relying on private wells and springs for domestic water supply generally perform minimal water treatment prior to use; therefore, source water concentrations are likely representative of the concentrations consumed in drinking water.

Given the prevalence (approximately one-third) of residents in the Commonwealth of Pennsylvania that utilize onsite wastewater treatment systems for domestic wastewater disposal (Pennsylvania Department of Environmental Protection, 1995) and private wells and springs for domestic drinking water supplies (Penn State Extension, 2007), documenting the occurrence of pharmaceuticals in groundwater is important for informing public health decisions related to septic systems and their potential impacts on well water quality. Previous studies have monitored the impact of septic systems on domestic wells in

102 targeted locations in Cape Cod, MA (Schaider et al., 2016; Schaider et al., 2014; Swartz et al., 2006), New York, and New England (Phillips et al., 2015); however, no such study has been performed in Pennsylvania.

The goal of this project was to screen for the presence and concentrations of selected pharmaceuticals in private domestic groundwater sources in central PA. The compounds selected for this study by request of the funding agency (PA Sea Grant) were six commonly used prescription antibiotics: ampicillin, ofloxacin, sulfamethoxazole, and trimethoprim; analgesics: acetaminophen and naproxen; and a stimulant, caffeine. The targeted compounds have a wide range of physicochemical properties and are therefore expected to be representative of a broader range of pharmaceuticals of interest. A solute transport modeling approach (Harman et al., 2011) was employed to predict vadose zone movement of pharmaceuticals compounds from an absorption unit to a groundwater limiting zone. The model was used to explore how the physicochemical characteristics of the pharmaceutical compounds contributed to their presence or absence in the 26 groundwater sites that were sampled. Additionally, the groundwater well concentrations were used to calculate human health risk quotients to better understand the potential risk that the presence of these compounds in the drinking water supplies may pose to residents.

Materials and Methods

A total of 26 homeowners who use groundwater for potable water supply (24 private wells and 2 springs) volunteered to participate in the study. The homeowners were recruited through the Master Well Owner Network (MWON), an Extension program

103 through the Pennsylvania State University (https://extension.psu.edu/programs/mwon). All sites were located in the West Branch sub-basin of the Susquehanna River Watershed

(Figure 4-1). Reported well depths ranged from 12 to 130 m and 96% of the households had active septic systems within the property (Table 4-5). All samples were collected in the winter of 2017 (January 29 – March 9). During the same sampling period, surface water samples were collected in January and March from the West Branch of the Susquehanna

River (Figure 4-1) to enable comparisons of pharmaceutical concentrations between surface and groundwater. All water samples were analyzed for seven pharmaceutical compounds (Table 4-1) selected to represent a wide range of physicochemical parameters.

9Figure 4-1. Map of private groundwater well sampling locations and surface water sampling location in the West Branch of the Susquehanna River Basin.

104

15Table 4-1: Physicochemical properties of selected pharmaceutical compounds. aMolar Pharmaceutical aChemical Mass alog % Half-life a Compound Formula (g/mol) pKa KOW Excreted in soil Antibiotics b a † Ampicillin C16H19N3O4S 349.40 2.5; 7.3 1.35 30-60 1 c f Sulfamethoxazole C10H11N3O3S 253.28 1.6; 5.7 0.89 30 39 d d g Ofloxacin C18H20FN3O4 361.37 5.97; 9.28 -0.39 70-98 4.1 yr c f Trimethoprim C14H18N4O3 290.32 7.12 0.91 80 62 d Analgesics b h Acetaminophen C8H9NO2 151.17 9.38 0.46 2-3 <1 d e b Naproxen C14H14O4 230.26 4.15 3.18 < 1 2 d Stimulant f gh i Caffeine C8H10N4O2 194.19 10.4 -0.07 -- 35 h Sample Collection Sampling kits containing two 250-mL trace-cleaned amber glass bottles, one 250- mL bottle of deionized water for creating a field blank, a pair of latex laboratory gloves, sample collection and handling procedures, ice packs, a brief survey about the sampled well, and a prepaid return shipment label were assembled and mailed to each household.

Participants collected their raw groundwater sample prior to any existing treatment in the household into one 250-mL trace-cleaned amber glass bottles. Participants then poured the shipped deionized water into the second 250-mL sample bottle to produce the field blank sample. This field blank was collected to understand any potential contamination that may have occurred at the time of sample collection. Any detected compound in field blanks

a https://pubchem.ncbi.nlm.nih.gov b Monteiro & Boxall, (2010) c Verlicchi et al., (2010) d Monk & Campoli-Richards, (1987) e Oosterhuis et al., (2013) f Wu et al., 2012 g Walters et al., (2010) hLi et al., 2014 iMartínez-Hernández et al., (2016) Ka = Acid dissociation constant; KOW = octanol-water partition coefficient; KOC=organic carbon partition coefficient; †Values for amoxicillin used due to lack of data for ampicillin

105 were used to censor groundwater concentrations at each site (Table 4-5). Detection frequencies in field blanks were generally lower than groundwater samples. Samples were shipped overnight on ice to the United States Department of Agriculture – Agricultural

Research Service (USDA-ARS) laboratory in University Park, PA and stored at 4°C until processing, which occurred within 48 h of collection.

Sample analysis

The analysis and quantification of pharmaceuticals was done using a high-resolution accurate mass (HRAM) Q Exactive Orbitrap mass spectrometer (ThermoFisher Scientific,

Bremen, Germany), interfaced to the chromatography system through a heated electrospray injection (HESI) source. Analytical methods have been described in detail by Kibuye et al.

(2019). In brief, water samples were filtered through a 0.22 µm polyethersulfone (PES) syringe filter after which samples were concentrated from a 500 µL volume to 20 µl volume using an inline concentrator column Hypersil Gold aQ 20x2.1 mm 12µm (ThermoFisher,

Sunnyvale, CA) then injected onto a 100 x 2.1mm 3µm Hypersil Gold analytical column.

For all tested analytes except ofloxacin, the method detection limit (MDL) was 0.01 μg/L

(signal-to-noise ratio of 3 over background), and the method quantification limit (MQL) was 0.1 μg/L (signal-to-noise ratio of 10 over background). The MDL and MQL for ofloxacin were 0.3 μg/L and 3 μg/L, respectively. Reporting limits for the measured analytes were set at one-half of the MQL as specified by the EPA Method 301 (USEPA,

2017) guidelines for determination of analytes below the quantification limit. The recoveries were determined from 12 samples for each compound for a calibration range of

106

0.1 – 500 µg/L. The percent recoveries were as follows: 90-95%, >95%, >85%, 80-85%,

70-80% and >75% for acetaminophen, caffeine, trimethoprim, sulfamethoxazole, naproxen, and ofloxacin, respectively.

Modeling Approach

As contaminants travel through the soil, they are subject to degradation and transformation. Half-lives for the selected pharmaceuticals range widely, with some shorter than one day (caffeine and acetaminophen; Table 4-1) and others as long as several years

(ofloxacin; Table 4-1). Compounds with long half-lives are persistent in soil and may contaminate groundwater. However, it is the combination of sorption to the soil, degradation during transport, and dilution that affect occurrence at a concentration above the MDL in groundwater.

Contaminant travel rate in the vadose zone can be estimated through the determination of retardation factors, which account for the diminished contaminant travel speed relative to the bulk groundwater. The retardation factor, R, is estimated using a compound’s soil-water partition coefficient (KD) and soil characteristics shown in Equation

1 below:

퐾 휌 푅 = 1 + 퐷 푏 (Equation 4-1) 휃 where ρb is the soil bulk density and θ is soil water content. KD is calculated as the product of the fraction of organic carbon in the soil and the organic carbon partition coefficient,

KOC (Table 4-2). Calculated R and KD values are summarized in Table 4-2.

107

16Table 4-2: Model parameters used for the contaminant transport calculations between septic tank leach field and groundwater well.

Pharmaceutical a Compound KD (L/kg) log KOC Retardation Factor (R) Acetaminophen 0.04 1.32 1.1 Ampicillin 0.2 2.00† 1.6 Caffeine 1.5 2.87 5.3 Naproxen 0.7 2.52 2.9 Ofloxacin 88.3 4.64 258.7 Sulfamethoxazole 0.1 1.86 1.4 Trimethoprim 0.2 1.88 1.4

A profile of a septic seepage bed designed to regulatory standards (PA Code Title

25 Chapter 73, 1997) is shown in Figure 4-2a. The absorption/aggregate layer is typically

0.3-0.9 m deep with a minimum of 0.3 m cover of backfill soil. A soil depth of at least 1.2 m, referred to as the suitable soil layer, between the bottom of the gravel absorption layer and a limiting zone of either seasonal high water table or bedrock is required. Applying these design characteristics, septic effluent containing a mixture of pharmaceuticals discharged through a lateral pipe in the absorption layer will travel a total of 1.2 m from the bottom of the absorption layer to an underlying water table that acts as the limiting zone. To predict vadose zone transport of the seven compounds of interest from a conventional septic absorption unit to a water table limiting zone, a solute transport modeling approach following the HEIST model by Harman et al. (2011) was used (Figure

4-2b).

a https://pubchem.ncbi.nlm.nih.gov †Values for amoxicillin used due to lack of data for ampicillin. KD = Koc x foc, where foc is the fraction of organic carbon in the soil and KOC is the organic carbon partition coefficient

108

10Figure 4-2: Front view of (a) a typical septic seepage bed per regulatory standards in Pennsylvania and (b) contaminant transport model schematic

In addition to the compound physicochemical characteristics, transport in the vadose zone is influenced by the distribution of infiltration events such as precipitation.

The range of storage in the vadose zone as function of hydrologic forcing is calculated as follows (Harman et al., 2011):

Z (휃 −휃 ) 훾 = 푓푐 푤푝 (Equation 4-2) α where Z is the total depth to groundwater (1.2 m), θfc is field capacity, θwp is the wilting point, and α is the mean storm depth, assumed to be 8.5 mm based on 15 years of public rainfall data within the Susquehanna River Basin in central PA

(https://www.wcc.nrcs.usda.gov). The dominant soil texture at the sampled sites was silt

109 loam and the corresponding soil properties used in the model are summarized in Table 4-

17Table 4-3. Silt loam soil properties used in model calculations. Parameter Value Units 3 Bulk Density (ρb) 1.33 g/cm a Fraction of Organic Carbon (foc) 0.002 g organic C/g soil Vadose Zone Depth (Z) 1200 mm b Field Capacity (θfc) 0.215 vol/vol b Wilting Point (θwp) 0.115 vol/vol b Residual Water Content (θr) 0.067 vol/vol b Saturated Water Content (θs) 0.454 vol/vol b Saturated Hydraulic Conductivity (Ksat) 622 mm/d aCalculated from average organic matter content from 1 m soil depth at the selected study sites. Organic matter values were obtained from the Web Soil Survey (https://websoilsurvey.sc.egov.usda.gov); bKarkanis 1983.

Assuming uniform flow through the soil, the mean and variance of the travel time that each compound front takes to reach underlying groundwater is given by (Harman et al.,

2011):

RF훾 2 2RF훾 µ푇 = , 휎 푇 = 2 (Equation 4-3) λ푝 λ푝

-1 where λp is the average rainfall frequency for humid climates of 0.3 d based on 15 years of public rainfall data within the Susquehanna River Basin in central PA

(https://www.wcc.nrcs.usda.gov). F is calculated by (θfc-θr)/(θfc-θwp), and θr is the residual water content (Harman et al., 2011). Since the compounds transported are subject to biodegradation, the delivery ratio (DR), which is the fraction of the initial mass that reaches groundwater, is a function of a compound’s first-order biodegradation rate, k (half-lives given in Table 4-1) and the average travel time in the vadose zone. This approach neglects the fact that some of the compounds are ionized under prevailing pH conditions, which could potentially influence their retention behavior. The mean and variance of the DR are given by (Harman et al., 2011):

110

RF훾 푘 2 2RF훾 푘 2RF훾 푘 µ퐷푅 = exp (− ), 휎 퐷푅 = exp (− ) − exp (− ) (Equation 4-4) λ푝+푘 λ푝+2푘 λ푝+푘

Once the compound reaches the water table limiting zone, no further concentration reduction due to biodegradation or sorption is expected because of low organic carbon content and reducing conditions. Thus, pharmaceuticals that persist through the vadose zone can travel with groundwater to impact a water supply well.

Risk Calculations

The occurrence of pharmaceuticals in the environment poses potential ecosystem and human health risks. For a septic system, pharmaceuticals dispersed in the vadose zone pose an underlying risk for soil organisms (Verlicchi & Zambello, 2015). The antimicrobial triclosan has been linked to decline in microbial biomass community in soil (Zaayman et al., 2017) and antibiotics in soil can increase the presence of antimicrobial-resistant bacteria and genes in the soil (Marti et al., 2013). The primary human exposure route to low concentration pharmaceuticals and other emerging contaminants is through drinking water. Because the sampled groundwater in this study are used as the primary drinking water sources in participating households, a human health risk assessment was conducted to evaluate if detected compounds occurred at levels that posed human health concern.

Assuming exposure to pharmaceuticals at average levels measured in domestic groundwater samples, a human health risk assessment was performed for adult population characterized by a 50th percentile body weight (BW) of 60 kg and daily drinking water intake (DWI) of 2 L/d for a frequency of exposure (FOE) of 0.96 (350 d/365) (de Jesus

Gaffney et al., 2015). Acceptable daily intakes (ADIs) of pharmaceuticals, which are the

111 recommended daily exposure levels that pose no adverse effects on a population (Schwab et al., 2005), were used as exposure thresholds. Drinking water equivalent levels (DWELs) were then estimated as shown below (Blanset et al., 2007):

ADI ×BW 퐷푊퐸퐿 (µ푔. 퐿−1푑−1) = (Equation 4-5) DWI×FOE

A risk quotient (RQ), which is calculated as the ratio of average pharmaceutical concentration in domestic groundwater and the estimated DWEL, was then used to characterize risk. Risk quotients greater than one suggest possible human health risk from drinking water, while quotients less than one indicate minimal risk (de Jesus Gaffney et al., 2015). These risk calculations are however limited as the impacts from mixtures of pharmaceuticals compounds and probable chronic effects are not addressed.

Results

Occurrence in groundwater and surface water

In the winter of 2017, samples were collected from 26 private groundwater sources in the West Branch of the Susquehanna and surface water at the watershed outlet. All groundwater samples contained at least one of the selected pharmaceutical compounds

(Table 4-5), with ofloxacin as the most frequently detected compound (Table 4-4).

Sulfamethoxazole was detected in 58% of the groundwater samples, with all levels above the MQL. The remaining compounds were detected in less than half of the samples collected, with caffeine and ampicillin detected in 46% of samples and acetaminophen and trimethoprim detected in 12% of samples. Naproxen, an anti-inflammatory drug, was not

112 detected in any of the groundwater samples. The most frequently detected compounds above the MDL were also detected at the highest concentrations (Figure 4-3), with ofloxacin, sulfamethoxazole, and caffeine quantified at concentrations as high as 122.7

μg/L, 32 μg/L, and 13.1 μg/L, respectively (Table 4-4). The mean concentrations detected in the groundwater samples were generally higher than the concentrations in surface water samples collected during the same period (winter 2017) at the outlet of the West Branch of the Susquehanna River (Table 4).

18Table 4-4: Summary of pharmaceutical concentrations (µg/L) in groundwater and surface water samples. Groundwater samples Surface water samples (n = 26) (n = 2) Pharmaceutical Std. % n > % n > Compound Mean Median Min. Max. Dev. MDL MQL January March Mean Acetaminophen 0.9 0.4 0.2 2.2 1.1 12% 12% 0.05 0.4 0.3 Ampicillin 0.4 0.4 0.2 0.7 0.2 46% 46% < MDL < MDL -- Caffeine 8.1 10.4 1.7 13.1 4.7 46% 46% 0.05 4.5 2.3 Naproxen MDL but

113

11Figure 4-3. Frequencies of detection and maximum concentrations of each compound of interest in groundwater well samples

114

19Table 4-5: Household source water characteristics and measured concentrations in groundwater and field blank samples

Concentration of Pharmaceutical Compounds (µg/L) in Groundwater and Field Blank Samples Site Characteristics Acetaminophen Ampicillin Caffeine Naproxen Ofloxacin Sulfamethoxazole Trimethoprim Well Depth Septic Field Field Field Field Field Field Field ID (m) System Groundwater Blank Groundwater Blank Groundwater Blank Groundwater Blank Groundwater Blank Groundwater Blank Groundwater Blank 1 50.29 Yes 2.18

†= Well depth not reported; ‡ = spring water source MDL = method detection limit; MQL= method quantification limit; Concentrations are reported as 0.05 μg/L or 1.5 μg/L (ofloxacin) for samples >MDL but

115

Predicted contaminant travel time and delivery ratios to groundwater

Vadose zone modeling was performed to explore the extent to which pharmaceutical physicochemical characteristics influence transport in the vadose zone and subsequent groundwater impact. Estimated travel times and delivery ratio of pharmaceuticals to a groundwater water table limiting zone 1.2 m below a septic absorption field are summarized in Table 4-6. Travel times were estimated as a function of retardation factors, range of water storage in the vadose zone and hydrologic variability. Average travel times varied between pharmaceuticals and ranged from less than 100 d for acetaminophen and sulfamethoxazole to several decades for ofloxacin.

In addition to sorption processes, pharmaceuticals in the vadose zone undergo microbial degradation that can further reduce contaminant mass delivered to groundwater.

The delivery ratio (DR), which is defined as the fraction of the original pharmaceutical mass (M0) expected to impact groundwater, was calculated using Equation 4-4, which accounts for pharmaceutical mass loss from the aqueous phase due to sorption and degradation processes in the vadose zone. The antibiotics trimethoprim and sulfamethoxazole had the highest delivery ratios, followed by ofloxacin, which was two orders of magnitude lower. The delivery ratios for the antibiotics indicate their likelihood to persist in the vadose zone and impact groundwater in comparison to the other evaluated pharmaceutical compounds.

116

20Table 4-6. Calculated mean travel and delivery ratio (M/M0) of pharmaceuticals to groundwater Travel Time to Delivery Ratio (M/M0) Pharmaceutical groundwater (d) to groundwater Compound Mean Std. Dev Mean Std. Dev Acetaminophen 78.1 22.8 7.9E-08 6.5E-05 Ampicillin 110.3 27.1 9.4E-11 1.2E-06 Caffeine 370.9 49.7 2.5E-30 4.4E-19 Naproxen 203.8 36.9 5.9E-15 5.4E-10 Ofloxacin 18,015 346.6 2.0E-04 3.2E-05 Sulfamethoxazole 98.9 25.7 0.2 8.3E-02 Trimethoprim 100.1 25.8 0.3 9.5E-02

Risk Calculations

Human health risk assessment was conducted by comparing average groundwater concentrations and calculated DWELs. All calculated RQs were less than 1, indicating minimal risk to human health (Table 4-7). The risk assessment is however limited as it does not address potential additive or synergistic effects from mixtures of pharmaceuticals in drinking water. Samples generally contained more than one target pharmaceutical compound, and likely other emerging contaminants that were not analyzed for in this study.

21Table 4-7: Acceptable daily intakes (ADIs), estimated drinking water equivalent levels (DWELs) and corresponding risk quotients for selected pharmaceuticals of interest. Pharmaceutical Compound ADI (µg/kg/day) DWEL (µg/L/day) Risk Quotient a Acetaminophen 340 10,428 0.00009 b Ampicillin 0.5 15 0.03 Caffeine c1.2 37 0.2 c Naproxen 7.1 218 0 Ofloxacin c5.7 175 0.05 a Sulfamethoxazole 130 3,987 0.004 a Trimethoprim 4.2 129 0.01

a Schwab et al., (2005) b Vragović et al., (2012) c Prosser & Sibley, 2015

117

Discussion

Comparison to other groundwater data

The trends for maximum concentrations and most frequently detected pharmaceuticals in groundwater samples are similar to other studies that have examined these compounds.

In a nationwide study in the United States evaluating pharmaceuticals and hormones in groundwater used as drinking water sources, the frequency of detection was higher in domestic wells than public supply wells and sulfamethoxazole, acetaminophen and caffeine were the most frequently detected compounds (Bexfield et al., 2019). For 20 domestic wells in Cape Cod, MA, Schaider et al. (2016) reported the highest detection frequencies for sulfamethoxazole (45%) and carbamazepine (25%) with corresponding highest maximum concentrations of 60 ng/L and 62 ng/L, respectively. Likewise, low detection frequencies were observed for trimethoprim (5%). At a spray-irrigation water reuse site in Central PA where wastewater effluent has been irrigated in forested and agricultural fields for over 40 years, Ayers et al. (2017) detected pharmaceuticals in 14 groundwater wells (depths ~ 50-100 m). Caffeine was the most frequently detected compound (55% of samples) while ofloxacin was present at the highest concentration (up to 116.4 μg/L). All other compounds were present in less than 45% of the samples and concentrations of individual pharmaceuticals were typically less than 10 μg/L (Ayers et al.,

2017). Caffeine is one of the most frequently detected compounds in groundwater samples

(Barnes et al., 2008; Fram & Belitz, 2011; James et al., 2016; Verstraeten et al., 2005).

Well depth is an important factor influencing the concentrations of pharmaceuticals in groundwater. Studies have reported higher concentrations and frequencies of detection in

118 shallow wells with depths typically < 23 m in comparison to deeper wells (Bexfield et al.,

2019; Barnes et al., 2008; Verstraeten et al., 2005). However, in the present study, no correlation was observed between well depth and concentrations of pharmaceuticals. This could be associated with different well selection strategies across studies. In the present study, well depth was not considered as a selection factor. Rather, homeowners volunteered for the study.

Hydrophobicity of a compound can influence its ability to sorb to organic matter and clay in the soil. Hydrophilic compounds are highly mobile and have the tendency to contaminate groundwater because of weaker retention by soil materials (Del Rosario et al.,

2014). Pharmaceuticals selected for this study were generally hydrophilic, with ofloxacin, the most frequently detected compound above MDL in all groundwater samples being the most hydrophilic (log KOW = -0.39; Table 4-1), followed by caffeine (log KOW = -0.07) that was detected in 46% of groundwater samples. Naproxen was the most hydrophobic compound of the selected pharmaceuticals and was not detected in any groundwater sample, potentially due to its high tendency to sorb to organic matter in soil. Similarly, other studies have reported emerging contaminants with high solubility, low log KOW and corresponding low log KOC to occur at high concentrations and frequencies in groundwater

(Bexfield et al., 2019; McEachran et al., 2017). Other factors such as compound ionization at ambient soil pH values can increase mobility of hydrophobic compounds in soil due to charge repulsion with the soil matrix. For instance, ibuprofen (log KOW = 3.97) is anionic and was mobile in negatively charged soils, resulting in its occurrence in groundwater (Del

119

Rosario et al., 2014). The role of compound physicochemical characteristics on their fate and transport in the vadose zone is further explored in the next section.

The one-time sampling of groundwater at the study sites through master well owner network volunteers did not allow for evaluation of seasonal variations. Although emerging contaminants in surface water sources typically vary seasonally in conjunction with consumer use patterns (Hedgespeth et al., 2012), seasonal variations in the groundwater are dampened through vadose zone attenuation processes and groundwater residence times

(Teijon et al., 2010). Furthermore, in a one-year monitoring study at a wastewater reuse site in Central PA, insignificant seasonal variations in impacted groundwater were observed (Kibuye et al., 2019). Biel-Maeso et al. (2018) reports lower biodegradation rates of pharmaceuticals in soil during colder seasons of the year due to reduced microbial activity, which can result in high vadose zone concentrations. Higher groundwater concentrations during such seasons are especially likely if transport is facilitated by the presence of preferential flow paths.

