Improving Water Reuse for a Healthier Potomac Watershed Improving Water Reuse for a Healthier Potomac Watershed

Home , Accotink Creek, Anacostia River, Antietam Creek, Broad Run (Maryland), Cabin John Creek, Conococheague Creek

PROJECT NO. 4790 Improving Water Reuse for a Healthier Potomac Watershed Improving Water Reuse for a Healthier Potomac Watershed

Prepared by: Sujay Kaushal Shuiwang Duan University of Maryland College Park Erik Rosenfeldt Hazen and Sawyer Amelia Flanery Adil Godrej Virginia Polytechnic Institute and State University Luke Iwanowicz United States Geological Survey Diana Aga University at Buffalo, The State University of New York

2020 The Water Research Foundation (WRF) is a nonprofit (501c3) organization which provides a unified source for One Water research and a strong presence in relationships with partner organizations, government and regulatory agencies, and Congress. The foundation conducts research in all areas of drinking water, wastewater, stormwater, and water reuse. The Water Research Foundation’s research portfolio is valued at over $700 million. The Foundation plays an important role in the translation and dissemination of applied research, technology demonstration, and education, through creation of research‐based educational tools and technology exchange opportunities. WRF serves as a leader and model for collaboration across the water industry and its materials are used to inform policymakers and the public on the science, economic value, and environmental benefits of using and recovering resources found in water, as well as the feasibility of implementing new technologies. For more information, contact: The Water Research Foundation 1199 North Fairfax Street, Suite 900 6666 West Quincy Avenue Alexandria, VA 22314‐1445 Denver, Colorado 80235‐3098 www.waterrf.org P 571.384.2100 P 303.347.6100 [email protected] ©Copyright 2020 by The Water Research Foundation. All rights reserved. Permission to copy must be obtained from The Water Research Foundation. WRF ISBN: 987‐1‐60573‐508‐5 WRF Project Number: 4790 This report was prepared by the organization(s) named below as an account of work sponsored by The Water Research Foundation. Neither The Water Research Foundation, members of The Water Research Foundation, the organization(s) named below, nor any person acting on their behalf: (a) makes any warranty, express or implied, with respect to the use of any information, apparatus, method, or process disclosed in this report or that such use may not infringe on privately owned rights; or (b) assumes any liabilities with respect to the use of, or for damages resulting from the use of, any information, apparatus, method, or process disclosed in this report. University of Maryland College Park, Virginia Polytechnic Institute and State University and Hazen and Sawyer This publication was developed under Assistance Agreement No. 83582501 awarded by the U.S. Environmental Protection Agency. It has not been formally reviewed by EPA. The views expressed in this document are solely those of the grantees and do not necessarily reflect those of the Agency. EPA does not endorse any products or commercial services mentioned in this publication. This document was reviewed by a panel of independent experts selected by The Water Research Foundation. Mention of trade names or commercial products or services does not constitute endorsement or recommendations for use. Similarly, omission of products or trade names indicates nothing concerning The Water Research Foundation's or EPA’s positions regarding product effectiveness or applicability.

ii The Water Research Foundation Acknowledgments

Research Team Principal Investigators: Sujay Kaushal, Ph.D. Shuiwang Duan, Ph.D. University of Maryland College Park Project Team: Erik Rosenfeldt, Ph.D., P.E. (Project Team Lead) Paul Knowles, Ph.D., P.E., ENV SP, LEED GA Hazen and Sawyer Diana S. Aga, Ph.D. Ping He, Ph.D. University at Buffalo, The State University of New York Amelia Flanery, M.S. Adil Godrej, Ph.D., P.E. Virginia Polytechnic Institute and State University Luke Iwanowicz, Ph.D. United States Geological Survey Sudhir Murthy, Ph.D., P.E., BCEE DC Water and Sewer Authority We would also like to acknowledge these additional stakeholders that supported the research team and PAC members during the October 2 Workshop. Lisa Ragain, MWCOG Justin Mattingly, PMP, U.S. EPA (Formerly WRF) Ann Spiesman, P.E., Washington Aqueduct Pam Kenel, P.E., Loudoun Water Greg Prelewicz, P.E., Fairfax Water Matt Ries, Ph.D., P.E., DC Water WRF Project Subcommittee or Other Contributors Susan T. Glassmeyer, Ph.D. U.S. Environmental Protection Agency Bob Angelotti, P.E. Upper Occoquan Service Authority (UOSA) Leita S. Bennett, P.E. GHD Steven Bieber Metropolitan Washington Council of Governments (MWCOG) Water Research Foundation Staff John Albert, MPA Chief Research Officer Lola Olabode, MPH Research Program Manager

Improving Water Reuse for a Healthier Potomac Watershed iii Abstract and Benefits

Abstract: The objectives of this study were to determine the ecological and human health benefits associated with improving reuse in the Potomac River Watershed. To address this objective, extensive sampling campaigns were performed to 1) assess presence and sources of conventional (i.e., nutrients) and emerging (i.e., endocrine disrupting compounds, pesticides) pollutants into the River, and 2) define potential benefits from pollution management strategies in four sectors: agriculture BMPs, urban stormwater BMPs, enhanced nutrient reduction at water reclamation facilities, and advanced water treatment for potable reuse. The results of the study indicated that agricultural inputs of pollutants accounted for the greatest share of the sectors, and that nutrient management strategies were effective in co-managing many CECs (constituents of emerging concern). Additionally, Urban stormwater management strategies were less effective for nutrient control (particularly nitrogen), and CEC control. Water reclamation and reuse strategies were both effective for nutrient reduction, and reuse was particularly effective at controlling CECs, however their input of these pollutants into the Potomac was significantly smaller than agriculture and urban stormwater inputs. Finally, a multi-criteria decision analysis framework was utilized to assess relative effectiveness of the four strategies for cost effectively and equitably improving ecological and human health in the environment. The results of the assessment indicate agriculture runoff management strategies were the most impactful and least costly of the four alternatives. Most recently, members of our team have developed a method using suspect screening analysis (SSA) that relies on accurate mass measurements and mass spectral fragmentation patterns obtained using liquid chromatography coupled to high-resolution mass spectrometry to achieve a wide-scope detection of micropollutants. A total of 103 micropollutants were detected in the samples, many of which were not previously monitored in the original target analysis method; 23 of these additional compounds have detection frequency of at least 50%. These compounds belong to various classes such as pharmaceuticals and personal care products (PPCPs), per- and polyfluoroalkyl substances (PFASs), organophosphate flame retardants (OPFRs), and various pesticide and their metabolites. In the future, further work is necessary to investigate these complex mixtures of micropollutants and the impacts of management on their concentrations and loads. Overall, our work demonstrates the potential for co- managing multiple pollutants in surface waters through an improved and integrated understanding of targeting shared sources, transport, and transformation of multiple contaminants in the environment. Benefits: • Presents a comprehensive dataset of conventional and emerging pollutant concentrations throughout the Potomac Watershed, with over 45 locations along the Potomac and major tributaries sampled quarterly in 2017 to assess “hotspots”, “hot-moments” and land-use correlations with pollutant loads, and another 17 locations monitored monthly in 2018 to assess impacts of sector-specific BMPs on load reduction. • Provides an estimate of relative loading of point and non-point sources to the Potomac Watershed, describing a larger contribution of nutrients and CECs from agricultural practices when compared to the flow contribution to the Watershed. • Demonstrates the effectiveness of sector-specific BMPs for nutrient and CEC co-management. • Multi-criteria decision analysis performed to evaluate four nutrient and CEC co-management strategies indicates that implementation of agricultural BMPs was the most cost-effective and equitable strategy for improving ecological and human health in the Potomac. Keywords: Nutrient management, constituents of emerging concern, Potomac River, Chesapeake Bay, multi-criteria decision analysis, best management practices.

iv The Water Research Foundation Contents

Acknowledgments ...... iii Abstract and Benefits ...... iv Tables ...... viii Figures ...... ix Acronyms and Abbreviations ...... xii Executive Summary ...... xiii

Chapter 1: Hotspots of EDC, Nutrients and Organic Carbon in the Potomac Watershed ...... 1 1.1 Potomac Watershed ...... 1 1.1.1 Watershed General Information ...... 1 1.1.2 Geological Setting ...... 2 1.1.3 Land Use...... 2 1.1.4 Population, Cities, and Industries ...... 3 1.2 Sampling Design ...... 4 1.2.1 Site Selection ...... 4 1.2.2 Sampling Frequency, Sample Collection, and Preparation ...... 5 1.3 Chemical Analyses ...... 7 1.4 Hotspots of EDC ...... 8 1.4.1 Hotspots of Estrogenic Activity ...... 8 1.4.2 Hotspots of Estrogen ...... 10 1.4.3 Hotspots of Herbicides...... 12 1.4.4 Linkage of Estrogenic Activity (BLYES) with EDCs ...... 13 1.5 Hotspots of Nutrient and Organic Carbon ...... 15 1.5.1 Hotspots of Nitrogen (N) ...... 15 1.5.2 N Source Tracking with Stable C and O Isotopes ...... 17 1.5.3 Hotspot of Phosphorus (P) ...... 19 1.5.4 Hotspot of Organic Carbon (C) ...... 21 1.5.5 Linking EDCs with Nutrients, Organic Carbon and Other Variables ...... 23 1.6 Source Estimation for Nutrients, DOC, and EDCs of the Potomac River ...... 24

Chapter 2: Impacts and Outcomes of Current Nutrient Management Strategies On Source Controls of EDCs at Maryland Sites ...... 27 2.1 Introduction ...... 27 2.2 Experiment Design ...... 27 2.2.1 Site Selection ...... 28 2.2.2 Water Sample Collection ...... 29 2.2.3 Chemical Analyses ...... 29 2.3 EDCs Data Synthesis ...... 31 2.3.1 Pesticides (or Synthetic Organic Compounds) ...... 31 2.3.2 Estrogen ...... 35 2.4 Effect of Agricultural BMPs on Nutrient and EDCs Reductions ...... 36 2.4.1 Description of Sampling Site and Agricultural BMPs ...... 36 2.4.2 Effect of Agricultural BMPs on Nutrient Reductions ...... 36 2.4.3 Effect of Agricultural BMPs on EDCs Reductions ...... 38 2.5 Effect of Urban BMPs on Nutrients and EDCs Reductions ...... 40

Improving Water Reuse for a Healthier Potomac Watershed v 2.5.1 Description of Sampling Site and Urban BMPs ...... 40 2.5.2 Effect of Urban BMPs on Nutrient Reductions ...... 40 2.5.3 Effect of Urban BMPs on EDCs Reductions ...... 41 2.6 Effect of WWTPs Discharges on Nutrients and EDCs of Receiving Waters ...... 44 2.6.1 Description of Seneca WWTP and Blue Plains WWTP ...... 44 2.6.2 Effect of WWTPs Discharges on Nutrients of the Receiving Waters ...... 44 2.6.3 Effect of WWTPs Discharges on EDCs of Receiving Waters ...... 48 2.7 Estimating Sources of Nutrients, DOC, and EDCs from Point and Nonpoint Sources to the Potomac River ...... 54 2.7.1 Fluxes of Nutrients, DOC, and EDCs from Point Sources ...... 54 2.7.2 Changes in Relative Contributions of Nutrients, DOC, and EDCs from Nonpoint Sources with BMPs Applications ...... 55

Chapter 3: Impacts and Outcomes of Current Nutrient Management Strategies On Source Controls of EDCs at Virginia Sites ...... 57 3.1 Experiment Design ...... 57 3.1.1 Site Selection ...... 57 3.1.2 Water Sample Collection ...... 58 3.1.3 Chemical Analyses ...... 59 3.2 Analyses of SOCs and Estrogen ...... 61 3.2.1 SOCs ...... 61 3.2.2 Estrogen ...... 65 3.3 Agriculture Sampling Sites Comparison ...... 66 3.3.1 Description of the Sampling Site and the Agricultural BMPs ...... 66 3.3.2 Effect of Agricultural BMPs on Nutrient and Organic Carbon Reductions ...... 66 3.3.3 Effect of Agricultural BMPs on Estrogen ...... 68 3.3.4 Effect of Agricultural BMPs on SOCs ...... 69 3.3.5 Evaluating Co-Management of Prometon and Nutrients ...... 70 3.4 Urban Sites Comparison ...... 71 3.4.1 Description of Sampling Site and Urban BMPs ...... 71 3.4.2 Effect of Urban BMPs on DOC and Nutrients ...... 72 3.4.3 Effect of Urban BMPs on Estrogen ...... 73 3.4.4 Effect of Urban BMPs on SOCs ...... 74 3.4.5 Evaluating Co-Management of 4-Nonylphenol and Nutrients ...... 76 3.5 UOSA Advanced Water Reclamation Facility ...... 77 3.5.1 Description of Sampling Sites ...... 77 3.5.2 Nutrients and DOC at UOSA AWRF Sampling Sites ...... 77 3.5.3 Estrogens at UOSA AWRF Sampling Sites ...... 79 3.5.4 SOCs at UOSA WWTP Sampling Sites ...... 79 3.5.5 UOSA SOCs Load Contribution to Bull Run and the Occoquan ...... 81 3.6 Water Treatment Plant Intake Load Comparisons ...... 84 3.6.1 Site Description ...... 84 3.6.2 DOC and Nutrients ...... 84 3.6.3 Estrogens ...... 85 3.6.4 SOCs ...... 86

vi The Water Research Foundation Chapter 4: Cost Benefit Analysis ...... 89 4.1 Cost Benefit Analysis Framework ...... 89 4.1.1 HazenConverge Tool Description...... 89 4.2 Stakeholder Criteria Development and Weighting Workshop ...... 92 4.2.1 Agenda ...... 93 4.2.2 Background on the Project to Date ...... 93 4.2.3 Introduction – MCDA Overview and Definition of Alternatives ...... 96 4.2.4 Category and Criteria Development Methods and Results ...... 98 4.2.5 Pairwise Comparison Methods and Results ...... 102 4.3 Defining the Metrics for Each Alternative ...... 105 4.3.1 Quantitative Metrics – Performance and Cost ...... 105 4.3.2 Qualitative Metrics ...... 109 4.4 Scoring Alternatives ...... 113 4.5 Conclusions of the Multi-Criteria Cost/Benefit Assessment ...... 120

Chapter 5: Conclusions ...... 121 5.1 Hot Spot and Hot Moment Analysis ...... 121 5.2 Contaminant Source Allocation ...... 121 5.3 Impacts of Best Management Practices ...... 122 5.4 Impact of Planned and Unplanned Indirect Potable Reuse ...... 122 5.5 Multi-Criteria Decision Analysis ...... 123

Appendix A ...... 125 Appendix B ...... 165

References ...... 167

Improving Water Reuse for a Healthier Potomac Watershed vii Tables

1-1 Land Uses of 32 Selected Sub-Watersheds of the Potomac River ...... 5 1-2 Correlation of Estrone, Estrogenic Activity, and Common Species of Herbicides with Nutrients, Organic Carbon, and Other Water Quality Parameters ...... 23 1-3 Estimated Annual Fluxes of Water, Nutrients, Organic Carbon, Estrogen, and Herbicides from Forest, Agricultural Runoff (Cropland and Pasture), Urban Runoff, and Urban Point Sources in the Potomac River ...... 24 2-1 Site Selected for Examining Effect of Agricultural BMPs, Urban BMPs, and Wastewater Treatment Plants on Nutrients and EDCs Retentions in Maryland and Virginia ...... 29 2-2 Estimates of Nutrient and DOC Yields from the Two Agricultural Streams Without or With Best Management Practices (BMPs) ...... 36 2-3 Estimates of EDCs Yields from the Two Agricultural Streams Without or With Best Management Practices (BMPs) ...... 38 2-4 Estimates of Nutrient and DOC Yields from the Two Urban Streams Without or With Best Management Practices (BMPs) ...... 41 2-5 Estimates of EDCs Yields from the Two Urban Streams Without or With Best Management Practices (BMPs) ...... 42 2-6 Flow-Averaged Concentrations of EDCs in the Effluents of Three Wastewater Treatment Plants ..... 54 2-7 Estimated Annual Fluxes of Water, Nutrients, Organic Carbon, Estrogen, and Herbicides from Forest, Agricultural Runoff (Cropland and Pasture), and Urban Runoff, and Urban Point Sources (WWTPs) in the Potomac River ...... 55 2-8 Effectiveness (%) in Reduction Nutrients, DOC, and EDCs with the Agricultural and Urban BMPs at the Maryland Sites and Virginia Sites ...... 56 2-9 Annual Fluxes of Nutrients, DOC, and EDCs from Point and Nonpoint Sources When Agricultural and Urban BMPs of This Study Applied to All the Potomac Watershed (Table 2-8) ...... 56 4-1 Example Pre-Loaded Criteria from the HazenConverge Tool ...... 91 4-2 Estimate of Pollutant Load Reduction with BMP Implementation ...... 95 4-3 Cost of BMP Implementation for Nutrient Control in the Chesapeake Bay ...... 96 4-4 Workshop Stakeholders ...... 97 4-5 Results of the Biased Representation Exercise ...... 98 4-6 Results of the Balanced Representation Exercise ...... 99 4-7 Similarity and Categorization ...... 100 4-8 Final Outcomes of the Criteria Identification Exercise ...... 101 4-9 Results of the Individual Stakeholder Weighting Exercise ...... 103 4-10 Potomac River Estimates of Percent Reduction to Achieve 2025 TMDL ...... 106 4-11 CEC Reduction Performance Summary ...... 106 4-12 Total Watershed Load Reductions Associated with Sector-Specific BMP Implementation ...... 107 4-13 Incremental Costs of BMP Implementation to Address TN, TP, and CECs ...... 108 4-14 Sector-Specific Aggregate Costs Required for Removing TN to Achieve 2025 TMDL ...... 108 4-15 Aggregate Costs of “Shared” TN Reduction to Achieve 2025 TMDL ...... 109 4-16 Summary of Criteria Scores, with Justification ...... 114

viii The Water Research Foundation Figures

1-1 Map of the Potomac River Watershed Showing Location and Primary Tributaries ...... 1 1-2 Generalized Geology and Physiography in the Potomac River Basin ...... 2 1-3 Map of Land Use in the Potomac River Basin ...... 3 1-4 Sampling Sites on the Map of Land Use in the Potomac River Basin ...... 4 1-5 Flow Conditions of the Potomac River When River Water Samples Were Collected ...... 6 1-6 Hotspots of Estrogenic Activity in the Potomac River Watershed During the Water Year 2016-2017 ...... 9 1-7 Hot Moments of Estrogenic Activity in the Hot Spots of the Potomac River Watershed During the Water Year 2016-2017 ...... 9 1-8 Correlations Between Estrogenic Activity (or BLYES) and Percent of Pasture/Hay Land Use in the Primary Tributaries of the Potomac River Watershed ...... 10 1-9 Hotspots of Total Estrogen (Estrone and Estrone-3S) in the Potomac River Watershed During the Water Year 2016-2017 ...... 11 1-10 Correlations Between Mean Total Estrogen and the Percent of Developed Land Uses (agricultural% + urban%) ...... 11 1-11 Hotspots of Total Herbicides in the Potomac River Watershed During the Water Year 2016-2017 ... 12 1-12 Seasonal Changes in Concentrations of Dominant Herbicide Species in the Selected Hotspot Sub-Watersheds of the Potomac River During the Water Year 2016-2017 ...... 13 1-13 Correlations Between Mean Total Herbicides and Percent of Agricultural Land Use (Cropland plus Pasture) in the Primary Tributaries of the Potomac River Watershed ...... 13 1-14 Correlations Between Estrogenic Activity and Total Estrogen, Total Pesticide, or Any Single EDC ..... 14 1-15 Spatial Distribution of Total Dissolved Nitrogen (TDN) Showing TDN Hotspots in the Potomac River Watershed During the Water Year 2016-2017 ...... 15 1-16 Correlations Between Mean Total Dissolved Nitrogen (TDN) Concentration and Percentage of Watershed Cropland Land Use (cropland %) ...... 16 1-17 Hot Moments of Total Dissolved Nitrogen Concentrations (TDN) in the Selected Hotspot Sub-Watersheds of the Potomac River During the Water Year 2016-2017 ...... 16 1-18 Longitudinal Changes in Nitrate Isotopic Compositions in the Main Stem and Primary Tributaries of the Potomac River During Low Flow Conditions and Possible Nitrate Sources from Dual Nitrate Isotopic Tracers ...... 18 1-19 Spatial Distribution of Soluble Reactive Phosphorus (SRP) Showing SRP Hotspots in the Potomac River Watershed During the Water Year 2016-2017 ...... 19 1-20 Correlations Between Soluble Reactive Phosphorus (SRP) and Percent of Cropland Land Use in the Primary Tributaries of the Potomac River Watershed ...... 20 1-21 Hot Moments of Soluble Reactive Phosphorus (SRP) in the Selected Hotspot Sub-Watersheds of the Potomac River During the Water Year 2016-2017 ...... 20 1-22 Spatial Distribution of Dissolved Organic Carbon (DOC) Showing DOC Hotspots in the Potomac River Watershed During the Water Year 2016-2017 ...... 21 1-23 Correlations Between Dissolved Organic Carbon (DOC) and Percent of Urban Land Use in the Primary Tributaries of the Potomac River Watershed ...... 22 1-24 Hot Moments of Dissolved Organic Carbon (DOC) in the Selected Hotspot Sub-Watersheds of the Potomac River Watershed During the Water Year 2016-2017 ...... 22 1-25 Relative Contributions of Water Flow, Nutrients, and DOC from Tributaries with Forest, Cropland, Pasture/Hay, and Urban/Suburban Land Use ...... 25