Comparison of groundwater samples to surface water samples

The concentrations in groundwater samples were higher than the concentrations at the watershed outlet (Table 4-4). These findings contrast with a nationwide study conducted in the United States that found concentrations of antibiotics and over-the-counter medications, among other emerging contaminants, to be present at higher concentrations in surface water than groundwater samples (Focazio et al., 2008). Concentrations of emerging contaminants in surface water fluctuate as a function of stream flow condition

120

(Reif et al., 2012). Therefore, flow conditions at the time of surface water sampling likely affected the concentrations observed, with groundwater concentrations potentially varying less than surface water concentrations. High discharge during the sampling period (January

– March) in the river may have resulted in a dilution effect that can lead to lower concentrations of pharmaceuticals that are predominantly present in wastewater effluent.

In Pennsylvania, pharmaceuticals have been observed at higher concentrations in surface water during low flow conditions due to a more dominant contribution of wastewater effluent to the total stream discharge compared to the dilution effect of that wastewater signal during higher flow conditions (Reif et al., 2012).

Predicted travel times and delivery ratios to groundwater

The modeling approach investigated the role of contaminant physicochemical properties in hydrologic-biogeochemical filtering in a septic absorption unit. Average travel times and delivery ratios to a water table limiting zone varied by compound depending on sorption potential and biodegradability. In a septic drainfield study, Yang et al. (2016) reported sorption and microbial degradation as the major mechanisms that limited the transport of pharmaceuticals. Therefore, physicochemical characteristics of pharmaceuticals that influence sorption and degradation extent are important factors in understanding occurrence in groundwater impacted by septic tanks.

Compounds with the highest retardation factors, such as ofloxacin (Table 4-2), were predicted to have the longest travel time, since their transport is slowed through sorption processes (Table -6) as opposed to pharmaceuticals with retardation factors close to unity.

121

However, due to slow biodegradation rates in soil, trimethoprim, sulfamethoxazole, and ofloxacin had the highest mean delivery ratios to groundwater. Consistently, in the current study, the highest detection frequencies and concentrations were exhibited by antibiotics sulfamethoxazole and ofloxacin. The transport of pharmaceuticals in the vadose zone can further be influenced by ionization processes at ambient soil pHs that can increase mobility of both hydrophobic and hydrophilic compounds in soil water (Lapworth et al., 2012). For instance, sorption studies by Srinivasan et al. (2013) highlighted pH dependent sorption of sulfamethoxazole such that cationic form of sulfamethoxazole sorbed more at experimental pH values near pKa1 and low sorption was observed when pH ≥ pKa2 since the anionic species were dominant. This modeling approach is thus limited as predicted groundwater delivery ratios do not incorporate pharmaceutical transport effects due to ionization characteristics.

In the studied private groundwater sources, concentrations and detection frequencies varied between sites, with detections at each site ranging from six out of the tested seven pharmaceuticals to only one detected compound (Table 4-5). Similarly, concentrations of pharmaceuticals at some sites were more than ten times higher than the average and median concentrations shown in Table 4-4. This can be due to varying household and/or site-specific characteristics that influence pharmaceutical transport to groundwater. Household-specific characteristics can include consumer pharmaceutical use patterns, number of household occupants, and volume of wastewater produced (Conn et al., 2006), which vary temporally within a household and spatially between households.

Some site-specific factors include varying degrees of pharmaceutical removal in septic

122 tanks (Godfrey et al., 2007) before dispersal to the absorption field and hydraulic loading rates to the absorption fields. High pharmaceutical concentration in septic effluents dispersed at low hydraulic loading rates may have lower mass loading to the soil absorption unit than low concentrations dispersed at higher hydraulic loading rates (Conn et al., 2006).

Soil physical, chemical and biological characteristics at the sites can impact attenuation processes as the septic effluent moves through the vadose zone. Sorption processes for antibiotics are higher in soils with high organic carbon contents and ionic strengths (Srinivasan et al., 2013) and degradation processes are higher in oxic conditions

(Carrara et al., 2008; Swartz et al., 2006). Estimations of average groundwater delivery ratios and travel times in the current study did not include varying the vadose zone organic carbon content since site specific averages obtained from Web Soil Survey

(https://websoilsurvey.sc.egov.usda.gov/) were used as model inputs. Other factors such as the presence of macropore flow in the soil profile can lower travel time in the vadose zone and increase contaminant mass delivered to groundwater. A 1.2 m distance between the bottom of the absorption unit and a water limiting zone is used in the model. Higher average delivery ratios and shorter travel time to underlying groundwater are expected when this distance is short as opposed to a longer separation distance with a deep suitable soil layer.

Compounds such as acetaminophen, ampicillin, caffeine, and naproxen are predicted to travel fast in the vadose zone, however due to their rapid biodegradation rates in soil, their average delivery ratios are lower (Table 4-6). Acetaminophen (0.9 µg/L) and ampicillin (0.4 µg/L) had the lowest average concentrations in the measured private groundwater while caffeine was among the most frequently detected with mean

123 concentrations of 8.1 µg/L (Table 4-4). These compounds may have high loading rates to the absorption unit due to a more frequent consumption relative to other prescription and over-the-counter pharmaceuticals. For instance, the average caffeine consumption rate per person in the United States is about 210 mg/d (Buerge et al., 2003). Naproxen was not detected in any groundwater samples. Since approximately 99% of naproxen is absorbed in the body with less than 1% excreted (Oosterhuis et al., 2013), concentrations in domestic waste are expected to be low. Biodegradation processes in septic tanks and soil can also lead to the transformation of naproxen into metabolites that were not evaluated in the current study. Other studies have detected naproxen in groundwater impacted by wastewater irrigation at concentrations ranging between 1.3 - 12 ng/L (McEachran et al.,

2017; McEachran et al., 2016), which are below the MQL for this study.

Human health risk assessment

Risk assessment calculations revealed that even the highest RQ values were several orders of magnitude lower than 1 (Table 4-7), which is the threshold used to differentiate between a low and high risk (de Jesus Gaffney et al., 2014). This suggests minimal risk through consumption of drinking water, as substantial margins exist between average concentrations in wells and the estimated DWEL values. The World Health Organization

(2012) concluded adverse human health effects were unlikely as a result of chronic exposure to pharmaceuticals in drinking water. Risk assessments by other studies have also concluded low human health impacts (de Jesus Gaffney et al., 2014).

124

The risk calculation may potentially underestimate overall risk since potential mixture and chronic effects are unaccounted for and because the risk assessment does not take into consideration exposure due to other potential pathways. Therefore, while there appears to be minimal risk to human health from the concentrations present in the wells sampled in this study, the results do suggest the potential for septic tanks to impact private well water quality. Samples were only collected once as part of this study, and therefore it is unknown how much concentrations may vary in groundwater over an extended period of time.

Routine monitoring may be desired to better understand the range of pharmaceutical concentrations present in private well water, in addition to other water quality parameters that have drinking water standards and better understood risks.

Conclusions

In winter 2017, the presence and concentrations of seven selected compounds, spanning a wide range of physicochemical parameters were evaluated in 26 private wells in the West

Branch of the Susquehanna River basin. The most commonly detected compounds were ofloxacin, sulfamethoxazole, ampicillin and caffeine. Average concentrations of each pharmaceutical were < 20 µg/L and were generally higher than concentrations observed at the watershed outlet during the same sampling period. A simple modeling approach was used to evaluate the influence of compound physicochemical properties on their fate and transport in the vadose zone. Estimations of average travel time in the vadose zone and delivery ratios to groundwater highlight that the extent of groundwater contamination by pharmaceutical compounds is controlled by both compound sorption potential and

125 biodegradability. Compounds with the highest mean delivery ratios including antibiotics such as ofloxacin, sulfamethoxazole, and trimethoprim were consistently among the most frequently detected compounds at high concentrations in the analyzed groundwater samples. Transport in the vadose zone and groundwater impact were majorly a strong function of retardation factors and biodegradation rates in soil, however, influences from compound ionization are not incorporated in the model. Risk assessment calculations using measured average groundwater concentrations suggested minimal human health risk from consumption of the private well water analyzed. Nevertheless, samples collected for extended periods of time during additional seasons to capture a range of environmental conditions would provide further insight into the occurrence and concentrations of these pharmaceuticals across the watershed in both surface and groundwater sources.

Data generated was summarized and shared with homeowners and was followed by an informational webinar that elaborated mechanisms on the fate and transport of emerging contaminants in groundwater. An Example of data communication template used is detailed in Appendix B.

References

Ayers, B., Elkin, K., Kibuye, F., & Gall, H. E. (2017). Pharmaceuticals at Penn State’s Living Filter: From Wastewater to Groundwater. ASABE Paper No. 1700255. St. Joseph, MI: American Society of Agricultural and Biological Engineering (ASABE).

126

Bexfield, L. M., Toccalino, P. L., Belitz, K., Foreman, W. T., & Furlong, E. T. (2019). Hormones and Pharmaceuticals in Groundwater Used As a Source of Drinking Water Across the United States. Environmental Science & Technology.

Biel-Maeso, M., Corada-Fernández, C., & Lara-Martin, P. A. (2018). Monitoring the occurrence of pharmaceuticals in soils irrigated with reclaimed wastewater. Environmental Pollution, 235, 312-321.

Buerge, I. J., Poiger, T., Müller, M. D., & Buser, H. R. (2003). Caffeine, an anthropogenic marker for wastewater contamination of surface waters. Environmental Science & Technology, 37(4), 691-700.

Conn, K. E., Barber, L. B., Brown, G. K., & Siegrist, R. L. (2006). Occurrence and fate of organic contaminants during onsite wastewater treatment. Environmental Science & Technology, 40(23), 7358-7366. de Jesus Gaffney, V., Almeida, C. M., Rodrigues, A., Ferreira, E., Benoliel, M. J., & Cardoso, V. V. (2015). Occurrence of pharmaceuticals in a water supply system and related human health risk assessment. Water Research, 72, 199-208.

Focazio, M. J., Tipton, D., Dunkle Shapiro, S., & Geiger, L. H. (2006). The chemical quality of self‐supplied domestic well water in the United States. Groundwater Monitoring & Remediation, 26(3), 92-104.

Focazio, M. J., Kolpin, D. W., Barnes, K. K., Furlong, E. T., Meyer, M. T., Zaugg, S. D., Larry, B. B., & Thurman, M. E. (2008). A national reconnaissance for pharmaceuticals and other organic wastewater contaminants in the United States—II) Untreated drinking water sources. Science of the Total Environment, 402(2-3), 201-216.

127

Glassmeyer, S. T., Furlong, E. T., Kolpin, D. W., Batt, A. L., Benson, R., Boone, J. S., Conerly, O., Donohue, M. J., King, D. N., Kostich, M. S. & Mash, H. E., 2017. Nationwide reconnaissance of contaminants of emerging concern in source and treated drinking waters of the United States. Science of the Total Environment, 581, 909-922.

Godfrey, E., Woessner, W. W., & Benotti, M. J. (2007). Pharmaceuticals in on‐site sewage effluent and ground water, Western Montana. Groundwater, 45(3), 263-271.

Harman, C. J., Rao, P. S. C., Basu, N. B., McGrath, G. S., Kumar, P., & Sivapalan, M. (2011). Climate, soil, and vegetation controls on the temporal variability of vadose zone transport. Water Resources Research, 47(10).

James, C. A., Miller-Schulze, J. P., Ultican, S., Gipe, A. D., & Baker, J. E. (2016). Evaluating contaminants of emerging concern as tracers of wastewater from septic systems. Water Research, 101, 241-251.

Karkanis, P. G. (1983). Determining field capacity and wilting point using soil saturation by capillary rise. Canadian Agricultural Engineering, 25, 19-21.

Katz, B. G., Griffin, D. W., McMahon, P. B., Harden, H. S., Wade, E., Hicks, R. W., & Chanton, J. P. (2010). Fate of effluent-borne contaminants beneath septic tank drainfields overlying a karst aquifer. Journal of Environmental Quality, 39(4), 1181-1195.

Kibuye, F. A., Gall, H. E., Elkin, K. R., Ayers, B., Veith, T. L., Miller, M., Jacob, S., Hayden, K. R., Watson, J. E., & Elliott, H. A. (2019). Fate of pharmaceuticals in a spray-irrigation system: From wastewater to groundwater. Science of The Total Environment, 654, 197- 208.

Lapworth, D. J., Baran, N., Stuart, M. E., & Ward, R. S. (2012). Emerging organic contaminants in groundwater: a review of sources, fate and occurrence. Environmental Pollution, 163, 287-303.

Li, J., Ye, Q., & Gan, J. (2014). Degradation and transformation products of acetaminophen in soil. Water Research, 49, 44-52.

Marti, R., Scott, A., Tien, Y. C., Murray, R., Sabourin, L., Zhang, Y., & Topp, E. (2013). The impact of manure fertilization on the abundance of antibiotic-resistant bacteria and frequency of detection of antibiotic resistance genes in soil, and on vegetables at harvest. Applied and Environmental Microbiology, AEM-01682.

128

McEachran, A. D., Shea, D., & Nichols, E. G. (2017). Pharmaceuticals in a temperate forest-water reuse system. Science of the Total Environment, 581, 705-714.

McEachran, A. D., Shea, D., Bodnar, W., & Nichols, E. G. (2016). Pharmaceutical occurrence in groundwater and surface waters in forests land‐applied with municipal wastewater. Environmental Toxicology and Chemistry, 35(4), 898-905.

Monk, J. P., & Campoli-Richards, D. M. (1987). Ofloxacin: A review of its antibacterial activity, pharmacokinetic properties and therapeutic use. Drugs, 33, 346-391.

Monteiro, S. C., & Boxall, A. B. (2010). Occurrence and fate of human pharmaceuticals in the environment. In Reviews of Environmental Contamination and Toxicology (pp. 53-154). Springer New York.

Penn State Extension. (2007). A quick guide to groundwater in Pennsylvania. Retrieved from http://extension.psu.edu/natural-resources/water/drinking-water/best-practices/a-quick- guide-to-groundwater-inpennsylvania

Pennsylvania Department of Environmental Protection (1995). Source water protection. Retrieved from http://www.sourcewaterpa.org/?page_id=2632

Pennsylvania Code. Title 25, Chapter 73. Standards for Onlot Sewage Treatment Facilities. (1997). Retrieved from https://www.pacode.com/secure/data/025/chapter73/chap73toc.html.

129

Schaider, L. A., Rudel, R. A., Ackerman, J. M., Dunagan, S. C., & Brody, J. G. (2014). Pharmaceuticals, perfluorosurfactants, and other organic wastewater compounds in public drinking water wells in a shallow sand and gravel aquifer. Science of the Total Environment, 468, 384-393.

Schwab, B. W., Hayes, E. P., Fiori, J. M., Mastrocco, F. J., Roden, N. M., Cragin, D., Meyerhoff, R.D., Vincent, J.D. & Anderson, P. D. (2005). Human pharmaceuticals in US surface waters: a human health risk assessment. Regulatory Toxicology and Pharmacology, 42(3), 296-312.

Smith, M., Cross, K., Paden, M., & Laban, P. (2016). Spring: Managing groundwater sustainably. International Union for Conservation of Nature (IUCN). Gland, Switzerland. https://portals.iucn.org/library/sites/library/files/documents/2016-039.pdf

Srinivasan, P., Sarmah, A. K., & Manley-Harris, M. (2013). Co-contaminants and factors affecting the sorption behaviour of two sulfonamides in pasture soils. Environmental pollution, 180, 165-172.

Swistock, B. R., Clemens, S., Sharpe, W. E., & Rummel, S. (2013). Water quality and management of private drinking water wells in Pennsylvania. Journal of Environmental Health, 75(6), 60-67.

U.S. Department of Housing and Human Development (2012). HUD HOC Reference Guide Water Systems: Individual Water Systems. Retrieved from https://archives.hud.gov/offices/hsg/sfh/ref/sfh1-21b.cfm

U.S. Environmental Protection Agency (2008). Retrieved June 1st, 2018, Private Drinking Water Wells, https://www.epa.gov/privatewells

130

U.S. Environmental Protection Agency (2008). Septic Systems Fact Sheet, US EPA. Office of Wastewater Management.

U.S. Environmental Protection Agency, (2010). Treating Contaminants of Emerging Concern, a Literature Review Database.

U.S. Environmental Protection Agency, (2017). Field Validation of Pollutant Measurement Methods from Waste Media. Retrieved from https://www.epa.gov/sites/production/files/2017-08/documents/method_301.pdf

Verlicchi, P., & Zambello, E. (2015). Pharmaceuticals and personal care products in untreated and treated sewage sludge: occurrence and environmental risk in the case of application on soil—a critical review. Science of the Total Environment, 538, 750-767.

World Health Organization. (2012). Pharmaceuticals in drinking-water. Retrieved from https://www.who.int/water_sanitation_health/publications/2012/pharmaceuticals/en/.

Zaayman, M., Siggins, A., Horne, D., Lowe, H., & Horswell, J. (2017). Investigation of triclosan contamination on microbial biomass and other soil health indicators. FEMS Microbiology Letters, 364(16).

131

Chapter 5

Seasonal variations of emerging organic contaminants (EOCs) in drinking water sources in the Susquehanna River Basin

Graphical Abstract:

Highlights: • 20 EOCs were monitored over a two-year period in six drinking water sources in PA. • Concentration-discharge relationships were used to classify transport dynamics. • EOCs were observed at higher concentrations in colder rather than warmer months. • Observed EOCs posed high risks to aquatic organisms but not to human health.

132

Abstract

Occurrence of emerging organic contaminants (EOCs) in surface water bodies can cause adverse effects on non-target organisms. When surface waters are used as drinking water sources, temporal variability in EOC concentrations can potentially impact drinking water quality and human health. To better understand spatiotemporal variability of EOCs in drinking water sources in Central Pennsylvania, EOCs were evaluated in six drinking water sources during a two-year study period (April 2016 – June 2018) in the Susquehanna

River Basin (SRB). The study was conducted in two phases: Phase I was a spatially distributed sampling approach within the SRB focusing on seven human pharmaceuticals and Phase II was a temporally intensive sampling regime at a single site focusing on a broader range of EOCs. Concentration-discharge relationships were utilized to classify

EOC transport dynamics and understand the extent to which hydrologic and anthropogenic factors, such as surface runoff and wastewater effluent may contribute to EOC occurrence.

Overall, EOCs were present at higher concentrations in colder seasons than warmer seasons. Thiamethoxam, a neonicotinoid insecticide, and caffeine exhibited accretion dynamics during high-flow periods, suggesting higher transport during surface runoff events. Human pharmaceuticals known to persist in wastewater effluent were inversely correlated with discharge, indicating dilution characteristics consistent with diminished wastewater signals during surface runoff events. Acetaminophen exhibited episodic transport dynamics indicating nonpoint source inputs during high-flow periods. Risk calculations revealed that although EOCs posed medium to high risk to fish and other aquatic organisms, human health risk through fish consumption was low.

133

Keywords: emerging contaminants, concentration-discharge relationships, seasonal variations, ecological and human health risk, surface water quality

Introduction

Understanding natural and anthropogenic drivers influencing surface water quality in drinking water sources is critical to protecting ecological and human health. The presence of emerging organic contaminants (EOCs) in these systems is affected by multiple natural and anthropogenic drivers, including hydroclimatic conditions, land-application of pesticides and human and livestock residuals, and discharge of treated wastewater effluent and combined sewer overflow (CSO) to surface water bodies. EOCs, including pharmaceuticals and personal care products (PPCPs) and some pesticides, currently lack water quality standards for wastewater effluent, surface water sources, and drinking water despite being known or suspected endocrine disruptors that pose adverse impacts on sensitive non-target organisms (Leet et al., 2011) and promote antibiotic resistance development (Boxall et al., 2012).

Widespread occurrence of EOCs in aquatic ecosystems is affected by proximity to human, agricultural, and industrial wastewater sources (Kolpin et al., 2002). Furthermore,

EOCs exhibit temporal variations associated with seasonal factors influencing PPCP use

(Yu et al., 2013), wastewater treatment plant (WWTP) removal efficiencies (Hedgespeth et al., 2012), and land applications of manure, biosolids, and pesticides (Bernot et al.,

2013). Observed EOC temporal fluxes in surface water are influenced by seasonal hydroclimatic patterns resulting in either dilution (Reif et al., 2012) or EOC mobilization

134 from diffuse nonpoint sources (Benotti & Brownawell, 2007). Therefore, EOC occurrence in surface water sources is not only influenced by point and nonpoint source distribution within a watershed, but also hydroclimatic variability throughout the watershed that can induce a range of contaminant transport responses.

Contaminant transport as a function of hydroclimatic conditions and subsequent impact on surface water quality is largely understood for conventional pollutants such as nutrients, suspended sediments, and pathogens (e.g., Cryptosporidium and Escherichia coli). However, few studies focus on EOC concentrations in surface water sources as a function of streamflow conditions (Benotti & Brownawell, 2007; Launay et al., 2013;

Pailler et al., 2009; Zhang et al., 2016) primarily due to higher costs associated with analyzing for EOCs as compared to nutrients and conventional water quality parameters.

Furthermore, studies seeking to examine EOC variability over seasons or hydrologic conditions generally conduct minimal sampling campaigns that are likely not fully representative of the range of seasonal and hydrologic variability (Cantwell, 2018; Loraine

& Pettigrove, 2006; Mijangos et al., 2018; Moreno-González et al., 2013), while studies conducting rigorous sampling tend to focus on trends at a few locations, potentially limiting the generalizability of findings (Gall et al., 2011; Pailler et al., 2009).

EOC transport dynamics during high and low streamflow conditions is compound specific and varies between study sites. Launay et al. (2013) investigated pollution dynamics in a catchment in Germany influenced by urban runoff and reported caffeine, N,

N-Diethyl-m-toluamide (DEET), 2-methylthio-benzothiazole (MTBT), and carbamazepine concentration increases during high-flow conditions associated with CSO

135 discharges. In an estuary in Jamaica Bay, NY, pharmaceutical levels before and after storm events varied by compound (Benotti & Brownawell, 2007); caffeine, codeine, cotinine, sulfamethoxazole, and trimethoprim exhibited lower concentrations following wet weather events, while acetaminophen and nicotine concentrations remained relatively constant in dry and wet weather conditions.

In comparison to groundwater, which is protected by soil layers acting as biogeochemical filters for EOCs (McEachran et al., 2017; Kibuye et al., 2019), surface water sources are dynamic and more vulnerable to point and nonpoint EOC sources.

Understanding dominant drivers of temporal variations of EOCs in impacted surface water is therefore essential for ecological and human health risk characterization, for estimating influent loads to drinking water treatment plants, and for designing best management practices to reduce EOC occurrence in the environment.

The objective of this study was to evaluate spatial and temporal variations of EOCs at six surface drinking water sources in the Susquehanna River Basin (SRB). This study was conducted in two phases: Phase I was a spatially distributed sampling approach within the SRB focusing on seven human pharmaceuticals, and Phase II was a temporally intensive sampling regime at a single site focusing on a broader range of EOCs. The study sites consisted of riverine and reservoir drinking water sources. Influences of hydrologic drivers on EOCs were evaluated through the development of concentration-discharge (C-

Q) relationships, depicting concentration variations as a function of varying hydrologic conditions. Results were used to understand the role that various anthropogenic sources, play in the occurrence of EOCs in aquatic environments. Risk assessments were conducted

136 to understand potential aquatic and human health risks posed by observed EOC concentrations.

Methodology

Site Description

In collaboration with PA American Water company, six drinking water sources located in the mid-to-upper and mid-to-lower regions of the Susquehanna River Basin

(SRB) were selected as study sites. The exact location of each sampling location is confidential, but A, B, and C are reservoir sources and D, E, and F are riverine sources.

The surface watersheds of the selected sites vary in land use composition (Figure 5-1) with sites E and F containing substantially more agricultural land than the other sites and sites

C, E, and F containing double the percentage of developed land as sites A, B, and D (Table

5-1). The selected sites serve as drinking water sources for 50,000 – 135,000 people.

22Table 5-1: Watershed characteristics of the selected study sites. Drinking Water Total Drainage Land Use Distribution Population Source Site Area (km2) Forest Developed Agricultural Open Water Served Reservoir Sources A 38 92.7% 5% 0.6% 1.8% 54,000 B 109 88% 4.4% 5.8% 1.8% 58,500 C 172 83.5% 10.1% 3.4% 3.4% 135,000 Riverine Sources D 17,339 82.2% 5.4% 11.9% 0.5% 30,000 E 1253 43.6% 14.5% 41.1% 0.8% 35,000 F 517 56.6% 11.5% 31.5% 0.4% 95,000

137

12Figure 5-1: Land use maps of the study sites in the Susquehanna River Basin. Sites A and C have two sub-watersheds serving as surface water inputs to the reservoirs.

Sample Collection

This study was conducted in two Phases. In Phase I (April 2016-December 2017), a spatially distributed sampling technique was implemented whereby grab water samples were collected at monthly and bimonthly scales from all six sites and samples from each set of sources (riverine and reservoir) were collected on the same day. All samples were collected in 250 mL amber glass bottles with polytetrafluoroethylene (PTFE)-lined caps

138 following EPA method 1694 (USEPA, 2007). Water quality indicators such as temperature, pH, conductivity, and dissolved oxygen were recorded in situ using calibrated handheld meters. In addition to EOCs, natural organic matter present in source water samples was characterized in terms of total organic carbon (TOC). TOC samples were collected in 250mL clear polyethylene terephthalate (PET) bottles and analyzed at the Penn

State Institutes of Energy and the Environment (PSIEE) Water Quality Laboratory in

University Park, PA.

Phase II (January-June 2018) was a temporally intensive sampling regime focusing spatially on site E which had the greatest percentages of developed land and agricultural land among the study sites. An automated ISCO 3700 sampler fitted with twenty-four

350mL glass bottles was programmed to collect stream water samples at time steps varying between 3 and 12h depending on the flowrate at the time sampling commenced. Sampling dates and times were determined using real-time United States Geological Survey (USGS) streamflow data from a USGS station located about 10km upstream from the sampling site.

A Campbell Scientific Inc. CR850 datalogger equipped with water level, temperature, dissolved oxygen, and pH sensors was deployed and programmed to scan each sensor every second and store measured averages every 60min. Water level measurements were correlated with reported USGS discharge data at the nearest monitoring station (R2=0.9).

Samples assessing concentrations during baseflow conditions were collected following a period of a short dry spell of no recorded precipitation (rainfall or snowfall) for at least 3d prior to the sampling date, as the average time between precipitation events in PA is 3d.

Post-storm samples were collected during high-flow events constituting the upper 20th

139 percentile of the full flow regime. All samples were preserved on ice during transportation to laboratories for processing and analysis and stored at 4°C before processing within 48 h of collection. Samples for EOCs were analyzed at the United States Department of

Agriculture-Agricultural Research Service (USDA-ARS) laboratory.

Sample Processing and Targeted Analysis of EOCs

In Phase I, all water samples were analyzed for seven human pharmaceutical compounds, including four antibiotics (ampicillin, sulfamethoxazole, ofloxacin, and trimethoprim), two analgesics (acetaminophen and naproxen), and a stimulant (caffeine).

These compounds have a wide range of physicochemical properties (Table 5-2) and are representative of a broader array of pharmaceutical and personal care product (PPCP) compounds. Caffeine and acetaminophen were included due to their common household usage and their utility as indicators of human wastewater sources. These two compounds, along with sulfamethoxazole and trimethoprim, are among the most frequently detected in

PA waters (Reif et al., 2012). However, the other three analytes of interest, ampicillin, naproxen and ofloxacin, remain understudied.

In Phase II, 13 more compounds, including human and veterinary antibiotics

(chlortetracycline, tetracycline, oxytetracycline, erythromycin, sulfadiazine, sulfadimethoxine, sulfamethazine, and tylosin), an antimicrobial (triclosan), an antihistamine (cimetidine), an antidiabetic (metformin), a caffeine metabolite

(theobromine), and a neonicotinoid insecticide (thiamethoxam) were targeted for analysis.

These compounds were selected based on standard availability and analytical capabilities,

140 frequency of use, and a hypothesized occurrence in surface water due to previously documented seasonal consumption and use patterns (Singer et al., 2014). Samples were analyzed and quantified using a high-resolution accurate mass (HRAM) Q Exactive

Orbitrap mass spectrometer (ThermoFisher Scientific, Bremen, Germany), interfaced to a chromatography system through a heated electrospray injection (HESI) source. The limit of detection (LOD) and quantification (LOQ) was 0.01µg/L (signal to noise ratio ≥3) and

0.1µg/L (signal to noise ratio ≥10), respectively, for all compounds except ofloxacin, which had an LOD and LOQ of 0.3µg/L and 3µg/L, respectively. Analytes measured between

LOD and LOQ were reported as LOQ/2. More details on methodology and high- performance liquid chromatography (HPLC) and mass spectrometer are summarized in

Kibuye et al. (2019).

37% of samples were field and travel blanks obtained during sampling trips and handled similarly as surface water samples. Field blanks were obtained by transporting two

500mL bottles of nanopure water to each sampling site and transferring to a 250mL amber glass sampling bottle. Each travel blank sample consisted of one 250mL nanopure water transported to sampling sites but not opened in the field. Detected EOCs in blanks were averaged and used to censor surface water concentrations.

141

23Table 5-2: Physicochemical characteristics of selected compounds Compound Chemical Molar Mass bLog a Formula (g/mol) pKa KOW Human Antibiotics Ampicillin C16H19N3O4S 349.40 2.5; 7.3 1.35 Sulfamethoxazole C10H11N3O3S 253.28 1.6; 5.7 0.89 Ofloxacin C18H20FN3O4 361.37 5.97; 9.28 -0.39 Trimethoprim C14H18N4O3 290.32 7.12 0.91 Antidiuretic Metformin HCL C4H12ClN5 165.63 12.4 -0.5 Antihistamine Cimetidine C10H16N6S 252.34 6.8 0.40 Analgesics Acetaminophen C8H9NO2 151.17 9.38 0.46 Naproxen C14H14O3 230.26 4.15 3.18 Stimulant Caffeine C8H10N4O2 194.19 10.4 -0.07 Metabolites Theobromine C7H8N4O2 180.17 9.9 -0.78 Antimicrobial Triclosan C12H7Cl3O2 289.54 7.9 4.73 Veterinary Antibiotics Chlortetracycline HCl C22H24Cl2N2O8 515.34 -- -- Erythromycin C37H67NO13 733.94 8.88 3.06 Tetracycline HCl C22H25ClN2O8 480.90 3.30 -1.37 Oxytetracycline HCl C22H25ClN2O9 496.90 9.5 -0.90 Sulfadiazine C10H10N4O2S 250.28 6.36 -- Sulfadimethoxine C12H14N4O4S 310.33 -- -- Sulfamethazine C12H14N4O2S 278.33 2.07; 7.49 0.14 Tylosin C46H77NO17 916.11 7.73 1.63 Neonicotinoid Insecticide Thiamethoxam C8H10ClN5O3S 291.71 -- -0.13

Statistical Analysis

Data analyses were performed using IBM SPSS Statistics version 23 (SPSS Inc.,

Chicago, Illinois, USA). A general linear model analysis of covariance (ANCOVA) was

a - Log of the acid dissociation constant b octanol-water partition coefficient Sources: https://pubchem.ncbi.nlm.nih.gov

142 used to test for fixed effects of season and source water type on concentrations of pharmaceuticals and on the influence of five covariate variables: pH, water temperature, dissolved oxygen, conductivity, and TOC. ANCOVA assumptions were checked using

Pearson’s correlation analysis. Three covariates-water temperature, dissolved oxygen, and conductivity were significantly correlated (α=0.01) thus only water temperature, pH, and

TOC were included in the ANCOVA model. To check for significance of source type and site differences, a one-way ANOVA test was performed. Further post-hoc analysis using

Tukey's Honestly Significant Difference (HSD) test was applied when there was a significant difference between means. All statistical hypotheses were tested at α=0.05 except for Pearson’s correlation tests, where α=0.05 and 0.01 were employed.

Concentration-Discharge (C-Q) Relationships and CVC/CVQ Ratios

Relationships between concentration (C) and discharge (Q) were employed to characterize contaminant transport dynamics using a power-law relationship C=aQb, where a and b are empirical constants (Vogel et al., 2005). When C and Q are plotted on a log- log scale, slopes of plotted relationships represent values of b that can be used to characterize the C-Q pattern (Vogel et al., 2005) as dilution (b<0), accretion (b>0), or chemostatic (b≈0) (Basu et al., 2010; Gall et al., 2015; Miller et al., 2019; Vogel et al.,

2005). C-Q relationships were developed using concentrations >LOQ for the most frequently detected compounds in Phase II. To explore concentration and flow variability in contaminant transport, a ratio of coefficients of variation (CV) for concentration (CVC) and flow (CVQ) was employed following Thompson et al. (2011). A high CVC/CVQ ratio

143 is expected when concentration variability is greater than flow variability, while a low ratio indicates a larger flow variability (Gall et al., 2013; Thompson et al., 2011). Thompson et al. (2011) further suggested a CVC/CVQ<0.3 as a nonparametric indicator of chemostatic dynamics (b≈0) and ratios >0.3 to indicate episodic transport where concentration and flow variability are important

Risk calculations

Ecological Risk Calculations

To estimate environmental risk posed by measured EOCs in aquatic systems, risk quotients (RQs) were calculated for representative trophic levels including algae, Daphnia, and fish. Predicted no effect concentration (PNEC) of EOCs were estimated using literature-derived acute toxicity data for algae, Daphnia, and fish by dividing literature values of EC50 or LC50 (the concentrations at which 50% of population exhibits a response or dies, respectively; Table 5-3) by an assessment factor of 1000. RQs associated with respective EOCs were then calculated by dividing measured environmental concentrations (MECs) by the corresponding estimated PNEC values (Sanderson et al.,

2003).

EC 푃푁퐸퐶 (µ푔 퐿−1) = 50 (Equation 5-1) 1000

푀퐸퐶 푅푄 = (Equation 5-2) P푁퐸퐶

Calculated risk quotients were summed to estimate an overall RQ. Relative risk contributions of individual EOCs to the total RQ index at any given time (T) were then

144 estimated, for the seven EOCs in Phase I and 20 EOCs in Phase II, as follows (Ginebreda et al., 2010):

푅푄푇 = ∑푗 푅푄푗 (Equation 5-3)

푅푄 퐸푂퐶 푗 푅푖푠푘 퐶표푛푡푟푖푏푢푡푖표푛 (%) = 푗 × 100 (Equation 5-4) 푅푄푇 where RQT represents the total RQ index for measured EOCs at any given time (T).

Human Health Risk Calculations

The primary human exposure route to EOCs in surface waters is through treated drinking water, especially when EOCs present in drinking water sources persist through water treatment processes. Other exposure routes can include consumption of fish from impacted surface water sources. The concentrations or dose of exposure for fish from EOCs in surface water can be estimated as follows (Muñoz et al., 2010):

퐶푓푖푠ℎ = C푊푎푡푒푟 × 퐵퐶퐹푓푖푠ℎ × 퐵푀퐹푓푖푠ℎ (Equation 5-5)

-1 -1 where Cfish is the estimated concentration in fish (µg kg wt), BCF (L kgwt ) is the bioconcentration factor in fish and BMF (dimensionless) is the biomagnification factor in fish. BCF can be estimated based on the guidelines by the European’s Commission’s

Technical Guidance Document on Risk Assessment (European Commission, 2003) that employs a compound log Kow using equation 5-6:

퐿표푔 퐵퐶퐹 = 0.85 Log K표푤 − 0.7 (Equation 5-6)

BMFs were also estimated using the approach based on log Kow of EOCs due to lack of data (European Commission, 2003). The estimated human exposure dose through the consumption of fish from the surface waters studied is calculated using equation 5-7:

145

퐶 × 퐼푛푡푎푘푒 퐷표푠푒 (µ푔 퐾푔 −1푑−1) = 푓푖푠ℎ 푓푖푠ℎ (Equation 5-7) 푏푤 BW

th where Cfish is the estimated concentration in fish and BW is the 50 percentile adult body weight of 60kg (de Jesus Gaffney et al., 2015). The average fish consumption rate in the

U.S. is 0.02 kg d-1 (National Marine Fisheries Service, 2015). Estimated dose exposure was compared with acceptable daily human intakes (ADI) of respective EOCs obtained from literature by dividing the dose with ADIs to get a risk quotient. Quotients below 1 suggest that adverse effects are not expected from the consumption of fish while quotients more than unity imply probable human health risks. Table 5-3 below summarizes the calculated

BCF and EC50/LC50 values used in risk calculations.

146

24Table 5-3: EC50 and LC50 (mg/L) used to calculate PNEC (by dividing EC50 by an assessment factor of 1000), for fish, algae and Daphnia EC50/ LC50 (mg/L) Compound aBCF Fish Algae Daphnia

Acetaminophen 0.49 b378 b132 b9.2 c Ampicillin 2.80 -- >1000 -- d e Caffeine 0.17 90 -- 16 Metformin 0.07 f*383.3 f320 f81.4 Naproxen 100.69 g34 g22 g15 Ofloxacin 0.09 h16 h4.74 i31.75 Sulfamethoxazole 1.14 g890 g51 j189.2 Trimethoprim 1.18 k92.66 k83.8 k8.21 Triclosan 2091.70 ------j j Chlortetracycline -- *78.9 3.1 -- Erythromycin 79.62 j*>100 j0.02 j22.45 Oxytetracycline 0.03 j*110-215.4 j4.5 j22.64

Tetracycline 0.01 -- j2.2 j44.8

Theobromine 0.04 ------

Thiamethoxam 0.15 j3.7 l81.8 l>100

Cimetidine 0.20 ------

Tylosin 4.85 -- j0.034 j680 Sulfadiazine -- -- j0.135 j221 sulfadimethoxine -- j*>100 j2.3 j248 Sulfamethazine 0.26 j*>100 -- j202

a Calculated Bioconcentration factors b Grung et al., (2008) c Eguchi et al., (2004) d Moore et al., (2008) e Lilius et al., 1995 f Lee, (2017) g Sanderson et al., (2003) h Ferrari et al., (2004) i Isidori et al., (2005) j Santos et al., (2010) k De Liguoro et al., (2012) l Finnegan et al., (2017) * LC50 values

147

Results and Discussion

Occurrence in the Susquehanna River Basin

EOC concentration spanned about one to two orders of magnitude throughout the sampling period. In Phase I, the most frequently detected compounds were sulfamethoxazole (54%), acetaminophen (42%), and caffeine (35%), while trimethoprim, naproxen, ofloxacin and ampicillin were detected in <35% of the samples (Figure 5-2).

Although not among the most frequently detected compounds, naproxen (31.38µg/L), ofloxacin (9.95µg/L) and trimethoprim (3.28µg/L) exhibited the highest average concentrations (Figure 5-2). Of the most frequently detected pharmaceuticals, caffeine was present at the highest average concentration (2.15µg/L), while sulfamethoxazole

(0.73µg/L) and acetaminophen (0.20µg/L) had lower average concentrations. Ampicillin

(0.26µg/L) was the least frequently detected compound with one of the lowest mean concentrations.

In Phase II, the most frequently detected compounds were acetaminophen (89%), theobromine (89%), caffeine (87%), and metformin (84%) (Figure 5-2). Thiamethoxam, naproxen and sulfamethoxazole were detected in 21%, 17%, and 16% of the analyzed samples, respectively, as opposed to <2% for ofloxacin, oxytetracycline, tetracycline, tylosin, sulfadimethoxine, and sulfamethazine. Some compounds including ampicillin, trimethoprim, triclosan, chlortetracycline, erythromycin, cimetidine and sulfadiazine were not detected in any sample. Concentrations measured in Phase II were lower than those recorded during Phase I; however, similar trends in average concentrations were observed whereby the least frequently detected compounds (<2%) including ofloxacin (3.16µg/L),

148 oxytetracycline HCl (0.41µg/L), tetracycline HCl (0.39µg/L), and tylosin (0.71µg/L) exhibited the highest concentrations (Figure 5-2). Apart from ofloxacin, compounds depicting the highest concentrations and lowest detection frequencies are veterinary antibiotics whose presence in rivers is likely attributed to intermittent precipitation events that mobilize EOCs from diffuse agricultural sources via surface runoff. Concentrations can further be influenced by factors such as timing of precipitation events in relation to manure or biosolid applications, EOC occurrence in applied manure, biodegradation rates or sorption potential, and timing of sampling events to capture these nonpoint source inputs

(Bernot et al., 2013).

13Figure 5-2: Detection frequencies and mean concentrations of EOC measured >LOD in Phase I (n=78) and Phase II (n=161) sampling periods. Error bars represent standard deviations and stars indicate compounds detected >LOQ only once during Phase II sampling period.

149

Spatial Variation and Influence of Source Water Type

To evaluate spatial variations of EOCs in Phase I, samples were collected on the same day from sites of similar source types. No significant variations (p>0.05) in concentrations among sites were recorded. Detection frequencies were however correlated with dominant land use types in the watersheds. There were comparatively fewer detections at sites in watersheds with high percentages (>83%) of forested land use and low (<5%) urban land uses (Figure 5-3). Similar findings were reported by Reif et al. (2012) and were associated with reduced point and nonpoint sources impacting water quality in highly forested watersheds with less agricultural and urban land uses. Despite being 82% forested, site D exhibited moderately high concentrations and detection frequencies. Predominantly forested watersheds may nevertheless have a high density of in situ domestic wastewater treatments, such as septic systems, capable of impacting groundwater (Schaider et al.,

2017; Kibuye et al., 2018) and being a source of PPCPs to nearby surface water sources

(Standley et al., 2008).

14Figure 5-3: Total EOC detection in samples collected in Phase I by percentage forested and developed land use in respective study site watersheds.

150

With regards to concentration variations due to source water type, detection frequencies were >5% higher in riverine sources than reservoir sources for acetaminophen, ampicillin, caffeine, and sulfamethoxazole. Detection frequencies for naproxen and ofloxacin differed by <5% between the two source water types, while trimethoprim was detected slightly more frequently in reservoir sources (33%) than riverine sources (28%) (Figure 5-4). As with detection frequency, average concentrations for ampicillin and trimethoprim were higher in riverine sources than reservoir sources. However, average concentrations of acetaminophen and caffeine, in addition to naproxen and ofloxacin, were similar across source types. Lower concentrations and detection frequencies in reservoir sources are attributed to higher forested land uses within the watersheds (Figure 5-3). Moreover, longer residence times in reservoirs promote environmental attenuation such as sorption to sediments and biodegradation (Glassmeyer et al., 2017), however such attenuation processes are dependent on EOC physicochemical characteristics and overall loading rates from contaminant sources. Interestingly, only sulfamethoxazole had a statistically significant (p<0.05) concentration variation between the two source types with higher average concentrations in reservoir than riverine sources. These results in conjunction with the detection frequencies suggest sulfamethoxazole is continually entering the surface water at low levels but, due to its slow biodegradation rates in a water-sediment system

(Xu et al., 2011), it builds up in lentic waters such as reservoirs.

151

15Figure 5-4: Detection frequencies >LOD and average concentrations in samples collected in Phase I by source water types (riverine and reservoirs).

Seasonal Variations

The EOCs investigated in Phase I were human pharmaceuticals that were likely present in surface water largely due to WWTP discharges. Accordingly, increased concentrations and detection frequencies were expected during low-flow periods because of reduced in stream dilution (Reif et al., 2012), higher pharmaceutical use (Yu et al.,

2013), and lower microbial and photolytic degradation rates (Vieno et al., 2005). Studied pharmaceuticals depicted varying seasonal patterns, but in general higher concentrations were recorded between fall through spring seasons while detection frequencies were higher in summer (Figure 5-5 & Table 5-4). Similarly, Reif et al. (2012) reported higher antibiotic detection frequencies in PA waters in summer and fall which are low baseflow seasons.

Lower concentrations in summer is attributed to faster in-stream degradation rates as opposed to lower degradation rates during peak consumer use in colder months (Vieno et al., 2005). Seasonal influence on concentrations were significant (p<0.05) for acetaminophen, caffeine, naproxen, and sulfamethoxazole, as opposed to insignificant

152

seasonal effects (p>0.05) observed for ampicillin and ofloxacin possibly due to infrequent

detection (<20%) in Phase I.

25Table 5-4: Summary statistics for seasonal variations in riverine and reservoir sources during Phase I (row 1: mean ± standard deviation, row 2: detection frequency). Spring Summer Fall Winter Compound Riverine Reservoir Riverine Reservoir Riverine Reservoir Riverine Reservoir Acetaminophen 0.08 ± 0.04

16Figure 5-4: Seasonal variations in mean concentrations in samples collected in Phase I by source water types (riverine and reservoirs). Concentrations between the limit of detection (LOD) and limit of quantification (LOQ) are reported as LOQ/2.

153

In Phase II, a wide variety of EOCs were analyzed (Figure 5-6). Significant seasonal influences (p<0.05) were observed for agricultural pesticide thiamethoxam which was predominantly detected in spring reflecting increased applications rates. Gómez et al.

(2012) reported significant increase in pesticide concentrations in May and June coinciding with peak application period in the watershed. The most frequently detected compounds acetaminophen (89%), theobromine (89%), caffeine (87%), and metformin (84%) exhibited insignificant seasonal variations (p>0.05) as concentrations and detection frequencies were similar in winter and spring (Table 5-5; Figure 5-6). Ofloxacin was detected once in spring and winter; however, it was present at a higher concentration in winter consistent with observations from Phase I. Oxytetracycline, tetracycline, tylosin, sulfadimethoxine, and sulfamethazine were only detected in winter and were present in

<2% of samples.

26Table 5-5: Summary statistics for seasonal variations at Site E during Phase II sampling period Concentration (µg/L) % Detection Frequencies

Compound Winter Spring Winter Spring Acetaminophen 0.14±0.09 0.14±0.06 100% 97% Caffeine 0.08±0.05 0.06±0.02 97% 93% Metformin 0.09±0.06 0.17±0.14 88% 86% Naproxen

Thiamethoxam

154

17Figure 5-6: Seasonal variations in detection frequencies greater than the limit of detection (>LOD) and mean concentrations in samples collected in Phase II. Standard deviations are shown as error bars for compounds detected >LOD in more than one sample.

Concentration-Discharge (C-Q) Relationships

Metformin and sulfamethoxazole depicted dilution responses (b<0), with inverse relationships between concentration and discharge (Figures 5-7, 5-8, 5-9, & 5-10).

Metformin and sulfamethoxazole are human pharmaceuticals with low WWTP removal efficiencies (Blair et al., 2013; Benotti & Brownawell, 2007; Kibuye et al., 2019).