Improving Water Reuse for a Healthier Potomac Watershed ix 1-26 Relative Contributions of Estrogen and Pesticides from Tributaries with Forest, Cropland, Pasture/Hay, and Urban/Suburban Land Use ...... 26 2-1 Water Quality Sampling Sites on the Land Use Land Cover Map of the Potomac River Basin ...... 28 2-2 Flow Conditions of the Potomac River and Selected Tributaries when River Water Samples Were Monthly Collected ...... 30 2-3 Pesticides (or Synthetic Organic Compounds) by Number of Detections and by Percent Detection .. 31 2-4 Frequent Pesticides Detected ...... 32 2-5 4-Nonylphenol Concentrations Present at Virginia Sampling Locations ...... 33 2-6 Atrazine Concentrations Present at Virginia Sampling Locations ...... 34 2-7 Estrone (E1, in red) and Estrone-3-Sulfate (E1-3S, in blue) Concentrations Present at Virginia Sampling Locations ...... 35 2-8 Comparison of Temperature (t), Dissolved Oxygen (DO), Specific Conductance (SC), Total Dissolved Nitrogen (TDN), Soluble Reactive Phosphorus (SRP), and Dissolved Organic Carbon (DOC) Between Two Agricultural Streams Without or With Best Management Practices (BMPs) ...... 37 2-9 A Herd of Cattle Accessed to the Unrestored Agricultural Stream on February 2018, Just Above the Water Sampling Site ...... 37 2-10 Comparison of Estrone (E1), Estrone Sulfate (E1-S), and Other Common Pesticides Between Two Agricultural Streams Without or With Best Management Practices (BMPs) ...... 39 2-11 Comparison of Temperature (t), Dissolved Oxygen (DO), Specific Conductance (SC), Total Dissolved Nitrogen (TDN), Soluble Reactive Phosphorus (SRP), and Dissolved Organic Carbon (DOC) Between Two Urban Streams Without or With Best Management Practices (BMPs) ...... 41 2-12 Comparison of Estrone (E1), Estrone Sulfate (E1-S), and Other Common EDCs (mainly pesticides) Between Two Urban Streams Without or With Best Management Practices (BMPs) ...... 43 2-13 Downstream Changes in Water Temperature (t), Dissolved Oxygen (DO), Specific Conductance (SC), Total Dissolved Nitrogen (TDN), Soluble Reactive Phosphorus (SRP), and Dissolved Organic Carbon (DOC) in the Seneca Creek After Effluent Discharges from Seneca WWTPs ...... 45 2-14 Estimated Contributions of Water, Total Dissolved Nitrogen (TDN), Soluble Reactive Phosphorus (SRP), and Dissolved Organic Carbon (DOC) from the Seneca WWTP or the Upstream Seneca Creek ...... 46 2-15 Downstream Changes in Water Temperature (t), Dissolved Oxygen (DO), Specific Conductance (SC), Total Dissolved Nitrogen (TDN), Soluble Reactive Phosphorus (SRP) and Dissolved Organic Carbon (DOC) in the Potomac River After Effluent Discharges from the Blue Plains WWTPs ...... 47 2-16 Estimated Contributions of Water, Total Dissolved Nitrogen (TDN), Soluble Reactive Phosphorus (SRP), and Dissolved Organic Carbon (DOC) from the Blue Plains WWTP or Upstream River in the Potomac River ...... 48 2-17 Downstream Changes in Estrone (E1), Estrone Sulfate (E1-S), and Other EDCs (Common Pesticides) in the Seneca Creek After Effluent Discharges from the Seneca WWTP ...... 50 2-18 Estimated Contributions of Estrone (E1), Estrone Sulfate (E1-S), and Other EDCs (Common Pesticides) from the Seneca WWTP or Upstream Stream in the Seneca Creek ...... 51 2-19 Downstream Changes in Estrone (E1), Estrone Sulfate (E1-S), and Other EDCs (Common Pesticides) in the Potomac River After Effluent Discharges from the Blue Plains WWTP ...... 52 2-20 Estimated Contributions of Estrone (E1), Estrone Sulfate (E1-S), and Other EDCs (Common Pesticides) from the Blue Plains WWTP or the Upstream River in the Potomac River ...... 53 3-1 Virginia Sampling Locations ...... 58 3-2 Griffith (ST01), Bull Run Downstream (ST45), Cub Run Downstream UMD SRP vs. All OWML SRP .... 60 3-3 SOCs Arranged by Detection Frequency and by Percent Detection ...... 61 3-4 Concentration of Frequently (> 50%) Detected SOCs ...... 62 3-5 Monthly 4-Nonylphenol Concentrations at Virginia Sampling Locations ...... 63

x The Water Research Foundation 3-6 Monthly Atrazine Concentrations at Virginia Sampling Locations ...... 64 3-7 Estrone (E1) and Estrone-3-sulfate (E1-3S) Concentrations Present at Virginia Sampling Locations ... 65 3-8 Map of Virginia Agricultural Sites ...... 66 3-9 Comparison of Water Quality and Nutrients As Paired Agriculture Sites Without BMP Implementation (Elk Run) and with BMPs (Furrs Run) ...... 67 3-10 Estrogen Concentrations at Elk Run (No BMPs) and Furrs Run (BMPs)...... 68 3-11 SOC Concentrations at Elk Run (No BMPs) and Furrs Run (BMPs) ...... 69 3-12 Comparing Nonylphenol concentration with Nutrients and Conventional Water Quality at Agriculture Sites ...... 70 3-13 Map of Virginia Stormwater Reuse Sites ...... 71 3-14 Water Quality, Nutrients, and DOC at the Upstream Cub Run (no BMPs) and Downstream Cub Run (BMPs) Urban Sites ...... 72 3-15 Cub Run Downstream Estrogens ...... 73 3-16 Percentages of SOC Detection at the Urban Sites for Individual Compound (upper) and for Monthly Variability (lower) ...... 74 3-17 Selected SOCs in Cub Run at Upstream and Downstream BMP Sites ...... 75 3-18 Comparing 4-Nonylphenol with Nutrients, DOC, and Conductivity at Stormwater Reuse Sites ... 76 3-19 Map of UOSA and Bull Run Sampling Sites ...... 77 3-20 Impact of UOSA Effluent On Water Quality, Nutrients, and DOC, in the Bull Run Receiving Stream .. 78 3-21 Estrone (E1) Concentrations at Planned Potable Reuse Sites ...... 79 3-22 Concentrations of Select SOCs at UOSA WWTP Sites ...... 80 3-23 UOSA Load Contribution to Bull Run Downstream ...... 82 3-24 UOSA Annual Load Contribution of SOCs to Reservoir Outflow, Compared with Unknown “Other” Sources Within the Watershed ...... 83 3-25 DOC, TDN, and SRP Concentrations at the Water Treatment Plant Intakes ...... 84 3-26 Estrogens at the Water Treatment Plant Intakes ...... 85 3-27 SOC Concentrations at Griffith and Corbalis Water Treatment Plant Intakes ...... 86 4-1 Relative Load Contributions of Conventional and Emerging Pollutants into the Potomac Watershed ...... 94 4-2 Results of the AHP Weighting Exercise for each Criterion and Summary by Category ...... 104 4-3 Screen Capture Describing Criteria Development and Weighting Input into the HazenConverge Tool ...... 115 4-4 Screen Capture Describing Raw Score Ranges and Scores Input into the HazenConverge Tool ...... 116 4-5 Results Summary Panel, with a Breakdown of Weighted Criteria Scores ...... 117 4-6 Alternative Scoring Arranged According to Unweighted Criteria ...... 118 4-7 Alternative Scoring Arranged According to Weighted Category ...... 119 4-8 Alternative Scoring Arranged According to Unweighted Category ...... 119

Improving Water Reuse for a Healthier Potomac Watershed xi Acronyms and Abbreviations

AWT Advanced water (reuse) treatment BMP Best management practice CECs Constituents of emerging concern DO Dissolved oxygen DOC Dissolved organic carbon BLYES Bioluminescent yeast estrogen screen EDC Endocrine-disrupting chemicals EEDC Estrogenic endocrine-disrupting chemicals GC-MS Gas chromatography-mass spectrometry LC-MS Liquid chromatography-mass spectrometry MCL Maximum contaminant level MCDA Multi-criteria decision analysis NO3-N Nitrate nitrogen POTW Publicly Owned Treatment Works SOC Synthetic organic compound SPE Solid phase extracted SRP Soluble reactive phosphorus TDN Total dissolved nitrogen U.S. EPA United States Environmental Protection Agency UMD University of Maryland College Park UOSA Upper Occoquan Service Authority VT Virginia Polytechnic Institute and State University WRF Water reclamation facility WTP Water treatment plant WWTP Wastewater treatment plant 15 - δ N (NO3 ) Nitrate nitrogen stable isotopic composition 18 - δ O (NO3 ) Nitrate oxygen stable isotopic composition

xii The Water Research Foundation Executive Summary

The presence of endocrine-disrupting compounds (EDCs) is of increasing concern in waterways in the United States and worldwide. The occurrence of intersex fish in the Potomac River and the Chesapeake Bay is believed to be a result of estrogenic EDC (EEDC) pollution in the Potomac watershed. The first year of this study aims to identify and track spatial and temporal variability in "hot spots" of contaminant sources (EDCs and nutrients) at the large watershed scale of the Potomac River. For this purpose, thirty-one primary tributaries of the Potomac River were sampled during 2016-2017 across seasonal flow regimes. Estrogenic activity (BLYES), estrogenic compounds (including 20 chemicals), and 15 other trace organic compounds (mainly pesticides and herbicides) were examined to gain insights into their potential sources and relative importance to the estrogenic water quality in the Potomac River. Hotspots of nutrients (nitrogen and phosphorus) and organic carbon were also examined to test if sources of nutrients or organic carbon are related to EDCs or the estrogenic character of the same water. The aims of the second year at the Maryland sites were to use paired watershed studies to evaluate impacts and outcomes of current reclamation, reuse, harvesting, and management strategies on source controls of pollutants. Monthly measurements of nutrients and endocrine disrupting chemicals (EDCs; mainly estrogen and pesticides) were taken at one pair of agricultural sites and another pair of urban sites (with or without best management practices) to examine effectiveness of retentions nutrients and EDCs. Effluent samples from two wastewater treatment plants (WWTPS) were also collected at Seneca and Blue Plains, and streams/rivers above and below the WWTPs to examine the effect of WWTPs on stream/river water quality. Results of the measurements were used to estimate inputs of nutrients and EDCs from point sources, and changes in non-point sources if agricultural and urban BMPs were applied to the entire watershed. It is found that hotspots of BLYES were primarily in sub-watersheds dominated by pasture/hay type land uses. The percent pasture/hay land use was positively correlated with BLYES. The “hot moment” of BLYES was during the spring high-flow period. The BLYES was not significantly correlated with estrogenic chemical compounds (exclusively estrone and estrone-3S). Herbicides were dominated by atrazine and metolachlor and were far higher in three subwatersheds than the others. At most sites, concentrations of atrazine and metolachlor were positively correlated with BLYES, and their hot moment was also during the spring high-flow period. Total dissolved nitrogen (TDN) hotspots occurred in sub-watersheds with larger percentages of cropland land use and TDN concentration was highly correlated with watersheds dominated by cropland. Nitrate isotopic tracers also suggested that soil nitrogen (N) (derived from fertilizers) was the primary N source within the Potomac River. The hot moment for elevated N concentration was during the summer low-flow period. Concentrations of soluble reactive phosphorus (SRP) increased with increasing degree of agricultural land use, and it was also high in watersheds containing a large percentage of wetlands. TDN was correlated with three common herbicides, while SRP was positively correlated with BLYES. Hotspots of dissolved organic carbon (DOC) were located in urban areas and DOC concentration was not correlated with BLYES or herbicides. This study identified EDC and BLYES hotspots and hot moments in the Potomac watershed, and provided insight into how herbicide use, including atrazine and metolachlor, are associated with BLYES. Our results further suggest that herbicides and TDN, and BLYES and SRP may have common sources. Thus, BLYES and CECs may be co-managed by applying best management practices for nutrients. The streams with BMPs were consistently lower in total dissolved nitrogen (TDN; only for agricultural BMP) and soluble reactive phosphors (SRP; for both agricultural and urban BMPs). Although EDCs concentrations in the streams with BMPs were not consistently lower than streams without BMPs, estimated export rates of estrogen (estrone or E1 and estrone sulfate or E1-S), and three common pesticides (atrazine, metolachlor and simazine) in the streams with BMPs were lower, indicating that

Improving Water Reuse for a Healthier Potomac Watershed xiii both BMPs for nutrient controls can decrease EDCs fluxes from watersheds. In general, discharges of WWTP effluent caused downstream decreases in dissolved oxygen, but increases in water temperature, specific conductance, SRP, DOC, and sometimes TDN (only at Blue Plains WWTP) in the receiving waters due to higher concentrations in the WWTP effluent than the stream/river upstream. WWTP discharges actually are improving the water quality of the River with respect to EDCs and pesticides. Point sources (mainly WWTPs) contributed 5-9% of nutrients and <2% of EDCs to the Potomac River, with the contributions of TDN and most EDCs lower than their contribution of water flow. Applying BMPs to agricultural runoff can significantly reduce the annual loads of nutrients, EDCs and pesticides/herbicides delivered to the River. The results of these analyses were used to develop a multi-criteria decision analysis (MCDA), to evaluate four alternatives for co-managing nutrients and contaminants of emerging concern. The four alternates considered included: • Co-managing nutrients and CECs through control of non-point source pollution with implementation of Agriculture Best Management Practices (BMPs). • Co-managing nutrients and CECs through control of non-point source pollution with implementation of Urban Stormwater Best Management Practices (BMPs). • Co-managing nutrients and CECs through control of point source pollution with implementation of Advanced Nutrient Control at publicly owned treatment works (POTWs). • Co-managing nutrients and CECs through control of point source pollution with implementation of Advanced Water Treatment technology for planned potable reuse at publicly owned treatment works (POTWs).

In order to define and rank the importance of criteria for the analysis, a stakeholder workshop was held which resulted in identification and prioritization of 10 specific criteria for analysis, grouped within five categories including cost, performance, equity, implementability, and risk. Scores were developed for each category, with quantitative scores developed for four criteria (cost/reduction N, CEC removal Efficiency, Capital Cost, and Potential Impact), and qualitative scores developed for six criteria (Affordability, Complexity of Implementation, Implementation Timeline to Benefits, Complexity of Policy/Regulation, Confidence of Performance, and Geographic Distribution of Benefits). When weighted scores were developed for each alternative, co-management of nutrients and CECs through implementation of Agriculture BMPs was the highest ranked options, collecting 71 total points (of 100). The next highest score was achieved by Urban BMPs (53 points), followed by point source management at Water Reclamation Facilities (44 points), and point source management through Advanced Water Treatment of Reclaimed Water (32 points). Most recently, members of our team have developed a method using suspect screening analysis (SSA) that relies on accurate mass measurements and mass spectral fragmentation patterns obtained using liquid chromatography coupled to high-resolution mass spectrometry to achieve a wide-scope detection of micropollutants. A total of 103 micropollutants were detected in the samples, many of which were not previously monitored in the original target analysis method; 23 of these additional compounds have detection frequency of at least 50%. These compounds belong to various classes such as pharmaceuticals and personal care products (PPCPs), per- and polyfluoroalkyl substances (PFASs), organophosphate flame retardants (OPFRs), and various pesticide and their metabolites. In the future, further work is necessary to investigate these complex mixtures of micropollutants and the impacts of management on their concentrations and loads. Overall, our work demonstrates the potential for co- managing multiple pollutants in surface waters through an improved understanding of shared sources, transport, and transformation of multiple contaminants in the environment.

xiv The Water Research Foundation CHAPTER 1

Hotspots of EDCs, Nutrients and Organic Carbon in the Potomac Watershed

1.1 Potomac Watershed 1.1.1 Watershed General Information The drainage area of the Potomac River includes 14,670 square miles in four states: Virginia, Maryland, West Virginia, Pennsylvania, and the District of Columbia (Figure 1-1). The Potomac River flows over 383 miles from Fairfax Stone, West Virginia to Point Lookout, Maryland. The major tributaries of the Potomac River include North Branch Potomac River, South Branch Potomac River, Cacapon River, Back Creek, Conococheague Creek, Antietam Creek, Shenandoah River, Monocacy River, Goose Creek, Seneca Creek, Rock Creek, Anacostia River, Occoquan River, Piscataway Creek, and Mattawoman Creek. The river is free-flowing above the US Geology Survey gauging station – Little Falls in Washington, D.C. It becomes a tidal river after the Fall Line.

Figure 1-1. Map of the Potomac River Watershed Showing Location and Primary Tributaries. Adapted from American Rivers https://www.americanrivers.org/river/potomac-river/.

Improving Water Reuse for a Healthier Potomac Watershed 1 1.1.2 Geological Setting The Potomac River basin lies in five geological provinces: the Appalachian Plateau, the Ridge and Valley, Blue Ridge, Piedmont Plateau, and Coastal Plain (Figure 1-2). For rock types, unconsolidated deposits (brown in map) occur mainly in the Coastal Plain and in isolated areas of alluvial deposits along stream channels in the Valley and Ridge (Figure 1-2). Crystalline rocks (pink) underlie only the Piedmont and Blue Ridge in the central and eastern parts of the basin. The entire basin west of the Blue Ridge is underlain by sedimentary rocks; these rocks are mainly siliciclastic (in green), with areas of extensive carbonate rocks (in blue), particularly in the Great Valley.

Figure 1-2. Generalized Geology and Physiography in the Potomac River Basin. (Adapted from U.S. Geological Survey Circular 1166) 1.1.3 Land Use Based on information from the 2011 National Land Cover Database (NLCD), the majority of the basin’s land area is covered by forests at 54.6% of the land area. Urban (developed) land makes up 14.1% of the basin’s land area, while agriculture covers 26.0%. Water and wetlands make up 5.9% of the basin’s land area.

Land use of the Potomac River Watershed is depicted by a spatial pattern in which forest, agricultural and urban land use is distributed successively from upriver to downriver like three belts (Figure 1-3). While these belts represent dominant land-use, there is still mixed land-use. Forest is mainly distributed in the Appalachian Plateaus Province, Appalachian Ridge, and Blue Ridge, while agricultural lands lie in Great Valley and part of Piedmont Province. Within this agricultural belt, Maryland has more cropland, while Virginia and West Virginia have relatively more pasture/hay (Figure 1-3). Urban lands are mainly located within the Washington, D.C. Metropolitan Area, although small amount of urban land is also distributed in forest or agricultural zones.

2 The Water Research Foundation

Figure 1-3. Map of Land Use in the Potomac River Basin. Data are from the NLCD 2011 database. 1.1.4 Population, Cities, and Industries The population of the basin is approximately 6.11 million (2010 estimated census) with people living in urban areas account for almost 81% of the population. Those living in rural areas make up 18.6% of the population, while those living on farms account for 0.7%. The Washington, D.C. metropolitan area has approximately 5.1 million residents, or 84%, of the basin’s population. Major industry with in the basin includes: agriculture and forestry distributed throughout the basin, coal mining and pulp and paper production along the North Branch Potomac River; chemical production and agriculture in the Shenandoah Valley; high-tech, service, and light industry, as well as military and government installations in the Washington, D.C. metropolitan area; and fishing in the lower Potomac estuary.

Improving Water Reuse for a Healthier Potomac Watershed 3 1.2 Sampling Design 1.2.1 Site Selection Water samples were collected from 19 main stem sites along the Potomac River and 31 primary tributaries (Figure 1-4). The sub-watersheds of these tributaries vary substantially in land use, which are listed in Table 1-1.

Figure 1-4. Sampling Sites on The Map of Land Use in the Potomac River Basin.

4 The Water Research Foundation Table 1-1. Land Uses of 32 Selected Sub-Watersheds of the Potomac River. Urban Cropland Pasture Forest Wetland Other Label Name Land Use (%) (%) (%) (%) (%) (%) 0 N Branch Potomac forest 6.1 0.2 9.7 79.0 1.6 3.3 1 Wills Creek forest 5.8 0.6 10.9 82.2 0.1 0.4 2 Patterson Creek forest 5.3 0.4 17.4 76.1 0.3 0.5 3 S Branch Potomac forest 3.9 0.4 14.6 80.3 0.5 0.2 4 Town Creek forest 3.9 0.9 11.0 83.7 0.1 0.4 5 Silderling Hill Creek forest 6.5 1.8 15.3 76.2 0.1 0.1 6 Cacapon River forest 3.6 0.1 11.4 84.2 0.5 0.2 7 Tonoloway Creek mix 7.9 4.1 23.7 64.1 0.1 0.2 8 Sleepy Creek forest 5.8 0.6 13.1 79.7 0.5 0.2 9 Licking Creek mix 6.1 6.2 16.8 70.4 0.4 0.1 10 Back Creek forest 6.8 0.9 17.1 74.2 0.6 0.4 11 Conococheague Creek cropland 14.5 24.5 24.0 35.4 1.5 0.1 12 Opequon Creek Pasture 20.7 2.5 46.8 29.0 0.7 0.3 13 Antietam Creek cropland 17.2 23.6 27.7 30.4 1.0 0.1 14 NF Shenandoah Pasture 6.4 1.9 28.7 62.4 0.4 0.2 15 SF Shenandoah Pasture 11.0 2.9 29.8 55.6 0.6 0.1 16 Shenandoah River Pasture 12.5 3.2 42.7 39.0 2.3 0.2 17 Catoctin Creek, MD Pasture 11.4 8.4 41.2 38.0 0.9 0.0 18 Catoctin Creek, VA Pasture 10.1 2.3 55.8 30.7 1.1 0.0 19 Monocacy River cropland 14.5 30.6 21.5 29.6 2.2 1.6 20 Goose Creek Pasture 11.5 4.1 45.4 37.6 0.9 0.6 21 Broad Run Suburban 54.7 10.7 6.0 19.5 7.1 1.9 22 Seneca Creek Suburban 33.2 15.8 17.0 27.9 4.6 1.5 23 Difficult Run Suburban 53.2 0.8 1.0 39.3 4.7 0.9 24 Cabin John Creek Suburban 72.7 0.0 0.1 26.0 0.8 0.4 25 Rock Creek Urban 71.4 1.0 4.6 20.1 1.9 1.0 26 Anacostia River Urban 72.2 1.5 4.0 18.1 3.4 0.7 27 Cameron Creek Urban 82.7 0.1 0.2 14.8 1.8 0.3 28 Piscataway Creek wetland 34.7 5.6 3.1 43.9 10.1 2.6 29 Accotink Creek Suburban 68.7 0.3 0.1 22.9 7.6 0.4 30 Occoquan River mix 27.5 10.8 14.4 38.4 6.6 2.2 31 Mattawoman Creek wetland 25.6 5.6 2.2 46.2 16.5 3.9 Headwater the Potomac River (North Branch of Potomac River) is considered as one sub-watershed. The sub- watersheds were classified to seven land use types (forest, mix, cropland, pasture, suburban, urban and wetland) according their land use composition. 1.2.2 Sampling Frequency, Sample Collection, and Preparation Grab water samples were collected quarterly for water quality analysis on August 22-26, 2016, November 30-December 2, 2016, March 1-3, 2017, and May 19-21, 2017, representing seasons of summer, fall, winter, and spring respectively. Flow conditions of the Potomac River change with the seasons with base flow occurring during summer, high flow in spring, and medium flow during the fall and winter months (Figure 1-5). Because the river is large, soluble and suspended particles may not be uniformly distributed. Sample site selections were based on criteria that minimized hydrodynamic bias.

Improving Water Reuse for a Healthier Potomac Watershed 5 Surface water collection sites were selected to favor well-mixed stream segments and samples were collected on convex banks. Back water conditions were avoided at the sites. This strategic site selection allowed for nonisokinetic sampling methods (dip sampling or weighted-bottle). Water samples were collected according to the Standard Methods for the Examination of Water and Wastewater wherever applicable. All samples collected in the laboratory and field were labeled clearly and legibly with date and time of sample collection, sample collection location, sample number and replicate number. Samples were transported in coolers on ice, and upon sample reception at the laboratory the integrity of the sample containers were assessed and the samples was either stored at 4ºC in darkness or processed immediately at the University of Maryland (UMD) sample clearinghouse. Trip blanks accompanied field sampling campaigns, and appropriate duplicate samples were collected for quality control. All samples collected were accompanied from the point of collection to the point of analysis with appropriate Chain of Custody paperwork. Sample processing and extraction was conducted at the UMD sample clearinghouse. The solid phase extraction (SPE) was performed at this single location as a quality control measure to ensure all samples were extracted under identical conditions such that analytical chemistry and bioassay analyses were conducted on as near identical samples as possible. This approach minimized costs while maximizing processing efficiency. It also offered an opportunity to ship samples on SPE columns to circumvent sample loss associated with shipping mishaps. In short, water samples (1 L) were filtered and solid phase extracted (SPE) at the UMD using Oasis HLBTM SPE cartridges as described previously (Batt and Aga 2005). This step is common for samples destined for analytical testing at the University at Buffalo (UB), or biological activity testing at the United States Geological Survey (USGS), Leetown Science Center. For analytical testing, a mixture of deuterium or 13C-labeled analogues of each target pharmaceutical was spiked into filtered samples as surrogate standards to facilitate quantification by isotope dilution mass spectrometry (Vanderford and Snyder 2006). This mixture of isotope-labeled surrogates was prepared by the Aga laboratory at UB and was shipped to UMD. To ensure quality control, standard operating procedures were developed for each preparation procedure, including proper cleaning methods and identification of proper storage and processing techniques. Spiked samples were not added to columns used for bioassays given the nature of the analysis.

Sampling dates with Potomac River streamflow 3500 3000 Spring 2017 ) 1 -

s 2500 high flow 3 2000 1500 winter 2017 1000 summer 2016 fall 2016 median flow Flow (m 500 median flow 0

Figure 1-5. Flow Conditions of the Potomac River When River Water Samples Were Collected. River flow data were collected from USGS Little Falls gauge station of Washington, D.C.