Therefore, higher inputs from WWTP effluent are expected relative to other pharmaceuticals. Consequently, metformin and sulfamethoxazole exhibited higher surface water concentrations during low-flow conditions when the wastewater effluent contribution to total streamflow is higher. Similar findings were reported in an estuary

155 where sulfamethoxazole concentrations decreased with increasing flow (Benotti &

Brownawell, 2007) and in a statewide study in PA where detection frequencies decreased during high-flow conditions (Reif et al., 2012). Furthermore, a model by Benotti &

Brownawell (2007) concluded pharmaceuticals with low removal extent in WWTPs are diluted during wet weather indicating wastewater dilution by storm runoff is a major process controlling transport.

Thiamethoxam, a neonicotinoid insecticide used to control a broad spectrum of insects in agricultural crops, and caffeine indicated accretion patterns (b>0) since concentrations increased with increasing flow (Figure 5-7). Nonpoint source contaminants, including veterinary pharmaceuticals and pesticides, occur in surface water sources at higher frequencies and concentrations during periods coinciding with increased application and precipitation (Gómez et al., 2012). Similarly, insect repellant DEET and fungicide

MTBT had the highest concentrations in an urban catchment in Germany following storm events (Launay et al., 2013). Observed accretion dynamics for nonpoint source contaminants were consistent with observations in other watersheds that have found surface runoff as the major transport pathway into surface water sources (Gómez et al.,

2012, Launay et al., 2013).

156

18Figure 5-7: Concentration-discharge (C-Q) relationships for EOCs present at concentrations above the limit of quantification (LOQ) at site E plotted with a flow duration curve. The coefficient of variation of conentrations (CVC) and coefficient of variation of flowrates (CVQ) represent coefficients of variation for EOC concentration and discharge, respectively.

Caffeine concentrations are reported to increase with flowrates (Benotti &

Brownawell, 2007; Buerge et al., 2006; Launay et al., 2013), which is consistent with depicted accretion dynamics (b>0) in this study. The typically high removal efficiency for caffeine through WWTPs during baseflow is reduced during high-flow conditions when the residence time through WWTPs is decreased (Phillips et al., 2012). Additionally,

157 stormwater dilution effects are offset by contributions from CSOs (Benotti & Brownawell,

2007) and potential caffeine inputs originating from agricultural sources during high-flow periods. With the goal of identifying EOC sources in a watershed in Southeastern

Minnesota, Karpuzcu et al. (2014) concluded potential caffeine contributions from runoff across biosolid amended soils since caffeine couldn’t be linked to one dominant land use in the watershed. The caffeine metabolite, theobromine, exhibited a weak C-Q relationship

(b=0.05) and a high CVC/CVQ ratio (0.82), suggesting that flow variability is the dominant driver of the observed transport dynamics (Thompson et al., 2011). As a caffeine metabolite, chemical and environmental factors affecting metabolite formation processes can further influence theobromine’s concentration.

19Figure 5-8. Fluxes of dissolved EOCs with discharge in January through March sampling period

158

20Figure 5-9. Fluxes of dissolved EOCs with discharge in April and May sampling period

No strong C-Q relationship was observed for acetaminophen, as reflected by a near- zero slope (Figure 5-7). The lowest CVC/CVQ ratio of the selected EOCs suggests that its transport was the least influenced by concentration variability. Other studies have reported increases in acetaminophen concentrations during high-flow periods (Benotti &

Brownawell, 2007; Buerge et al., 2006; Launay et al., 2013; Reif et al., 2012). Since acetaminophen is efficiently removed in WWTPs, inputs from untreated sources including

CSOs (Benotti & Brownawell, 2007) and runoff from biosolid amended soils (Karpuzcu et al., 2014) during high-flow conditions likely contribute to maintain acetaminophen’s concentrations at relatively constant concentrations.

159

21Figure 5-10. Fluxes of dissolved EOCs with discharge in June sampling period

Relationships with Water Quality Indicators

In general, sulfamethoxazole (r=-0.22), caffeine (r=-0.40), and theobromine (r=-

0.39) were significantly inversely correlated with temperature (p<0.01). Increased temperatures generally correspond with increased solar UV radiation, which increases biodegradation and photodegradation rates. In a water-sediment system, sulfamethoxazole’s biodegradation rates have been observed to increase by over 40% at elevated temperatures independent of initial concentrations (Xu et al., 2011). Metformin

(r=0.38), thiamethoxam (r=0.32), and naproxen (r=0.49) depicted significant (p<0.01) positive correlation with temperature. For thiamethoxam and naproxen, correlations may be representative of temporal variability especially since concentrations were higher in

160 samples collected during warmer temperatures. Metformin concentrations decreased with increasing DO concentration (p<0.01; r=-0.30). Although metformin is not readily biodegradable in aquatic environments (Scheurer et al., 2012), biodegradation rates in soil have increased (80-92%) under aerobic conditions (Mrozik & Stefańska, 2014) indicating a likelihood to degrade in water-sediment interphase under high DO levels. Naproxen showed similar patterns with DO (p<0.01; r=-0.28), however as a moderately hydrophobic compound, sorption to sediment can reduce aqueous concentrations (Kunkel & Radke,

2008).

Bernot et al. (2013) reported total dissolved carbon in surface water as a significant aqueous PPCP concentration regulator. In this study, trimethoprim (r=-0.16), ampicillin

(r=-0.14), and naproxen (r=-0.12), had weak inverse relationships with total organic carbon

(TOC). Increasing TOC may provide a sink for concentration as EOCs can combine with organic matter to reduce aqueous concentrations, especially for hydrophobic compounds.

Sorption in natural environments is further influenced by compound characteristics such as structure and pKa and by aqueous phase characteristics such as pH and temperature.

Hydrophilic compounds, ofloxacin (r=-0.26) and caffeine (r=-0.15), also had weak negative correlations with TOC. Heterocyclic-N group in caffeine increases its ability to form hydrogen bonds to increase sorption potential (Nam et al., 2014) and sorption capacity for quinolones e.g. ofloxacin are pH dependent, with highest sorption potential near neutral pH (Fu et al., 2017). This pH dependence is demonstrated in part with increasing ofloxacin concentrations as water pH increases (r=0.14).

161

Ecological and Human Health Risk Assessment

In Phase I, naproxen and ofloxacin posed high risk (RQ>1) to fish, Daphnia, and algae in riverine and reservoir sources, while acetaminophen and sulfamethoxazole generally posed low risk (RQ<0.1). Moreover, highest RQs were observed in fall, winter, and spring, in similarity with relatively high concentrations measured during these seasons

(Table 5-6). In Phase II, calculated RQs per compound for fish, Daphnia, and algae were low (RQ<0.01) consistent with relatively lower concentrations measured in comparison to

Phase I. The highest RQs for fish, Daphnia, and algae were for ofloxacin (RQ=0.023), thiamethoxam (RQ=0.021), acetaminophen (RQ=0.02), and tylosin (RQ=20.74); respectively (Table 5-7). Although using seasonal averages is an adequate indicator of relative risk posed by EOCs, the RQ index doesn’t provide information on the degree of potency on a temporal scale.

162

27Table 5-6: Calculated PNEC and seasonal risk quotients (RQ) for fish, algae and Daphnia during phase I sampling period Riverine Sources PNEC (µg/L) Risk Quotients (RQ) Compound Fish Fall Spring Summer Winter Acetaminophen 378 3.0E-04 2.0E-04 6.0E-04 6.0E-04 Ampicillin ------Caffeine 90 3.6E-03 5.3E-02 1.2E-03 3.7E-02 Naproxen 34 6.8 1.7E-01 1.8E-02 1.5E-01 Ofloxacin 16 1.12 4.7E-01 1.1E-01 1.9E-01 Sulfamethoxazole 890 1.0E-04 1.0E-04 2.0E-04 1.3E-03 Trimethoprim 92.66 0 2.4E-01 1.2E-02 3.0E-03 Daphnia Fall Spring Summer Winter Acetaminophen 9.2 1.14E-02 8.2E-03 2.63E-02 2.38E-02 Ampicillin ------Caffeine 16 2.04E-02 2.95E-01 7.00E-03 2.05E-01 Naproxen 15 1.54E+01 3.84E-01 4.09E-02 3.39E-01 Ofloxacin 31.75 5.63E-01 2.39E-01 5.52E-02 9.66E-02 Sulfamethoxazole 189.2 3.0E-04 4.0E-04 8.0E-04 6.10E-03 Trimethoprim 8.21 0 2.76 1.37E-01 3.38E-02 Algae Fall Spring Summer Winter Acetaminophen 132 8.0E-04 6.0E-04 1.8E-03 1.7E-03 Ampicillin 1000 1.0E-04 7.0E-04 4.0E-04 1.0E-04 Caffeine ------Naproxen 22 1.05E+01 2.62E-01 2.79E-02 2.31E-01 Ofloxacin 4.74 3.77 1.6 3.7E-01 6.47E-01 Sulfamethoxazole 51 1.2E-03 1.3E-03 3.1E-03 2.28E-02 Trimethoprim 83.8 0 2.71E-01 1.34E-02 3.3E-03 Reservoir Sources PNEC (µg/L) Risk Quotients (RQ) Compound Fish Fall Spring Summer Winter Acetaminophen 378 4.0E-04 0 1.0E-04 1.0E-03 Ampicillin ------Caffeine 90 9.0E-04 7.0E-04 2.7E-02 3.2E-02 Naproxen 34 2.69 3.9E-02 1.5E-03 0 Ofloxacin 16 8.52E-01 7.13E-02 0 5.6E-01 Sulfamethoxazole 890 1.0E-04 1.0E-04 3.0E-04 5.5E-03 Trimethoprim 92.66 6.1E-03 4.5E-03 3.0E-03 5.2E-03 Daphnia Fall Spring Summer Winter Acetaminophen 9.2 1.69E-02 0 5.4E-03 3.99E-02 Ampicillin ------Caffeine 16 5.2E-03 3.8E-03 1.52E-01 1.8E-01 Naproxen 15 6.11 8.84E-02 3.3E-03 0 Ofloxacin 31.75 4.3E-01 3.59E-02 0 2.82E-01 Sulfamethoxazole 189.2 3.0E-04 6.0E-04 1.4E-03 2.57E-02 Trimethoprim 8.21 6.86E-02 5.1E-02 3.36E-02 5.83E-02 Algae Fall Spring Summer Winter Acetaminophen 132 1.2E-03 0 4.0E-04 2.8E-03 Ampicillin 1000 1.0E-04 0 0 1.0E-04 Caffeine ------Naproxen 22 4.16 6.03E-02 2.3E-03 0 Ofloxacin 4.74 2.88 2.41E-01 0 1.89 Sulfamethoxazole 51 1.0E-03 2.3E-03 5.0E-03 9.52E-02 Trimethoprim 83.8 6.7E-03 5.0E-03 3.3E-03 5.7E-03 RQ= 0 when surface water concentrations were

163

28Table 5-7: Calculated PNEC and seasonal risk quotients for fish, algae and Daphnia during phase II sampling period Risk Quotients (RQ) for fish Compound PNEC (µg/L) January February March May June Acetaminophen 378 3.0E-04 4.0E-04 4.0E-04 3.0E-04 4.0E-04 Ampicillin ------Caffeine 90 1.4E-03 7.0E-04 6.0E-04 4.0E-04 7.0E-04 Metformin 383.3 3.0E-04 4.0E-04 1.0E-04 2.0E-04 7.0E-04 Naproxen 34 2.0E-04 0 0 1.0E-04 1.2E-03 Ofloxacin 16 2.3E-02 0 0 0 4.0E-04 Sulfamethoxazole 890 0 0 0 0 1.0E-04 Trimethoprim 92.66 0 0 0 0 0 Triclosan ------Chlortetracycline 78.9 0 0 0 0 0 Erythromycin 100 0 0 0 0 0 Oxytetracycline 162.7 3.0E-04 0 0 0 0 Tetracycline ------Theobromine -- Thiamethoxam 3.7 0 0 0 1.23E-02 2.05E-02 Cimetidine ------Tylosin ------Sulfadiazine ------Sulfadimethoxine 100 2.0E-04 0 0 0 0 Sulfamethazine 100 4.0E-05 0 0 0 0 Risk Quotients (RQ) for Algae PNEC (µg/L) January February March May June Acetaminophen 132 Ampicillin 1000 8.0E-04 1.0E-03 1.0E-03 1.0E-03 1.1E-03 Caffeine -- 0 0 0 0 0 Metformin 320 ------Naproxen 22 4.0E-04 5.0E-05 1.0E-04 3.0E-04 8.0E-04 Ofloxacin 4.74 3.0E-04 0 0 1.0E-04 1.8E-03 Sulfamethoxazole 51 7.75E-02 0 0 0 1.2E-03 Trimethoprim 83.8 0 0 0 1.0E-04 2.5E-03 Triclosan -- 0 0 0 0 0 Chlortetracycline 3.1 ------Erythromycin 0.02 0 0 0 0 0 Oxytetracycline 4.5 0 0 0 0 0 Tetracycline 2.2 1.01E-02 0 0 0 0 Theobromine -- 1.99E-02 0 0 0 0 Thiamethoxam 81.8 ------Cimetidine -- 0 0 0 6.0E-04 9.0E-04 Tylosin 0.034 ------Sulfadiazine 0.034 0 0 0 0 0 Sulfadimethoxine 0.135 0 0 0 0 0 Sulfamethazine 2.3 1.0E-01 0 0 0 0

164

Risk Quotients (RQ) for Daphnia PNEC (µg/L) January February March May June Acetaminophen 9.2 1.0E-02 2.0E-02 2.0E-02 1.4E-02 1.53E-02 Ampicillin ------Caffeine 16 8.1E-03 3.8E-03 3.2E-03 3.0E-03 4.0E-03 Metformin 81.4 1.5E-03 2.0E-04 5.0E-04 1.0E-03 3.0E-03 Naproxen 15 0 0 0 2.0E-04 3.0E-03 Ofloxacin 31.75 0 0 0 0 2.0E-04 Sulfamethoxazole 189.2 0 0 0 1.0E-05 7.0E-04 Trimethoprim 8.21 0 0 0 0 0 Triclosan ------Chlortetracycline ------Erythromycin 22.45 0 0 0 0 0 Oxytetracycline 22.64 2.0E-03 0 0 0 0 Tetracycline 44.8 1.0E-03 0 0 0 0 Theobromine ------Thiamethoxam 100 0 0 0 5.0E-04 8.0E-04 Cimetidine ------Tylosin 680 0 0 1.0E-03 0 0 Sulfadiazine 221 0 0 0 0 0 Sulfadimethoxine 248 1.0E-04 0 0 0 0 Sulfamethazine 202 2.0E-05 0 0 0 0 RQ= 0 when surface water concentrations were

For Phase I, naproxen and ofloxacin contributed to most of the total risk to all three non-target species (Figure 5-11). For Phase II, the major contributions to risk for fish,

Daphnia, and algae were attributable to thiamethoxam, acetaminophen, and tylosin, respectively. For both sampling phases, percent contributions for each EOC varied seasonally, reflecting seasonal risk dynamics to aquatic species. During Phase II, the major contributions for fish were attributable to thiamethoxam and were highest in May and June.

This suggests a high relative risk during the spawning and early developmental period for fish populations in the Northeastern U.S. Laboratory toxicity tests have revealed thiamethoxam posed early life stage toxicity with endpoints including egg hatching, mortality and growth rate (Finnegan et al., 2017). Higher exposure concentrations and

165 subsequent risks are also anticipated during this period as high concentrations are transported to streams via surface runoff during high precipitation in spring.

22Figure 5-11: % Risk contributions of EOCs to overall risk quotients for fish, Daphnia and algae.

Human health RQs were only calculated for Phase I samples and most EOCs posed low human health risk (RQ<0.01) through fish consumption (Table 5-8). The highest RQ corresponds to naproxen (RQ=0.2), with the highest bioaccumulation factor in fish

-1 (100Lkgwt ) and was measured at the highest concentrations in surface water samples collected in Phase I.

166

29Table 5-8: Predicted concentrations in fish, predicted human dose and human health risk quotients Compound Predicted Predicted Human Acceptable Daily Human Health Risk Concentration in Dose Intake (ADI) Quotients from fish -1 -1 -1 -1 Fish (µg kg wt d ) (µg Kg d ) consumption -1 (µg kg wt) Acetaminophen 0.07 2.48E-05 340a 7.3E-08 b Ampicillin 0.5 2.0E-04 0.5 3.0E-04 Caffeine 0.3 1.0E-04 30c 3.36E-06 d Naproxen 4214 1.4 5.7 2.0E-01 d Ofloxacin 0.6 2.0E-04 7.1 3.0E-05 a Sulfamethoxazole 1.0 3.0E-04 130 2.46E-06 Trimethoprim 3.8 1.3E-03 4.2a 3.0E-04

Conclusion

20 EOCs were evaluated to determine how seasonality, hydrologic conditions, and source water type influenced concentrations during a two-year period in the SRB.

Acetaminophen, caffeine, and sulfamethoxazole, three human pharmaceuticals commonly associated with WWTP inputs, were spatiotemporally ubiquitous as they were consistently among the most abundant across seasons and source water types, while the least frequently detected human pharmaceuticals and veterinary antibiotics occurred at the highest average concentrations. No significant variations in concentration among sites were observed, however EOC occurrences were lower in reservoir than riverine sources likely due to higher forested land use and increased residence times for natural attenuation in reservoir source sites.

Seasonality was a significant factor influencing concentrations; higher concentrations were recorded in colder seasons corresponding to high consumer use and

167 application and lower environmental attenuation rates. WWTP-associated EOCs with consistent loading to aquatic systems including metformin and sulfamethoxazole depicted dilution dynamics since WWTP signals are diminished during high-flow periods, while

EOCs from agricultural sources such as thiamethoxam exhibited accretion responses as the dominant pathway is via surface runoff. Although caffeine and acetaminophen are often associated with WWTP influences, their accretion and episodic transport dynamics, respectively, are patterns of greater loading during high-flow periods suggesting surface runoff pathways in addition to WWTP inputs. There was variability in correlations between

EOCs and water quality parameters likely due to diverse EOC structural and physicochemical characteristics, however further studies are necessary to comprehensively characterize influences of such factors on EOC fate and transport. Although EOCs in aquatic systems posed minimal human health risk through fish consumption, risk quotients were mostly high (>0.1) for aquatic organisms and depicted temporal variations highlighting potential aquatic organism health impacts during seasons of important developmental stages.

168

References

Basu, N.B., Destouni, G., Jawitz, J.W., Thompson, S. E., Loukinova, N.V., Darracq, A., Zanardo, S., Yaeger, M., Sivapalan, M., Rinaldo, A., & Rao, P.S. C. (2010). Nutrient loads exported from managed catchments reveal emergent biogeochemical stationarity. Geophysical Research Letters, 37(23).

Benotti, M.J., & Brownawell, B.J. (2007). Distributions of pharmaceuticals in an urban estuary during both dry-and wet-weather conditions. Environmental Science & Technology, 41(16), 5795-5802. Bernot, M.J., Smith, L., & Frey, J. (2013). Human and veterinary pharmaceutical abundance and transport in a rural central Indiana stream influenced by confined animal feeding operations (CAFOs). Science of the Total Environment, 445, 219-230.

Blair, B.D., Crago, J. P., Hedman, C.J., & Klaper, R. D. (2013). Pharmaceuticals and personal care products found in the Great Lakes above concentrations of environmental concern. Chemosphere, 93(9), 2116-2123.

Boxall, A.B., Rudd, M.A., Brooks, B.W., Caldwell, D.J., Choi, K., Hickmann, S., Innes, E., Ostapyk, K., Staveley, J.P., Verslycke, T. & Ankley, G.T., (2012). Pharmaceuticals and personal care products in the environment: what are the big questions? Environmental Health Perspectives, 120(9), 1221.

Buerge, I.J., Poiger, T., Müller, M.D., & Buser, H.R. (2006). Combined sewer overflows to surface waters detected by the anthropogenic marker caffeine. Environmental Science & Technology, 40(13), 4096-4102.

De Liguoro, M., Di Leva, V., Dalla Bona, M., Merlanti, R., Caporale, G., & Radaelli, G. (2012). Sublethal effects of trimethoprim on four freshwater organisms. Ecotoxicology and Environmental Safety, 82, 114-121.

Eguchi, K., Nagase, H., Ozawa, M., Endoh, Y.S., Goto, K., Hirata, K., Miyamoto, K. & Yoshimura, H. (2004). Evaluation of antimicrobial agents for veterinary use in the ecotoxicity test using microalgae. Chemosphere, 57(11), 1733-1738.

Ferrari, B., Mons, R., Vollat, B., Fraysse, B., Paxēaus, N., Giudice, R.L., Pollio, A. & Garric, J. (2004). Environmental risk assessment of six human pharmaceuticals: are the current environmental risk assessment procedures sufficient for the protection of the aquatic environment? Environmental Toxicology and Chemistry, 23(5), 1344-1354.

Finnegan, M.C., Baxter, L.R., Maul, J.D., Hanson, M.L., & Hoekstra, P. F. (2017). Comprehensive characterization of the acute and chronic toxicity of the neonicotinoid insecticide thiamethoxam to a suite of aquatic primary producers, invertebrates, and fish. Environmental Toxicology and Chemistry, 36(10), 2838-2848.

169

Fu, H., Li, X., Wang, J., Lin, P., Chen, C., Zhang, X., & Suffet, I. M. (2017). Activated carbon adsorption of quinolone antibiotics in water: Performance, mechanism, and modeling. Journal of Environmental Sciences, 56, 145-152. Gall, H.E., Sassman, S.A., Lee, L.S., & Jafvert, C.T. (2011). Hormone discharges from a Midwest tile-drained agroecosystem receiving animal wastes. Environmental Science & Technology, 45, 8755-8764.

Gall, H.E., Park, J., Harman, C.J., Jawitz, J.W., & Rao, P.S.C. (2013). Landscape filtering of hydrologic and biogeochemical responses in managed catchments. Landscape ecology, 28(4), 651-664.

Gall, H.E., Sassman, S.A., Jenkinson, B., Lee, L.S., & Jafvert, C.T. (2015). Comparison of export dynamics of nutrients and animal-borne estrogens from a tile-drained Midwestern agroecosystem. Water Research, 72, 162-173.

Glassmeyer, S.T., Furlong, E.T., Kolpin, D.W., Batt, A.L., Benson, R., Boone, J.S., Conerly, O., Donohue, M.J., King, D.N., Kostich, M.S., & Mash, H.E. (2017). Nationwide reconnaissance of contaminants of emerging concern in source and treated drinking waters of the United States. Science of the Total Environment, 581, 909-922.

Gómez, M.J., Herrera, S., Solé, D., García-Calvo, E., & Fernández-Alba, A.R. (2012). Spatio- temporal evaluation of organic contaminants and their transformation products along a river basin affected by urban, agricultural and industrial pollution. Science of the Total Environment, 420, 134-145. Grung, M., Källqvist, T., Sakshaug, S., Skurtveit, S., & Thomas, K.V. (2008). Environmental assessment of Norwegian priority pharmaceuticals based on the EMEA guideline. Ecotoxicology and Environmental Safety, 71(2), 328-340.

Hedgespeth, M.L., Sapozhnikova, Y., Pennington, P., Clum, A., Fairey, A., & Wirth, E. (2012). Pharmaceuticals and personal care products (PPCPs) in treated wastewater discharges into Charleston Harbor, South Carolina. Science of the Total Environment, 437, 1–9.

Karpuzcu, M.E., Fairbairn, D., Arnold, W.A., Barber, B.L., Kaufenberg, E., Koskinen, W.C., Novak, P.J., Rice, P.J., & Swackhamer, D.L. (2014). Identifying sources of emerging organic contaminants in a mixed use watershed using principal components analysis. Environmental Science: Processes & Impacts, 16(10), 2390-2399. Kibuye, F.A., Elkin, K.R., Gall, H.E., Elliott, H.A., Watson, J.E., & Swistock, B. (2018). Occurrence and concentrations of pharmaceutical compounds in private wells in Central

170

Pennsylvania. In 2018 ASABE Annual International Meeting (p. 1-14). American Society of Agricultural and Biological Engineers.

Kibuye, F.A., Gall, H.E., Elkin, K.R., Ayers, B., Veith, T.L., Miller, M., Jacob, S., Hayden, K.R., Watson, J.E.& Elliott, H.A. (2019). Fate of pharmaceuticals in a spray-irrigation system: From wastewater to groundwater. Science of the Total Environment, 654, 197-208.

Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D., Barber, L.B. & Buxton, H.T., (2002). Pharmaceuticals, Hormones, and Other Organic Wastewater Contaminants in U.S. Streams, 1999 - 2000: A National Reconnaissance. Environmental Science & Technology, 36(6), 1202–1211.

Kunkel, U., & Radke, M. (2008). Biodegradation of acidic pharmaceuticals in bed sediments: insight from a laboratory experiment. Environmental Science & Technology, 42(19), 7273- 7279. Launay, M., Dittmer, U., Steinmetz, H., & Kuch, B. (2013). Temporal variations of organic pollutants levels during storm events in an urban receiving water body. NOVATECH 2013.

Lee, S.Y. (2017). Assessment of aquatic toxicity and endocrine disruption potential of metformin, an anti-diabetic drug (Doctoral dissertation, Seoul National University).

Leet, J.K., Gall, H.E., & Sepúlveda, M.S. (2011). A review of studies on androgen and estrogen exposure in fish early life stages: effects on gene and hormonal control of sexual differentiation. Journal of Applied Toxicology, 31(5), 379-398.

Loraine, G.A., & Pettigrove, M.E. (2006). Seasonal variations in concentrations of pharmaceuticals and personal care products in drinking water and reclaimed wastewater in southern California. Environmental Science & Technology, 40(3), 687-695.

McEachran, A.D., Shea, D., & Nichols, E.G. (2017). Pharmaceuticals in a temperate forest-water reuse system. Science of the Total Environment, 581, 705-714.

Mijangos, L., Ziarrusta, H., Ros, O., Kortazar, L., Fernández, L.A., Olivares, M., Zuloaga, O., Prieto, A., & Etxebarria, N. (2018). Occurrence of emerging pollutants in estuaries of the Basque Country: Analysis of sources and distribution, and assessment of the environmental risk. Water Research, 147, 152-163.

Miller, M.D., Gall, H.E., Buda, A.R., Saporito, L. S., Veith, T.L., White, C.M., Williams, C.F., Brasier, K.J., Kleinman, P.J., & Watson, J.E. (2019). Load-discharge relationships reveal the efficacy of manure application practices on phosphorus and total solids losses from agricultural fields. Agriculture, Ecosystems & Environment, 272, 19-28.

171

Moore, M.T., Greenway, S.L., Farris, J.L., & Guerra, B. (2008). Assessing caffeine as an emerging environmental concern using conventional approaches. Archives of Environmental Contamination and Toxicology, 54(1), 31-35.

Moreno-González, R., Campillo, J.A., García, V., & León, V.M. (2013). Seasonal input of regulated and emerging organic pollutants through surface watercourses to a Mediterranean coastal lagoon. Chemosphere, 92(3), 247-257.

Mrozik, W., & Stefańska, J. (2014). Adsorption and biodegradation of antidiabetic pharmaceuticals in soils. Chemosphere, 95, 281-288.

Nam, S.W., Choi, D.J., Kim, S.K., Her, N., & Zoh, K.D. (2014). Adsorption characteristics of selected hydrophilic and hydrophobic micropollutants in water using activated carbon. Journal of Hazardous Materials, 270, 144-152.

Pailler, J.Y., Guignard, C., Meyer, B., Iffly, J.F., Pfister, L., Hoffman, L., & Krein, A. (2009). Behaviour and fluxes of dissolved antibiotics, analgesics, and hormones during flood events in a small heterogenous catchment in the Grand Duchy of Luxembourg. Water, Air and Soil Pollution, 203, 79-98.

Reif, A.G., Crawford, J. K., Loper, C.A., Proctor, A., Manning, R., & Titler, R. (2012). Occurrence of pharmaceuticals, hormones, and organic wastewater compounds in Pennsylvania waters, 2006-09. US Department of the Interior, US Geological Survey.

Sanderson, H., Johnson, D.J., Wilson, C.J., Brain, R.A., & Solomon, K.R. (2003). Probabilistic hazard assessment of environmentally occurring pharmaceuticals toxicity to fish, daphnids and algae by ECOSAR screening. Toxicology Letters, 144(3), 383-395. Santos, L.H., Araújo, A.N., Fachini, A., Pena, A., Delerue-Matos, C., & Montenegro, M.C. B.S. M. (2010). Ecotoxicological aspects related to the presence of pharmaceuticals in the aquatic environment. Journal of Hazardous Materials, 175(1-3), 45-95.

Schaider, L.A., Ackerman, J.M., & Rudel, R.A. (2016). Septic systems as sources of organic wastewater compounds in domestic drinking water wells in a shallow sand and gravel aquifer. Science of the Total Environment, 547, 470-481.

Scheurer, M., Michel, A., Brauch, H.J., Ruck, W., & Sacher, F. (2012). Occurrence and fate of the antidiabetic drug metformin and its metabolite guanylurea in the environment and during drinking water treatment. Water Research, 46(15), 4790-4802. Singer, A.C., Järhult, J.D., Grabic, R., Khan, G.A., Lindberg, R.H., Fedorova, G., Fick, J., Bowes, M.J., Olsen, B. & Söderström, H. (2014). Intra-and inter-pandemic variations of antiviral, antibiotics and decongestants in wastewater treatment plants and receiving rivers. PLoS One, 9(9), e108621.

Standley, L.J., Rudel, R.A., Swartz, C.H., Attfield, K.R., Christian, J., Erickson, M., & Brody, J.G. (2008). Wastewater‐contaminated groundwater as a source of endogenous hormones and pharmaceuticals to surface water ecosystems. Environmental Toxicology and Chemistry, 27(12), 2457-2468.

172

Thompson, S.E., Basu, N.B., Lascurain, J., Aubeneau, A., & Rao, P.S. C. (2011). Relative dominance of hydrologic versus biogeochemical factors on solute export across impact gradients. Water Resources Research, 47(10).

Vieno, N.M., Tuhkanen, T., & Kronberg, L. (2005). Seasonal variation in the occurrence of pharmaceuticals in effluents from a sewage treatment plant and in the recipient water. Environmental Science & Technology, 39(21), 8220-8226.

Vogel, R.M., Rudolph, B.E., & Hooper, R.P. (2005). Probabilistic behavior of water-quality loads. Journal of Environmental Engineering, 131(7), 1081-1089.

Xu, B., Mao, D., Luo, Y., & Xu, L. (2011). Sulfamethoxazole biodegradation and biotransformation in the water–sediment system of a natural river. Bioresource Technology, 102(14), 7069-7076.

Yu, Y., Wu, L., & Chang, A.C. (2013). Seasonal variation of endocrine disrupting compounds, pharmaceuticals and personal care products in wastewater treatment plants. Science of the Total Environment, 442, 310-316.

173

Chapter 6

Occurrence and Seasonal Variations of Pharmaceutical Compounds in Source and Treated Drinking Water

Abstract

The occurrence of pharmaceuticals in various surface water sources raises concern over their removal through conventional drinking water treatment processes. A monitoring study of 7 pharmaceutical compounds (acetaminophen, ampicillin, caffeine, naproxen, ofloxacin, sulfamethoxazole, and trimethoprim) in 2 conventional drinking water treatment plants (DWTPs) was performed. Monthly to bimonthly scale sampling was conducted during a one-year study period from raw untreated waters used as drinking water sources, intermediate treatment steps and finished drinking water distributed to consumers. At least one pharmaceutical compound was detected in 88% of source water samples, 80% of samples collected from intermediate treatment steps, and 64% of finished drinking water samples. Average concentrations in source water samples ranged between 0.12-14.66 µg/L to 0.05 µg/L-7.87 µg/L in finished drinking water. Detection frequencies and concentrations during drinking water treatment varied by compound such that patterns of concentration increase, and decrease were observed while some compounds occurred at the same levels in source and finished water samples. Seasonal variations in concentrations were observed, with higher concentrations in the source waters recorded in the fall, winter and spring seasons while lower concentrations occurred in the summer. Seasonal variations

174 in the intermediate treatment steps and finished drinking water samples reflected seasonal patterns in source waters.

Environmental risk assessment for the targeted organisms reveal that the studied pharmaceuticals pose medium to high risks to the aquatic organisms. Although pharmaceutical compounds were quantified in finished drinking water samples, human health risk assessments predict that risks to consumers are low.

Keywords. emerging contaminants, surface water, drinking water, seasonal variations, risk assessment

Introduction

There is increasing pressure in our water resources due to a higher demand for drinking water in adequate quantity and quality. This is mainly because many water sources are presently threated by increased anthropogenic pollution of domestic, agricultural, and industrial sources. Pharmaceutical compounds including prescription drugs, over-the- counter drugs, veterinary drugs, as well as others identified as contaminants of emerging concern are ubiquitous in aquatic systems used as drinking water sources (Barnes et al.,

2008; Focazio et al., 2008; Kolpin et al., 2002). Pharmaceuticals are introduced into the environment throughout their manufacture, consumer usage, and disposal cycle, however their primary point sources into aquatic systems are effluents from wastewater treatment plants (WWTPs) (Verlicchi et al., 2012). As more pharmaceuticals compounds are manufactured to meet a growing population’s health needs, there is a likelihood of higher

175 concentrations and a broader range of pharmaceuticals in drinking water sources due to increased input rates that can offset natural environmental attenuation processes such as degradation and dilution.

In 2002, the first large-scale investigation of emerging contaminants in U.S. surface water sources was performed across 139 streams impacted by multiple sources including

WWTPs (Kolpin et al., 2002). Collected samples were tested for a diverse range of emerging contaminants including pharmaceuticals, hormones, plasticizers, pesticides, and other wastewater organic compounds with 80% of the sampled sites testing positive for at least one contaminant. From 2006-2009, the U.S. Geological Survey (USGS) and

Pennsylvania (PA) Department of Environmental Protection (DEP) collected data on the occurrence of emerging contaminants in PA waters (Reif et al., 2012). Sampling sites included streams used as drinking water sources that tested positive for wide range of emerging contaminants, but pharmaceuticals were the most frequently detected class.

Pharmaceuticals are designed to directly affect biochemical and physiological functioning in microorganisms, animals and humans (Jjemba, 2006), and they remain active even at low concentrations (ng/L-µg/L) hence are capable of evoking various responses. Some impacts of pharmaceuticals in the environment include development of inherent or acquired resistances in microbial and bacterial species, a major public health problem (Boxall et al., 2012) because it can result in the development of treatment-resistant illnesses. Furthermore, the presence of mixtures of pharmaceuticals in one sample at a point in time, inherently creates concerns about potential synergistic effects that are currently not known. Human exposure to low concentrations of pharmaceuticals and the resulting

176 impacts is thus a growing public health concern especially because most aquatic systems used as drinking water sources are impacted either directly or indirectly by WWTP effluent. Most emerging contaminants lack regulatory standards and there is no requirement for their removal during water treatment, hence they are not frequently monitored. Pharmaceuticals and hormones were, however, included in the U.S.

Environmental Protection Agency (EPA) Contaminant Candidate List 3 (CCL 3) as contaminants expected to occur in public water systems that may need regulation under safe drinking water Act (SDWA) (USEPA, 2010).

Since WWTPs are major contributors to emerging contaminants in the environment, many studies have performed monitoring campaigns for emerging contaminants in WWTPs and impacted aquatic systems. In contrast, limited scientific attention has been given to their fate in drinking water treatment plants (DWTPs). The efficiencies of water treatment unit processes on the removal of various emerging contaminants have been evaluated in simulated and pilot scale studies (Broséus et al., 2009;

Li et al., 2017; Ma et al., 2017; Rozas et al., 2017; Westerhoff et al., 2005). However, these studies might not capture the effect of external environmental factors such as occurrence of complex mixtures of pharmaceuticals in source waters as well as concentration fluxes in source water due to seasonal and hydroclimatic influences. At the same time, compared to WWTPs, fewer studies have holistically assessed the occurrence of pharmaceuticals in drinking water sources and treated drinking water with the first nationwide reconnaissance study in U.S. DWTPs conducted recently by Glassmeyer et al. (2017).

177

In this study, seasonal variations of concentrations of seven selected pharmaceuticals including four antibiotics (ampicillin, sulfamethoxazole, ofloxacin, and trimethoprim), two anti-inflammatory medications (acetaminophen and naproxen), and a stimulant (caffeine) in drinking water sources and treated drinking water were assessed in two DWTPs located in the Susquehanna River Basin. Preliminary ecological and human health risk assessments were also conducted using toxicological data gleaned from literature.

Materials and Methods

Site Descriptions

Two conventional drinking water treatment plants (DWTPs), DWTP A and B located within the Susquehanna River Basin were used as study sites. DWTP A uses a riverine source while DWTP B uses a reservoir source for potable water production. DWTP

A is in a watershed of almost equal forested (44%) and agricultural (41%) land uses with

15% developed land use while DWTP B is in a highly forested watershed (88%), with minimal developed (4%) and agricultural (6%) land uses. DWTP A treats around 42 million liters per day (MLD) of water to serve a population of about 35,000 people and the drinking water treatment steps consist of pretreatment comprising of the addition of powdered activated carbon (PAC), chlorine, sulfuric acid (H2SO4), sodium permanganate

(NaMnO4), and a ferric blend to promote coagulation. Water is clarified prior to granular activated carbon (GAC) filtration, after which it is chlorinated followed by ammonia post treatment. DWTP B produces about 25 MLD of potable water for nearly 58,500 people.

178

Treatment steps include potassium permanganate (KMnO4) pretreatment, followed by PAC and lime treatment, and coagulation and flocculation steps. As a direct filtration system, water is then passed through a mixed media filtration system after which filtered water is chlorinated and stored in a clear well. Schematics of the DWTPs are summarized in Figure

6-1.

23Figure 6-1: Schematic diagram showing drinking water treatment at DWTP A and B. Post filtration and clear well samples are referred to as ‘combined filter effluent (CFE)’ and ‘clear well effluent (CWEFF)’, respectively, while samples collected at the final treatment steps were called ‘finished water’.

179

Sample Collection

To assess the occurrence of pharmaceuticals in source waters and during drinking water treatment at different stages of drinking water treatment, grab samples were collected in source water prior to any treatment and after each treatment process to the final step prior to consumer distribution. Schematics for the DWTPs and sample collection points are summarized in Figure 6-1. Seasonal variation in the concentrations of pharmaceuticals in source water and treated water was evaluated through monthly and bi-monthly grab sampling technique. All samples were collected in 250mL amber glass bottles fitted with

Teflon-lined caps following EPA method 1694.

Collected samples were chilled on ice during transportation to the United States

Department of Agriculture-Agricultural Research Service (USDA-ARS) laboratory in

University Park, PA and processed within 48 h of collection. Aliquots of each sample were used in situ to determine water quality parameters such as pH, temperature, dissolved oxygen (DO) and total organic carbon (TOC). For quality assurance and control, field and travel blank samples were collected and handled using similar protocols as DWTP samples.

Any detected compounds in blanks were averaged and used to censor DWTP samples.

Targeted Pharmaceutical Analysis

180 stimulant (caffeine). These seven pharmaceuticals exhibit a wide range of physicochemical properties (Table 6-1) and are therefore representative of a broader array of compounds.

Additionally, acetaminophen, naproxen, sulfamethoxazole, and trimethoprim are ranked among compounds to be prioritized in the monitoring of emerging contaminants in aquatic systems and drinking water based on their frequency of occurrence in the environment, ecological effects, and potential human health impacts (Kumar & Xagoraraki 2010).

Procedures on sample preparation and analysis are described in depth in Chapter 3 and

Appendix A. In brief, analysis and quantification of pharmaceuticals was done using a high-resolution accurate mass (HRAM) Q Exactive Orbitrap mass spectrometer

(ThermoFisher Scientific, Bremen, Germany). The limit of detection (LOD) for all pharmaceuticals was 0.01 g/L (signal:noise ratio ≥ 3) and the limit of quantification

(LOQ) was 0.1 g/L (signal:noise ratio ≥ 10) except for ofloxacin with an LOD and LOQ of 0.3 g/L and 3g/L, respectively. In all cases, the reporting limits of the analytes measured between the LOD and LOQ were determined as one half of the LOQ as stated in

EPA Method 301 (USEPA, 2018).

181

30Table 6-1: Physicochemical properties of selected pharmaceuticals. Pharmaceutical Chemical Molar Mass a b c Compound Formula (g/mol) pKa log KOW Log KOC Antibiotics d Ampicillin C16H19N3O4S 349.40 2.50; 7.30 1.35 2.00 Sulfamethoxazole C10H11N3O3S 253.28 1.60; 5.70 0.89 1.86 Ofloxacin C18H20FN3O4 361.37 5.97; 9.28 -0.39 4.64

Trimethoprim C14H18N4O3 290.32 7.12 0.91 1.88 Analgesics Acetaminophen C8H9NO2 151.17 9.38 0.46 1.32 Naproxen C14H14O3 230.26 4.15 3.18 2.52 Stimulant Caffeine C8H10N4O2 194.19 10.40 -0.07 2.87

Risk calculations

Ecological Risk Assessment

To estimate environmental risk posed by measured pharmaceuticals in drinking water sources, risk quotients (RQs) were calculated for representative trophic levels including algae, Daphnia, and fish. Predicted no effect concentration (PNEC) of pharmaceuticals were estimated using acute toxicity data for algae, Daphnia and fish by dividing EC50 values obtained from literature (Table 6-2) which represent concentration at which 50% of population exhibits a response by an arbitrary assessment factor of 1000.

RQs associated with respective compounds were then calculated by dividing measured environmental concentrations (MECs) and the corresponding estimated PNEC values. RQs greater than unity indicate high risk to the aquatic organisms (Sanderson et al., 2003).

a Ka =Acid dissociation constant b KOW = octanol-water partition coefficient c KOC=organic carbon partition coefficient d values for amoxicillin used due to lack of data for ampicillin Sources: https://pubchem.ncbi.nlm.nih.gov

182

31Table 6-2: EC50 and LC50 (mg/L) used to calculate PNEC (by dividing EC50 by an assessment factor of 1000), for fish, algae and Daphnia; and Acceptable daily intakes (ADI) used to calculate Drinking water equivalent levels (DWELs) for human health risk assessment EC50 (mg/L) Human ADI Compound Fish Algae Daphnia (µg kg-1 d-1) Acetaminophen a378 a132 a9.2 j340 Ampicillin -- b>1000 -- k0.5 Caffeine c90 -- d16 l30 Naproxen e34 e22 e15 m5.7 Ofloxacin f16 f4.74 g31.75 m7.1 Sulfamethoxazole e890 e51 h189.2 j130 Trimethoprim i92.66 i83.8 i8.21 j4.2klm

Human Health Risk Assessment

The primary route of human exposure to pharmaceuticals is through drinking water.

To assess potential human health risks, average concentrations measured in finished drinking water were used as a conservative representative of likely exposure level. Risk estimations were calculated for adult populations with a 50th percentile body weight (BW) of 60 kg and daily drinking water intake (DWI) of 2.04 L d-1 for a frequency of exposure

(FOE) of 350 d yr-1(350 d/365 d, =0.96) (de Jesus Gaffney et al., 2015). Acceptable daily

a Grung et al., (2008) b Eguchi et al., (2004) c Moore et al., (2008) d Lilius et al., (1995) e Sanderson et al., (2003) f Ferrari et al., (2004) g Isidori et al., (2005) h Santos et al., (2010) i De Liguoro et al., (2012) j Schwab et al., (2005) kVragović et al., (2012) l McEachran et al., (2017) m Prosser & Sibley, (2015)

183 intakes (ADIs) for respective pharmaceuticals obtained from literature (Table 6-2) were used to estimate exposure thresholds since ADIs represent daily exposure levels without adverse effects on a population (Schwab et al., 2005). Drinking water equivalent levels

(DWELs) were then estimated as shown below (Equation 6-1).

ADI ×BW 퐷푊퐸퐿 (µ푔 퐿−1푑푎푦−1) = (Equation 6-1) DWI×FOE

A human health risk quotient, which is a ratio of finished water concentration and the estimated DWEL was then used to characterize risk whereby a quotient greater than unity suggests possible risk from drinking water.

Results and Discussion

Overview of Occurrence in Drinking Water Sources

Source water samples were collected at the intake of the DWTPs to evaluate levels of selected pharmaceuticals in DWTP influent. DWTP A uses a creek source while DWTP

B uses a reservoir source. In DWTP A, sulfamethoxazole, caffeine and acetaminophen were the most frequently detected and were present in 82%, 64% and 55% of collected samples, respectively. In DWTP B, lower detection frequencies typically ≤36% were observed. Caffeine, trimethoprim and acetaminophen present in 36% of samples were the most frequently detected. A summary of the results is shown in Tables 6-3 and 6-4 below.

184

32Table 6-3: Concentrations of pharmaceuticals (µg/L) and detection frequencies in samples collected from DWTP A Mean Concentrations (µg/L) and (% >LOD) Coagulation & Post Finished Source Water Flocculation Clarification CFE CWEFF water Acetaminophen 0.20±0.15 (55) 0.21±0.19 (55) 0.25±0.30 (55) 0.23±0.29 (45) 0.34±0.07 (18) 0.15±0.13 (45) Ampicillin 0.22±0.05 (18) 0.05 (9) 0.11 (9) 0.05 (9) 0.05 (9) 0.05 (9) Caffeine 1.46±2.65 (64) 0.77±1.64 (55) 0.63±1.17 (36) 1.04±1.65 (27) 0.97±1.59 (29) 1.41±2.00 (36) Naproxen 3.69±2.51 (27) 2.26±2.96 (27) 18.89±25.94 (18) 18.49±29.15 (27) 2.24±2.65 (18) 1.84±0.83 (27) Ofloxacin 11.19±7.24 (27) 10.11±5.30 (18) 5.84±3.79 (18) 7.30 (9) 5.05 (9) 7.87±4.33 (27)

Sulfamethoxazole 0.76±1.88 (82) 0.88±2.10 (64) 0.14±0.14 (45) 0.14±0.14 (45) 0.18±0.18 (45) 1.47±2.74 (36) Trimethoprim 14.66±26.35 (36) 2.52±2.25 (36) 4.35±4.27 (36) 2.25±2.14 (36) 1.31±1.16 (36) 0.89±0.70 (36) n=11; Values in parentheses represent % detection frequencies of pharmaceuticals; Summaries calculated for concentrations >LOD for all compounds. LOD= Limit of Detection (0.01 µg/L) for all compounds except ofloxacin (0.3 µg/L). Concentrations are reported as 0.05 µg/L or 1.5 µg/L (ofloxacin) when concentrations are above the LOD but less than the LOQ which is 0.1 µg/L for all compounds except ofloxacin 3 µg/L. CFE-Combined filter effluent; CWEFF-clear well effluent.