6 The Water Research Foundation 1.3 Chemical Analyses Water temperature, pH, and conductivity were measured in situ, using a Waterproof Multiparameter Testr™ 35-Series Pocket Meter. The precision of temperature and pH measurements were ± 0.1 and ±0.5°C, respectively. Concentrations of DOC and total dissolved nitrogen (TDN) were measured on a Shimadzu Total Organic Carbon Analyzer (TOC-L CPH/CPN) (Álvarez-Salgado and Miller 1998). Three injections (with a maximum of five times) were run for each sample to obtain a standard derivation less - - than 0.2. Nitrate (NO3 ), nitrite (NO2 ) and soluble reactive phosphorus (SRP) was automatically measured on a QuikChem 8500 Series 2 FIA System (Gordon et al. 1993). Blank and standards were run every 15 samples to ensure accuracy of the analyses. Isotopic tracers of nitrate were analyzed for N source tracking. Samples were processed and prepared at 15 - 18 - the University of Maryland prior to δ N (NO3 ) and δ O (NO3 ) analysis, using the denitrifier method. Denitrifying bacteria (Pseudomonas auroeofaciens) convert nitrate to gaseous nitrous oxide for isotopic analysis (Casciotti et al. 2002). Gas samples are then analyzed with an isotope ratio mass spectrometer 15 - (IRMS). Sample duplicates typically have an average standard deviation of 0.2ppm for δ N (NO3 ) and 18 - 0.7ppm for δ O (NO3 ). For QA/QC, samples are prepared and analyzed in batches and each batch contains approximately 12 vials of three reference materials. Standard deviation should be ≤0.25‰ for δ15N, and ≤ 0.5 ‰ for δ18O, use mean delta. Pharmaceuticals and metabolites were measured using liquid chromatography with tandem mass spectrometry (LC-MS/MS) (AgilentTM 6410 series MSD; Batt and Aga 2005). Typical detection limits in the low pictogram per μL are achievable with these methods (typically 1 to 10 ng/L) under multiple reaction monitoring (Tso and Aga 2010). The analysis was achieved using an optimized solid-phase extraction (SPE), followed by an LC-MS/MS method that does not require enzymatic cleavage of the conjugated metabolites, nor derivatization of the free estrogens (Tso et al. 2011). Our optimized LC-MS/MS methods are capable of analyzing pharmaceuticals with typical detection limits between 1 and 10 ng/L, and provide simultaneous analysis of conjugated metabolites and free estrogens at the sub nanogram per liter concentrations. These methods have been demonstrated to be effective in the analysis of these compounds in run-off water and soil from fields where poultry liter has been applied as fertilizer (Dutta et al. 2012). Pesticides were analyzed with GC-MS/MS based on proven methods. GC-MS/MS analysis was performed on a Thermo Trace GC Ultra system coupled to a TSQ Quantum™ XLS mass spectrometer (Thermo Fisher Scientific, San Jose, CA). Chromatographic separation was performed using a DB-5HT capillary GC column (0.25 mm i.d. × 15 m length, 0.1 µm film thickness) from Agilent Technologies (Santa Clara, CA). Helium was used as the carrier gas at a 1 mL min-1 constant flow rate. The column temperature program was as follows: the initial temperature of 60⁰C was held for 1 min, and then temperature was ramped at 10⁰C min-1 to 130⁰C, then at 3⁰C min-1 to 180⁰C, and finally at 60⁰C min-1 to 330⁰C, where it was held for 2 min. The total run time was 29 min. The source temperature was kept at 200⁰C. The transfer line temperature was 350⁰C. EI mode was performed with 70 eV electron energy. The injection volume was 1 mL. The mass spectrometer was operated under unit resolution in the time-scheduled SRM mode. In scheduled SRM, the acquisition occurs over a defined period around the retention time of every compound, with a time window and a defined cycle time. The cycle time was set to 0.9 s, with a time window of 2 min before and after the retention time of each analyte peak. In GC-MS/MS analysis, the molecular ion of the target analyte is typically weak because of the high energy applied during ionization causing significant fragmentation of the molecule. Therefore, the most abundant fragment ion (base peak) was chosen as the precursor ion for SRM. Each individual selected precursor ion/qualifier ion was optimized for high sensitivity and selectivity, of which the more abundant ion was used for quantification and the other as a qualifier for confirmation. To maintain the performance of the LC-MS/MS and GC-MS/MS instrument calibration are done on a

Improving Water Reuse for a Healthier Potomac Watershed 7 monthly basis using specific manufacturer-prescribed tuning mix. These mixes calibrate and tune the mass spectrometer from the low mass to a higher mass range. A manufacturer provided method is used to align, correct, and optimize instrument parameters for optimum performance. A report is generated and results are compared to existing ones to determine whether the instrument is performing consistently compared to previous months of usage. Manual adjustments are made in cases when certain parameters are out of specification. Tune files are updated for each individual method after calibration. If the level of an analyte exceeds that of the highest calibration curve level, the sample is diluted and rerun. Some samples are run twice, both diluted and undiluted to ensure that levels of all analytes fall within the quantification ranges. If any irregularities are observed including signal suppression greater than 30% or pressure fluctuations exceeding 1%, the instrument undergoes maintenance either by lab members, or by Agilent service technicians and use is discontinued until the abundancies and pressure profile are back within the acceptable ranges and a tune has been passed in both positive and negative mode. All tuning files are stored on the computer, and the report is printed and placed in a binder. Logbooks are kept recording when tuning and maintenance has been performed and what the outcome was. Estrogenic activity (BLYES) of OASIS HLB extracts was determined using the bioluminescent yeast estrogen screen (Sanseverino et al. 2005) as described for environmental water samples by Ciparis et al. (2012). The reporting limit for this assay of BLYES is approximately 0.31 ng/L. For continuity BLYES was conducted on extracts from SPE using the optimized protocols utilized for analytical chemistry. The bioluminescence-based bioassay is conducted on solid bottom white places. Every plate contains a full dynamic range standard curve of 17β-estradiol and a solvent control blank. In this manner every analytical plate read is calibrated to a replicated, 12-point standard curve including a solvent blank. Standards are run in duplicate on every plate and all sample are run in triplicate. If the coefficient of variation (%CV) of sample reads exceed 20% data are rejected and samples are re-run. 1.4 Hotspots of EDCs 1.4.1 Hotspots of Estrogenic Activity Estrogenic activity measured using a bioluminescent yeast estrogen screen (BLYES) in a net sum of all estrogen analytes in a given sample. Unlike analytical chemistry measures, to use the assay it is not necessary to know target compounds of interest in order to assess the composite estrogenic activity potential. Hotspots of BLYES in the Potomac River watershed were identified Catoctin Creek, MD, Shenandoah River and Piscataway Creek, with mean values > 1.5 ng L-1 (Figure 1-6). The sub-watershed of the first two have large percent of pasture/hay land use. The mean concentrations in sub-watersheds of Catoctin Creek VA, Monocacy River, and NF-Shenandoah were all > 1 ng L-1. This finding on BLYES hotspots is also consistent with previous report that intersex fish were found in the Shenandoah River, NF-Shenandoah, and the Monocacy River (Chesapeake Bay News August 09, 2012; https://www.chesapeakebay.net/news/blog/intersex_fish_widespread_in_potomac_river_basin). Additionally, forest percentage presented opposite of the others (Figure 1-6) because greater forest coverage decreases estrogenic activity potential. Seasonal data showed hot moments of the three hotspots (Catoctin Creek, MD, Shenandoah River, and Piscataway Creek) to be during spring high flows (Figure 1-7). This observation suggests that BLYES may largely be exported from the Potomac River watershed during spring high-flow periods due to the highest BLYES concentrations and the largest river discharges that occur at that time of the year. These exports may occur as a result of intense erosion during spring high-flow periods that carry soil downstream along with absorbed organic compounds that express estrogenicity. Summer was also a hot moment for some tributaries (Shenandoah River, SF_ Shenandoah and Monocacy River; Figure 1-7).

8 The Water Research Foundation 3.5 Urban Forest Agricultural 3

) 2.5

1 Ranch - Cropland

g L 2

n Wetland >10% Forest 1.5 Urban/suburban

BLYES ( BLYES 1

0.5

0 Broad Run Back Creek Back Rock CreekRock Wills Creek Wills Town Creek Town Difficult Run Goose Creek Goose Sleepy Creek Sleepy Silderling Hill… Silderling Licking Creek Licking Seneca Creek Mattawoman Cacapon River Accotink Creek SF Shenandoah Anacostia River Anacostia Cameron Creek Cameron Antietam Creek Opequon CreekOpequon Occoquan River Occoquan NF Shenandoah Conococheague Monocacy River Patterson Creek Patterson Cabin John Creek John Cabin Tonoloway Creek Piscataway Creek Piscataway Catoctin Creek VA S S PotomacBranch Shenandoah River N Branch Potomac Catoctin Creek MD Figure 1-6. Hotspots of Estrogenic Activity in the Potomac River Watershed During the Water Year 2016-2017. The blue arrows indicate increasing percentages of forest, agricultural and urban land covers.

4.5 4.5 Catoctin Creek MD 4 4 Catoctin Creek VA Shena0oah River 3.5 3.5 Monocacy River

) Piscataway Creek 1 - 3 3 SF Shena0oah River

g L 2.5 2.5 n ( 2 2 1.5 1.5 BLYES 1 1 0.5 0.5 0 0 summer fall winter spring summer fall winter spring

Figure 1-7. Hot Moments of Estrogenic Activity in the Hot Spots of the Potomac River Watershed During the Water Year 2016-2017.

Improving Water Reuse for a Healthier Potomac Watershed 9 BLYES was correlated with land uses of sub-watersheds, and found that BLYES increased with increasing watershed pasture/hay land use (Figure 1-8). This relationship suggests that land use associated with animal industries (e.g., cattle) is one of the better predictors of BLYES in the Potomac River. Notably, intersex in smallmouth bass and estrogenic activity have previously been associated with agricultural land use and animal feeding operations in the watershed (Blazer et al. 2012).

3.5 Catoctin Creek MD 3

2.5 ) y = 0.0122x + 0.733 1 - R² = 0.2592 2 Piscataway Creek Shenandoah River 1.5

BLYES (ng L 1

0.5

0 0 10 20 30 40 50 60 Ranch (%)

Figure 1-8. Correlations Between Estrogenic Activity (or BLYES) and Percent of Pasture/Hay Land Use in the Primary Tributaries of the Potomac River Watershed. 1.4.2 Hotspots of Estrogen Estrone and estrone-3S were the only estrogenic chemicals detectable in the primary tributaries of the Potomac River watershed using SPE and LC-MS/MS. Actual measured estrone and estrone-3 concentrations ranged from below detection to 1.6 ng L-1. Relative to BLYES, the distribution of estrone and estrone-3s was more evenly distributed (Figure 1-9). The sum of these two estrogenic compounds was ubiquitous throughout the Potomac watershed. Somewhat elevated concentrations of these two estrogenic compounds were found in agricultural sub-watersheds (Catoctin Creek, MD), urban sub- watersheds (Broad Run and Anacostia River) as well as a forested sub-watershed (Tonoloway Creek) (Figure 1-9). It was also observed that estrone plus estrone-3s concentrations were always lower in agricultural sub-watersheds dominated by cropland (in purple) compared to those agricultural sub- watersheds dominated by pasture/hay (in brown). This result may suggest that the presence of animal agriculture contributes more to the occurrence of estrone and estrone-3S in runoff from agricultural land uses. It has previously been reported that estrogenicity is associated with livestock-based agriculture (Ciparis et al. 2012). Measured estrone and estrone-3S concentrations varied considerably over time throughout the sub- watersheds. No consistent seasonal pattern could be identified that might suggest hot moments for these two compounds (data not shown).

10 The Water Research Foundation 1.4 Urban Forest Agricultural 1.2 Cropland )

1 1 - Wetland >10% Urban/suburban g L Forest Ranch

n 0.8

0.6

0.4 Estrogen ( Estrogen 0.2

0 Broad RunBroad Back Creek Rock Creek Wills Creek Town Creek Difficult Run Goose Creek Sleepy Creek Sleepy Licking Creek Licking Seneca Creek Mattawoman Cacapon River Accotink Creek Accotink SF ShenandoahSF Anacostia River Cameron Creek Cameron Antietam Creek Opequon Creek Occoquan River NF Shenandoah Conococheague Monocacy River Patterson Creek Patterson Cabin Creek John Tonoloway Creek Piscataway Creek Catoctin Creek VA S Branch Potomac Branch S Shenandoah River N Branch Potomac Catoctin Creek MD Silderling Creek Hill Figure 1-9. Hotspots of Total Estrogen (Estrone and Estrone-3S) in the Potomac River Watershed During the Water Year 2016-2017. The blue arrows indicate increasing percentages of forest, agricultural and urban land covers.

Neither estrone, estrone-3S, nor their sum (total estrogen) were correlated with any single type of land use at a significant level. However, total estrogen concentrations increased with increasing percent of developed land uses (agricultural land plus urban; Figure 1-10), suggesting the importance of both agricultural and urban sources of these two compounds.

1.4

1.2 y = 0.0043x + 0.2261 1 R² = 0.2228

0.8

0.6

Estrogen (mg/L) 0.4

0.2

0 0 10 20 30 40 50 60 70 80 90 Urban + agricultural (%)

Figure 1-10. Correlations Between Mean Total Estrogen and the Percent of Developed Land Uses. (agricultural% + urban%)

Improving Water Reuse for a Healthier Potomac Watershed 11 1.4.3 Hotspots of Herbicides Herbicides that were widely detected in the Potomac watershed included atrazine, metolachlor, prometon, and simazine. They are all selective herbicide for pre-emergence and preplant weed control in corn, soybeans, peanuts, sorghum, pod crops, potatoes, cotton, safflower, and woody ornamentals. They can be sprayed on croplands before crops start growing and after they have emerged from the soil. They are usually used in the spring and summer months. The application of these compounds to crops as herbicides accounts for almost all of the herbicides that enters the environment, but some may be released from manufacture, formulation, transport, and disposal. The former two (atrazine and metolachlor) are still widely used in the United States, while the latter two are banned. In general, the concentrations of total herbicides (up to 300 ng L-1) were several orders of magnitude higher than that of estrone and estrone-3 (Figure 1-11). There were three hotspots of total herbicides – Monocacy River (cropland), Catoctin Creek (pasture), and Occoquan River (mixed 27% urban, 25% agriculture, 38% forest), with mean values > 250 ng L-1, much higher than that of other sub-watersheds (< 100 ng L-1; Figure 1-11). Total herbicides concentrations were significantly higher in agricultural than forest dominated lands (one-way ANOVA, p <0.05).

600 Urban Forest Agricultural 500 ) 1 - 400

g L Cropland n ( 300 Forest Urban/suburban 200 Ranch

100 Wetland >10% Herbicides

0 Broad Run Back Creek Back Rock CreekRock Wills Creek Wills Town Creek Town Difficult Run Goose Creek Goose Sleepy Creek Sleepy Licking Creek Licking Seneca Creek Mattawoman Cacapon River Accotink Creek SF Shenandoah Anacostia River Anacostia Cameron Creek Cameron Antietam Creek Opequon CreekOpequon Occoquan River Occoquan NF Shenandoah Conococheague Monocacy River Patterson Creek Patterson Cabin John Creek John Cabin Tonoloway Creek Piscataway Creek Piscataway Catoctin Creek VA S S PotomacBranch Shenandoah River N Branch Potomac Catoctin Creek MD Silderling Hill Creek Hill Silderling Figure 1-11. Hotspots of Total Herbicides in the Potomac River Watershed During the Water Year 2016-2017. The blue arrows indicate increasing percentages of forest, agricultural and urban land covers.

The highest concentration of herbicides was observed in Catoctin Creek. The most abundant herbicides were atrazine, and to a lesser extent, metolachlor. Herbicide concentrations were generally highest during spring and summer, while concentrations were relatively lower in fall and winter (Figure 1-12). The dominance of atrazine and metolachlor is not surprising given that they are commonly used herbicides associated with corn, and soybeans. They are typically applied to land in the spring, which explains the observed seasonal pattern. This herbicide composition and seasonal pattern were also observed in other agricultural rivers (e.g., the Conococheaque Creek). At other hotspot sites, the hot moment of the herbicide was different. For example, herbicides (mainly atrazine) concentrations were highest during fall in the Monocacy River. At the Occoquan River, all pesticides were far higher in summer than other seasons. This may be the result of hold-up and delayed delivery of herbicides coming from cropland high up in the watershed to the main stem of the Occoquan River which was sampled downstream of the Occoquan Reservoir. The hydraulic residence time of the reservoir may have impacted the timing of peak herbicide concentrations downstream.

12 The Water Research Foundation Conococheague Creek 1500 Catoctin Creek VA 300 Atrazine 200 Metolachlor 1000 Prometon 100 Simazine 500 ) 1 - 0 0 (ng L 800 1000 Monacay Occoquan River River 600 800 Pesticides 600 400 400 200 200

0 0 Summer Fall Winter Spring Summer Fall Winter Spring Figure 1-12. Seasonal Changes in Concentrations of Dominant Herbicide Species in the Selected Hotspot Sub-Watersheds of the Potomac River During the Water Year 2016-2017.

Total herbicide was correlated with land uses of sub-watersheds and found that concentrations of total herbicides generally increased with increasing watershed agricultural land use (cropland plus pasture) (Figure 1-13). This relationship suggests the importance of agricultural herbicide inputs from both cropland and pastureland use.

600 )

1 500 -

(ng L 400 Monocacy River Catoctin Creek MD 300 Occoquan River

200 y = 2.3826x + 1.9355 Herbicides R² = 0.3047 100

0 0 10 20 30 40 50 60 70 Agricultural (%)

Figure 1-13. Correlations Between Mean Total Herbicides and Percent of Agricultural Land Use (Cropland plus Pasture) in the Primary Tributaries of the Potomac River Watershed. 1.4.4 Linkage of Estrogenic Activity (BLYES) With EDCs Data were analyzed to identify correlations between BLYES, estrone, estrone-3s, ad total estrogen, or individual herbicides to identify which of these were best related to estrogenic activity. Results showed that there were no significant correlations between BLYES and estrone, or estrone-3S or estrone + estrone-3S. Very few of the estrogenic compounds were detected in the samples, and only estrone and estrone-3s were detectable at such concentrations close to the detection limit and there may be some uncertainty in these measured results. Figure 1-14 also shows no well-defined correlation between BLYES and measured concentrations of estrone plus estrone-3S. Therefore, estrogen contributes only partially to observed estrogenic activity.

If three outliers (Monocacy River, Catoctin Creek, MD, and Occoquan River) are excluded, there is a slight positive correlation between BLYES and total pesticide, or atrazine, or metolachlor (R2 = 0.15 –

Improving Water Reuse for a Healthier Potomac Watershed 13 0.20; n = 88, p < 0.01). This suggests that BLYES is likely associated by a compound that co-occurs with atrazine and metolachlor. Higher surface water concentrations of atrazine and metolachlor are generally associated with agricultural land use runoff. Additionally, literature reports observations of reproductive system abnormalities, hermaphraditism and even complete sex reversal resulting from significant environmental exposures of aquatic vertebrate species to atrazine (Hayes et al. 2003). However, the mechanism causing these abnormalities has not been conclusively determined and may not be an endocrine disruption that results in estrogenic expression as measured by the BLYES assay.

5 5

4 y = 0.0047x + 0.754 4 y = 0.2729x + 0.8639 3 R² = 0.1736 3 R² = 0.0133 2 2 1 1 0 0 0 200 400 600 800 1000 1200 0 0.5 1 1.5 2 pesticides (ng/L) Estrogen (ng/L)

5 5 4 4 y = 0.0086x + 0.7699 y = 0.0255x + 0.7646 3 R² = 0.1988 3 R² = 0.1439 2 2

BLYES (ng/L) 1 1 0 0 0 200 400 600 800 1000 1200 0 100 200 300 400 Atrazine (ng/L) Metolachlor (ng/L)

5 5 4 4 y = 0.0065x + 1.0845 3 3 y = 0.0453x + 0.8961 R² = 0.0445 2 R² = 0.0204 2 1 1 0 0 0 20 40 60 80 100 120 0 100 200 300 400 Prometon (ng/L) Simazine (ng/L)

Figure 1-14. Correlations Between Estrogenic Activity and Total Estrogen, Total Pesticide, or Any Single EDC. The red dots stand for three outliers – Monocacy River, Catoctin Creek, MD, and Occoquan River.

14 The Water Research Foundation 1.5 Hotspots of Nutrients and Organic Carbon 1.5.1 Hotspots of Nitrogen (N) Total dissolved nitrogen (TDN) includes dissolved inorganic nitrogen (DIN; nitrate, nitrite, and ammonium) and dissolved organic nitrogen (DON). TDN concentrations were in a range of 0.16-5.42 mg L-1. TDN concentration peaked in the agricultural zone in the middle section of the Potomac River across seasons (Figure 1-15). The hotspots of TDN were the agricultural watersheds with large areas of cropland (e.g., Conococheague Creek, Antietam Creek, and Monocacy River). Agricultural watersheds with large areas of Pasture/hay (e.g., Opequon Creek, NF Shenandoah River, and Catoctin Creek, MD) and suburban watersheds used to be croplands (e.g., Seneca Creek) had intermediate levels, and forested watersheds in the upper Potomac River had the lowest. TDN concentrations of some urban tributaries in the tidal zone near Washington DC (Cameron Creek, Piscataway Creek, Accotink Creek, and Mattawoman Creek) were also low (< 0.69 mg L-1), probably due to intensive transformation in entrenched zone or riparian wetland.

Conococheague Antietam Monocacy

Figure 1-15. Spatial Distribution of Total Dissolved Nitrogen (TDN) Showing TDN Hotspots in the Potomac River Watershed During the Water Year 2016-2017.

Improving Water Reuse for a Healthier Potomac Watershed 15 TDN concentrations were correlated with land uses of sub-watersheds, and found that TDN concentrations were all positively correlated with percent of cropland land use (r2 = 0.68, p < 0.01; Figure 1-16). The coefficients of correlation decreased when percentages of pasture/hay were included with cropland. This finding is further demonstrated by Figure 1-15 that shows the sub-watersheds having large areas of cropland as those with the highest TDN concentrations. The reason for the correlation between N hotspots and cropland is best explained by the extensive use of chemical fertilizers on croplands (Ator et al. 1998).

5 4.5 Conococheague 4 y = 0.0928x + 0.6283 Antietam 3.5 R² = 0.6818 3 2.5 2 TDN (mg/L) 1.5 Monocacy 1 0.5 0 0 5 10 15 20 25 30 35 Cropland (%)

Figure 1-16. Correlations Between Mean Total Dissolved Nitrogen (TDN) Concentration and Percentage of Watershed Cropland Land Use (cropland %)

In two of the three TDN hotspot sub-watersheds (Antietam Creek and Monocacy River), the hot moments of TDN occurred in the summer when river flow was lowest (Figure 1-17). The TDN hot moment was found to be the spring high-flow period for Conococheague Creek. In other sub- watersheds with moderate TDN levels, the hot moment of TDN was also the summer base flow period.

6 6 Conococheague Opequon Creek Antietam Creek 5 NF Shenandoah Monocacy River 5 Catoctin Creek MD ) 1 - 4 4 Seneca Creek

3 3

2 TDN (mg L 2

1 1

0 0 summer fall winter spring summer fall winter spring Figure 1-17. Hot Moments of Total Dissolved Nitrogen Concentrations (TDN) in the Selected Hotspot Sub-Watersheds of the Potomac River During the Water Year 2016-2017.

16 The Water Research Foundation 1.5.2 N Source Tracking With Stable C and O Isotopes Sources of nitrate may roughly represent the sources of total N because nitrate is the main, active nitrogen species of N in the Potomac River and its tributaries. Nitrogen and oxygen isotopic compositions of nitrate in the Potomac watershed were in the ranges of +4 to +22‰ and -5 to +11‰ during summer low flow conditions. There was a trend of longitudinal increases in both nitrogen and oxygen isotopic delta values along the Potomac River from forest, agricultural, suburban, to urban land use (Figure 1-18). Development of dual isotopic tracer of nitrate with differentiate nitrate source signatures make to possible to track sources of nitrogen and potential transformation in streams and rivers (Kaushal et al. 2011). The plots of δ18O versus δ15N allow differentiating the inputs of precipitation, fertilizers, soil N, and manure/septic tank (Kendall and McDonnell 1998; Figure 1-18). Meanwhile, the linearity of δ18O versus δ15N with slope between 1:1 to 1:2 indicates occurrence of denitrification. In this study, the nitrate isotopic compositions of the agricultural watersheds lumped together in the regions of soil N and manure, suggesting soil N and manure were the most likely sources (Figure 1-18). Because these agricultural watersheds were highest in nitrate concentration, soil N (probably derived from fertilizers) and manure was the most important source of nitrogen of the Potomac River – consistent with our previous conclusions from hotspot maps or correlation analysis (Figure 1-16). Relative to values of agricultural watersheds, nitrate levels in the forest of the upper Potomac and the two lower watersheds with large area of wetland (Piscataway Creek and Mattawoman Creek) were more depleted (Figure 1-18), suggesting more inputs from natural sources that are characteristic of negative values (Kendall and McDonnell 1998). Instead, nitrate isotopic compositions in three urban/suburban watersheds (Broad Run, Seneca Creek and Occoquan River) were more enriched, suggesting inputs from wastewater or occurrence of denitrification (Kendall and McDonnell 1998).

Improving Water Reuse for a Healthier Potomac Watershed 17 NO3 in precipitation

NO3 in fertilizer

NH4 in Soil N Denitrification fertilizer & Manure & rain septic waste

Figure 1-18. Longitudinal Changes in Nitrate Isotopic Compositions in the Main Stem and Primary Tributaries of the Potomac River During Low Flow Conditions and Possible Nitrate Sources from Dual Nitrate Isotopic Tracers.

18 The Water Research Foundation 1.5.3 Hotspot of Phosphorus (P) Concentrations of soluble reactive phosphorus (SRP), the most reactive P species was in the range of 0.02 to 96 µg L-1 in the Potomac watershed. Hotspots of SRP (> 40 µg/L) were agricultural sub- watersheds in the middle Potomac River in Maryland (Antietam Creek and Catoctin Creek, MD), and the two wetland sub-watersheds in the lower Potomac (Piscataway Creek and Mattawoman Creek) (Figure 1-19). Other agricultural sub-watersheds in the middle Potomac watershed like the Conococheague Creek, the Opequon Creek, and the Monocacy River were also high in SRP, while SRP in forested sub- watersheds in the upper Potomac River were lowest (Figure 1-19). The reason for the two agricultural SRP hotspots was attributed to chemical fertilizer use in cropland as an important P source to rivers and streams. The reason for the hotspots of the two wetland streams is not clear, but it is known that P is readily released in organic, anoxic environments such as wetlands (Golterman 2001).

Conococheague Antietam Monocacy Catoctin Creek MD Opequon

Piscataway Mattawoman

Figure 1-19. Spatial Distribution of Soluble Reactive Phosphorus (SRP) showing SRP Hotspots in the Potomac River Watershed During the Water Year 2016-2017.

Improving Water Reuse for a Healthier Potomac Watershed 19 Like TDN, SRP concentrations also increased with an increasing percentage of watershed croplands (Figure 1-20), suggesting the contribution from fertilizer inputs on cropland. However, there were two groups of sub-watersheds. The first group had lower slope (1.09), and included mostly forested sub- watersheds and cropland sub-watersheds. The second group had a larger slope (4.75), and included a few pasture-dominated sub-watersheds, a few urban sub-watersheds and two wetland sub-watersheds. The relationship of the second group suggests that SRP may have additional sources other than those from agriculture (such as wetland and wastewater).

60 y = 4.7906x + 8.2096 R² = 0.6713 50

40 g/L)

µ y = 1.0545x + 2.8761 30 R² = 0.7755 SRP (

0 0 5 10 15 20 25 30 35 Cropland (%)

Figure 1-20. Correlations Between Soluble Reactive Phosphorus (SRP) and Percent of Cropland Land Use in the Primary Tributaries of the Potomac River Watershed.

The hot moment of SRP was usually the summer when river flow was lowest, and spring high-flow when SRP was in the second place (Figure 1-21). The occurrence of SRP hotspots during summer can be attributed to higher temperature, because SRP is more readily released at higher temperatures (Duan and Kaushal 2013).

150 150 Piscataway Creek Conococheague Catoctin Creek MD 120 120 Opequon Creek Antietam Creek Monocacy River ) 1 - Mattawoman Creek 90 90 g L µ 60 60 SRP (

30 30

0 0 summer fall winter spring summer fall winter spring Figure 1-21. Hot Moments of Soluble Reactive Phosphorus (SRP) in the Selected Hotspot Sub-Watersheds of the Potomac River During the Water Year 2016-2017.

20 The Water Research Foundation 1.5.4 Hotspot of Organic Carbon (C) Flow-averaged concentration of dissolved organic carbon (DOC) was in the range of 1.2-8.1 mg L-1. Different from N and P, hotspots of DOC occurred in the urban/suburban tributaries of the lower Potomac watershed near the Washington, D.C. (Figure 1-22), rather than the agricultural tributaries in the middle Potomac Watershed. These hotspots included Broad Run (4.9 mg L-1), Rock Creek (4.3 mg L-1), Anacostia River (5.1 mg L-1), Accotink Creek (4.6 mg L-1), Occoquan River (5.1 mg L-1) and Mattawoman Creek (9.0 mg L-1) (Figure 1-22).

Rock Broad Accotink Anacostia Occoquan

Mattawoman

Figure 1-22. Spatial Distribution of Dissolved Organic Carbon (DOC) showing DOC Hotspots in the Potomac River Watershed During the Water Year 2016-2017.

DOC concentrations were positively correlated with the percentage of watershed urban land use (Figure 1-23). Although the correlation was not strong (R2 = 0.17, n = 31, P <0.05), the relationship became more significant if two outliers (Occoquan River and Mattawoman Creek) were excluded (R2 = 0.38, n = 29, P <0.01). One of the outliers is Mattawoman Creek – a river that drains a large area of wetland. The reason for this wetland DOC hotspot, because the highest DOC concentrations are usually found in wetlands (e.g., Duan et al. 2017). There are two reasons for higher DOC concentrations in urban watersheds: 1) there is higher DOC loss due to a wide occurrence of impervious surface cover, and 2) production of new organic carbon due to anthropogenic inputs (e.g., wastewater), and enhanced algal production as a result of eutrophication (Kaushal et al. 2014). These reasons may also explain the high DOC concentration in our second outlier (Occoquan River). At first, algal production could be a significant new DOC source because the Occoquan River is dammed. Secondly, there is a wastewater treatment plant flowing to the river, which could contribute a large amount DOC (see Chapter 3).

Improving Water Reuse for a Healthier Potomac Watershed 21 10 9

8 y = 0.022x + 2.4419 7 R² = 0.1686 6 5 4 DOC (mg/L) 3 2 1 0 0 10 20 30 40 50 60 70 80 90 Urban (%)

Figure 1-23. Correlations Between Dissolved Organic Carbon (DOC) and Percent of Urban Land Use in the Primary Tributaries of the Potomac River Watershed. The value of R2 would increase to 0.38 if two outliers were excluded.

In the hotspot sub-watersheds, the hot moments of DOC were usually in the fall (in Rock Creek and Accotink Creek), the spring (in Mattawoman Creek and Occoquan River), or both (Anacostia River) (Figure 1-24). The hot moment in fall can be attributed to leaf falls, which release large amounts of DOC from the watershed to the rivers (Duan et al. 2014). The hot moment of DOC in spring may be due to erosion during high-flow period. Since this study only took place over a single year, further investigation is needed.

10 9 8 7 ) Rock Creek 1 - 6 Anacostia River 5 Accotink Creek 4

DOC (mg L Occoquan River 3 2 Mattawoman Creek 1 0 summer fall winter spring

Figure 1-24. Hot Moments of Dissolved Organic Carbon (DOC) in the Selected Hotspot Sub-Watersheds of the Potomac River Watershed During the Water Year 2016-2017.

22 The Water Research Foundation 1.5.5 Linking EDCs With Nutrients, Organic Carbon and other Variables Correlation analyses were conducted between estrogenic activity, estrone (one of the estrogenic compounds), herbicides, nutrients, and organic carbon to determine if their hotpots are the same. Estrogenic activity was used as a proxy for all estrogenic EDCs (EEDCs) in this analysis. Results of the analyses showed that there was no correlation between estrogenic activity (or estrone) and TDN or DOC, suggesting different N, organic carbon, or EDC sources (Table 1-2). Nitrogen was from chemical fertilizers and manure on cropland, DOC was mainly from urban watersheds, while estrogen came mainly from pasture/hay (see above). However, a positive relationship was found between estrogenic activity and SRP (p < 0.01), suggesting the possibility that they come from similar sources. This may be a result from both estrogen and SRP being related to agricultural land uses. Therefore, it may be possible to co-manage estrogen and SRP in the Potomac River watershed by focusing on agricultural land BMPs. Three common herbicides (atrazine, metolachlor, and prometon) were all positively correlated with TDN, especially nitrate, suggesting a common source from croplands (Table 1-2). One of the herbicides, prometon, was also positively correlated with SRP. Therefore, it may be possible to co-manage herbicides and nutrients (nitrogen and phosphorus) in the Potomac River watershed by focusing on applying appropriate BMPs to crop production type agricultural land uses. Table 1-2. Correlation of Estrone, Estrogenic Activity, and Common Species of Herbicides With Nutrients, Organic Carbon, and Other Water Quality Parameters

t pH SC SRP DOC NO3-N Estrone 0.00 -0.27 0.00 0.08 0.23 -0.10 Estrogenic activity 0.54 0.29 -0.07 0.34 -0.08 0.12 Atrazine 0.22 0.12 0.07 0.17 0.01 0.39 Metolachlor 0.22 0.11 0.08 0.09 0.12 0.29 Prometon 0.31 0.50 0.38 0.43 0.02 0.37 Simazine 0.19 0.16 -0.07 0.02 0.00 0.16

Correlation analyses were conducted between EDCs (estrogenic activity and herbicides) and other quality variables (e.g., water temperature, pH and conductivity) to test the likely control on EDC. A significant positive correlation was found between estrogenic activity (and prometon) and water temperature (p < 0.05; Table 1-2). The reason for this correlation is not clear. One possible reason is that animals in pasture/hay may be more active and thus have more access to stream water in warm months than in cold months. Another possibility may be that estrogenic activity is related to herbicide use and herbicide use increases during the warmer period of the year which aligns with the normal growing season. Estrogenic activity and prometon were also positively correlated with pH (Table 1-2), but the reason for these correlations is not clear.

Improving Water Reuse for a Healthier Potomac Watershed 23 1.6 Source Estimation for Nutrients, DOC, and EDCs of the Potomac River The concentrations of nutrients, DOC and EDCs of the tributaries representing typical land use type were used to estimate loads from different types of land use – forest, agricultural with croplands, agricultural with pasture/hay, urban/suburban. Typical forest tributaries are those with 80% of forested land use in watersheds including the N Branch Potomac, Wills Creek, Patterson Creek, S Branch Potomac, Town Creek, Silderling Hill Creek, Cacapon River, Sleepy Creek, and Back Creek. Typical agricultural tributaries with cropland include Conococheague Creek, Antietam Creek, and Monocacy River (cropland > 24%). Typical agricultural tributaries with pasture/hay are Opequon Creek, Shenandoah River, Catoctin Creek, MD, Catoctin Creek VA, Goose Creek with (pasture/hay > 42%). Five tributaries (Difficult Run, Cabin John Creek, Rock Creek, Anacostia River, and Cameron Creek with urban > 54%) were selected as urban/suburban types. Flow-averaged concentration was collected for each tributary, thereby including the effect of stream flow. In each type of tributaries, the flow-averaged concentrations were normalized by their drainage area to obtain area-averaged concentration for this type of land use. In the end, the fluxes and estimated annual loads were calculated for each types of tributaries (Table 1-3).

Table 1-3. Estimated Annual Fluxes of Water, Nutrients, Organic Carbon, Estrogen and Herbicides from Forest, Agricultural Runoff (Cropland and Pasture), Urban Runoff, and Urban Point Sources in the Potomac River. The concentrations of nutrients, DOC, estrogen, and herbicides were flow-averaged measurements conducted during year 2016-2017 at subwatersheds with dominant land use of forest, cropland, pasture, and urban. Forest Cropland Pasture Urban Water 109m3/yr 4.19 0.49 1.63 0.07 TDN 106kg/yr 5.16 21.15 8.51 5.51 SRP 106kg /yr 0.02 0.18 0.14 0.07 DOC 106kg /yr 15.29 16.16 15.69 29.22 E1 103kg/yr 0.0021 0.0027 0.0032 0.0040 BLYES 103kg/yr 0.0044 0.0079 0.0156 0.0058 Atrazine 103kg/yr 0.01 1.35 0.47 0.18 Metolachlor 103kg/yr 0.01 0.20 0.11 0.11 Simazine 103kg/yr 0.00 0.20 0.30 0.09 Prometon 103kg/yr 0.00 0.02 0.03 0.01 Total herbicides 103kg/yr 0.03 1.76 0.90 0.43

24 The Water Research Foundation Results showed that forest, cropland, pasture/hay and urban tributaries contributed approximately 66%, 8%, 25%, and 1% of the total water source (Figure 1-25). Agricultural tributaries accounted for most of the TDN and SRP loads (73% of TDN and 78% of SRP), with the most contributions from tributaries with more croplands (53% of TDN and 43% of SRP). This is consistent with numerous previous studies showing that agricultural land use is the main source of nutrients in the Potomac River. Urban tributaries were the No. 1 source of DOC (38%) although they accounted for 1% of water flow. Forest tributaries contributed least DOC (20%) although they contributed most of water flow (66%). The unexpected high contribution of DOC from urban tributaries may be related to impervious cover and high DOC export rate, as well as DOC inputs from waste and elevated primary production in streams (Kaushal et al. 2014).

water TDN 1% 14% 13%

25% forest forest

cropland cropland 21% pasture pasture 8% 66% urban urban 52%

DOC SRP

4% 20% 18% forest forest

38% cropland cropland

43% pasture pasture 21% urban 35% urban

21%

Figure 1-25. Relative Contributions of Water Flow, Nutrients, and DOC from Tributaries With Forest, Cropland, Pasture/Hay, and Urban/Suburban Land Use.

Improving Water Reuse for a Healthier Potomac Watershed 25 Results also show that estrogenic activity (BLYES) were mostly contributed from pasture/hay tributaries (46%) while urban/suburban tributaries were an important input of estrone (33%; Figure 1-26), indicating sources of animal industry. Tributaries influenced by cropland were the main contributor of atrazine and metolachlor (67% and 47%). The main source of the other two banned pesticides (simazine and prometon) were tributaries influenced by pasture/hay (Figure 1-26).

EA E1

13% 17% 18% forest forest 33% cropland cropland 24% pasture 22% pasture

urban urban 46% 27%

Atrazine Metolachlor

9% 1% 3%

forest 25% forest

23% cropland cropland

47% pasture pasture

67% urban urban 25%

Simazine Prometon

0% 0% 15% 22% forest forest 34% 32% cropland cropland

pasture pasture

urban urban 51% 46%

Figure 1-26. Relative Contributions of Estrogen and Pesticides from Tributaries With Forest, Cropland, Pasture/Hay, and Urban/Suburban Land Use.

26 The Water Research Foundation CHAPTER 2

Impacts and Outcomes of Current Nutrient Management Strategies on Source Controls of EDCs at Maryland Sites

2.1 Introduction Results from Year 1 showed that hotspots of nutrients (nitrogen (N) and phosphorus (P)) in the Potomac River were the agricultural sub-watersheds especially those that drain large percentages (>23%) of cropland. Hotspots of estrogen (using estrogen activity as a proxy) and pesticides (mainly herbicides) were also agricultural tributaries especially those with larger percentages of pasture/hay land use, as well as some suburban tributaries (i.e., Occoquan River). Because the hotspots of the nutrients and EDCs had many overlaps, it is hypothesized that best practices (BMPs) that are currently used to manage nutrients from point and nonpoint sources can be used to reduce EDCs as well. The objectives of this project in the second year includes two parts: a) Use paired watershed studies to evaluate impacts and outcomes of current reclamation, reuse, harvesting, and management strategies on source controls of pollutants and b) focused study on the comparative impact of planned potable reuse. The University of Maryland conducted research at Maryland sites for the first objective (this chapter), while Virginia Polytechnic Institute and State University was in charge of research for both objectives at Virginia sites (Chapter 3). These groups collaborated on sample collection, sample preparation and chemical analyses. 2.2 Experiment Design Here, associations of two types of watershed nutrient management approaches (agricultural BMPs and urban BMPs) on EDCs reduction relative to the engineered targeted nutrient reduction were examined. Concentrations of EDCs and nutrients in point and nonpoint sources were measured across seasons and stream flow for a year. The impacts of several nutrient management strategies on nutrient and EDCs retention were evaluated using different approaches. In the paired series designs (Bishop et al. 2005) for agriculture and urban BMPs, two streams within the same sub-watershed were sampled, each with similar land use patterns but different degrees of best management practice (BMP) implementation. Then, the difference in fluxes of EDCs and nutrients between restored and unrestored streams was the EDC and nutrient retention of the BMPs. Conversely, upstream-downstream changes in water quality were monitored to examine the effect of point source pollutions (wastewater treatment plants, or WWTPs). Mass balances were conducted to make sure the upstream-downstream changes were due to the inputs from the discharges from WWTPs. Other water quality parameters – water temperature, pH, specific conductance and dissolved organic carbon (DOC) were measured because BMPs that target DOC may show promise in also controlling estrogenic activity. Specific conductance (SC) also was used as a surrogate for wastewater:streamflow ratios when mass balances could not be conducted due to lack of stream flow. In addition, changes in nutrients and total estrogen were examined across the wastewater treatment profile to identify the treatment barriers to nutrient or total estrogen removal through the WWTP. Finally, all the data were used to estimate the relative importance of several point versus nonpoint sources of EDCs in the Potomac River watershed. Results from this research will help inform watershed approaches to efficiently co-manage multiple pollutants.

Improving Water Reuse for a Healthier Potomac Watershed 27 2.2.1 Site Selection Site selection were all based on the results of the first year’s hotspot identification. In order to examine the effect of agricultural BMPs, we selected one pair of tributaries in Bens Run in the Monocacy watershed and another pair of sites (Furrs Run and Elk Run) in the Occoquan watershed (Table 2-1), both of which were high in nutrients and EDCs (see Chapter 1). To examine the effect of urban BMPs, we selected another two pairs of sites - one pair of sites (Paint Branch/Briar Ditch) in the Anacostia Watershed and another pair (in Cub Run) in the Occoquan watershed (Table 2-1). We collected wastewater effluent from three WWTPs (Seneca, Blue Plains and Upper Occoquan Service Authority (UOSA)) to examine the effect of wastewater treatment plants with advanced compared to conventional reclamation. Stream water samples up- and downstream of the effluents were collected for the comparisons (Table 2-1). All the sampling sites in Maryland are presented in Figure 2-1.

Agricultural (Bens Run) Tributary 1: W/O BMP Tributary 2: With BMP

Monocacy River

Seneca Urban With BMP WWTP Paint Branch

Great Seneca Creek

Urban W/O BMP Brier Ditch Potomac River Anacostia River

Hoof Run Blue Plains WWTP

Figure 2-1. Water Quality Sampling Sites on the Land Use Land Cover Map of the Potomac River Basin.

28 The Water Research Foundation Table 2-1. Site Selected for Examining Effect of Agricultural BMPs, Urban BMPs, and Wastewater Treatment Plants on Nutrients and EDCs Retentions in Maryland and Virginia. Virginia U. of Category Sampling Site Tech Maryland Agricultural (BMPs) Ben's Run BMP x Ben's Run no BMP x Furrs Run BMP x Elk Run no BMP x Urban (BMPs) Paint Branch BMP x Briar Ditch no BMP x Cub Run above BMP x Cub Run below BMP x Wastewater treatment plants Seneca WWTP effluent x Seneca Upstream WWTP x Seneca downstream WWTP x Blue Plains WWTP effluent x Potomac upstream WWTP x Potomac downstream WWTP x UOSA WWTP effluent x Bull Run Upstream WWTP x Bull Run downstream WWTP x

2.2.2 Water Sample Collection Samples were collected on a monthly basis during 2017-2018. According to hydrographs, the period can be separated into three stages: a low-flow period from October-January, an intermediate flow period in February-April, and a high-flow period between May-June (Fig. 2-2). The temporal changes in the flow of wastewater were minor relative to changes in the stream/river hydrograph, but wastewater discharge flow peaks matched well with the stream peak flows (Fig. 2-2). At each site, 1,500 mL of unfiltered water was collected by immersing three 500-ml amber glass bottles under the water surface without disturbing bed sediment. For sampling of point source inputs (the Seneca WWTP), the downstream sites were at least 200 m below the confluence in order to ensure complete mixing between the WWTP effluent and stream water. The upstream site above the Blue Plains WWTP was selected at Hains Point of Washington DC, where the Anacostia River flows into the Potomac River. The downstream site below Blue Plains WWTP was located in National Harbor of Fort Washington, Maryland, where the effect of bank inputs was likely minimal. Water temperature and conductivity were measured in situ, using a Waterproof Multi-parameter Testr™ 35-Series Pocket Meters. The instrumental precision of temperature and conductivity measurements were ± 0.1 °C and ± 0.1°C, respectively. Water samples were stored on wet ice during transportation. Water samples were filtered through pre-weighed Whatman GF/F filters on the same day of collection. The filtrates were either stored at 4°C until extraction of estrogens, or at -20°C for the other analyses (e.g., nutrients and DOC). Sample collection and processing were generally completed within 48 hours. 2.2.3 Chemical Analyses Concentrations of DOC, total dissolved nitrogen (TDN), and soluble reactive phosphorus (SRP) were analyzed by the University of Maryland (UMD) using the methods described in Chapter 1. Sample processing and extraction for EDCs was conducted at the University of Maryland (UMD) sample clearinghouse. Sample analyses for estrogenic activity were conducted at the USGS, Leetown Science Center. Estrogenic compounds and pesticides were analyzed at the University of Buffalo (see Chapter 1).

Improving Water Reuse for a Healthier Potomac Watershed 29

5000 Potomac at Little Fall, DC 4000 3000 2000 1000 0 1000 750 Agricultural_Monocay 500 ) 1 - 250 S 3 0 (m 100 Urban_NE Branch Anacostia

Flow 75 50 25 0

80 1 Seneca Creek and WWTP effluent (brown) 60 0.8 0.6 40 0.4 20 0.2 0 0

Figure 2-2. Flow Conditions of the Potomac River and Selected Tributaries When River Water Samples Were Monthly Collected. River flow data were collected from USGS gauge stations, and the dash lines stand for the days when water samples were collected.

In January 2018, the method for pesticides analysis were switched from the GC-MS/MS to a LC-MS/MS so that more pesticides could be analyzed. Results of January 2018 showed the two methods are comparable (Ping and Aga 2019). LC-MS/MS analysis was performed using a TSQ Quantum™ Ultra triple quadrupole mass spectrometer (Thermo Fisher Scientific, San Jose, CA) equipped with an Agilent 1200 HPLC system (Agilent Technologies, Santa Clara, CA). Chromatographic separation was performed using an Xselect CSH™ C18 HPLC column (130 A pore size, 3.5 ˚ mm particle size, 2.1 mm i.d. × 150 mm length) from Waters (Mildford, MA) equipped with an XSelect CSH™ C18 VanGuard Cartridge (130 A pore size, 3.5 ˚mm particle size, 2.1 mm i.d. × 5 mm length). Gradient elution at a flow rate of 200 mL min-1 was carried out using mobile phase A [water–methanol (96 : 4, v/v) with 5 mM ammonium hydroxide] and mobile phase B [acetonitrile–methanol–water (80 : 10 : 10, v/v/v) with 5 mM ammonium hydroxide]. The total run time for each sample was 35 min. The injection volume was 20 mL. The mass spectrometer was operated under unit resolution in the time scheduled SRM mode. By allowing monitoring of analytes' SRM transitions only around their retention time, the mass spectrometer spends more time on each transition. Two transitions were monitored for each compound with the more abundant ion for quantification and a second ion as the qualifier for confirmation.