The least frequently detected pharmaceutical compound in both plants was ampicillin that was present in 18% of samples in DWTP A while was

B samples. Samples collected from source and treated waters in a DWTP in New Jersey by

Stackelberg et al. (2007) similarly reported no detection of ampicillin in source water samples. A study in Southeastern U.S. analyzed samples from a river used as a drinking water source and found higher detection frequencies for analyzed compounds most notably for trimethoprim (100%), sulfamethoxazole (88%), and caffeine (50%) while acetaminophen and naproxen were present in 13% of samples (Padhye et al., 2014). Reif et al. (2012) collected samples from a range of study sites that included streams used as drinking water sources in the state of Pennsylvania and the most frequently detected pharmaceuticals were caffeine (71%), sulfamethoxazole (40%), acetaminophen (25%), and trimethoprim (8%).

185

33Table 6-4: Concentrations of pharmaceuticals (µg/L) and detection frequencies in samples collected from DWTP B Mean Concentrations (µg/L) and (% >LOD)

KMnO4 Source Water Pre-Treatment Flocculation CFE Finished Water Acetaminophen 0.12±0.16 (36) 0.13±0.13 (43) 0.20±0.33 (36) 0.09±0.09 (36) 0.23±0.30 (36) Ampicillin LOD for all compounds. LOD= Limit of Detection (0.01 µg/L) for all compounds except ofloxacin (0.3 µg/L). Concentrations are reported as 0.05 µg/L or 1.5 µg/L (ofloxacin) when concentrations are above the LOD but less than the LOQ which is 0.1 µg/L for all compounds except ofloxacin 3 µg/L. CFE-Combined filter effluent

Average pharmaceutical concentrations ranged between 0.20-14.66 µg/L and 0.12-

7.81 µg/L in DWTP A and B, respectively. The catchment in DWTP A presented higher

mean concentrations for all the pharmaceutical compounds which agrees with the fact that

the creek is impacted by several upstream discharges of wastewater effluent. Lower

concentrations and detection frequencies in the reservoir source can be attributed to a

higher forested land use and low urban/developed land uses within the watershed (Reif et

al., 2012). Moreover, lesser concentrations in reservoirs has been attributed to increased

environmental attenuation due to longer residence times in reservoirs (Glassmeyer et al.,

2017).

The highest average concentrations in DWTP A were observed for trimethoprim

(14.66 µg/L), ofloxacin (11.19 µg/L), and naproxen (3.69 µg/L), while for DWTP B, the

greatest mean values were depicted by ofloxacin (7.81 µg/L), sulfamethoxazole (1.61

µg/L) and caffeine (1.85 µg/L). In both DWTPs, maximum concentrations observed from

all source water samples were as follows: trimethoprim (54.16 µg/L), ofloxacin (17.83

186

µg/L), caffeine (7.07 µg/L), sulfamethoxazole (6.10 µg/L), naproxen (5.18 µg/L), acetaminophen (0.40 µg/L), and ampicillin (0.25 µg/L). Concentrations of pharmaceuticals measured in the present study are in the same ranges to values reported in the statewide study (Reif et al., 2012) with maximum values for sulfamethoxazole (1.34 µg/L) within the same order of magnitude with the present dataset. However, maximum values for ofloxacin

(0.33 µg/L), and trimethoprim (0.26 µg/L) were two orders of magnitude lower than those in the present study. Similarly, concentrations measured by Padhye et al. (2014) were at least two orders of magnitude lower in comparison to the current study with pharmaceutical concentrations ranging between 0.9-132.9 ng/L.

Occurrence during Drinking Water Treatment

Removal of pharmaceuticals during water treatment is influenced by both physical and chemical characteristics of the compound and the treatment processes used. This section summarizes occurrence within each treatment step and a discussion on concentration variation due to associated properties is included.

KMnO4 Pre-oxidation

The first treatment step in DWTP B consisted of KMnO4 oxidation that is widely used in water treatment to remove soluble manganese (II) and iron (II) as well as taste and order compounds (Hu et al., 2010). Targeted pharmaceuticals were detected at least once during this oxidative treatment step (Figure 6-2) with average concentrations ranging

187 between 0.13-6.54 µg/L which did not vary significantly with mean concentrations measured in untreated source water samples (0.12-7.81 µg/L). In comparison to source water samples, all the compounds had lower average concentrations during this oxidation step except for acetaminophen (Table 6-4). Measured concentrations are majorly influenced by individual compound reactivity with KMnO4. A study by Hu et al. (2010) examined KMnO4 oxidation of widely detected pharmaceuticals and found that sulfamethoxazole was unreactive with KMnO4 as opposed to ciprofloxacin, lincomycin, and trimethoprim that depicted high reactivities. KMnO4 oxidation rates are also influenced by other factors such as pH and the presence or organic matter. Oxidation rates for trimethoprim increased with decreasing pH (Hu et al., 2010) and although oxidation rates were low for sulfamethoxazole, Gao et al. (2014) reported increasing oxidation rates with increasing pH and lower rates with increasing initial sulfamethoxazole concentrations.

Furthermore, the presence of humic acids inhibited the reactivity with sulfamethoxazole hypothetically because humic acids can consume more KMnO4 and compete for secondary oxidants that are generated during oxidation process (Gao et al., 2014).

188

24Figure 6-2: Concentrations of pharmaceutical compounds >LOD in samples collected at each treatment step in DWTP B. Each open circle dot represents one sample >LOD. Darker circles represent overlapping concentration from multiple samples. Ampicillin was not detected in any sample and is not represented in this figure.

Coagulation, Flocculation and Sedimentation

Coagulation with ferric and alum coagulants were used in DWTP A and B, respectively. However, both plants added other reagents such as powdered activated carbon

(PAC), caustic soda, KMnO4, and H2SO4 in DWTP A as opposed to PAC and lime in

DWTP B. Pre-oxidation steps with KMnO4 or its addition during treatment can amplify the efficiency of the coagulation and flocculation steps and the addition of PAC at the early

189 stages of treatment is effective for the removal of natural organic matter (NOM) (Ezwald,

2010). H2SO4, caustic soda and lime are added to optimize pH levels during treatment.

During coagulation and flocculation treatment, dissolved hydrophobic pharmaceuticals can sorb to suspended sediments and natural organic matter (Adams et al.,

2002) and ionic compounds can also have an electrostatic interactions with floc particles

(Vieno et al., 2007). Therefore, pharmaceuticals can potentially be simultaneously removed with suspended sediments and colloids during settling processes in sedimentation tanks. Sorption to PAC and oxidation by KMnO4 is also an important removal route. PAC adsorption is recognized as the most effective technology for removal of organic pollutants from water (Aksu & Tunç, 2005; Westerhoff et al., 2005).

Bench-scale studies have highlighted the inadequacy of coagulation and flocculation treatment method in eliminating pharmaceuticals in water. Westerhoff et al.

(2005) reported comparable removal efficiencies (20-50%) for both alum and ferric salts at similar dosages and noted that initial pharmaceutical concentration in raw water had no influence on the removal efficiencies. Relatively hydrophilic compounds such as acetaminophen (33%) and sulfamethoxazole (60%) were moderately removed and was associated with a possible weak acid-base hydrolysis influence from ferric chloride. In a study by Hu et al. (2017), sulfamethoxazole, ofloxacin and trimethoprim were averagely removed (43-56%) due to probable electrostatic interactions between the compounds and flocs. Similarly, in the present study, when post coagulation and flocculation samples were compared with source water samples in DWTP A, average concentrations ranged from

190

0.05 µg/L-10.11 µg/L and detection frequencies ranged between 27%- 64% which were lower than ranges in source water (Table 6-3; Figure 6-3).

In DWTP B, coagulation and flocculation water samples were compared with pre- oxidation samples. Although the overall detection frequencies in the coagulation and flocculation were lower than those in the KMnO4 oxidation step, average concentrations were higher (0.14 µg/L-8.88 µg/L) (Table 6-4). Concentrations of sulfamethoxazole and caffeine were reduced after the coagulation and flocculation treatment in 14% of samples while only 7% of samples were lower in acetaminophen, naproxen and trimethoprim levels

(Figure 6-2). For both DWTPs, sulfamethoxazole and acetaminophen depicted lower concentrations after coagulation and flocculation. In a full scale DWTP study, Stackelberg et al. (2007) reported moderate removal of hydrophilic compounds such as sulfamethoxazole and acetaminophen and associated it to potential acid base hydrolysis resulting from ferric chloride coagulation, which may be the case for DWTP A as ferric salts are used.

As a direct filtration system, DWTP B has no clarifiers, in comparison to DWTP A that includes a clarification process. There was minimal concentration reduction during the clarification step in DWTP A with most of the samples having similar if not higher concentrations in comparison to the coagulation and flocculation steps (Table 6-3; Figure

6-3). Only about 18% of clarification step samples exhibited reduced concentrations of acetaminophen, caffeine, ofloxacin, and sulfamethoxazole in comparison to the coagulation and flocculation steps. Naproxen exhibited substantially higher average

191 concentrations in settled water samples (18.89 µg/L) in comparison to coagulation and flocculation steps (2.29 µg/L).

25Figure 6-3: Concentrations of pharmaceutical compounds >LOD in samples collected in source water and after each treatment step and finished drinking water in DWTP A. Each open circle dot represents one sample >LOD. Darker circles represent overlapping concentration from multiple samples. CFE-combined filter effluent; CWEFF-clear well effluent.

192

Filtration

Pharmaceutical compounds that are sorbed to floc particles can be removed simultaneously with floc during filtration processes, where filter media can act as a strainer, or compounds can adhere to previously deposited floc. This removal mechanism is efficient for hydrophobic compounds as opposed to polar compounds (Kim et al., 2007).

Pharmaceutical compounds can also be removed through sorption to filter media especially in a granular activated carbon (GAC) filter. Adsorption on GAC is primarily controlled by hydrophobic interactions between the sorbent and solute. Contaminant Kow can therefore be used as a predictor of its removal by GAC such that higher adsorptive removal are predicted for compounds with high Kow (Westerhoff et al., 2005).

DWTP A employs a GAC filtration system while DWTP B uses a mixed media filter. For DWTP A, average concentrations (0.05 µg/L – 18.49 µg/L) as well as detection frequencies were slightly lower than those in the clarification step. Apart from trimethoprim and acetaminophen that had lower concentrations in 36 % and 18% the combined filter effluent (CFE) samples, respectively, most compounds depicted similar and, in some cases, higher concentrations after filtration in comparison to the clarification step (Figure 6-3). Since DWTP A employs intermediate chlorine during filtration process, some concentration reduction can be attributed to reactivity with chlorine. Trimethoprim and acetaminophen react readily with free available chlorine (Dodd & Huang 2007;

Xagoraraki et al., 2008). Removal during GAC processes can also be influenced by specific

GAC characteristics as well as DWTP operational conditions and source water characteristics. Snyder et al. (2007) studied GAC performance in two DWTPs. In one

193

DWTP naproxen and caffeine removal were as low as 6% and 16% respectively, but moderate to high removal for trimethoprim (64%), sulfamethoxazole (83%), and caffeine

(99%) were observed. In contrast, moderate removal for caffeine > 41% was reported in the second DWTP. Other factors such as desorption of previously sorbed pharmaceuticals in filter media can also result in higher concentrations in post-filtration samples. This phenomenon is seen for sulfamethoxazole following GAC filtration in a DWTP study by

Hu et al. (2017). In DWTP B, minimal concentration variations were observed in the combined filter effluent (CFE) samples in comparison to the coagulation and flocculation steps since average concentration in CFE were around the same ranges (0.09 µg/L- 9.19

µg/L) as the previous step. Most of the pharmaceuticals have similar and higher concentrations in CFE samples with only 7% of samples having lower concentrations of acetaminophen, sulfamethoxazole and trimethoprim.

Chlorination

Chlorine oxidation is an important removal mechanism particularly for hydrophilic pharmaceuticals that fail to sorb to suspended sediments and filter media during water treatment (de Jesus Gaffney et al., 2015). The efficacy of chlorination on pharmaceutical compounds is influenced by free available chlorine species, the speciation characteristics of respective pharmaceuticals (neutral, cationic, or anionic) (Qiang et al., 2006), as well as functional groups existing on their benzene ring (Rivera-Utrilla et al., 2013).

Post-treatment chlorination was performed in DWTP A at the clear well, thus clear well effluent samples were compared with CFE samples. A lower concentration range

194

(0.05 µg/L -5.05 µg/L) was observed and of all the clear well effluent samples collected,

27% and 18% had lower trimethoprim and caffeine concentrations, respectively, while the concentrations for other compounds were similar and higher than those in CFE samples, with the highest concentrations observed for ofloxacin. With a potentially longer chlorine contact time, lower concentrations were observed when DWTP A effluent samples were compared to clear well effluent samples. Trimethoprim concentrations were lower in 36% of the effluent samples while other compounds depicted minimal concentration differences. Majority of compounds were measured at the same levels as the clear well effluent samples, while higher levels were seen for ofloxacin and sulfamethoxazole.

Concentrations differences in plant effluent samples can also be due to chloro amines resulting from ammonia addition.

While most compounds exhibited unchanged and higher concentrations after chlorination in DWTP B, trimethoprim, acetaminophen and caffeine exhibited lower concentrations in 7% of effluent samples after post-chlorination. Average concentrations ranged from 0.23 µg/L – 14.51 µg/L), ranges that are higher than those in CFE step with substantial average concentration increases for ofloxacin, naproxen and acetaminophen.

Minimal concentration variations in pre and post chlorination samples can be attributed to the influence of factors such as compound reactivity and aqueous media pH. The typical pH measured during the study period ranged between 6-8.5. Trimethoprim is readily reactive with chlorine with the highest reactivities observed at pH 7 essentially, due to the presence of a nucleophilic form of trimethoprim at this pH that is more reactive than its protonated form (Dodd & Huang, 2007). Naproxen is more reactive with chlorine at weak

195 acidic to neutral pH values (Boyd et al., 2005), thus the wide pH ranges measured in the study period may have influenced reactivity. Ofloxacin is very reactive with chlorine at a pH of 7.2 with a half-life of 7.7 s., but oxidation rates were dependent upon initial concentrations (Yassine et al., 2017), while acetaminophen is reactive with chlorine at pHs higher than 9 (Xagoraraki et al., 2008). Caffeine is, however, unreactive with chlorine resulting in very low oxidative removal (Yang et al., 2016). Furthermore, Padhye et al.

(2014) pointed out that most compounds studied in a full scale DWTP did not follow their reactivity trends reported in bench scale studied possibly due to more complex conditions in DWTPs that can influence reactivity and are generally not accounted for in such experiments.

Occurrence in Finished Water

Finished water samples were collected at the final step in storage tanks from which water was pumped to consumers. Levels in these samples were therefore used as conservative predictors of what consumers are exposed to through tap water. However, further degradation is anticipated from chlorine added to maintain residual chlorine levels in the distribution systems.

At least one pharmaceutical compound was detected >LOD in 64% of finished water samples. The detection of these compounds in finished water suggests that they resist removal or undergo partial removal through the conventional drinking water treatment processes evaluated within the study. In DWTP A, acetaminophen was the most frequently detected in finished water and was present in 45% of samples while caffeine,

196 sulfamethoxazole and trimethoprim were present in 36% of finished water sample.

Naproxen and ofloxacin were present in 27% of DWTP effluent samples while ampicillin occurred at the least frequency of 9%. In DWTP B, the most frequently detected compounds were acetaminophen, caffeine and trimethoprim that were present in 36% of effluent samples while sulfamethoxazole and naproxen were detected in 29% of the samples and ofloxacin was only detected in 14% of the effluent samples. Ampicillin was not detected in any finished water samples in DWTP B.

In general, most finished water concentrations of studied pharmaceutical compounds were similar to or less than concentrations in the associated source water samples (Figure 6-4), though some were quantified at higher concentrations. Average concentrations for all compounds in finished water were lower than those in source water in DWTP A, as opposed to DWTP B where only sulfamethoxazole and trimethoprim exhibited lower mean concentrations in finished water (Table 6-4). Lower concentrations in finished waters imply that most of the compound were partially removed during treatment. However, since DWTP residence times were not considered in this sampling campaign, the higher and/or lower concentrations in finished than source water samples could be due to lack of staggering of samples.

197

2627Figure 6-4: Concentrations of pharmaceutical compounds >LOD in source and finished water in DWTP A and B.

Average pharmaceutical concentrations ranged between 0.05-7.87 µg/L and 0.23-

7.51 µg/L in finished waters of DWTP A and B respectively (Tables 6-3 & 6-4). The highest average concentrations in DWTP A were observed for ofloxacin (7.87µg/L), naproxen (1.84 µg/L) and sulfamethoxazole (1.47 µg/L), while for DWTP B the greatest mean values were depicted by ofloxacin (14.51 µg/L), caffeine (2.99 µg/L) and naproxen

(2.76 µg/L). Maximum concentrations measured from all finished water samples were as follows: ofloxacin (28.44 µg/L), naproxen (8.85 µg/L), caffeine (7.91 µg/L), sulfamethoxazole (5.92 µg/L), trimethoprim (1.94 µg/L), acetaminophen (0.73 µg/L), and ampicillin (0.05 µg/L). Except for ampicillin, it was also observed that majority of the maximum values were from DWTP B. This implies that concentrations in finished water could be dependent on source water quality and associated treatment at the DWTP.

Maximum values measured in finished water samples in general were lower than those in source water samples apart from compounds such as ofloxacin, naproxen and caffeine.

198

Other studies in DWTPs have reported similar findings of detectable levels of pharmaceuticals in finished drinking water with concentrations typically <0.5 µg/L

(Stackelberg et al., 2004). Lower concentrations of similar pharmaceuticals generally <100 ng/L were reported by Glassmeyer et al. (2017) in a nationwide study while even lower concentrations of <20 ng/L for acetaminophen, caffeine, naproxen, sulfamethoxazole and trimethoprim in finished drinking water is reported by Padhye et al. (2014). In a DWTP located in South eastern Spain (Azzouz & Ballesteros, 2013) concentrations were as low as <1 ng/L. Variations in finished water concentrations can be attributed to varying source water characteristics as well as DWTP treatment mechanisms and operational characteristics that can influence removal extent during water treatment. Furthermore, variations in analytical procedures and detection limits can result in inconsistencies in reported values.

Seasonal Variations

Seasonal Variations in Drinking Water Sources

Figures 6-5 and 6-6 show the seasonal variations of pharmaceuticals compounds in source water and through each treatment step for the studied DWTPs. Concentrations in source water samples differed markedly between sampling season and target compound.

Overall, higher source water concentrations were recorded in the fall, winter and spring seasons while lower concentrations and detections were recorded in the summer. Higher detections in the fall through spring seasons correspond to higher consumer use of

199 pharmaceuticals (Yu et al., 2013), lower microbial and photolytic degradation rates (Vieno et al., 2005) and lower stream flow conditions (Reif et al., 2012).

In DWTP A (Figure 6-5), minimal seasonal variations were observed for acetaminophen, but was notably not detected in the spring season and was present at relatively similar levels (0.2 µg/L) throughout the other seasons. Naproxen and caffeine were highest in the spring (5.18 µg/L and 3.56 µg/L, respectively) and winter (5.09 µg/L and 1.35 µg/L, respectively) seasons. The antibiotic sulfamethoxazole depicted the highest concentrations in the winter (1.95 µg/L) as opposed to spring for trimethoprim (54.16

µg/L). Ofloxacin indicated highest levels in the fall (17.83 µg/L) while ampicillin was mostly detected in the spring and summer, but at lower levels (0.18-0.25 µg/L) in comparison to other antibiotics.

In DWTP B, generally lower concentrations were recorded in comparison to DWTP

A (Figure 6-6). Acetaminophen was detected throughout the sampling period at levels

Elevated caffeine concentrations were observed in the summer (3.39 µg/L) followed by winter season (2.32 µg/L) in comparison to levels

200

28Figure 6-5: Seasonal variations in pharmaceutical concentrations in source water, intermediate treatment steps and treated finished water in DWTP A. CFE-combined filter effluent; CWEFF-clear well effluent.

Similar findings have been reported by other researchers. Reduced flow, dilution rates, and slower biochemical degradation of pharmaceuticals were speculated to cause higher concentration and detection frequencies during winter sampling in Missouri surface water (Mu et al., 2017). Loraine & Pettigrove, (2006) found high concentrations in fall season in southern California whereby concentrations were higher in the dry season in

August through November samples than those collected between January and June during the wet season. Furthermore, in southeast U.S., Padhye et al. (2014) reported spikes in the total concentrations of pharmaceuticals in early spring and late summer.

201

29Figure 6-6: Seasonal variations in pharmaceutical concentrations in source water, intermediate treatment steps and treated finished water in DWTP B. CFE-combined filter effluent.

Seasonal Variations During Water Treatment

As opposed wastewater treatment where biological treatment is the most important removal mechanism of pharmaceutical compounds, drinking water treatment plants mostly employ physical and chemical treatment processes. Therefore, while removal efficiencies in WWTPs vary based on seasonal variations in temperature that influence biodegradation rates (Vieno et al., 2005; Hedgespeth et al., 2012), seasonal influences in drinking water removal efficiencies can be due to temperature influences on chemical oxidation rates.

Photolytic degradation can also be a seasonally influenced attenuation route especially in

202 outdoor coagulation, flocculation and sedimentation basins. Though seasonal variations in drinking water treatment for pharmaceuticals is understudied, research by Azzouz &

Ballesteros, (2013) in a full scale DWTP indicate that the removal extent during KMnO4 oxidation and alum-aided coagulation treatment step were three times greater in the hottest period than the coldest period. Similarly, more compounds were detected after chlorine oxidation in the fall and winter periods in comparison to summer months. Since pharmaceutical reactivity can be dependent on initial concentrations, observed seasonal variations during drinking water treatment water can reflect seasonal variations in source waters that are used for potable water production.

Pharmaceutical compounds detected in source water were also detected during the intermediate treatment steps. Seasonal variations of pharmaceutical compounds varied between treatment steps and between compounds. However, in general, seasonal concentrations reflected those patterns in associated source waters (Figures 6-5 and 6-6).

Differences were observed when compounds were

203

Seasonal Variations in Finished Water

Seasonal variations in finished water were comparable to overall patterns in source water samples. This implies that human exposure through drinking water varies throughout the year and is influenced by concentrations in source waters.

In DWTP A (Figure 6-5), highest concentrations of acetaminophen are observed in spring and winter (0.17 µg/L) while ampicillin was only detected in the summer (0.05

µg/L). Similarly, naproxen and caffeine were highest in the spring (2.35 µg/L; 0.64 µg/L, respectively) and winter (2.31 µg/L; 4.30 µg/L, respectively) seasons, and sulfamethoxazole levels peaked in the winter (5.57 µg/L). Though similar seasonal detections are depicted for trimethoprim in source and finished water, the highest levels in finished were recorded in the summer as opposed to spring in source water. Ofloxacin exhibited different patterns from those in source water samples such that it was detected in the spring, summer and fall seasons with the highest season being in the spring (12.59

µg/L) as opposed to spring, fall and winter for source water samples.

In DWTP B, generally lower concentrations were noted in comparison to DWTP

A (Figure 6-6). Acetaminophen was detected throughout the sampling period at levels

204 ofloxacin was highest in the fall (28.44 µg/L). Ampicillin was not detected during the sampling period.