30 The Water Research Foundation 2.3 EDCs Data Synthesis 2.3.1 Pesticides (or Synthetic Organic Compounds) Figure 2-3 shows pesticides that were commonly present throughout the sampling period. The upper panel shows the total number of detections for each pesticide in order from most to fewest. Since the list of pesticides being analyzed changed slightly in January 2018, it was important to know the percent detection, meaning the number of detections out of the total number of times the compound was tested. These figures, as well the organic compounds’ toxicity, help determine which specific compound patterns to further analyze. According to Figure 2-3, the following pesticides were commonly detected (n > 43, percent > 72%): metolachlor, atrazine, prometon, simazine, 4-nonylphenol, clothianidin, imidacloprid, fipronil, acetamiprid, dinotefuran). Some pesticides, including butylat, coumaphos, cycloate, dieldrin, erin aldehyde, fenchlorphos, methyl parathion, parathion, and tributylphosphorothithiote, were not detected at all. They were not included in any further analysis.

Synthetic Organic Compound by numbers of detections

100 90 80 70 60 50 40 30 20 10

Numbers of detection 0

Synthetic Organic Compound by percent of detections

100 90 80 70 60 50 40 30 20 10 0 Percent detections (%)

Figure 2-3. Pesticides (or Synthetic Organic Compounds) by Number of Detections and by Percent Detection. The sum of the number of detections for each compound in order from most to fewest. The number of detections was divided by the number of times the compound was analyzed to get the percent detection. Non-detect data were not presented graphically, since the concentrations were between 0 and the detection limit, and thus unknown. The GC-MS and LC-MS analyzed data were combined into one data set, with the month of January being an average of the two methods per pesticides since both analyses were conducted.

Improving Water Reuse for a Healthier Potomac Watershed 31 The most frequently detected pesticides were analyzed to determine if there was a pattern among them (Figure 2-4). 4-Nonylphenol tended to have the highest concentrations, especially during the period of January to April. In May and June, other pesticides such like atrazine also increased to comparable levels to 4-nonylphenol. Concentrations of the pesticides increased in May through June, with May being the greatest. Interestingly, chronic exposures to atrazine and nonylphenol concentrations have resulted in observed amphibian or fish feminizations and gonadal transition responses (Mackenzie et al. 2003). Comparison of chronic aquatic toxicity levels of these pesticides compared to the concentrations observed can provide insight into the environmental relevance of the concentrations observed in water samples, but may be beyond the scope of this particular study.

10000 Oct Nov Dec 1000

100

0.1 Concentration (ng/L) BP eff BP eff BP eff BP UP BP UP BP UP Wp eff Wp eff Wp eff WP up WP up WP up Agr BMP Agr BMP Agr BMP Urb BMP Urb BMP Urb BMP BP Down BP Down BP Down WP down WP down WP down Agr no BMP Agr no BMP Agr no BMP Urb no BMP Urb no BMP Urb no BMP

10000 Jan Feb Mar 1000

100

1 Concentration (ng/L) Concentration 0.1 BPdn BPdn Bpup Bpup Bpeff Bpeff BP eff BP UP Wpup Wpup Wpeff Wpeff Wp eff UrbNO UrbNO WP up AgrBNO AgrBNO AgrBMP AgrBMP UrbBMP UrbBMP Agr BMP Urb BMP BP Down Wpdown Wpdown WP down Agr no BMP Urb no BMP

10000 Apr May Jun 1000

100

Concentration (ng/L) 0.1 BPdn BPdn BPdn Bpup Bpup Bpup Bpeff Bpeff Bpeff Wpup Wpup Wpup Wpeff Wpeff Wpeff UrbNO UrbNO UrbNO AgrBNO AgrBNO AgrBNO AgrBMP AgrBMP AgrBMP UrbBMP UrbBMP UrbBMP Wpdown Wpdown Wpdown

Metolachlor Atrazine Prometon Simazine 4-Nonylphenol Clothianidin Imidacloprid Fipronil Acetamiprid Dinotefuran

Figure 2-4. Frequent Pesticides Detected. The frequent pesticides were determined by finding the pesticides with the most number of detections, and that had a greater than 70% percent detection rate (Figure 2-3).

32 The Water Research Foundation 4-Nonylphenol and atrazine concentrations were analyzed further (Figure 2-5). They were graphed by the concentration at each sampling location per month. 4-Nonylphenol is a persistent, hydrophobic industrial compound (U.S. EPA 2019) that was tested for from January to September 2019. 4- Nonylphenol was consistently detected in samples and was comparatively high in concentration. High levels are seen especially in May (up to 1000 ng/L).

10000 Jan Feb Mar 1000

100

10 Nonylphenol (ng/L) Nonylphenol - 4 1 BPdn BPdn Bpup Bpup Bpeff Bpeff BP eff BP UP Wpup Wpup Wpeff Wpeff Wp eff UrbNO UrbNO WP up AgrBNO AgrBNO AgrBMP AgrBMP UrbBMP UrbBMP Agr BMP Urb BMP BP Down Wpdown Wpdown WP down Agr no BMP Urb no BMP

10000 Apr May Jun

1000

100

10 Nonylphenol (ng/L) - 4 1 BPdn BPdn BPdn Bpup Bpup Bpup Bpeff Bpeff Bpeff Wpup Wpup Wpup Wpeff Wpeff Wpeff UrbNO UrbNO UrbNO AgrBNO AgrBNO AgrBNO AgrBMP AgrBMP AgrBMP UrbBMP UrbBMP UrbBMP Wpdown Wpdown Wpdown Figure 2-5. 4-Nonylphenol Concentrations Present at Maryland Sampling Locations. 4-Nonylphenol concentrations for each site plotted for the period the compound was tested for, from January to June 2018.

Improving Water Reuse for a Healthier Potomac Watershed 33 1000 Oct Nov Dec 100

1 Atrazine (ng/L)

0.1 BP eff BP eff BP eff BP UP BP UP BP UP Wp eff Wp eff Wp eff WP up WP up WP up Agr BMP Agr BMP Agr BMP Urb BMP Urb BMP Urb BMP BP Down BP Down BP Down WP down WP down WP down Agr no BMP Agr no BMP Agr no BMP Urb no BMP Urb no BMP Urb no BMP

1000 Jan Feb Mar 100

1 Atrazine Atrazine (ng/L)

0.1 BPdn BPdn Bpup Bpup Bpeff Bpeff BP eff BP UP Wpup Wpup Wpeff Wpeff Wp eff UrbNO UrbNO WP up AgrBNO AgrBNO AgrBMP AgrBMP UrbBMP UrbBMP Agr BMP Urb BMP BP Down Wpdown Wpdown WP down Agr no BMP Urb no BMP

1000 Apr May Jun

100

10 Atrazine (ng/L)

1 BPdn BPdn BPdn Bpup Bpup Bpup Bpeff Bpeff Bpeff Wpup Wpup Wpup Wpeff Wpeff Wpeff UrbNO UrbNO UrbNO AgrBNO AgrBNO AgrBNO AgrBMP AgrBMP AgrBMP UrbBMP UrbBMP UrbBMP Wpdown Wpdown Wpdown Figure 2-6. Atrazine Concentrations Present at Maryland Sampling Locations. Atrazine concentrations for each site plotted for the entire sampling period.

Atrazine concentrations increased in the months of May and June (Figure 2-6), which is not surprising since it is applied as an herbicide in the spring. The MCL for atrazine specified by the U.S. EPA is 0.003 mg/L, or 3000 ng/L. The maximum concentration of atrazine detected is 827 ng/L, which was about 1/3 to ¼ of the MCL.

34 The Water Research Foundation 2.3.2 Estrogen A total of 18 estrogens were assessed in each sample, but only two were ever present. Concentrations of estrone (E1) and estrone-3-sulfate (E1-3S), the two estrogens detected, are presented in Figure 2-7. Estrogens were not detected at all sampling site locations. The main trend visible was an extremely high value (up to a few hundreds ng/L) in E1 and E1-3S concentrations in Febuary 2018 at one site. E1-3S and E1 concentrations were also higher than 1 ng/L in October 2018 and April 2019.

1.6 Oct Nov Dec 1.4 1.2 1 0.8 0.6 0.4 Estrogen (ng/L) 0.2 0 BP eff BP eff BP eff BP UP BP UP BP UP Wp eff Wp eff Wp eff WP up WP up WP up Agr BMP Agr BMP Agr BMP Urb BMP Urb BMP Urb BMP BP Down BP Down BP Down WP down WP down WP down Agr no BMP Agr no BMP Agr no BMP Urb no BMP Urb no BMP Urb no BMP

1000 Jan Feb Mar 100 10 1 0.1

Estrogen (ng/L) Estrogen 0.01 0.001 BPdn BPdn Bpup Bpup Bpeff Bpeff BP eff BP UP Wpup Wpup Wpeff Wpeff Wp eff UrbNO UrbNO WP up AgrBNO AgrBNO AgrBMP AgrBMP UrbBMP UrbBMP Agr BMP Urb BMP BP Down Wpdown Wpdown WP down Agr no BMP Urb no BMP

2.5 Apr May Jun 2

1.5

Estrogen (ng/L) 0.5

0 BPdn BPdn BPdn Bpup Bpup Bpup Bpeff Bpeff Bpeff Wpup Wpup Wpup Wpeff Wpeff Wpeff UrbNO UrbNO UrbNO AgrBNO AgrBNO AgrBNO AgrBMP AgrBMP AgrBMP UrbBMP UrbBMP UrbBMP Wpdown Wpdown Wpdown

Figure 2-7. Estrone (E1, in red) and Estrone-3-Sulfate (E1-3S, in blue) Concentrations Present at Virginia Sampling Locations. The estrone (E1) and estrone-3-sulfate (E1-3S) concentrations that met criteria plotted for each sample site over the sampling period.

Improving Water Reuse for a Healthier Potomac Watershed 35 2.4 Effect of Agricultural BMPs on Nutrient and EDCs Reductions 2.4.1 Description of Sampling Site and Agricultural BMPs The restored stream is a small headwater tributary surrounded by a 1.8-km2 crop and livestock farm with about 150 head of Black Angus cattle on about 0.32-km2 of pasture. Prior to BMP installation, the stream’s riparian zone was part of a cattle pasture that provided the livestock with unrestricted access to the stream. The BMPs included more than 2,680-m of fencing, three spring developments to replace in-stream cattle watering, improvements to areas heavily used by livestock, including two stream crossings as well as plantings of 0.04-km2 of cool-season grasses (Shanks and Soehl 2008). As a result of this effort, streambank stability is much improved and the percentage of a desirable in-stream gravel substrate (material that forms streambed) has increased, while the percentage of the sand/mud substrate has decreased. Phosphorus concentrations also declined, indicating a decrease in the amount of sediment entering the stream from the surrounding pasture. The unrestored stream is a nearby small, unnamed headwater tributary with similar land use and geomorphology. 2.4.2 Effect of Agricultural BMPs on Nutrient Reductions Results of water temperature, conductivity, nutrients and DOC measurements of the agricultural streams with/without BMPs are summarized in Figure 2-8 and Table 2-2. Despite apparent seasonal pattern of water temperature and dissolved oxygen (DO), no significant difference was observed in water temperature, conductivity or TDN between streams with and without BMPs (p > 0.05, one-way ANOVA). The restored agricultural streams were consistently lower in total dissolved nitrogen (TDN) and soluble reactive phosphorus (SRP) than that in the unrestored streams (Figure 2-8). The fluxes of TDN and SRP of the restored streams were estimated to be 85 and 0.66 kg/km2/yr, showing reductions by 24% and 61% relative to the fluxes of unrestored streams (85 and 0.66 kg/km2/yr; Table 2-2). Although DOC concentrations at the restored site was not consistently lower than the unrestored stream, DOC flux from the restored stream was approximately lower than the unrestored stream by 26% (Figure 2-3 and Table 2-2). The consistent lower concentrations and lower fluxes of TDN and SRP than the unrestored site suggest that the agricultural BMPs can actually retain both N and P. Prior study at this restored site showed P concentrations have also been declining since the installation of the BMPs (Shanks and Soehl 2008). The highest values of TDN, SRP and DOC occurred at the unrestored site in February 2018 when the stream was disturbed by a herd of cattle (observed by Co-PI when he collected water samples; Figure 2- 9). This observation directly showed that the agricultural BMPs for stopping in-stream cattle watering actually works for nutrient reductions. Without the BMPs, the cattle can go directly to streams and disturb bed sediment, causing mobilization of dissolved, particulate nutrients, and organic carbon. Moreover, nutrients and organic carbon in the animal waste may discharge directly to the stream as the cattle drink in the stream, without any pre-reduction or recycling. Meanwhile, the cattle also disturbed sediments in the banks, and nutrients and organic carbon from bank sediment were washed into the streams especially during storm events. By installation of the BMPs, nutrients and organic carbon from bed sediment, animal waste and bank sediment can be greatly reduced, shown by significantly lower fluxes and concentrations of TDN, SRP and DOC concentrations in the restored stream than the unrestored stream. Table 2-2. Estimates of Nutrient and DOC Yields from the Two Agricultural Streams Without or With Best Management Practices (BMPs). BMP No BMP Reduction (%) TDN (kg/km2/yr) 112 85 24 SRP (kg/km2/yr) 1.75 0.68 61 DOC (kg/km2/yr) 50.0 37.2 26

36 The Water Research Foundation In addition, it was observed that TDN concentrations were higher during winter, and SRP and DOC concentrations were also high in May to June (besides February) in both restored and unrestored streams (Figure 2-3). Reasons for these seasonal patterns warrant further investigation.

25 No BMP_Agr 8 ) 1 20 BMP_Agr - 6

C) 15 ⁰ 4

t ( 10 2

5 L (mg TDN 0 0

20 180 ) ) 1 - 1

15 - 120 10 60

5 SRP (g L DO (mg L DO (mg 0 0

600 5 ) ) 1 1 4 - - 400 3

200 2 SC (µS cm DOC (mg DOC (mg L 1 0 0 Oct Nov Dec Jan Feb Mar Apr May June Oct Nov Dec Jan Feb Mar Apr May June Figure 2-8. Comparison of Temperature (t), Dissolved Oxygen (DO), Specific Conductance (SC), Total Dissolved Nitrogen (TDN), Soluble Reactive Phosphorus (SRP) and Dissolved Organic Carbon (DOC) Between Two Agricultural Streams Without or With Best Management Practices (BMPs). The unrestored stream was not sampled in January 2018 due to no flow in the frozen stream.

Figure 2-9. A Herd of Cattle Accessed to the Unrestored Agricultural Stream on February 2018, Just Above the Water Sampling Site.

Improving Water Reuse for a Healthier Potomac Watershed 37 2.4.3 Effect of Agricultural BMPs on EDCs Reductions Figure 2-5 show extremely high values of E1 and E1-S concentrations (36.3 ng/L and 428.2 ng/L) at the unrestored agricultural site on February 2018 when a herd of cattle accessed the stream for water drinking (Figure 2-9). Otherwise, there was no significant difference in their concentration of E1 or E1-S between the two streams (p = 0.18 and 0.45, one-way ANOVA). Because of the huge difference in February 2018, the fluxes of E1 and E1-S of the restored streams (0.001 and 0.004 g/km2/yr) were far lower than that of the unrestored stream (0.144 and 1.68 g/km2/yr), resulting in almost >99% reductions of the estrogen compounds due to the agricultural BMP (Table 2-3). The results above suggest that stopping in-stream cattle watering was the main reason for reduction in estrogen in the restored stream. The cattle can discharge high dose of estrogen directly into stream with animal wastes (urine and stool), and high concentrations of estrogen were observed in cattle wastes (Hanselman et al. 2003). By stopping in-stream cattle watering, animal waste would be recycled or partially decomposed before reaching stream in the restored stream. Because of the low background of estrogen, the pulses of estrogen input from animal waste (once to several times a day) probably contribute much higher fluxes of estrogen than what we estimated here. Different from estrogen, no peak concentrations of herbicides and insecticides were observed in February 2018 when a herd of cattle accessed the stream for water drinking (Figure 2-9). Instead, the peak values of atrazine, metolachlor and Clothianidin, which are currently used in agriculture and showed higher levels here, occurred in May to June. This timing matched well with application of these herbicides and insecticide in this region (Dernoeden 2005). 4-Nonylphenol was consistently detected and its concentration did not change much but also had a peak value in May. Concentrations of prometon and simazine (two common herbicides but are currently banned) and other three insecticide (Imidacloprid, Fipronil and Acetamiprid) were relatively lower than that of atrazine and metolachlor. Prometon, simazine and Imidacloprid were higher in March through June, while Fipronil and Acetamiprid were detected only in January and February. There was no significant difference in the concentrations of these insecticides and herbicides between the agricultural streams with and without BMPs (p > 0.05). However, the stream with BMPs was lower in the fluxes of 4-nonylphenol (by 87%), atrazine (69%), Imidacloprid (98%), Fipronil (30%) and metolachlor (15%). The fluxes of prometon, simazine, acetamiprid, and clothianidin of the restored stream were higher or close to that of the unrestored stream. This suggests the BMPs in this study did not enhance retention of these pesticides. Table 2-3. Estimates of EDCs Yields from the Two Agricultural Streams Without or With Best Management Practices (BMPs). No BMP BMP Reduction (%) E1 (g/km2/yr) 0.144 0.001 99 E1-S (g/km2/yr) 1.68 0.004 100 4-Nonylphenol (g/km2/yr) 22.30 2.85 87 Atrazine (g/km2/yr) 6.26 1.93 69 Metolachlor(g/km2/yr) 1.01 0.86 15 Prometon (g/km2/yr) 0.007 0.021 -213 Simazine (g/km2/yr) 0.078 0.076 2 Imidacloprid (g/km2/yr) 0.011 0.0002 98 Fipronil (g/km2/yr) 0.008 0.005 30 Dinotefuran (g/km2/yr) BDL BDL Acetamiprid (g/km2/yr) 0.023 0.021 9 Clothianidin (g/km2/yr) 0.374 0.656 -75

38 The Water Research Foundation

5 No BMP_Agr 5 ) 1 )

4 - 4

1 BMP_Agr - 3 3 ng L

2 S (ng L 2 - E1 (

1 E1 1 0 0

10000 800 ) ) 1 1 - - 1000 600 100 400 10 200 Atrazine (ng L (ng Atrazine

Nonylph. (ng L (ng Nonylph. 0

- 1 4

150 ) 40 ) 1 1 - - 30

(ngL 100 20 50 10

Metolach. Metolach. 0 0 Clothianid. (ng (ng L Clothianid.

8 ) 1

- 8 ) 1 - 6 6

(ng L 4 4 2 2 Acetamipr. (ng L (ng Acetamipr.

Simazine 0 0 ) 1

- 2.5 6 ) 1 - 2 4 (ng L 1.5 1 2

Fipronil (ng L 0.5

0 Prometon 0 ) ) 1 -

0.8 1

- 1 0.6 0.8 0.6 0.4 0.4 Imidaclo. (ng L (ng Imidaclo.

0.2 (ng L Dinotef.. 0.2 0 0 Oct Nov Dec Jan Feb Mar Apr MayJune Oct Nov Dec Jan Feb Mar Apr MayJune Figure 2-10. Comparison of Estrone (E1), Estrone Sulfate (E1-S), and Other Common Pesticides Between Two Agricultural Streams Without or With Best Management Practices (BMPs). E1 and E1-S concentrations in February were 36.3 ng/L and 428.2 ng/L, respectively.

Improving Water Reuse for a Healthier Potomac Watershed 39 2.5 Effect of Urban BMPs on Nutrients and EDCs Reductions 2.5.1 Description of Sampling Site and Urban BMPs The paired urban streams with/without BMPs (Paint Branch vs. Brier Ditch) are located in the Anacostia watershed of Maryland Montgomery and Prince George’s County (Figure 2-1). Both watersheds are dominated by urban land use but only Paint Branch was restored with urban BMPs. In 1995, the Montgomery County Council designated the upper Paint Branch watershed as a Special Protected Area, under a county law that provides additional protections to designated sensitive areas (Montgomery County Department of Environmental Protection 2002). The Maryland National Parks and Planning Commission purchased undeveloped land along the stream to maintain shaded habitat and leaf litter inputs and implemented regulations that limited impervious area to 10% of all new development in the watershed. The Briar Ditch watershed has not implemented BMPs, and is of the most densely populated regions of the area with the fewest stormwater management controls. Only 13% of the stream has adequate riparian forest buffer compared to Paint Branch’s 53% (http://www.anacostia.net/Subwatershed/Brier_Ditch.html). 2.5.2 Effect of Urban BMPs on Nutrient Reductions Results of water temperature, conductivity, nutrients and DOC measurements of the agricultural streams with/without BMPs are summarized in Figure 2-11 and Table 2-4. No significant difference was observed in water temperature, conductivity or TDN between streams with and without BMPs (p > 0.05, one-way ANOVA), despite of apparent seasonal patterns (lower water temperature but higher dissolved oxygen and conductivity in winter). The higher conductivity in winter (not observed in agricultural streams) was probably due to use of deicers. The restored agricultural streams were consistently lower in SRP and DOC (one exception) than that in the unrestored streams (Figure 2-11). The fluxes of SRP and DOC of the restored streams were estimated to be 0.65 and 40.8 kg/km2/yr, showing reductions by 84% and 30% relative to the fluxes of unrestored streams (0.10 and 28.7 kg/km2/yr; Table 2-4). The consistent lower concentrations and lower fluxes of SRP and DOC than the unrestored site suggest that the BMPs at this agricultural site can actually retain both P and organic C. Different from SRP or DOC, TDN concentrations at the restored site was comparable to that of the unrestored stream, and TDN flux from the restored and the unrestored stream was almost the same (Figure 2-11 and Table 2-4). This lack of difference in TDN between the two streams suggests that the urban BMP in this study did not increase TDN retention, probably due to lower TDN levels in urban streams (0.5-1 mg/L) relative to the agricultural streams (3-8 mg/L).

40 The Water Research Foundation 25 No BMP_Urban 1.5 ) 1 20 BMP_Urban - 1

C) 15 ⁰

t ( 10 0.5

5 L (mg TDN 0 0

20 120 ) ) 1 - 1

15 - 80 10 40

5 SRP (g L DO (mg L DO (mg 0 0

1200 8 ) ) 1 1 - - 6 800 4 400

SC (µS cm 2 DOC (mg DOC (mg L

0 0 Oct Nov Dec Jan Feb Mar Apr May June Oct Nov Dec Jan Feb Mar Apr May June Figure 2-11. Comparison of Temperature (t), Dissolved Oxygen (DO), Specific Conductance (SC), Total Dissolved Nitrogen (TDN), Soluble Reactive Phosphorus (SRP) and Dissolved Organic Carbon (DOC) Between Two Urban Streams Without or With Best Management Practices (BMPs).

Table 2-4. Estimates of Nutrient and DOC Yields from the Two Urban Streams Without or With Best Management Practices (BMPs). No BMP BMP Reduction (%) TDN (kg/km2/yr) 8.95 9.08 -1 SRP (kg/km2/yr) 0.65 0.10 84 DOC (kg/km2/yr) 40.8 28.7 30

2.5.3 Effect of Urban BMPs on EDCs Reductions Figure 2-12 shows that there was no significant difference in their concentration of E1 or E1-S between the two streams (p > 0.05, one-way ANOVA). However, the fluxes of E1 and E1-S of the restored streams were lower that of the unrestored stream by 34% and 17% (Table 2-5). The effect of the urban BMPs on estrogen retention warrants further investigations by improving the estrogen analytical method. At the unrestored urban stream (without BMPs), the highest values of the three herbicides (atrazine, metolachlor and simazine) occurred in May, followed by June (Figure 2-12). This timing matched well with application of these herbicides in this region (Dernoeden 2005). In the restored urban stream, however, such a concentration peak was not observed. As a result, the fluxes of the atrazine, metolachlor and simazine from the restored stream (0.04, 0.22 and 0.015 ng/L) were lower than that from the unrestored stream (0.76, 0.82 and 0.13 ng/L), showing decreases by 94%, 73% and 88%, respectively (Table 2-5). Therefore, these three herbicides were largely retained at the restored site mainly during the period of May to June. Only in the restored stream, a peak of metolachlor or prometon was observed in October (165 and 168 ng/L), and high values occurred throughout until December (Figure 2-12). As a result, the fluxes of prometon at the restored site was even higher than that at the unrestored site (Table 2-5). The reason for the high values of metolachlor and prometon during this period warrants further examination.

Improving Water Reuse for a Healthier Potomac Watershed 41 Table 2-5. Estimates of EDCs Yields from the Two Urban Streams Without or With Best Management Practices (BMPs). No BMP BMP Reduction (%) E1 (g/km2/yr) 0.0011 0.0007 34 E1-S (g/km2/yr) 0.0020 0.0017 17 4-Nonylphenol (g/km2/yr) 4.996 1.114 78 Atrazine (g/km2/yr) 0.76 0.04 94 Metolachlor(g/km2/yr) 0.82 0.22 73 Prometon (g/km2/yr) 0.051 0.214 -318 Simazine (g/km2/yr) 0.133 0.015 88 Imidacloprid (g/km2/yr) 0.217 0.129 41 Fipronil (g/km2/yr) 0.043 0.026 39 Dinotefuran (g/km2/yr) 0.014 0.019 -35 Acetamiprid (g/km2/yr) 0.019 0.013 33 Clothianidin (g/km2/yr) 0.023 0.018 24

4-Nonylphenol and all four insecticides were constantly detected in urban streams, and the concentrations of imidacloprid and Dinotefuran were significantly higher relative to that in the agricultural streams (Figure 2-12). This suggests that more midacloprid and dinotefuran were used in urban subwatersheds than in agricultural ones. In general, there was no apparent seasonal pattern in the insecticides, except that 1) highest levels of imidacloprid, fipronil and clothianidin occurred in May at the unrestored site, while Acetamiprid were highest in January and February at both sites. The reasons for the former probably was related to the timing of using these insecticides. No significant difference was observed in the concentrations of 4-nonylphenol or pesticides between the urban streams with and without BMPs (p > 0.05). However, the stream with BMPs was lower in the fluxes of 4-nonylphenol (by 78%), atrazine (94%), imidacloprid (41%), fipronil (39%), metolachlor (73%), simazine (88%), acetamiprid (33%), and clothianidin (24%). The fluxes of prometon and dinotefuran of the stream with BMPs were higher than that of the stream without. This suggests the BMPs in this study enhanced retention of 4-nonylphenol and most of the pesticides.

42 The Water Research Foundation 2 No BMP_Urban 2 ) 1 ) - 1 1.5 BMP_Urban 1.5 -

ng L 1 1 S (ng L -

E1 ( 0.5 0.5 E1 0 0

1000 400 ) ) 1 1 - - 800 300 600 200 400 100 200 Atrazine (ng L (ng Atrazine

Nonylph. (ng L (ng Nonylph. 0

- 0 4

200 400 ) ) 1 1 - - 300 150 (ngL (ng L 200 100 100 50 Metolach. Metolach. 0 Prometon 0

50 4 ) 1 ) - 1 - 40 3

(ng L 30 2 20 1 10 Simazine 0 Clothianid. (ng (ng L Clothianid. 0 ) ) 1 1 - - 8 5 6 4 3 4 2 2 Fipronil (ng L 1

0 L (ng Acetamipr. 0 ) ) 1 -

40 1

- 2.5 30 2 1.5 20 1 Imidaclo. (ng L (ng Imidaclo.

10 (ng L Dinotef.. 0.5 0 0 Oct Nov Dec Jan Feb Mar Apr MayJune Oct Nov Dec Jan Feb Mar Apr MayJune Figure 2-12. Comparison of Estrone (E1), Estrone Sulfate (E1-S), and Other Common EDCs (mainly Pesticides) between Two Urban Streams Without or With Best Management Practices (BMPs).

Improving Water Reuse for a Healthier Potomac Watershed 43 2.6 Effect of WWTPs Discharges on Nutrients and EDCs in Receiving Waters 2.6.1 Description of Seneca WWTP and Blue Plains WWTP Two WWTPs of different sizes (Blue Plains WWTP of Washington DC and Seneca WWTP of Germantown, Maryland) were sampled for point sources of EDCs and nutrients. The Blue Plains (BP) WWTP serves over two million customers with a collection area of Washington DC and surrounding suburbs of Maryland and Virginia. It is the largest treatment plant in the Potomac River watershed and the largest treatment facility of its kind in the U.S., with a rated capacity of 384 million gallons per day (MGD). The treatment process utilizes preliminary and primary treatment, secondary treatment, nitrification/denitrification, effluent filtration, chlorination- dechlorination and post aeration (https://www.dcwater.com/about/facilities.cfm). In the last several years, the plant has dramatically upgraded and improved its liquid processing systems through implementation of anaerobic digestion (https://www.dcwater.com/sites/default/files/Blue_Plains_Plant_brochure.pdf). Consequent improvements in water quality of the Potomac River were recently reported by Pennino et al. (2016). The Seneca WWTP is located in Germantown, Maryland and discharges effluent to Seneca Creek of Maryland. Compared with the Blue Plains WWTP, it is much smaller (20 MGD). The Seneca WWTP was also updated to increase nutrient removal, and the new wastewater treatment plant uses state-of-the- art biological and chemical processes that remove 64% more nitrogen along with 77% more phosphorous than the original 5-MGD plant (http://www.specializedengineering.com/index.php/project/seneca-wastewater-treatment-plant/). 2.6.2 Effect of WWTPs Discharges on Nutrients of the Receiving Waters In general, the effluents of the Seneca WWTP were higher in water temperature (t) and specific conductance (SC) but lower in dissolved oxygen (DO) than the Seneca Creek up (p <0.05, one-way ANOVA). As a result, the discharges of the WWTP effluents caused downstream increases in water temperature and conductivity in the receiving water – Seneca Creek (Figure 2-13). The exceptions for temperature and DO occurred in May and June, when stream temperatures were comparable to that of the effluent. The exception of SC occurred during February, when SC of the stream was even higher due to use of deicers in watershed for snow melting. Concentrations of SRP and DOC in effluents from the Seneca WWTP were consistently higher than that in the Seneca Creek at the upriver site, while concentrations of TDN were not (Figure 2-13). As a result, SRP and DOC concentrations in Seneca Creek increased by 24 to 1395% and 15 - 152% from upriver to downriver sites, suggesting significant effect of the WWTP effluent on the phosphorus and organic carbon of the Seneca Creek. In contrast, downriver changes in TDN were inconsistent but usually lied between the values of the WWTP effluent and the upstream values (Figure 2-13), suggesting relative minor effect of WWTP on the Seneca Creek relative to SRP or DOC.

44 The Water Research Foundation 25 3

up Effluent down ) 1

20 - 2

C) 15 ⁰

t ( 10 1

5 (mg TDN L 0 0

20 160 ) ) 1 - 15 1 120 - 10 80 SRP (g L (g SRP

DO (mg L (mg DO 5 40 0 0

1200 6 ) ) 5 1 1 - - 800 4 3 400 2 SC (µS cm DOC L DOC (mg 1 0 0 Oct Nov Dec Jan Feb Mar Apr MayJune Oct Nov Dec Jan Feb Mar Apr May June Figure 2-13. Downstream Changes in Water Temperature (t), Dissolved Oxygen (DO), Specific Conductance (SC), Total Dissolved Nitrogen (TDN), Soluble Reactive Phosphorus (SRP) and Dissolved Organic Carbon (DOC) in the Seneca Creek after Effluent Discharges from Seneca WWTPs.

Improving Water Reuse for a Healthier Potomac Watershed 45 By collecting streamflow data and flow of WWTP effluent, fluxes of water, nutrients and DOC during the study period were estimated, and percentage of inputs to the downstream segment from the upstream segment and the WWTP were calculated. Results show that Seneca Creek upstream and the Seneca WWTP contributed approximately 60% and 40% of the Seneca flow at this site (Figure 2-14). The contributions of TDN from upstream and the WWTP were similar to that of water, because the TDN concentrations of the upstream and the WWTP effluent were comparable. The WWTP contributed more SRP (79%) and DOC (60%) to the Seneca Creek relative to water flow (40%) at this site, due to consistently higher SRP and DOC concentrations in the WWTP effluents (Figure 2-13).

water TDN

Effluent Effluent 40% 38%

Stream Stream 60% 62%

SRP DOC Stream 21%

Stream 40%

Effluent 60%

Effluent 79%

Figure 2-14. Estimated Contributions of Water, Total Dissolved Nitrogen (TDN), Soluble Reactive Phosphorus (SRP), and Dissolved Organic Carbon (DOC) from the Seneca WWTP or the Upstream Seneca Creek.

46 The Water Research Foundation The effluents of the Blue Plains WWTP were consistently higher in water temperature (t), SC, TDN, SRP and DOC but lower in DO than the Potomac River upriver. In general, we observed very small downstream differences (a few exceptions) in the above water quality variable (or downstream decrease in DO) were observed in the receiving water - the Potomac River (Figure 2-15). This suggests that the Blue Plains WWTP effluent has a minor impact on water quality of the Potomac River.

25 5

up Effluent down ) 1

20 - 4

C) 15 3 ⁰

t ( 10 2

5 (mg TDN L 1 0 0

20 120 ) ) 1 - 15 1 - 80 10 40 SRP (g L (g SRP

DO (mg L (mg DO 5 0 0

1200 8 ) ) 1 1 - - 6 800 4 400 SC (µS cm

DOC L DOC (mg 2

0 0 Oct Nov Dec Jan Feb Mar Apr May June Oct Nov Dec Jan Feb Mar Apr May June

Figure 2-15. Downstream Changes in Water Temperature (t), Dissolved Oxygen (DO), Specific Conductance (SC), Total Dissolved Nitrogen (TDN), Soluble Reactive Phosphorus (SRP) and Dissolved Organic Carbon (DOC) in the Potomac River After Effluent Discharges from the Blue Plains WWTPs.

Improving Water Reuse for a Healthier Potomac Watershed 47 By collecting streamflow data and flow of WWTP effluent, fluxes of water, nutrients and DOC during the study period were estimated, and percents of inputs from the upstream Potomac and the Blue Plains WWTP were calculated. Results show that Potomac upstream and the Blue Plains WWTP contributed approximately 92% and 8% of the Potomac River flow at this site (Figure 2-16). The Blue Plains WWTP contributed more TDN (13%), SRP (14%) and DOC (12%) to the Potomac River relative to its water flow (8%), due to consistently higher TDN, SRP and DOC concentrations in the WWTP effluents than the upstream Potomac (Figure 2-15).

water Effluent TDN Effluent 8% 13%

River River 87% 92%

SRP Effluent DOC Effluent 14% 12%

River River 86% 88%

Figure 2-16. Estimated Contributions of Water, Total Dissolved Nitrogen (TDN), Soluble Reactive Phosphorus (SRP), and Dissolved Organic Carbon (DOC) from the Blue Plains WWTP or Upstream River in the Potomac River.

2.6.3 Effect of WWTPs Discharges on EDCs of Receiving Waters Only estrone (E1) and estrone-sulfate (E1-3S) of estrogenic EDCs (EEDCs) were detected in the WWTP effluents and the receiving streams. The concentrations of E1 and E1-3S were (< 1.5-2 ng/L) close to detection limits, and the effect of Seneca WWTP effluents on EEDCs on the receiving water is not clear (Figure 2-17). Atrazine, metolachlor, prometon, and simazine were four common herbicides detected in the Seneca Creek WWTP. For the first three pesticides, these concentrations were much higher in May, during which their concentrations in the WWTP effluent were usually higher than the Seneca Creek and discharge of the Seneca WWTP effluent caused downstream increases in their concentration in the Seneca Creek. During October through April, the concentration of atrazine and metolachlor (currently used) were usually lower in the WWTP effluent than in the Seneca Creek, and the discharge of the Seneca WWTP effluent caused downstream decreases in these concentrations in the Seneca Creek (Figure 2-17). This suggests that Seneca WWTP was not an important source of atrazine and metolachlor of the Seneca Creek, and stormwater may carry large amount of atrazine, metolachlor and prometon into the WWTP during the May high-flow period and discharged into the Seneca Creek. Monthly variability of prometon was not as large as the three others. Prometon and simazine were consistently higher in WWTP effluent than the Seneca Creek across seasons, and the discharge of the WWTP effluent usually caused downstream increases in their concentrations in the Seneca Creek (Figure 2-17).

48 The Water Research Foundation Like prometon, monthly variability of 4-nonylphenol and all the insecticides (Clothianidin, Acetamiprid, Dinotefuran, Fipronil, and Imidacloprid) in the Seneca Creek and WWTP were relatively minor. Their concentrations were consistently higher (one exception for 4-nonylphenol) in WWTP effluent than the Seneca Creek across seasons, and the discharge of the WWTP effluent usually caused downstream increases in their concentrations in the Seneca Creek (Figure 2-17). Fluxes of EDCs and the four pesticides during the study period were estimated, and percent of inputs from the upstream segment of Seneca Creek and the Seneca WWTP was calculated. Results show that Seneca WWTP contributed 39%-47% of E1-3S, atrazine and 4-nonylphenol (close to its percent of water 40%) (Figure 2-18), suggesting minor effect of the WWTP on these EDCs (Figure 2-18). On the other hand, Seneca WWTP contributed more E1 (68%), prometon (63%), simazine (76%), metolachlor (96%), clothianidin (53%), Acetamiprid (72%), Dinotefuran (89%), Fipronil (94%), and Imidacloprid (73%) to the stream than water flow (40%) (Figure 2-18), suggesting that Seneca WWTP was an important source of these EDCs to the Seneca Creek. Relative more inputs of these EDCs came from the Seneca WWTP than from the upstream segment. This was consistent with higher concentrations in WWTP effluent relative to the upstream segment of Seneca Creek.

Improving Water Reuse for a Healthier Potomac Watershed 49 2 Up 1.5 ) 1 ) - 1 1.5 - Effluent 1 Down

ng L 1 S (ng L (ng S - 0.5

E1 ( 0.5 E1 0 0

800 500 ) ) 1 1 - - 600 400 300 400 200 200 100 Atrazine(ng L

Nonylph.(ng L 0 - 0 4

6 500 ) ) 1 1 - - 400 4 (ngL 300 L (ng 200 2 100 Metolach. 0 Prometon 0

200 20 ) ) 1 - 1 - 150 15 (ng L (ng 100 10

50 5 Simazine Clothianid.(ng L 0 0 ) ) 1 1 - - 50 30 40 20 30 20 10

Fipronil L (ng Fipronil 10

0 Acetamipr. (ng L 0 ) ) 1 -

100 1 - 30 80 25 60 20 15 40 10 Dinotef. (ng L (ng Dinotef. Imidaclo. (ng (ng L Imidaclo. 20 5 0 0 Oct Nov Dec Jan Feb Mar Apr MayJune Oct Nov Dec Jan Feb Mar Apr MayJune Figure 2-17. Downstream Changes in Estrone (E1), Estrone Sulfate (E1-S), and Other EDCs (Common Pesticides in the Seneca Creek after Effluent Discharges from the Seneca WWTP.

50 The Water Research Foundation E1 E1-S 4-Nonylphenol Imidacloprid

River Stream 27% 32% Effluent Effluent 39% 44% Stream 56% River Effluent 61% 68% Effluent 73%

Atrazine MetolachlorStream Clothianidin Fipronil River 4% 6%

Effluent River Stream 46% 47% Effluent 54% 53%

Effluent Effluent 96% 94%

Prometon Simazine Acetamiprid Dinotefuran River Stream 11% River 24% Stream 28% 37%

Effluent 63% Effluent Effluent 76% 72% Effluent 89%

Figure 2-18. Estimated Contributions of Estrone (E1), Estrone Sulfate (E1-S), and Other EDCs (Common Pesticides) from the Seneca WWTP or Upstream Stream in the Seneca Creek.

For Blue Plains WWTP, the concentrations of E1 and E1-S were also close to detection limits (<1.5 ng/L), and the effect of Seneca WWTP effluents on EEDCs on the receiving water is not clear (Figure 2-19). Monthly variability of 4-nonylphenol in the Blue Plains WWTP was not apparent, and its concentration in Blue Plains WWTP were not consistently higher than that in the Potomac River (Figure 2-19). This suggests Blue Plains WWTP was not an important source to the Potomac River. Similar to Seneca Creek and WWTP, three herbicides (atrazine, metolachlor, and simazine) in Blue Plains WWTP and Potomac River showed apparent seasonal pattern – extremely higher during May and June high-flow period (Figure 2-19). This corresponded to the timing of herbicide use. However, their concentrations in the WWTP effluent were not consistent higher (or lower) than in the receiving water, and therefore no consistent downriver changes were did not observed in the Potomac River (Figure 2-19). Prometon and all insecticides (clothianidin, acetamiprid, dinotefuran, fipronil, and imidacloprid) was consistent higher in the Blue Plains WWTP effluent than the receiving water, and there were consistent downstream increases in the Potomac River (Figure 2-19). No apparent seasonal pattern was observed in the concentrations of prometon or any insecticides.

Improving Water Reuse for a Healthier Potomac Watershed 51 2 Up 0.6 ) 1 ) - 1 1.5 - Effluent 0.4 Down

ng L 1 S (ng L (ng S - 0.2

E1 ( 0.5 E1 0 0

400 250 ) ) 1 1 - - 300 200 150 200 100 100 50 Atrazine(ng L

Nonylph.(ng L 0 - 0 4

100 200 ) ) 1 1 - - 150 80 (ngL (ng L (ng 60 100 40 50 20 Metolach. 0 Prometon 0

200 15 ) ) 1 - 1 - 150 10 (ng L (ng 100 5 50 Simazine Clothianid.(ng L 0 0 ) ) 1 1 - - 25 15 20 15 10 10

Fipronil L (ng Fipronil 5 5 Acetamipr. (ng L 0 0 ) ) 1 -

80 1 - 40

60 30

40 20 Dinotef. (ng L (ng Dinotef. Imidaclo. (ng (ng L Imidaclo. 20 10

0 0 Oct Nov Dec Jan Feb Mar Apr MayJune Oct Nov Dec Jan Feb Mar Apr MayJune Figure 2-19. Downstream Changes in Estrone (E1), Estrone Sulfate (E1-S), and Other EDCs (Common Pesticides) in the Potomac River after Effluent Discharges from the Blue Plains WWTP.

52 The Water Research Foundation Blue Plains WWTP contributed less percent of E1, atrazine, metolachlor, simazine and 4-nonylphenol to the Potomac River (4%-7%) than the percentages of water (8%) (Figure 2-20), suggesting minor effect of the Blue Plains WWTP discharge on the Potomac River. The percentages of E1-3S (18%), prometon (33%) and all insecticides (14-79%) were higher than that of water flow (Figure 2-20), suggesting that Blue Plains WWTP was an important source of these EDCs to the Potomac River. This is consistent with consistently higher concentrations of prometon and all insecticides in the Blue Plains WWTP effluent than in the Potomac River. In particular, the inputs of dinotefuran, fipronil, and imidacloprid from the Blue Plains WWTP effluent accounted for 50-79% of their loads in the Potomac River (Figure 2-20), far higher than water contribution of the Blue Plains WWTP (8%). Reason for the higher contributions of prometon and all insecticides in the Blue Plains WWTP relative to the Potomac River warrant further investigation.

E1 Effluent E1-S 4-NonylphenolEffluent Imidacloprid 5% Effluent 7% 18%

River Effluent 50% 50%

River River 82% River 95% 93%

Atrazine Effluent Metolachlor Effluent Clothianidin Fipronil Effluent 4% 4% 14% River 25%

Effluent River 75% River River 86% 96% 96%

Prometon Simazine Effluent Acetamiprid Dinotefuran 5% River 21% Effluent Effluent 33% 31%

River River 67% 69% Effluent 79% River 95%

Figure 2-20. Estimated Contributions of Estrone (E1), Estrone Sulfate (E1-S), and Other EDCs (Common Pesticides) from the Blue Plains WWTP or the Upstream River in the Potomac River.

Improving Water Reuse for a Healthier Potomac Watershed 53 2.7 Estimating Sources of Nutrients, DOC, and EDCs from Point and Nonpoint Sources to the Potomac River 2.7.1 Fluxes of Nutrients, DOC, and EDCs from Point Sources Flow-averaged concentrations of nutrients, DOC and EDCs in the effluents of the 3 WWTPs are listed in Table 2-6. UOSA was far higher in TDN, while the highest SRP, E1, and E1-S values were observed in Seneca WWTP. In general, 4-nonylphenol, most herbicides (atrazine, metolachlor, and simazine) and two insecticides (Acetamiprid and Fipronil) were highest in Seneca WWTP, followed by Blue Plains WWTP, and lowest in UOSA. The rest of pesticides (prometon, Clothianidin, Dinotefuran and Imidacloprid) were highest in Blue Plains WWTP. Flow-averaged concentrations for the total, using mean daily effluent discharge of the WWTPs, are also listed. Because the Blue Plains WWTP dominates the water flow, the flow-averaged concentrations were close to that of the Blue Plains WWTP. Table 2-6. Flow-Averaged Concentrations of EDCs in the Effluents of Three Wastewater Treatment Plants. UOSA refers to Upper Occoquan Service Authority. Seneca Blue Plains Flow-Average Units WWTP WWTP UOSA for Total Flow (MGD) 20 384 32 TDN (mg/L) 1.97 2.61 10.0 3.13 SRP (ug/L) 93.9 56.2 69.8 58.9 DOC (mg/L) 4.66 5.97 3.32 5.71 E1 (ng/L) 0.22 0.06 0.11 0.07 BLYES (ng/L) 0.45 0.33 - 0.34 Total pesticides (ng/L) 179.6 75.2 23.0 76.1 Atrazine (ng/L) 63.8 22.6 10.90 23.6 Metolachlor (ng/L) 62.2 21.2 5.16 21.9 Simazine (ng/L) 51.4 14.0 3.63 15.0 Prometon (ng/L) 2.20 17.4 3.34 15.7 4-Nonylphenol (ng/L) 190 125 84 125 Clothianidin (ng/L) 10.3 11.4 10.0 11.2 Acetamiprid (ng/L) 7.4 6.6 2.6 6.3 Dinotefuran (ng/L) 14.0 26.6 21.1 25.6 Fipronil (ng/L) 30.3 8.8 0.5 9.2 Imidacloprid (ng/L) 60.6 61.6 33.3 59.5

It is assumed the flow-averaged concentrations of nutrients, DOC and EDCs of the three WWTPs represented all WWTPs in the Potomac watershed, and used the data to calculate annual fluxes of nutrients, DOC and EDCs from point sources, assuming WWTP effluents dominated the point sources. Results show that WWTP effluents accounted for 9% of the total water flow (Figure 2-7). TDN, SRP and DOC from point sources (mainly WWTPs) accounted for 5-9% of the total inputs, comparable or less than the contribution of water flow (9%). Contributions of E1, estrogenic activities (BLYES), and three common pesticides (atrazine, metolachlor, and simazine) and total pesticides from point sources were even less (<2%), once again, suggesting point sources were not an important to nutrients, DOC or EDCs of the Potomac River. The only exception, prometon from point sources accounted for 14% of the total inputs.