Risk Assessment

Ecological Risk Assessment

An ecological risk assessment was performed using PNEC values estimated for fish, Daphnia and algae by dividing EC50 values obtained from literature by an arbitrary safety factor (Sanderson et al., 2003). Risk quotients were derived by comparing measured average concentrations in each source water with the PNEC. Risk calculations do not assess the effect on mixtures of pharmaceutical compounds to these organisms or the effects from chronic exposure. Figure 6-7 summarizes calculated risks for each of the taxa. It is observed that ofloxacin and trimethoprim posed high risk algae and Daphnia respectively in source water at DWTP A. Ofloxacin posed medium risk to fish (>0.1 RQ <1) in both source water types, while naproxen exhibits medium risk to all taxa (>0.1 RQ <1) in DWTP

A source water. In both source types, acetaminophen and sulfamethoxazole depicted the lowest risk to the aquatic organisms. Risk quotients in DWTP A source water are higher than those in DWTP B since concentrations were higher in DWTP A source waters. Other researchers (Archana et al., 2017; Komori et al., 2013; Wu et al., 2014) report similar ranges in risk quotients for surface waters have been reported. Since average concentrations were used in the risk assessment, actual risks to aquatic organisms may be higher as concentrations in water varies temporally and could be higher than average values used.

Furthermore, some of the pharmaceuticals can bioaccumulate in organism tissues (Muñoz et al., 2010) thereby posing potential chronic effects that were not evaluated in the

205 assessment. Besides, there may be synergistic effects resulting from the fact that aquatic organisms are impacted by several other pharmaceutical compounds not measured in the study among other emerging contaminants that are endocrine disruptors.

30Figure 6-7: Risk Quotients for fish, Daphnia, and algae from exposure levels corresponding to average pharmaceutical concentrations in source water.

Human Health Risk Assessment

Results of the risk characterizations for each pharmaceutical compound in finished water are shown in Figure 6-8. DWELs were calculated assuming an exposure rate of 2 L d-1 through drinking water for a frequency of exposure of nearly one year. This assessment assumes that exposure levels are comparable to average concentrations measured in finished drinking water, though levels of exposure may be lower since there is a likely concentration reduction from residual chlorine during distribution. As observed in the previous sections, finished water concentrations fluctuated significantly throughout the sampling period thus exposure levels may sometimes be higher than the average concentrations used. Though the assessment focuses on drinking water exposure for adult populations based on an average body weight of 60 kg, individuals could also be exposed

206 through other environmental sources such as agricultural products as well as from fish consumption. Furthermore, risk estimations were only calculated for one compound at a time and did not assess any effects from mixtures as well as risk resulting from chronic exposure.

As shown in Figure 6-8, calculated risk quotients are well below the risk threshold of 1. The antibiotics ofloxacin and trimethoprim and anti-inflammatory drug naproxen depicted the highest risk quotients that were still <0.1, while the lowest risk quotients

(<0.001) are seen for acetaminophen and sulfamethoxazole. These risk quotient values imply that pharmaceuticals measured in the study do not pose any risk to consumer health when present at average levels in the current dataset. This finding agrees with conclusions from other risk assessments (Cunningham et al., 2009; de Jesus Gaffney et al., 2015;

Schwab et al., 2005)

31Figure 6-8: Human Health Risk Quotients for adult population (age>18) from exposure levels corresponding to average pharmaceutical concentrations in finished water.

207

Conclusion

Studies have highlighted the widespread occurrence of pharmaceutical compounds in aquatic systems. This study conducted in two DWTPs with different treatment processes shows that pharmaceuticals present in drinking water sources are also present at levels

>LOQ in intermediate drinking water treatment steps as well as in finished drinking water.

This indicates that conventional treatment methods may not succeed in eliminating pharmaceutical compounds from water.

At least one pharmaceutical compound was detected in source water samples with sulfamethoxazole (82%), caffeine (64%) and acetaminophen (55%) as the most frequently detected in DWTP A, and caffeine, trimethoprim and acetaminophen present in 36% of samples were the most frequently detected in DWTP B. Average pharmaceutical concentrations ranged between 0.20-14.66 µg/L and 0.12-7.81 µg/L in DWTP A and B respectively. Concentration and detection frequencies were higher in DWTP A source water as it is a stream influenced with several WWTP effluent discharges.

Pharmaceuticals depicted different patterns throughout the treatment such that pharmaceutical concentrations increased, decreased or remained at the same level in comparison to untreated source waters and preceding treatment steps. This observation may be due to the fact that the sampling regime did not incorporate DWTP residence times, though this observation is consistent with previous findings (Azzouz & Ballesteros 2013;

Stackelberg et al., 2007; Padhye et al., 2014).

At least one pharmaceutical compound was detected in finished water samples suggesting that the studied pharmaceuticals resist removal or undergo partial removal

208 through the conventional drinking water treatment plants of this study. Detection frequencies varied by compound and by plant, and in general detection frequencies were lower if not similar to than those in source water samples. In DWTP A, detection frequencies were as follows: acetaminophen was the most frequently detected (45%) followed by caffeine, sulfamethoxazole, and trimethoprim (36%); naproxen and ofloxacin

(27%) and ampicillin that was detected at the least frequency of 9%. In DWTP B, the most frequently detected compounds were acetaminophen, caffeine and trimethoprim that were present in 36% of finished water samples while sulfamethoxazole and naproxen were detected in 29% of the samples and ofloxacin was only detected in 14% of the effluent samples. Average pharmaceutical concentrations ranged between 0.05-7.87 µg/L and 0.23-

7.51 µg/L in finished waters of DWTP A and B respectively. Average concentrations for all compounds in finished water were lower than those in source water in DWTP A, as opposed to DWTP B where most compounds had higher means in finished water.

We examined the influence of seasonal variations on the concentrations of the target compounds in the source and treated water. Higher source water concentrations were recorded in the fall, winter and spring seasons while lower concentrations and detections were recorded in the summer. Seasonal variations during water treatment reflected seasonal patterns in source water implying that higher levels in finished water are expected during the colder seasons of the year as opposed to warmer seasons.

Environmental risk assessment for the targeted organisms reveal that at average concentrations in surface water sources, pharmaceuticals pose medium to high risks to the aquatic organisms. Highest risk quotients were observed from ofloxacin to fish, Daphnia

209 and algae, and from trimethoprim to Daphnia. Although pharmaceutical compounds were quantified in the finished drinking water samples, human health risk assessments reveal that risks to consumers is low based on existing toxicity data.

References

Aksu, Z., & Tunç, Ö. (2005). Application of biosorption for penicillin G removal: Comparison with activated carbon. Process Biochemistry, 40(2), 831–847.

Archana, G., Dhodapkar, R., & Kumar, A. (2017). Ecotoxicological risk assessment and seasonal variation of some pharmaceuticals and personal care products in the sewage treatment plant and surface water bodies (lakes). Environmental Monitoring and Assessment, 189(9), 446.

Boxall, A.B., Rudd, M.A., Brooks, B.W., Caldwell, D.J., Choi, K., Hickmann, S., Innes, E., Ostapyk, K., Staveley, J.P., Verslycke, T. & Ankley, G. T. (2012). Pharmaceuticals and personal care products in the environment: what are the big questions? Environmental Health Perspectives, 120(9), 1221.

Boyd, G. R., Zhang, S., & Grimm, D. A. (2005). Naproxen removal from water by chlorination and biofilm processes. Water Research, 39(4), 668–676.

Cunningham, V. L., Binks, S. P., & Olson, M. J. (2009). Human health risk assessment from the presence of human pharmaceuticals in the aquatic environment. Regulatory Toxicology and Pharmacology, 53(1), 39–45. de Jesus Gaffney, V., Almeida, C. M. M., Rodrigues, A., Ferreira, E., Benoliel, M. J., & Cardoso,

210

V. V. (2015). Occurrence of pharmaceuticals in a water supply system and related human health risk assessment. Water Research, 2, 1–10.

Dodd, M. C., & Huang, C. H. (2007). Aqueous chlorination of the antibacterial agent trimethoprim: reaction kinetics and pathways. Water Research, 41(3), 647-655.

Edzwald, J. K. (2010). Water Quality and Treatment A Handbook on Drinking Water. McGrawHill.

Gao, S., Zhao, Z., Xu, Y., Tian, J., Qi, H., Lin, W., & Cui, F. (2014). Oxidation of sulfamethoxazole (SMX) by chlorine, ozone and permanganate—a comparative study. Journal of Hazardous Materials, 274, 258-269.

Hu, L., Martin, H. M., & Strathmann, T. J. (2010). Oxidation kinetics of antibiotics during water treatment with potassium permanganate. Environmental Science & Technology, 44(16), 6416-6422.

Hu, X. L., Bao, Y. F., Hu, J. J., Liu, Y. Y., & Yin, D. Q. (2017). Occurrence of 25 pharmaceuticals

211

in Taihu Lake and their removal from two urban drinking water treatment plants and a constructed wetland. Environmental Science and Pollution Research, 24(17), 14889- 14902.

Jjemba, P. K. (2006). Excretion and ecotoxicity of pharmaceutical and personal care products in the environment. Ecotoxicology and Environmental Safety, 63(1), 113–130.

Kim, S. D., Cho, J., Kim, I. S., Vanderford, B. J., & Snyder, S. a. (2007). Occurrence and removal of pharmaceuticals and endocrine disruptors in South Korean surface, drinking, and waste waters. Water Research, 41(5), 1013–1021.

Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D., Barber, L.B. & Buxton, H.T., (2002). Pharmaceuticals, Hormones, and Other Organic Wastewater Contaminants in U. S. Streams, 1999 - 2000: A National Reconnaissance. Environmental Science & Technology, 36(6), 1202–1211.

Komori, K., Suzuki, Y., Minamiyama, M., & Harada, A. (2013). Occurrence of selected pharmaceuticals in river water in Japan and assessment of their environmental risk. Environmental Monitoring and Assessment, 185(6), 4529-4536.

Loraine, G. A., & Pettigrove, M. E. (2006). Seasonal variations in concentrations of pharmaceuticals and personal care products in drinking water and reclaimed wastewater in Southern California. Environmental Science and Technology, 40(3), 687–695.

Ma, L., Li, J., & Xu, L. (2017). Aqueous chlorination of fenamic acids: Kinetic study, transformation products identification and toxicity prediction. Chemosphere, 175, 114- 122.

McEachran, A. D., Shea, D., & Nichols, E. G. (2017). Pharmaceuticals in a temperate forest-water reuse system. Science of the Total Environment, 581, 705-714.

Moore, M. T., Greenway, S. L., Farris, J. L., & Guerra, B. (2008). Assessing caffeine as an

212

emerging environmental concern using conventional approaches. Archives of Environmental Contamination and Toxicology, 54(1), 31-35.

Muñoz, I., Bueno, M. J. M., Agüera, A., & Fernández-Alba, A. R. (2010). Environmental and human health risk assessment of organic micro-pollutants occurring in a Spanish marine fish farm. Environmental Pollution, 158(5), 1809-1816.

Padhye, L. P., Yao, H., Kung’u, F. T., & Huang, C.-H. (2014). Year-long evaluation on the occurrence and fate of pharmaceuticals, personal care products, and endocrine disrupting chemicals in an urban drinking water treatment plant. Water Research, 51, 266–276.

Santos, L. H., Araújo, A. N., Fachini, A., Pena, A., Delerue-Matos, C., & Montenegro, M. C. B. S. M. (2010). Ecotoxicological aspects related to the presence of pharmaceuticals in the aquatic environment. Journal of Hazardous Materials, 175(1-3), 45-95.

Schwab, B.W., Hayes, E.P., Fiori, J.M., Mastrocco, F.J., Roden, N.M., Cragin, D., Meyerhoff, R.D., Vincent, J.D. and Anderson, P. D. (2005). Human pharmaceuticals in US surface waters: A human health risk assessment. Regulatory Toxicology and Pharmacology, 42(3), 296–312.

213

Snyder, S.A., Adham, S., Redding, A.M., Cannon, F.S., DeCarolis, J., Oppenheimer, J., Wert, E.C. and Yoon, Y. (2007). Role of membranes and activated carbon in the removal of endocrine disruptors and pharmaceuticals. Desalination, 202(1–3), 156–181.

Stackelberg, P. E., Furlong, E. T., Meyer, M. T., Zaugg, S. D., Henderson, A. K., & Reissman, D. B. (2004). Persistence of pharmaceutical compounds and other organic wastewater contaminants in a conventional drinking-water-treatment plant. Science of the Total Environment, 329(1-3), 99-113.

U.S. EPA, 2010. Drinking Water Contaminant Candidate List. https://www.epa.gov/ccl/contaminant-candidate-list-3-ccl-3 (accessed 10.29.18.).

Vieno, N. M., Härkki, H., Tuhkanen, T., & Kronberg, L. (2007). Occurrence of Pharmaceuticals in River Water and Their Elimination in a Pilot-Scale Drinking Water Treatment Plant. Environmental Science & Technology, 41(14), 5077–5084.

Vragović, N., Bažulić, D., & Zdolec, N. (2012). Dietary exposure assessment of ß-lactam antibiotic residues in milk on Croatian market. Croatian journal of Food Science and Technology, 4(1), 81-84.

Wu, C., Huang, X., Witter, J. D., Spongberg, A. L., Wang, K., Wang, D., & Liu, J. (2014). Occurrence of pharmaceuticals and personal care products and associated environmental risks in the central and lower Yangtze river, China. Ecotoxicology and Environmental Safety, 106, 19-26.

Xagoraraki, I., Hullman, R., Song, W., Li, H., & Voice, T. (2008). Effect of pH on degradation of acetaminophen and production of 1, 4-benzoquinone in water chlorination. Journal of Water Supply: Research & Technology-AQUA, 57(6).

Yang, X., Sun, J., Fu, W., Shang, C., Li, Y., Chen, Y., Gan, W. and Fang, J. (2016). PPCP degradation by UV/chlorine treatment and its impact on DBP formation potential in real

214

waters. Water Research, 98, 309–318.

Yassine, M. H., Rifai, A., Hoteit, M., Mazellier, P., & iskandarani, M. Al. (2017). Study of the degradation process of ofloxacin with free chlorine by using ESI-LCMSMS: Kinetic study, by-products formation pathways and fragmentation mechanisms. Chemosphere, 189, 46– 54.

215

Chapter 7

Conclusion

The goal of this study was to evaluate the occurrence, fate and transport of selected emerging organic contaminants in surface and groundwater sources impacted by wastewater discharges and to understand their occurrence and removal during both wastewater treatment and drinking water treatment systems. Risk assessments for aquatic ecosystem health and human health were also performed using concentrations measured in surface water and water used for human consumption. Findings from each research objectives are summarized in Chapters 3-6. In this concluding chapter, a summary of key findings from each objective is presented. The impacts of water reuse practices on the different aquatic systems are compared and finally directions for future research are recommended.

Summary of Key Findings and Implications

The first objective (Chapter 3) was to understand the fate of seven commonly used pharmaceutical compounds in a wastewater treatment plant and impacted groundwater wells at a forested and cropped spray-irrigation site. Water samples were collected during a 14-month study period from the WWTP influent, effluent, and groundwater monitoring wells at the spray-irrigation site. Consistent with other studies, acetaminophen and caffeine were generally better removed (>88%) during wastewater treatment in comparison to antibiotics and the anti-inflammatory drug naproxen. Concentrations varied seasonally

216 whereby higher concentrations were recorded in winter and spring seasons corresponding to both high consumer use and reduced biodegradation rates during low temperatures.

Detection frequencies and concentrations of PPCPs in groundwater were lower than effluent concentrations by at least one order of magnitude, posing minimal risk to human health. In general, findings imply that soil acts as an effective biogeochemical filter for

PPCPs in wastewater effluent, however the impact of long-term wastewater irrigation on groundwater is apparent as non-irrigated wells impacted by flow from irrigated sites depicted similar and, in some cases, higher levels of PPCPs than irrigated wells. The Living

Filter site performs an ecosystem service by mitigating an ecosystem risk to aquatic organisms as WWTP effluent posing medium to high risk to aquatic organisms is spray irrigated to allow for natural attenuation of PPCPs before recharging groundwater. These results indicate that the ecosystem services provided by a wastewater reuse system is a feasible option to reduce the risk posed by PPCPs in wastewater effluent.

Since the occurrence of emerging contaminants in private wells are less studied, in the second objective (Chapter 4), the occurrence, range of concentrations, and potential human health risks of pharmaceutical compounds in private wells in central PA were evaluated. In the winter of 2017, samples were collected from 26 households in the West

Branch of the Susquehanna River and from the watershed outlet. At least one pharmaceutical compound was detected in each well sample, with ofloxacin, sulfamethoxazole and caffeine being the most frequently detected and present at the highest concentrations. Since studied wells were in highly forested land use systems, findings suggest that nearby septic systems are acting as important sources of pharmaceuticals to

217 groundwater. Concentrations of pharmaceuticals varied between sites with levels as low as

<1µg/L to as high as 100 µg/L and likewise detection frequencies varied from one to six compounds detected in one ground water sample indicating varying groundwater impacts per household. Further studies are however necessary to inform management practices to reduce loads of pharmaceutical to groundwater from these systems. Though risk calculations based on the measured mean concentrations suggest a minimal human health risk, concentrations in groundwater were generally higher than those measured in the watershed outlet implying that exposure of pharmaceuticals from private wells may be slightly higher at the studied groundwater sites than public water supply especially if groundwater is used without further treatment. As a onetime monitoring method, environmental and anthropogenic factors influencing levels of pharmaceuticals in these sites were not assessed. Therefore, more frequent monitoring of these systems may be desired to better understand temporal variations, long-term impacts, and risks to private wells from pharmaceuticals deposited into nearby septic systems.

When concentrations in private wells were compared to groundwater concentrations in the Living Filter reuse system that has been spray irrigated for over 40 years, average concentrations in both systems were within the same orders of magnitude except for sulfamethoxazole (17.12 µg/L) and trimethoprim (1.34 µg/L) where mean concentrations in private wells were higher than Living filter wells (2.13 µg/L and 0.61

µg/L for sulfamethoxazole and trimethoprim respectively). Similarly, average concentrations of ofloxacin were higher in Living Filter wells (13.53 µg/L) in comparison to private wells (8.13 µg/L) by around a factor of two and naproxen was below limits of

218 detection in private wells in comparison to the highest concentrations in the Living filter wells of 37.7 µg/L. Concentrations in Living Filter wells may be higher due to a higher population density, seasonally dynamic effluent concentrations, and potentially higher loading rates through irrigation in comparison the single-household set ups for private wells. Septic systems have poor water quality effluent (Garcia et al., 2013) and up to >100

µg/L of pharmaceuticals have been detected in septic tank effluent (Godfrey et al., 2007).

As opposed to WWTPs where emerging contaminants can be removed through various physical, chemical and biological processes, septic systems are relying mostly on septic tank biodegradation and soil’s filtering capacity as wastewater plumes percolate to underlying groundwater. Therefore, shallow groundwater in private wells may be more impacted especially if systems are not functioning properly or have outlived design life span among other natural and anthropogenic factors that may influence attenuation processes.

In the third objective (Chapter 5), the goal was to evaluate the occurrence of 20

EOCs in surface water used as drinking water sources for potable water production and to understand factors influencing the fate and transport of selected compounds. Samples were collected from 6 surface water sites located in the Susquehanna River Basin during a 2-yr monitoring study. Sampling regime entailed a monthly-bimonthly scale sampling and high temporal inter- and intra-day sampling protocol during varying streamflow conditions.

Findings indicate that drinking water treatment plants using reservoir sources had higher quality source water in compassion to riverine sources as concentrations and detection frequencies of studied EOCs were higher in riverine sources. As expected, watersheds with

219 lower agricultural and urban land uses that tend to have fewer point and non-point contributions of EOCs into aquatic systems had lower concentrations and frequencies of occurrence of EOCs. Seasonality was a significant factor influencing concentrations with higher concentrations recorded between the fall through spring seasons corresponding with high pharmaceutical consumption periods, seasonal use and lower environmental attenuation rates. Concentrations of EOCs were dependent on hydroclimatic factors such as precipitation. An agricultural pesticide, thiamethoxam, followed hydrograph trends signifying that high flow events following periods of high precipitation events contribute to the majority of loads in receiving surface water. In contrast, human prescription drugs such as metformin and sulfamethoxazole depicted dilution characteristics such that concentrations were highest during low streamflow conditions. Since the studied site has several upstream wastewater discharges, high flow resulting from precipitation events dampens the wastewater signal thereby resulting in lower in stream levels. Therefore, higher concentrations of EOCs of wastewater origin are expected in drinking water treatment plant influents in the colder seasons of the year during low flow conditions as opposed to high flow conditions for EOCs originating from agricultural sources.

As systems impacted with wastewater inputs, concentrations of seven common pharmaceuticals in both groundwater impacted with wastewater irrigation (Objective 1) and surface water (Objective 3) studies were compared. In general surface water quality is more dynamic than groundwater hence concentrations ranges in surface water were wider with clearer seasonal patterns. There were no significant variations between average concentrations measured in groundwater at the reuse site and average concentrations at the

220 six surface water sites studied. Concentrations were within the same orders of magnitude apart from trimethoprim that was five times higher in surface water (3.28 µg/L) in comparison to living filter wells (0.61 µg/L) and ofloxacin (13.53 µg/L) and sulfamethoxazole (2.13 µg/L) that were slightly higher in groundwater sites than surface water sites (0.73 µg/L and 9.95 µg/L, respectively). Spray irrigation of effluent diverts direct wastewater inputs to surface water sources thereby mitigating potential high acute ecotoxicological impacts by allowing naturally attenuation of effluent during infiltration processes. Given the persistent nature of some pharmaceutical compounds, and that some attenuation processes such as sorption to soil may be reduced at certain depths, there may be underlying risks from pharmaceutical compounds that persist in groundwater following prolonged irrigation.

To assess occurrence during drinking water treatment, in objective 4 (Chapter 6), a monitoring study of 7 pharmaceutical compounds was conducted in 2 conventional drinking water treatment plants (DWTPs) for a 1-yr period. Samples were collected at a monthly- bimonthly scale from drinking water sources, intermediate treatment steps and in finished drinking water. DWTP B with a reservoir source had lower concentrations of pharmaceuticals in source water in comparison to DWTP A employing a riverine source, however finished water concentrations were lower in DWTP A in comparison to DWTP

B. Treatment processes in DWTP A constituted a granular activated carbon filtration that may have resulted in better removal in comparison to DWTP B where a mixed media filtration system is employed. Sorption to activated carbon is considered as the most effective treatment for organic compounds (Aksu & Tunç, 2005; Westerhoff et al., 2005).

221

In both DWTPs, pharmaceutical compounds were detected during intermediate treatment steps and in finished water whereby some compounds depicted patterns of increasing, decreasing or constant concentrations. This may be an indication of pharmaceutical persistence or partial degradation during water treatment processes. Nevertheless, since the sampling technique employed in this study is limited by the collection of grab samples, concentration differences in intermediate treatment steps and in source and finished water samples may be due to lack of sample staggering to incorporate treatment plant residence times. Temporal variations in occurrence of pharmaceuticals were observed such that source water concentrations were higher in colder sampling months in comparison to warmer months. Similar patterns were observed for detections and concentrations in the intermediate steps and finished water samples. This implies that human exposure to pharmaceuticals through drinking water fluctuates seasonally to reflect concentration fluxes in source waters. This is relevant in risk assessment studies to establish chronic impacts from the exposure of pharmaceutical compounds. Current risk assessments using conservative exposure levels at the average concentrations in DWTP effluent suggest minimal risk, though apart from temporally fluctuating chronic exposure, risk assessments need to also incorporate the occurrence of mixtures of emerging contaminants in water.