54 The Water Research Foundation Moreover, agricultural inputs dominated over other sources nutrients and pesticides, including 70-71% of nutrients, 40-49% of E1 and BLYES, and 84% of total herbicides (Table 2-7). Inputs of nutrients and EDCs from urban runoff were also important (9-33%), considering its 1% of water input. Future studies on nutrients and EDCs control should focus on nonpoint sources.

Table 2-7. Estimated Annual Fluxes of Water, Nutrients, Organic Carbon, Estrogen and Herbicides from Forest, Agricultural Runoff (Cropland and Pasture), and Urban Runoff, and Urban Point Sources (WWTPs) in the Potomac River. Forest Cropland Pasture Urban WWTP WWTP% Agr.% Urban% Water 109m3/yr 4.19 0.49 1.63 0.07 0.65 9 30 1 TDN 106 kg/yr 5.16 21.15 8.51 5.51 2.03 5 70 13 SRP 106 kg/yr 0.02 0.18 0.14 0.07 0.04 9 71 16 DOC 106 kg/yr 15.29 16.16 15.69 29.22 3.7 5 40 36 E1 103kg/yr 0.0021 0.0027 0.0032 0.004 0.00005 0 49 33 BLYES 103kg/yr 0.0044 0.0079 0.0156 0.0058 0.0002 1 69 17 Total 103kg/yr 0.03 1.76 0.9 0.43 0.05 2 84 14 herbicides Atrazine 103kg/yr 0.01 1.35 0.47 0.18 0.02 1 90 9 Metolachlor 103kg/yr 0.01 0.2 0.11 0.11 0.01 2 70 25 Simazine 103kg/yr 0 0.2 0.3 0.09 0.01 2 83 15 Prometon 103kg/yr 0 0.02 0.03 0.01 0.01 14 71 14

1. Total herbicides include 18 herbicides that were dominated by atrazine, metolachlor, simazine and prometon, which is different from total pesticides listed in Table 2-6. 2. Data for all nonpoint sources were from year one report of this project. 3. The concentrations of nutrients, DOC, estrogen, and herbicides were flow-averaged measurements conducted during year 2016-2017 at subwatersheds with dominant land use of forest, cropland, pasture, and urban. 4. Water discharge of the Potomac River was from daily flow measurement of the Potomac River at Little Falls by USGS. Water discharges from each type of land use were Estimated from land use percent of the Potomac watershed (http://www.washingtonpost.com /wp-srv/metro/daily/ 111307/fullreport.pdf), and assumed surface runoff was evenly distributed across land use. 5. WWTP flow of the Potomac watershed was estimated from http://www.washingtonpost.com/wp- srv/metro/daily/111307/fullreport.pdf.

2.7.2 Changes in Relative Contributions of Nutrients, DOC, and EDCs from Nonpoint Sources With BMPs Applications Table 2-8 list effectiveness of the agricultural and urban BMPs at Maryland site, Virginia site and mean values. Because only two sites were collected for each type of land use and not all BMPs were examined in this study, there could be large errors in the mean effectiveness. Using the effectiveness of above table, the annual fluxes of nutrients, DOC and EDCs from point and nonpoint sources were recalculated (Table 2-9). For negative values that indicate the BMPs did not work, the fluxes of contaminants were not changed. Results showed the contribution from agricultural inputs of nutrients, DOC and pesticides (except prometon) (in bond font) if the BMPs of this study were applied to the whole Potomac watershed. Nutrients inputs from agriculture would decrease from 70- 71% to 59-60%, and E1, Atrazine Simazine Prometon inputs from agriculture would decrease from 49%, 90%, 70% and 83% to 27%, 77%, 60% and 78%, respectively. The relative contributions from urban inputs would not decrease with application of the BMPs except slight decrease in SRP input.

Improving Water Reuse for a Healthier Potomac Watershed 55 Table 2-8. Effectiveness (%) in Reduction Nutrients, DOC and EDC With the Agricultural and Urban BMPs at the Maryland Sites, the Virginia Sites. Negative values refer to the case that values at site with BMPs were higher than the site without BMPs, or suggest the BMPs did not work in reducing contaminants. Agricult. BMP Urban BMP MD VT mean MD VT mean TDN 24 52 38 -1 -26 -14 SRP 61 40 51 84 6 45 DOC 26 22 24 30 5 18 E1 99 39 69 34 28 31 E1-S 100 82 91 17 -37 -10 Atrazine 69 85 77 94 2 48 Metolachlor 15 85 50 73 -4 34 Prometon -213 59 -77 -318 -70 -194 Simazine 2 56 29 88 -241 -77

Table 2-9. Annual Fluxes of Nutrients, DOC and EDCs from Point and Nonpoint Sources When Agricultural and Urban BMPs of This Study Applied to All the Potomac Watershed (Table 2-8). Numbers with * indicate decrease in contributions with application of the BMPs of this study to the whole watershed. Forest Cropland Pasture Urban Point Point% Agr.% Urban% Water 109m3 4.19 0.49 1.63 0.07 0.65 9 30 1 TDN 106kg/yr 5.16 13.11 5.28 5.51 2.03 7 59* 18 SRP 106kg/yr 0.02 0.09 0.07 0.04 0.04 16 61* 15* DOC 106kg/yr 15.3 12.3 11.9 24.0 3.7 6 36* 36 E1 103kg/yr 0.002 0.0008 0.001 0.003 0.00005 1 27* 41 Atrazine 103kg/yr 0.01 0.311 0.11 0.09 0.02 4 77* 17 Metolachlor 103kg/yr 0.01 0.10 0.055 0.083 0.01 4 60* 32 Simazine 103kg/yr 0 0.14 0.21 0.09 0.01 2 78* 20 Prometon 103kg/yr 0 0.02 0.03 0.01 0.01 14 71 14

56 The Water Research Foundation CHAPTER 3 Impacts and Outcomes of Current Nutrient Management Strategies on Source Controls of EDCs at Virginia Sites

3.1 Experiment Design 3.1.1 Site Selection In order to examine the effect of agricultural and urban BMPs, one pair of agricultural streams (Furrs Run and Elk Run) and two sites of an urban stream (in Cub Run) in the Occoquan watershed were selected. Wastewater effluent from one WWTP - UOSA was collected to examine the effect of wastewater treatment plants with advanced reclamation. Stream water samples were collected upstream and downstream of the UOSA contribution to Bull Run. All the sampling sites in Virginia are presented in Figure 3-1. At each of the 10 locations shown in Figure 3-1, water samples were collected and analyzed for nutrients and endocrine disrupting compounds (EDCs), which includes estrogens, and a number of synthetic organic compounds (SOCs) including herbicides, SOCs, 4-nonylphenol and BPA. In situ water quality parameters, such as temperature, were also measured when sampling.

Improving Water Reuse for a Healthier Potomac Watershed 57

Figure 3-1. Virginia Sampling Locations.

3.1.2 Water Sample Collection Water sampling was completed monthly from October 2017 to September 2018, for a total of 12 months. Details for water sampling were described in Chapter 2. Once water samples were collected, 500 mL of the samples were sent to University of Maryland for analyses of nutrients and organic carbon in the same day. Water samples were stored on wet ice during transportation. Water samples were filtered through pre-weighed Whatman GF/F filters on the same day of collection. The filtrates were either stored at 4°C until extraction of estrogens, or at -20°C for the other analyses (e.g., nutrients and DOC). Sample collection and processing were generally completed within 48 hours. Samples for EDC analyses were prepared at Virginia Tech’s (VT’s) Occoquan Watershed Monitoring Laboratory (OWML), and then sent to the University at Buffalo and USGS directly. Two trip blanks and one field blank were taken while sampling. The trip blanks were sample bottles filled with Milli-Q® water carried throughout the sampling events, unopened, in April and August. The field blank was collected in September. A sample bottle filled with Milli-Q® water was carried throughout the sampling event close, but was opened at the Bull Run Downstream sampling location. No estrogens

58 The Water Research Foundation were detected in the blanks. SOCs were detected in the blanks, but the concentrations of most of the SOCs detected were very low except for 4-nonylphenol. The SOC concentrations found in the blanks were compared to the sample results to confirm the validity of the results. The April trip blank indicated error as it is ~15% of the total SOC concentrations that month, mainly in 4-nonylphenol. The 4- nonylphenol concentration found in the trip blank in April is higher than all other sample concentrations. However, the associated lab blank (field blank) at the same time was fine. Moreover, no 4-nonlyphenol concentrations were detected in the trip blanks in September. 3.1.3 Chemical Analyses The methods for water quality analyses for nutrients (including total dissolved nitrogen and soluble reactive phosphorus), estrogen and SOCs are described in Chapter 2. Soluble reactive phosphorus (SRP) concentrations results by the UMD lab were compared to the SRP results produced by the OWML. The comparison was done for three sampling locations: the Griffith Intake which corresponded with OWML ST01, the Bull Run Downstream site which corresponded with OWML ST45, and the Cub Run Downstream site which corresponded with OWML ST50 (Figure 3-2). This comparison was performed to check unusually high SRP values that were reported by the UMD lab in some samples collected from October to January. These samples were rerun by UMD and the graphs show that the rerun SRP values throughout the entire sampling period were more in line with those reported by OWML. The samples of OWML and UMD were not collected at the same day, so some small differences exist.

Improving Water Reuse for a Healthier Potomac Watershed 59

Figure 3-2. Griffith (ST01), Bull Run Downstream (ST45), Cub Run Downstream UMD SRP vs. All OWML SRP.

60 The Water Research Foundation 3.2 Analyses of SOCs and Estrogen 3.2.1 SOCs In order to understand the SOCs that were commonly present throughout the sampling period Figure 3-3 is presented. The upper panel shows the total number of detections for each SOC in order from most to fewest. Since the list of SOCs being analyzed changed slightly in January 2018, it was important to understand the percent detection. A 50% detection frequency was used as a cutoff for further SOC evaluation. Alachlor, butylat, coumaphos, dieldrin, endrin aldehyde, and methyl parathion, were never observed above detection and were not included in any further analysis. Non-detect data were not presented graphically, and were removed from further data analysis, so average concentrations presented throughout do not included non-detect samples. At least one duplicate was taken per monthly sampling effort and averaged for presentation. The GC-MS and LC-MS analyzed data were combined into one data set, with the month of January being an average of the two methods per SOC since both analyses were conducted.

Figure 3-3. SOCs Arranged by Detection frequency and by Percent Detection.

Improving Water Reuse for a Healthier Potomac Watershed 61

Figure 3-4. Concentration of Frequently (> 50%) Detected SOCs. The data are presented on a log scale.

62 The Water Research Foundation Because of their presence in the samples collected and of interest in the Chesapeake Bay region, 4- nonylphenol and atrazine concentrations were analyzed further (Figure 3-5 and 3-6). They were graphed by the concentration at each sampling location per month. 4-Nonylphenol is a persistent, hydrophobic industrial compound (U.S. EPA 2019) that was tested for from January to September 2019. 4- Nonylphenol was consistently detected in samples and was comparatively high in concentration. High levels are seen especially in the summer months. Atrazine concentrations increased in the months of May and June, which is not surprising since it is applied as an herbicide in the spring. The MCL for atrazine specified by the U.S. EPA is 0.003 mg/L, or 3000 ng/L. The maximum concentration of atrazine detected is 394 ng/L, which is a factor of 10 less than the MCL. A summary study on vertebrates indicated that atrazine feminizes 5-50% of male fish at concentrations ranging between 21 and 2160 µg/L (higher than the EPA MCL), between 10-50% of amphibians at concentrations between 0.1 and 100 µg/L, and at 0.5 µg/L it feminizes about 10% of reptiles at a water temperature of 26°C but 40% of reptiles at a temperature of 29.2°C (Hayes et al. 2011).

Figure 3-5. Monthly 4-Nonylphenol Concentrations at Virginia Sampling Locations.

Improving Water Reuse for a Healthier Potomac Watershed 63 Atrazine Concentrations Continued

100 90 80 70 60 50 40 30 20

Atrazine (ng/L) Atrazine 10 0 UOSA UOSA UOSA Griffith Griffith Griffith Elk Run Elk Run Elk Run Corbalis Corbalis Corbalis Furrs Run Furrs Run Furrs Broad Run Broad Broad Run Broad Broad Run Broad Bull Run Upstream Run Bull Upstream Run Bull Upstream Run Bull Cub Run Upstream Cub Run Upstream Cub Run Upstream Cub Run Bull Run Downstream Bull Run Downstream Bull Run Downstream Cub Run Downstream Cub Run Downstream Cub Run Downstream Cub Run July August September Figure 3-6. Monthly Atrazine Concentrations at Virginia Sampling Locations. The following samples are not seen in the plotted range: Griffith in May was 328 ng/L, Griffith in June was 134 ng/L, and Griffith in July was 394 ng/L.

64 The Water Research Foundation 3.2.2 Estrogen A total of 18 estrogens were tested for in each sample, but only two were ever present. The estrogen peaks met two out of three criteria (see criteria in raw data from UBuffalo). The same analysis techniques for duplicates and non-detects that were used for SOCs were used for estrogens as well. Concentrations of estrone (E1) and estrone-3-sulfate (E1-3S), the two estrogens detected, are presented in Figure 3-7. No estrogenic compounds were detected in the UOSA effluentNo estrogenic compounds were detected in the UOSA effluent, and thus these are not shown in the graphs. The main trend visible is the increase in Estrone concentrations in January. The Estrone-3-sulfate (E1-3S) concentrations show an increase in February into March. Otherwise only the agriculture sites had an E1-3S concentration detected. Estrogen concentrations were found to be very low, if found at all, so they may not be good for further analysis. Estrogen concentration were not correlated with any of the other parameters of in situ measurements or the nutrients evaluated.

Figure 3-7. Estrone (E1) and Estrone-3-sulfate (E1-3S) Concentrations Present at Virginia Sampling Locations.

Improving Water Reuse for a Healthier Potomac Watershed 65 3.3 Agriculture Sampling Sites Comparison 3.3.1 Description of the Sampling Site and the Agricultural BMPs Two sampling sites shown in Figure 3-8 were chosen to compare managed and unmanaged streams surrounded by cropland. The Furrs Run site had a riparian buffer, and so was chosen as the BMP site. It runs through a watershed of 1600 acres of farmed land. From the sampling location, the watershed area was determined to be 2.6 square mile, with 60% being forests/shrub and almost 30% as cultivated crops and hay (USGS 2019). Elk Run was the unmanaged cropland stream running through a watershed of 3700 acres of thoroughly farmed land. The watershed area from the sampling point was 5.9 square miles, with almost 50% cultivated crops and hay. About 40% of that area was forests/shrub (USGS 2019).

Figure 3-8. Map of Virginia Agricultural Sites.

3.3.2 Effect of Agricultural BMPs on Nutrient and Organic Carbon Reductions Furrs Run and Elk Run nutrients data were compared. Generally, DOC, TDN, and SRP concentrations were less in the Furrs Run as compared with those of Elk Run (Figure 3-9). Elk Run SRP concentrations usually were higher than those of Furrs Run except March and April 2018, when the opposite occurred (Figure 3-9). If the differences in concentrations between the two agricultural streams, one with BMPs (Furrs Run) and one without BMPs (Elk Run) were attributed to the use of BMPs, we can conclude that BMPs can consistently decrease concentrations of DOC, TDN and SRP. Moreover, the concentrations in stream with BMPs showed less variability compared to the Furrs Run (without BMPs).

66 The Water Research Foundation

Figure 3-9. Comparison of Water Quality and Nutrients As Paired Agriculture Sites Without BMP Implementation (Elk Run) and With BMPs (Furrs Run)

Improving Water Reuse for a Healthier Potomac Watershed 67 3.3.3 Effect of Agricultural BMPs on Estrogen The estrogens detected at Elk Run and Furrs Run are shown in Figure 3-10. Estrogens were detected six time at the Elk Run site, where four of which were E1-3S. Furrs Run had a total of four times when estrogens were detected, twice each for E1 and E1-3S. Generally, levels observed at Elk Run (0.5-4.5 ng/L no BMPs) were higher than at Furrs Run (03 – 0.9 ng/L with BMPs), however because the compounds were not detected in most samples for both locations, definitive conclusions cannot be made regarding the effect of BMPs on estrogen concentrations in agricultural runoff at these locations.

Figure 3-10. Estrogen Concentrations at Elk Run (No BMPs) and Furrs Run (BMPs).

68 The Water Research Foundation 3.3.4 Effect of Agricultural BMPs on SOCs The SOCs detected at the agriculture sites, Elk Run and Furrs Run, were plotted on the timeline of the sampling period, respectively (Figure 3-11). At both sampling sites, the 4-nonylphenol concentrations were noticeably high compared to the other organic compounds. Although both sites had high levels of 4-nonylphenol, Elk Run had higher concentrations of other compounds detected than Furrs Run. A comparison of Figure 3-11 confirmed that Furrs Run (with BMPs) had lower concentrations of SOCs than Elk Run, probably due to the riparian buffer present along Furrs Run acting as a BMP. Therefore, the agricultural BMP selected for the Virginia portion of this study (riparian buffer) can lead to decreases in SOC concentrations.

Figure 3-11. SOC Concentrations at Elk Run (No BMPs) and Furrs Run (BMPs).

Improving Water Reuse for a Healthier Potomac Watershed 69 Specifically, the Furrs Run site had lower concentrations of nonylphenol than Elk Run until July through September. The Elk Run atrazine concentration values were higher than the Furrs Run values. This confirms the hypothesis that the site with the BMP would provide lower values of SOCs. The MCL for atrazine according to the U.S. EPA is 0.003 mg/L, or 3000 ng/L. Concentrations observed at these sites were approximately two orders of magnitude less than the MCL (see Section 3.2.1 for more information on atrazine concentrations causing feminization in vertebrates). The Furrs Run site was consistent at a lower concentration of imidacloprid, metolachlor, and simazine throughout the sampling period than Elk Run. Once again, this confirms the hypothesis that the site with the BMP would provide lower values of EDCs. Moreover, each of the SOCs (imidacloprid, metolachlor, atrazine, and simazine) showed significant increase in a late spring or summer month, potentially linked to product application. (Figure 3-11). 3.3.5 Evaluating Co-Management of Prometon and Nutrients A comparison of nutrient management (TDN, SRP, DOC) with several SOCs was performed. For illustrative purposes, the relationship of Prometon with TDN, SRP, DOC, and conductivity is shown in Figure 3-12. Prometon is a nonselective herbicide (O’Neil 2006), was consistently detected at low levels at both agriculture locations. Strong linear correlations were not observed between prometon and nutrients, DOC, and conductivity, although generally the BMPs (Furrs Run) showed correlations of lower nutrients and Prometon than the location without BMPs. This observation speaks to the effectiveness of agriculture BMPs for managing both nutrients and pesticides such as Prometon. Similar trends were observed for atrazine and metolachlor, although 4-nonylphenol showed stronger linear correlations between individual parameters and the compound, and less broad management correlations.

Figure 3-12. Comparing Nonylphenol concentration With Nutrients and Conventional Water Quality at Agriculture Sites.

70 The Water Research Foundation 3.4 Urban Sites Comparison 3.4.1 Description of Sampling Site and Urban BMPs Two stormwater reuse sites, Cub Run Upstream and Cub Run Downstream, were chosen in the Occoquan Watershed to compare to the effect of stormwater reuse (Figure 3-13). These sites represent managed and unmanaged urban stormwater stream sites. The Cub Run Upstream site is located at the intersection of Braddock Road and Old Lee Road, in a highly developed portion of Fairfax County, and is south of the Dulles International Airport. The Cub Run Downstream sampling site is located where the OWML ST50 stream gaging station is located. In between the two sites is the Chantilly National Golf and Country Club, the Cub Run Stream Valley Park, and residential area. Because of these land uses and management practices, the Upstream site is considered the unmanaged site (no BMPs) and the Downstream site would be considered the managed site (BMPs).

Figure 3-13. Map of Virginia Stormwater Reuse Sites.

Improving Water Reuse for a Healthier Potomac Watershed 71 3.4.2 Effect of Urban BMPs on DOC and Nutrients Cub Run Upstream (no BMPs) and Cub Run Downstream (BMPs) nutrients were compared for Dissolved Organic Carbon (DOC), Total Dissolved Nitrogen (TDN), and Soluble Reactive Phosphorus (SRP), in addition to conventional water quality parameters. All graphs showed a common pattern between the two sites (Figure 3-14). DOC, SRP, and TDN concentrations were equivalent or slightly higher at the Downstream site. Although it cannot be accurately claimed that the “managed” portion of the Cub Run removed nutrients, the observations support minimal additional inputs to the system through this urbanized watershed. A possible reason for the small increases observed could be additional N inputs from the golf course, parks, or fertilizers in residential areas.

Figure 3-14. Water Quality, Nutrients, and DOC at the Upstream Cub Run (no BMPs) and Downstream Cub Run (BMPs) Urban Sites.

72 The Water Research Foundation 3.4.3 Effect of Urban BMPs on Estrogen No estrogenic compound concentrations were measured above their quantification limit at the Cub Run Upstream site. E1 (Estrone) and E1-3S (Estrone-3-sulfate) were plotted for Cub Run Downstream in Figure 3-15. There were only three samples that included free estrogens, and the concentrations were less than 1 ng/L. Because of the scarcity of detections, no further analysis was performed.

Figure 3-15. Cub Run Downstream Estrogens.

Improving Water Reuse for a Healthier Potomac Watershed 73 3.4.4 Effect of Urban BMPs on SOCs An initial comparison of SOCs at the urban sites was done by plotting percent SOCs detection compared by SOC (Figure 3-16 upper), and by aggregate SOCs temporally (Figure 3-16 lower). The monthly comparison shows a periodic relationship in SOCs detection, with detection frequencies exceeding 50% only in the wetter spring and early summer months (February – June). Generally, the unmanaged Cub Run Upstream site tended to have more detections than the Downstream site. The following compounds were detected more than 50% of the time; 4-nonylphenol, atrazine, clothianidin, dinotefuran, fipronil, imidacloprid, metolachlor, prometon, and simazine.

Figure 3-16. Percentages of SOC Detection at the Urban Sites for Individual Compound (upper) and for Monthly Variability (lower).

74 The Water Research Foundation Select SOCs were plotted along the timeline of the sampling period (Figure 3-17), comparing levels at the Cub Run Upstream (no BMPs) and Cub Run Downstream (BMPs) sites. Although the Cub Run Upstream site had more SOC detections, the concentration magnitudes tended to be lower than those of the Downstream site. For example, simazine and imidacloprid had lower concentrations at the Upstream site than at the Downstream site. Possible reasons for this downstream increase could be additional SOC inputs from the golf course, parks, or products used in residential areas. A pattern of imidacloprid between the two sites is present: an increase in spring and summer, with decreasing concentration in July. Simazine remained fairly constant except for the increase at the Cub Run Downstream site in late summer. 4-Nonylphenol, atrazine and metolachlor concentration had no distinct similarities between the Cub Run sites (Figure 3-17). Atrazine and metolachlor concentration values were similar for each data sampled, with a common pattern presenting itself. There was an increase in concentrations in spring and into the early summer months, corresponding to likely seasonal application of these herbicides.

Figure 3-17. Selected SOCs in Cub Run at Upstream and Downstream BMP Sites.

Improving Water Reuse for a Healthier Potomac Watershed 75 3.4.5 Evaluating Co-Management of 4-Nonylphenol and Nutrients Relationships between several SOCs and nutrients, DOC, and conductivity measurements at the Cub Run sampling sites were examined and, because of its constant presence in the water samples, those for 4-nonylphenol were plotted (Figure 3-18). For the Urban sites, the relationships between implementation of BMPs and lower levels of SOCs and water quality is not observed. However, more obvious positive relationships can be observed between nutrients and SOCs, indicating the ability for successful implementation of co-management strategies. Uniquely, 4-nonylphenol appears to have a reasonably strong negative correlation with conductivity, indicating that levels of this compound may be concentrated in the stream and diluted with flushes of low conductivity rainwater.