222

Recommendations for Future Research

1. Based on the findings from this study, aquatic organisms are at higher risk from the

exposure of studied EOCs in comparison to humans. Concentrations in surface

water sources varied seasonally with highest concentrations observed in the colder

seasons especially in the winter and spring. Thus, aquatic organisms such as fish

may be at risk during pre-spawn and spawn periods which are sensitive

developmental stages. The occurrence of EOCs in the environment and associated

ecological impacts can be lowered through reduction of loads from point and non-

point sources. Since WWTP removal efficiencies are also lower during colder

seasons, studies on improving WWTP operational and treatment efficacy will be

beneficial in alleviating the ecotoxicological stressors for aquatic organisms.

Furthermore, control of sources of EOCs will consistently reduce loads to drinking

water treatment plants and exposure levels to humans through drinking water.

2. Current ecotoxicological and human health risk assessment approaches are limited

to one contaminant at a time, possibly underestimating potential effects of mixture

interactions. Studies on the effect of mixtures of pharmaceuticals and the

toxicological endpoints are necessary to improve risk assessment processes. The

impacts of chronic exposures are also essential to characterize long term impacts.

3. More studies are needed to characterize physico-chemical properties of emerging

contaminants including octanol-water coefficients, solid and soil distribution

coefficients. Since it may be costly to monitor a wide array of emerging

contaminants as needed, these characteristics are important in fate and transport

223

modeling approaches as well as predictions on attenuation in wastewater and water

treatment, and characterization of ecotoxicological and human health impacts.

4. Studies investigating the efficiencies of existing drinking water treatment

technologies should also investigate how factors such as source water

concentration, seasonal temperature variations, reagent dosages, and treatment

plant residence times influence removal extent of EOCs.

5. The detection of EOCs in finished water and intermediate treatment processes

suggest that existing conventional treatment technologies may not be adequate to

eliminate EOCs from water. Considering that some treatment technologies are

superior than others in removing EOCs, it is necessary to examine the efficiency of

combined technologies in both bench scale as well as in scaled up pilot studies.

6. Studies on the fate of pharmaceuticals in wastewater and drinking water treatments

should assess the dynamics and risks of associated metabolites and/or degradation

by-products.

7. Climate change impacts threatens both water quantity and quality. The potential

impacts of increased pollutant load into aquatic systems due to higher frequencies

of extreme events, alterations in aquatic properties such as temperature and pH on

the fate and transport of EOCs should be assessed. Evaluating potential

accompanying ecotoxicological impacts will be helpful as the susceptibility of

aquatic organisms due to all these stressors may be increased.

224

References

Aksu, Z., & Tunç, Ö. (2005). Application of biosorption for penicillin G removal: Comparison with activated carbon. Process Biochemistry, 40(2), 831–847.

Godfrey, E., Woessner, W. W., & Benotti, M. J. (2007). Pharmaceuticals in on‐site sewage effluent and ground water, western Montana. Groundwater, 45(3), 263-271.

225

APPENDIX A

Supplementary Information

Fate of Pharmaceuticals in a Spray-Irrigation System: From Wastewater to Groundwater Sample Preparation and Analysis

Sample and Reagent Preparation

The mass spectrometer was tuned and calibrated weekly, prior to analysis, using Peirce

LTQ ESI positive and negative ion solutions (ThermoFisher Scientific, Rockford, IL). All reagents used for preparation of the mobile phase (Ammonium Formate, Formic Acid and

Methanol) were LC-MS optima grade (Fisher Scientific, Waltham, MA). Water was prepared in house using a modified 8 stage 0.22µm filtered reverse-osmosis and deionization system (Apec Water, City of Industry, CA). Pharmaceutical secondary standards were procured from Sigma Aldrich (St. Louis, MO). Liquid Chromatography vials were 8-425, 1.8 ml capped glass vials (Fisher Scientific, Waltham, MA), and the filters used were 25 mm, 0.22µm polyethersulfone (PES) (VWR International, Radnor,

PA).

Based on solubility of PPCPs of interest, individual stock standards were prepared at a concentration of 5 mg/L in water miscible solvent. A working standard was made to a concentration of 1mg/L and used to make 7 subsequent diluted calibration standards with final concentrations of 500, 100, 10, 5, 1, 0.5, and 0.1 µg/L. Additionally, a fortified matrix sample was spiked with 10 µg/L of the target analytes. All working and stock solutions were stored at 4˚C. Mobile phase eluents were made by preparing buffer solutions of 4 mM ammonium formate and 0.1% formic acid in LC-MS grade methanol and in water.

226

HPLC Parameters

The chromatography system consisted of an ICS-5000+ (ThermoFisher Dionex,

Sunnyvale, CA) equipped with a 4-port proportioning gradient pump as well as a temperature thermostated column compartment (set at 30˚C). For each injection, 500µl of sample was loaded onto a Hypersil Gold aQ (20x2.1 mm 12μm) polar endcapped in-line concentrator column (Thermo Scientific, Waltham, MA). Once the sample was loaded onto the concentrator column, the samples were separated using a Hypersil Gold aQ (100x2.1 mm 3μm) polar endcapped analytical column (Thermo Scientific, Waltham, MA), prior to introduction into the mass spectrometer.

Table A-1 gives the linear gradient using aqueous buffer (eluent A) and Methanol buffer (eluent B), with an isocratic flow rate of 0.4 ml/min and total run time of 21 minutes.

Following the completion of each sample analysis, the inline concentrator and analytical columns were rinsed with neat methanol or acetonitrile, followed by mobile phase buffer to condition and re-equilibrate the columns. The operation and timing of the system was controlled through the Chromeleon 7.2 Data System (ThermoFisher Dionex, Sunnyvale,

CA).

Table A-1: Linear gradient profile of eluents Time (min) Eluent A% Eluent B% 0 100 0 10 100 0 11 80 20 18 0 100 21 0 100

227

Mass Spectrometer Parameters

The Q Exactive instrument was operated in either Full MS or ddMS2 mode. In Full MS mode, the instrument performed a full spectrum from 80-440 m/z and tuned for maximum ion throughput. The automatic gain control (AGC) target was set to 1.0 x106 with a maximum inject time of 247 ms (conditions shown in Table A-2). All quantitative data collected during this study were acquired using the full MS mode without high-energy collision dissociation (HCD) fragmentation and alternating ESI polarities.

Identification and confirmation were completed using the ddMS2 mode of operation, using a predefined inclusion list in Full MS mode followed by an MS2 scan. If a targeted compound was detected within the 10-ppm mass error range and was greater than the intensity threshold (5.0 x 104), the precursor ion was isolated by the quadrupole and sent to the HCD cell prior to mass analysis. The ddMS2 analysis, with an automatic gain control

(AGC) target, was set to 1.0 x105, with a maximum inject time of 45 ms, isolation window of ± 1.0 m/z, underfill ratio at 10%, apex trigger at 3-5 s, and dynamic exclusion at 10 s.

Table A-2: Heated Electrospray Injection source conditions. Ion source polarity Positive ion mode Negative ion mode Spray voltage 4000 V -2750 V Ion transfer tube temp. 300˚ C S-lens level 50.0 Sheath gas pressure 30 arbitrary units Auxiliary gas pressure 15 arbitrary units Sweep gas pressure 2 arbitrary units HCD cell energy Stepped (N)CE 15, 30, 45 eV MS resolution 70000 @ m/z 200 ddMS2 resolution 17500 @ m/z 200

228

Figure A-1. Groundwater elevation map for the State Gameland site. Provided by the Penn State Office of Physical Plant.

229

Figure A-2. Groundwater elevation map for the Astronomy site. Provided by the Penn State Office of Physical Plant.

230

Table A-3: Summary statistics of concentrations in groundwater wells during the one-year sampling period. Mean Median Minimum Maximum Standard % n (µg/L) (µg/L) (µg/L) (µg/L) Deviation > LOD (µg/L) P1 Acetaminophen 0.05 0.05 0.05 0.05 0 45% Ampicillin 0.36 0.36 0.36 0.36 9% Caffeine 1.27 0.15 0.05 5.85 2.56 45% Naproxen 8.79 8.79 8.79 8.79 -- 9% Ofloxacin

231

Trimethoprim 0.21 0.21 0.05 0.38 0.23 18% W6 Acetaminophen 0.09 0.05 0.05 0.22 0.09 36% Ampicillin 0.27 0.27 0.27 0.27 -- 9% Caffeine 1.03 0.15 0.05 5.11 2.01 55% Naproxen 32.49 38.84 6.95 51.69 23.03 27% Ofloxacin 3.21 3.21 3.21 3.21 -- 9% Sulfamethoxazole 1.63 0.20 0.10 10.37 3.85 64% Trimethoprim 0.17 0.17 0.05 0.29 0.17 18% W5 Acetaminophen 0.26 0.05 0.05 0.93 0.38 46% Ampicillin 0.20 0.20 0.20 0.20 -- 9% Caffeine 2.19 0.16 0.08 12.36 4.98 56% Naproxen 63.74 63.74 29.09 98.39 49.01 18% Ofloxacin 1.50 1.50 1.50 1.50 -- 9% Sulfamethoxazole 2.85 0.18 0.07 16.27 6.58 56% Trimethoprim 0.11 0.11 0.05 0.17 0.09 18% W1 Acetaminophen 0.08 0.05 0.05 0.17 0.06 36% Ampicillin 0.65 0.65 0.65 0.65 -- 9% Caffeine 3.68 0.24 0.08 14.15 6.98 36% Naproxen 33.41 33.41 21.47 45.35 16.88 18% Ofloxacin 1.50 1.50 1.50 1.50 -- 9% Sulfamethoxazole 3.82 0.05 0.05 26.42 9.97 64% Trimethoprim 0.09 0.09 0.05 0.12 0.05 18% P2 Acetaminophen 3.93 0.05 0.05 15.58 7.76 36% Ampicillin 2.09 2.09 0.48 3.69 2.27 18% Caffeine 4.58 3.24 0.09 11.74 5.59 36% Naproxen 18.90 18.76 13.13 24.80 5.84 27% Ofloxacin 58.22 58.22 1.50 114.94 80.22 18% Sulfamethoxazole 1.12 0.05 0.05 9.54 3.16 82% Trimethoprim 3.53 3.53 0.11 6.95 4.84 18% F3 Acetaminophen 0.07 0.05 0.05 0.15 0.05 36% Ampicillin 0.37 0.37 0.37 0.37 -- 9% Caffeine 2.32 0.22 0.09 8.76 4.29 36% Naproxen 19.93 19.93 14.93 24.93 7.07 18% Ofloxacin

232

Naproxen 37.93 37.93 34.71 41.15 4.56 18% Ofloxacin

233

APPENDIX B

Master Well Owner Network (MWON) Volunteer Communication Templates

WATER SAMPLE COLLECTION KIT BOX AND HANDLING INSTRUCTIONS

You have received a Uline shipping box

Inside the box, there is an insulated foam container and this instruction manual on top of the lid. Open the lid of the insulated foam container to view packed items.

Inside the insulated foam container, the following items have been packed for sampling purposes: Packed Item Image A sample collection procedure sheet (printed on green paper)

A short survey about your water source (printed on turquoise paper)

234

Black Marker

2 pairs of nitrile gloves

Bubble wraps and bubble wrap envelopes

4 Ice Packs

50ml Colorless plastic bottle filled with nano-pure water labeled ‘nanopure water’

25 ml Amber glass bottle labeled ‘field blank’

50ml Amber glass bottle labeled ‘well/spring water sample’

Control sample (taped to the inside of the sampling kit)

235

Prepaid return UPS overnight shipment sticker

Handling Instructions • Please follow the instructions on the sample collection procedure outlined in the “Sample Collection Procedure” sheet provided, and the handling directions provided below. • We are interested in sampling RAW water from your private well or spring i.e. water directly from the well or spring prior to any purification steps (e.g. water softening, filtration, reverse osmosis, disinfection etc.). The study is analyzing for emerging organic contaminants in water. • Avoid the use of antibiotic soap, hand sanitizers etc. before sample collection. These personal cleaning items may contain triclosan, which is an active ingredient, tested in this study. • A control sample has been taped to inside of the sampling kit. Please do not remove or tamper with it. This control sample is used for laboratory purposes only to study the transformation of the emerging organic compounds during transportation. • Freeze the icepacks after receiving this package. Remember to repack them into the insulated foam container prior to mailing in your collected samples. It is important that the samples remain chilled during shipment to maintain sample integrity and prevent possible degradation and transformation • The compounds being tested for have short half-lives and can degrade if return shipment is delayed. This can result in misleading findings in your well/spring water sample. We therefore recommend shipment within 24 hours of sample collection using the enclosed prepaid UPS overnight shipping sticker. We recommend shipping on Monday, Tuesday, Wednesday and Thursday. This will ensure sample delivery during the week and enables prompt sample analysis. • The sampling kit is insulated and designed to minimize direct impact on sampling bottles. We request that you repack all collected samples and icepacks back in this kit, reseal and mail it back to us using enclosed UPS overnight shipping sticker. • If you have any questions regarding the sample collection and handling procedures, please contact Faith Kibuye via email at [email protected] or a phone call at (814)- 863-8233.

236

SAMPLE COLLECTION PROCEDURE

STEP 1: Collecting ‘Field Blank’ sample I. Wear one pair of the enclosed gloves II. Transfer the water from the colorless plastic bottle labeled ‘Nanopure water’ into the 125ml amber glass bottle labeled ‘Field Blank’ filling it completely leaving no headspace. Secure the bottle lid tightly.

Pour the nanopure water from the plastic bottle into the ‘Field Blank’ amber glass bottle

III. Record the sample collection time and date on the ‘Field Blank’ white label using provided marker IV. Wrap the ‘Field Blank’ bottle in a bubble wrap envelope and add the bottle to the insulated foam kit box. V. Pour out the remaining water in the colorless plastic ‘Nanopure water’ bottle, secure the lid tightly and return the bottle to the insulated foam kit box.

STEP 2: Collecting your private well or spring water I. Identify the tap where you will take your sample. - If you do not have any water treatment devices or equipment e.g water softener, water filters etc. you can use your kitchen sink tap. - If you do have any water treatment devices and/or equipment, identify a tap before any treatment devices. Sometimes the outside faucet bypasses the treatment equipment. If you use the outside faucet then please remove any attached hoses from the faucet before you take your sample. Another option may be the tap on the bottom the pressure tank. II. The water source should be purged for 8 minutes to get a representative sample. - Let the tap run for 8 minutes. - If you are using the faucet on the pressure tank you can run a faucet at the sink for 7.5 minutes and then just run the pressure tank faucet for 30 seconds before collecting your sample. III. Wear the other pair of enclosed clean gloves once you are ready to collect the sample IV. Rinse out the sample bottle 125ml amber glass bottle labeled ‘Well/Spring water’ three times with your raw well or spring water before collecting the actual sample.

237

V. When collecting the sample, completely fill the ‘Well/Spring water sample’ bottle with your raw well or spring water. Ensure the bottle is completely filled with water leaving no headspace. Secure the bottle lid tightly.

VI. Record the sample collection time and date on the label on the bottle using provided marker VII. Securely wrap the ‘Well/Spring water sample’ bottle using provided bubble wraps and add the bottle to the insulated foam kit box. STEP 3: Water Source Survey I. Remove the survey sheet from the plastic wrap II. Fill out the short survey on your water source (private well or spring). III. Return the survey in the protective plastic wrap after filling it out and add it to the insulated foam kit box. STEP 4: Sample Kit Box Shipping I. Pack the ‘field blank’ and ‘well/spring water’ sample bottles full of water samples; empty plastic ‘nanopure water’ bottle; filled out survey and frozen icepacks into the insulated foam kit box. II. Check the boxes below to ensure you have packed everything. Well or spring water sample, with secure lid, wrapped with bubble wrap Field Blank Sample, with secure lid, wrapped in bubble wrap Completed survey Empty nanopore water plastic container Frozen Icepacks

III. Secure the lid/top cover of the insulated foam kit box with its lid to ensure samples do not shift during shipment. IV. Replace the original shipment sticker on the Uline shipping box with the enclosed prepaid return UPS sticker (You can tape the return sticker over the used sticker). - Remove the return shipment sticker from the plastic wrap. - Fold the paper along the line written ‘Fold here’

238

- Replace the old sticker by taping it to the box. Ensure its four edges are securely taped.

V. Reseal the outside Uline shipping box using some tape to secure it. VI. Drop off the sealed kit box for shipment at the nearest UPS store or arrange for pickup by UPS. Due to a short holding time for sample processing, we recommend shipping on Mondays, Tuesdays, Wednesdays, and Thursday. Please avoid shipping on Fridays and Saturdays.

239

SOURCE WATER SURVEY Site No. xx This is a short survey to enable the project team to know more about you water source i.e private well or spring. 1. What is the source of your home water supply?

Private well

Spring

2. If you are using a private well for your home water supply, what is the depth of your well?

Feet

I do not know the depth of my well

3. Is there a septic tank within your property?

Yes

4. If you answered ‘Yes’ to question 3: Is the septic tank in active use? Yes

5. Any comments:

240

RESULT COMMUNICATION TEMPLATE

Occurrence of Emerging Organic Contaminants in Groundwater Sample ID: Site No. xx Date Sampled: 2/2/2017 Date Received: 2/2/2017 Type of analysis: Emerging organic compounds (EOCs) including acetaminophen, ampicillin, caffeine, naproxen, ofloxacin, sulfamethoxazole, and trimethoprim. Background on EOCs EOCs include pharmaceutical compounds, ingredients in personal care products, and hormones. They are referred to as emerging contaminants since they have just recently been quantified in the environment at trace concentration in the nanogram and microgram per liter range (ng-µg/L). EOCs do not have any drinking water regulation standards and therefore lack legal concentration limits. Since EOCs analyzed for in this study are human pharmaceutical compounds, their potential source in your groundwater include septic tanks, septic drain fields and or leakage from associated piping. Sample Analysis All samples were analyzed using a Thermo Scientific Q Exactive Orbitrap Liquid Chromatography-Mass Spectrometry (LC-MS/MS) adapting EPA Method 1694. • This instrument allowed us to detect compounds with concentrations as low as 0.01 µg/L. In Table 1, we have reported any compound that could not be detected as less than the method detection limit (

Table 1: Concentrations of analyzed EOCs in your well water sample Compound Commercial Use Concentration in Well Water (µg/L) Acetaminophen Analgesic 0.40 Caffeine Stimulant 0.27 Ampicillin Antibiotic 0.46 Naproxen Nonsteroidal anti-inflammatory

241

Sulfamethoxazo Antibiotic 9.32 le Trimethoprim Antibiotic 3.24 Comparison with water samples from all participants We analyzed a total of 26 well and spring water samples. Table 2 below summarizes the mean, minimum and maximum concentrations that were quantified (greater than or equal to the LOQ, 0.1 µg/L) in well and spring water samples collected as part of this study. Samples with concentrations less than method detection limit (0.01 µg/L) and method quantification limits (0.1 µg/L) are not included in this summary. Naproxen was not detected in any sample.

Table 2: Summary of concentrations of EOCs in analyzed well and spring water samples Average Maximum Minimum Concentration Concentration Concentration Compound (µg/L) (µg/L) (µg/L) Acetaminophen 0.94 2.77 0.24 Caffeine 0.44 0.73 0.20 Ampicillin 8.11 13.07 1.69 Naproxen Not Detected in any well/spring sample Ofloxacin 8.13 122.93 1.5 Sulfamethoxazole 17.12 32.04 0.10 Trimethoprim 1.34 3.24 0.10

Figure 1 below represents a comparison of the concentrations from your well water sample with the overall mean concentrations summarized in table 1.

242

Figure 1: Comparison of concentrations in your well sample with average concentrations of all analyzed well and spring water samples

Some compounds were more frequently detected than others in the 26 well and spring water samples analyzed in this study. The most detected compounds were ofloxacin, detected in all samples, sulfamethoxazole detected in 58% of the samples and ampicillin and caffeine detected in 46% of analyzed samples. It is important to note that compounds that were

Figure 2: Detection frequencies of EOCs in analyzed well and spring water samples

243

What can be done? In order to minimize environmental concentrations of EOCs including pharmaceutical compounds: • do not flush drugs down the toilet unless the label explicitly instructs you to do so • take advantage of pharmaceutical take-back programs. Find out from your municipal/county government office and pharmacist if they have any drug take- back programs available. • More information on pharmaceuticals in water can be found at Penn State’s Extension site: http://extension.psu.edu/natural-resources/water/drinking-water/water- testing/pollutants/pharmaceutical-disposal-and-water-quality

If you would like to reduce the concentrations of EOCs in your tap water, faucets can be fitted with activated carbon or reverse osmosis filters. When drawing water from a filter fitted faucet, allow the water to trickle slowly. This increases the residence time with the filter. However, EOCs do not have drinking water standards and therefore no guidance as to “safe” concentrations can be provided by Penn State.

Curriculum vitae

Faith Awino Kibuye

EDUCATION Ph.D., BioRenewable Systems, Pennsylvania State University, University Park, PA. Dissertation 2018: Effects of Natural and Anthropogenic Drivers on Emerging Organic Contaminants in Wastewater and Drinking Water Systems: Occurrence and Removal. Advisor: Heather E. Gall, Ph.D. BSc, Environmental Health Science GPA: 3.94/4.0, Benedict College, Columbia, SC, May 2015 Thesis: Heavy Metal Adsorption Using Pinecone Derived Adsorbents Advisor: Samuel Darko, Ph.D.

PROFESSIONAL AND ACADEMIC EXPERIENCE Graduate Research Assistant Fall 2015-Current Pennsylvania State University, Agricultural and Biological Engineering Department Graduate Teaching Assistant; Pennsylvania State University Fall 2016 Course: ASM / ERM 309: Measurement and Monitoring of Hydrologic Systems Research Intern, Oak Ridge National Laboratory Summer 2014 Research area: Geochemistry - Organic Matter Adsorption to Selected Mineral Soil Fractions, Atomic Force Microscopy (AFM) measurements and analysis Undergraduate Research Assistant Fall 2011-Summer 2015 Benedict college Advanced Materials Laboratory Research area: Heavy metal remediation from water using biomass derived adsorbents Undergraduate Teaching Assistant Spring 2012-Fall 2014 Benedict College, Centre for Teaching and Learning; Courses: Biology, Physics & French

FELLOWSHIPS, GRANTS, AND SCHOLARSHIPS Penn State Agricultural & Biological Engineering Departmental Fellowship (2015-2019) Penn State College of Agriculture Competitive Grants, December 2017, $3,000 Educational Testing Service (ETS) Presidential HBCU Scholarship (2013-2014)-$16,000 Benedict College International Academic Scholarship Award (2011-2015)

PEER-REVIEWED PUBLICATIONS Kibuye, F.A., H.E. Gall, K.R. Elkin, B. Ayers, T.L. Veith, M. Miller, S. Jacob, K.R. Hayden, J.E. Watson, and H.A. Elliott. 2019. Fate of pharmaceuticals in a spray-irrigation system: From wastewater to groundwater. Science of the Total Environment, 654: 197- 208. DOI: 10.1016/j.scitotenv.2018.10.442 Kibuye, F.A., H.E. Gall, K.R. Elkin, B. Swistock, T.L. Veith, J.E. Watson, and H.A. Elliott. 2019. Occurrence, concentrations, and risks of pharmaceutical compounds in private wells in Central Pennsylvania. Journal of Environmental Quality. DOI: 10.2134/jeq2018.08.0301