Figure 3-18. Comparing 4-Nonylphenol With Nutrients, DOC, and Conductivity at Stormwater Reuse Sites.

76 The Water Research Foundation 3.5 UOSA Advanced Water Reclamation Facility 3.5.1 Description of Sampling Sites The sampling sites in the Occoquan Watershed being analyzed for their impact on reuse are Bull Run Upstream (upstream of UOSA), UOSA (product water from within UOSA), and Bull Run Downstream (downstream of UOSA effluent) (Figure 3-19). In 1978, the UOSA Regional Water Reclamation Plant, located on 470 acres in western Fairfax County and serving four jurisdictions, commenced operations and replaced eleven small secondary treatment plants in the region. Since that time, water quality in the Occoquan Reservoir has steadily improved and the reliable, high-quality effluent produced by UOSA has increased the safe yield of the Reservoir. Through several expansions, the initial 10 million gallons per day (mgd) capacity of UOSA was increased to 27 mgd, then 32 mgd, and most recently a major expansion to 54 mgd was completed. After 40 years of highly successful operations, UOSA reclaimed water is an increasingly important component of the drinking water supply strategy for the Washington metropolitan area.

Figure 3-19. Map of UOSA and Bull Run Sampling Sites.

3.5.2 Nutrients and DOC at UOSA AWRF Sampling Sites Nutrients, DOC, and water quality (temperature, DO, and conductivity) impacts of UOSA on the Bull Run receiving stream was evaluated through the plots presented in Figure 3-20. Generally, the results speak to the high quality of the UOSA product water and strategic process control employed by UOSA to maintain the quality of the Occoquan reservoir and the associated planned indirect potable reuse cycle. DOC discharged from UOSA was typically at levels less than that of Bull Run, with all effluent measurements less than 4 mg/L. This difference suggests UOSA, which employs activated carbon to maintain low levels of organic carbon discharge to the receiving stream, was not a significant source of DOC to Bull Run. With the consistent DOC concentrations in the effluent, the UOSA discharge serves to moderate swings in carbon in Bull Run, significantly lowering DOC in Bull Run during elevated DOC events. Also evident in the TDN concentration data is a unique Occoquan Water Quality improvement control strategy that UOSA has employed. Levels of TDN in UOSA product water were always greater than the concentration in Bull Run, and significantly increased TDN in the Bull Run receiving water. These observations do not demonstrate mismanagement of nutrient control at the UOSA facility, rather it - displays purposeful release of nitrate (NO3 ) in its product water, designed to improve water quality in the upper reaches of the Occoquan reservoir in the summer. Historically, during the first summer of UOSA’s operation, it discharged nitrified product water to the waters of the Occoquan because its denitrification units were not operational. The waters of Bull Run,

Improving Water Reuse for a Healthier Potomac Watershed 77 where the UOSA discharge was, were generally cooler (hence denser) than the summertime surface waters of the reservoir, and thus this nitrified discharge entered the hypolimnion of the stratified reservoir. In monitoring and analysis performed by the Occoquan Watershed Monitoring Laboratory (OWML), an important discovery was made that the released nitrate, in the absence of dissolved oxygen, acted as an alternate terminal electron acceptor, prevented the system from going anaerobic, and poised the oxidation-reduction potential high enough to maintain oxidizing conditions and thus prevented the release of phosphorus, iron and manganese that typically occurs under reduced conditions. Due to this beneficial effect of nitrate during summer, UOSA continues to discharge nitrified waters in the summer, and thus helps maintain the water quality in the reservoir. The nitrate is, therefore, denitrified, typically to nitrogen gas, and escapes to the atmosphere. It can also be noted that the UOSA SRP was consistently higher than the Bull Run stream sites and led to modest increases in the SRP of the Bull Run receiving stream. However, at levels well below 0.1 mg/L, the reactive phosphorous released into Bull Run suggests little effect from the UOSA contribution on the SRP concentration downstream.

Figure 3-20. Impact of UOSA Effluent On Water Quality, Nutrients, and DOC, in the Bull Run Receiving Stream.

78 The Water Research Foundation

3.5.3 Estrogens at UOSA AWRF Sampling Sites Estrone (E1) was detected only three times throughout the sampling period at the Bull Run sites (Figure 3-21). It was never detected at UOSA. No Estrone-3-sulfate (E1-3S) was detected at the UOSA, Bull Run Downstream, or Bull Run Upstream locations. This speaks to the high level of treatment employed at the advanced UOSA treatment system. Since no estrogens were detected at UOSA, no load calculations were performed.

Figure 3-21. Estrone (E1) Concentrations at Planned Potable Reuse Sites.

3.5.4 SOCs at UOSA WWTP Sampling Sites SOC concentrations from UOSA, Bull Run Upstream, and Bull Run Downstream were compared to gain an understanding of the impact of UOSA on downstream water quality. As described previously, UOSA employs advanced treatment including activated carbon which can significantly reduce the presence of organic compounds in the effluent discharge. Comparison of the concentrations were initially performed to observe impacts of UOSA effluent on the Bull Run receiving stream. Generally, for most SOCs, levels were not significantly higher in the UOSA effluent than in the background Bull Run, so levels were not observed to increase significantly after discharge. A few exceptions to this include atrazine, dinotefuran, and seasonal imidacloprid. Atrazine concentrations tended to be higher at UOSA than in Bull Run, and as result, there were downstream increases in atrazine along the stream (Figure 3-22). The exception was in May, when atrazine concentrations in Bull Run spiked significantly, and UOSA effluent reduced background levels. The UOSA concentration of dinotefuran was consistently higher than that of the Bull Run sites, and UOSA contributed to the Bull Run concentrations. UOSA contributed to the imidacloprid concentrations in the Bull Run Downstream site in the winter and early spring. Beginning in May, UOSA did not seem to contribute to the downstream site imidacloprid concentration as much as the Bull Run Upstream site did. The exceptions were July and September. The highest concentrations of metolachlor occurred in May, with the source being the Bull Run Upstream site, not UOSA. Prometon concentrations remained relatively consistent between the three sites. Concentrations stayed within less than a 5 ng/L range throughout the sampling period.

Improving Water Reuse for a Healthier Potomac Watershed 79

Figure 3-22. Concentrations of Select SOCs at UOSA WWTP Sites.

80 The Water Research Foundation 3.5.5 UOSA SOCs Load Contribution to Bull Run and the Occoquan Bull Run Receiving Stream Concentrations are not the most applicable measure of the potential downstream impact of a point source. Therefore, the impact of the UOSA discharge on levels of SOCs in Bull Run and the outflow of the Occoquan Reservoir was analyzed by estimating SOC load contributions. The loads of the SOCs were calculated by multiplying the average concentrations in each flow stream over the period of observation by the associated average UOSA or Bull Run Downstream flow rate for the same observation period. Once the observation period loads for UOSA and Bull Run Downstream were calculated, the load of UOSA was divided by that of Bull Run Downstream to estimate the percent SOC load coming from UOSA that is in the Bull Run Downstream load during the monitoring period. Figure 3-23 displays the load contribution of UOSA to the Bull Run receiving stream for water and 10 regularly observed SOCs. The results highlight the significant average load contribution of UOSA for some SOCs, including atrazine, metolachlor, simazine, dinotefuran, and imidacloprid. Interestingly, UOSA did not appear to contribute the majority of 4-nonylphenol, a regularly observed wastewater indicator, to Bull Run. When the load contribution was extended for all SOCs where the load percentage was greater than 0% at least once during the sampling period, it is notable that eight out of 26 SOCs had no load contribution, because concentrations of the SOCs at UOSA and/or Bull Run were under detection limits.

Improving Water Reuse for a Healthier Potomac Watershed 81

Figure 3-23. UOSA Load Contribution to Bull Run Downstream.

82 The Water Research Foundation Occoquan Reservoir The UOSA product water impact on the Occoquan Reservoir outflow was analyzed as an annual load contribution. Daily load comparisons are improper because there is about a 20-day residence time in the Reservoir. An annual average was calculated from the daily loads of each pesticide at each site. The annual load contribution was then compared as a share contribution from UOSA as compared to other sources in the Occoquan system (Figure 3-24).

Figure 3-24. UOSA Annual Load Contribution of SOCs to Reservoir Outflow, Compared With Unknown “Other” Sources Within the Watershed.

Improving Water Reuse for a Healthier Potomac Watershed 83 The load comparison speaks volumes to the contributions to the Occoquan system of UOSA. By contributing 8% of the flow to the Occoquan reservoir at the dam in a relatively wet year, the benefits of the urban water reuse cycle can be immediately observed. Additionally, the load contributions of SOCs are all less than the water contribution, speaking to the outsized benefits of the system. The sole exception to this rule is for dinotefuran, for which UOSA contributed more than a third of the load to the Occoquan system. 3.6 Water Treatment Plant Intake Load Comparisons 3.6.1 Site Description Two sampling sites were used to study the impacts of planned and unplanned indirect potable reuse. The sites in the Potomac Watershed were considered unplanned reuse, and the Occoquan Watershed sites were considered planned potable reuse. The Griffith site, located at the dam of the Occoquan Reservoir, is the location of the Frederick P. Griffith Water Treatment Plant intake. This site is compared, as a planned potable reuse site, to the James J. Corbalis Water Treatment Plant, or Corbalis unplanned reuse site (Figure 3-1). The Corbalis WTP is located on the Potomac River, with known upstream POTWs employing advanced nutrient control but not necessarily advanced treatment for removal of trace organics. 3.6.2 DOC and Nutrients Levels of nutrients and DOC detected at the water treatment plants were compared (Figure 3-25). The DOC loads for the water treatment plants also follow a similar pattern, but the Griffith site was consistently higher than the Corbalis site. The Corbalis TDN was higher than the Griffith TDN load, except for October through December 2017. The SRP loads had a similar pattern between Griffith and Corbalis. From May to September, the Corbalis SRP concentrations are higher than the Griffith site SRP concentrations. The SRP at Griffith was initially higher than the Corbalis site from November to March. The levels observed speak to the impacted nature of the Potomac watershed.

Figure 3-25. DOC, TDN, and SRP Concentrations at the Water Treatment Plant Intakes.

84 The Water Research Foundation 3.6.3 Estrogens Estrogens were detected in the water intake samples four times. The daily loads of these estrogens were calculated (Figure 3-26), but since there were so few detections no major conclusions or further analysis was done.

Figure 3-26. Estrogens at the Water Treatment Plant Intakes.

Improving Water Reuse for a Healthier Potomac Watershed 85 3.6.4 SOCs SOC concentrations from the intakes (raw water) can be found in Figure 3-27 for the Griffith Intake and the Corbalis Intake. Generally, levels at the Griffith intake generally seemed to have a higher average load of the organic compounds than detected at Corbalis. An important distinction between the two sites was the much higher levels of atrazine and metolachlor at the Griffith intake in the summer months (Figure 3-27). More analysis is needed to understand the differences between the SOC loads at the two water treatment plants.

Figure 3-27. SOC Concentrations at Griffith and Corbalis Water Treatment Plant Intakes.

Both WTP intakes had high concentrations of industrial compound 4-nonylphenol, and herbicides atrazine, metolachlor, and simazine occur during the spring and summer months. Corbalis WTP intake had higher average concentrations of industrial compounds, 4-nonylphenol and 4-tert-octylphenol; insecticides, acetamiprid, dichlorvos, and fenchlorphos; and triazine herbicide simazine. Griffith WTP intake had higher concentrations of the other triazine herbicides, atrazine and prometon; neonicotinoids, clothianidin, dinotefuran, and imidacloprid; industrial compound bisphenol A; and herbicide metolachlor. It must also be noted that no raw water intake SOC levels were above their corresponding U.S. EPA ALB levels nor U.S. EPA MCL values. SOCs with MCLs include atrazine, heptachlor epoxide, and simazine. The WTPs had about the same number of detections of all SOCs monitored (50% at each). This suggests that both planned and unplanned IPR scenarios are equally effective, although magnitude of SOC concentrations are an important factor that is considered as well.

86 The Water Research Foundation While it was not possible from this dataset to determine a watershed or sewershed source upstream of the WTP intakes for the SOCs, it is apparent that the water supply upstream of Griffith tended to have higher concentrations of SOCs. The Occoquan Reservoir has two sources of water: one from Bull Run and the other from the Broad Run/Occoquan Creek system. The Broad Run/Occoquan Creek system drains less-developed areas, including significant agricultural areas. Summertime spikes of atrazine have been noted in those waters in monitoring done by the OWML. Future studies should determine the upstream sources of SOCs in planned vs. unplanned IPR systems.

Improving Water Reuse for a Healthier Potomac Watershed 87

88 The Water Research Foundation CHAPTER 4 Cost Benefit Analysis

In order to address the project objective of developing a quantitative assessment of costs, benefits, and impact of advanced reclamation, reuse, and management practices on human and ecological health in the Potomac Watershed, the project team employed a multi-criteria decision analysis (MCDA) framework. Through this process, four alternatives were ranked for their ability to cost-effectively and equitably co-manage nutrient and CEC pollution according to stakeholder-developed criteria. The four alternatives included: • Co-managing nutrients and CECs through control of non-point source pollution with implementation of Agriculture Best Management Practices (BMPs). • Co-managing nutrients and CECs through control of non-point source pollution with implementation of Urban Stormwater Best Management Practices (BMPs). • Co-managing nutrients and CECs through control of point source pollution with implementation of Advanced Nutrient Control at publicly owned treatment works (POTWs). • Co-managing nutrients and CECs through control of point source pollution with implementation of Advanced Water Treatment technology for planned potable reuse at publicly owned treatment works (POTWs). 4.1 Cost Benefit Analysis Framework Multi-criteria decision analysis (MCDA), provides a framework for comparing alternatives through a series of important criteria, similar to a triple bottom line (TBL) evaluation. However, using the MCDA framework provides flexibility in the analysis and an opportunity for a stakeholder panel of experts to develop and weight the importance of specific criteria for the evaluation. Hazen facilitated a MCDA process to address the questions posed in this evaluation, by: • Developing descriptions of the four alternatives evaluated in this project for co-managing nutrient and CEC pollution to the Potomac River. • Convening a group of stakeholders comprised of academic, utility, the EPA, and NGO representatives with expertise in Potomac River pollution in order to develop and rank criteria for the evaluation • Developing quantitative and qualitative information related to the developed criteria to facilitate scoring • Providing an “Engineer’s Estimate” scoring of the four alternatives according to the top 10 criteria developed in the stakeholder workshop • Providing a spreadsheet-based tool (HazenConverge), to be utilized as an organized, defensible framework for comparing the impact and costs of the four evaluated Potomac River Watershed pollution management alternatives for improving ecological health of the Potomac. 4.1.1 HazenConverge Tool Description HazenConverge is an evaluation tool that can allow stakeholders to compare alternatives across various decision-making criteria. Criteria are developed and weighted, creating a custom framework for evaluating reuse alternatives and watershed protection measures that match stakeholder priorities. This is accomplished through identifying, weighting, and quantitatively or qualitatively enumerating criteria that prioritized by the stakeholders, then normalized to rank and compare alternatives.

Improving Water Reuse for a Healthier Potomac Watershed 89 The tool is being utilized to compare directly costs and benefits of four alternatives studied within this project for the Potomac Watershed. The alternatives for analysis have been selected based upon “BMPs” selected for Year 2 monitoring, as these provide the necessary data for evaluation, and include: • Co-managing nutrients and CECs through control of non-point source pollution with implementation of Agriculture Best Management Practices (BMPs). • Co-managing nutrients and CECs through control of non-point source pollution with implementation of Urban Stormwater Best Management Practices (BMPs). • Co-managing nutrients and CECs through control of point source pollution with implementation of Advanced Nutrient Control at publicly owned treatment works (POTWs). • Co-managing nutrients and CECs through control of point source pollution with implementation of Advanced Water Treatment technology for planned potable reuse at publicly owned treatment works (POTWs). The HazenConverge Tool requires definition and weighting of decision-making criteria against which each of these alternatives will be compared. The HazenConverge tool comes preloaded with a variety of criteria from which selections can be made, in a variety of categories, and also allows for user input. Table 4-1 displays example criteria preloaded into the HazenConverge tool. Although there is no limit to the number of criteria which can be selected, it is important to ensure uniqueness of selected criteria to avoid “double counting” of strengths and weaknesses.

90 The Water Research Foundation Table 4-1. Example Pre-Loaded Criteria from the HazenConverge Tool. CATEGORY: IMPLEMENTATION Ease of integration Utility crossings Accessibility Lead time requirement Prevalence of confined spaces Ease of regulatory approval Consistency with other initiatives Permitting requirements Ease of construction Sampling/monitoring requirements Ease of expansion Success of previous installations Ease of maintenance Implementation timeline Ease of operation

CATEGORY: ECONOMY Capital cost Cost risk factors O&M cost Impact to surrounded economic growth Life cycle cost

CATEGORY: ENVIRONMENT Resiliency to climate change Construction energy use Drought susceptibility Renewable energy use Emergency reliability Energy generation Carbon footprint Raw water withdrawal Greenhouse gas emissions Operational water use Air pollutant emissions Construction water use Energy footprint Disturbance of prime farmland Climate change vulnerability Impact to wetlands and/or buffers Air quality impacts Disturbance of greenfield Ease of waste disposal Reclamation of brownfield Impact to biodiversity Stormwater runoff Impact to ecologically sensitive sites Use of pesticides Spill potential Use of fertilizers Use of recycled materials Use of liquid/gaseous chemicals Landfill diversion of construction waste Construction in natural floodplains Landfill diversion of operational waste Control of invasive species Earthwork balance Soil protection/restoration Operational energy use

CATEGORY: COMMUNITY Public acceptance Community green space Public safety Impact to public space Staff safety Impact to views and local character Construction safety Habitat enhancement Easement requirements Traffic impacts Impact of failure Use of shared/mass transportation Job creation Water resources impacts

Improving Water Reuse for a Healthier Potomac Watershed 91 Noise pollution Visual impact Odor generation Equitable and social impacts Light pollution Preservation of historic and/or cultural resources

CATEGORY: OTHERS Certainty of information Resilience to natural and human threats Availability of materials Efficiency increase to existing infrastructure Collaboration potential Repurposing of existing infrastructure Contractual requirements Adaptability/flexibility Redundancy System recovery requirement Vulnerability Co-benefits and synergies Durability

While some of these criteria are quantitative in nature and are currently being quantified (i.e., capital, O&M, life cycle costs, along with percent reduction (concentration or load) of contaminants), many are more qualitative and require input from stakeholders. Our project team and the Project Advisory Committee provide a cross section of potential stakeholders, from academics, regulators, and practitioners, to serve as the basis of a stakeholder committee. The team participated in a criteria selection and weighting workshop. 4.2 Stakeholder Criteria Development and Weighting Workshop A workshop was held October 3, 2019 at the Metropolitan Washing Council of Governments offices in Washington, D.C., from 9 am to 3 pm EST. The goal of the workshop was to develop criteria and rankings/weightings for the HazenConverge Analysis. The workshop was facilitated by Hazen staff, Paul Knowles and Erik Rosenfeldt. As managing PI on the project, Mr. Rosenfeldt provided project context to the analysis, developed the group of stakeholders, and facilitated the meeting. As the corporate HazenConverge expert, Dr. Knowles led stakeholders through the critical criteria selection and pairwise comparison process for criteria weighting. Stakeholders from project collaborators participated in the workshop, representing the Project Team, the Project Advisory Committee, and Potomac region utilities (water and wastewater focused). Professional categories represented at the workshop included management, research, consultant, policy, academy, government, utility, private and publicly owned companies. The diverse group was assembled to ensure the criteria development and ranking exercises received input from a wide group of stakeholders, although it is recognized that the group was limited in size and results from the analysis should be viewed as a “case study” of the process.

92 The Water Research Foundation 4.2.1 Agenda The agenda for the October 3, 2019 meeting is presented below. Location: Metropolitan Washington Council of Governments 777 North Capitol Street NE, Suite 300 Washington, D.C. 20002 Time: 9:00am - 3:00pm EST

Subject: EPA STAR STAR_WR1SG16: Improving Water Reuse for a Healthier Potomac Watershed Multi-Criteria Decision Framework Workshop Agenda

9:00 – 10:00 Background on the Project to Date* 10:00 – 10:30 Introduction - MCDA overview and definition of alternatives* 10:30 – 10:45 Break 10:45 – 12:00 Exercise on category and criteria development 12:00 – 1:00 Lunch 1:00 – 1:15 Intro to pairwise comparison theory 1:15 – 2:30 Pairwise comparison exercise 2:30 – 3:00 Overflow time and Q&A *Presentation slides from the workshop are presented in Appendix A.

A summary of key workshop activities, findings, and conclusions from the workshop are presented below. 4.2.2 Background on the Project to Date Erik Rosenfeldt provided an introductory overview of the workshop and summary of work to date, including the alternatives that had been identified. The presentation provided information regarding land use within the Potomac watershed and summarized relevant preliminary findings with respect to sources of nutrients and endocrine disrupting compounds in the watershed. Figure 4-1 summarizes the contribution (separated by land use) to the Potomac Watershed for conventional and emerging pollutants. These contributions were estimated from a compilation of data collected during monthly sampling campaigns for 2017.

Improving Water Reuse for a Healthier Potomac Watershed 93

Figure 4-1. Relative Load Contributions of Conventional and Emerging Pollutants Into the Potomac Watershed.

94 The Water Research Foundation The focus of the second year of study was to examine the ability of Best Management Practices for controlling nutrient loads and co-managing EDCs and other contaminants in the Potomac Watershed. This “performance testing” occurred by monthly monitoring of point and non-point sources, which were chosen in co-located watersheds experiencing differing degrees of BMP implementation. The focus of this testing gave us a dataset for which to define performance, but also cost metrics, for four alternatives for managing nutrients and emerging contaminants in the Potomac Watershed. The four alternatives included: 1. Implementation of Non-point Agriculture BMPs (nutrient control). 2. Implementation of Non-point Urban Stormwater BMPs (nutrient control). 3. Implementation of Point Source BMPs (i.e., advanced nutrient removal technology). 4. Implementation of Point Source Advanced Reuse BMPs (emerging contaminant control). To provide context for the options, preliminary performance estimates were developed and presented for agriculture BMPs, urban BMPs, and advanced nutrient removal. Performance for removal of nutrients, bulk water quality, as well as measurable EDCs and pesticides with point and non-point source BMPs were provided to stakeholders and are summarized in Table 4-2.

Table 4-2. Estimate of Pollutant Load Reduction With BMP Implementation.

In addition, preliminary costs were collected from applicable sources, where available, to present a preliminary understanding of costs of implementation of non-point source nutrient control strategies, presented in Table 4-3. These costs were all presented as very generic numbers, to provide order of magnitude understanding of resources required for widespread BMP implementation. It should be repeated that these quantitative measures were presented merely for background information, and not meant to influence stakeholder’s decision making process regarding important criteria. The intention is to develop final quantifiable estimates for loads, benefits, and cost of implementation after the workshop, to finalize the analysis according to stakeholder preferences.

Improving Water Reuse for a Healthier Potomac Watershed 95 Table 4-3. Cost of BMP Implementation for Nutrient Control in the Chesapeake Bay.

4.2.3 Introduction – MCDA Overview and Definition of Alternatives Paul Knowles explained the goal of Multi-Criteria-Decision-Analysis, which is to develop a framework for evaluating non-cost performance criteria, such as social and environmental impacts, alongside traditional cost-metrics for cost performance. Multi-Criteria-Decision-Analysis is conducted through a series of steps: 1. Define alternatives. 2. Identify quantitative and qualitative criteria. 3. Categorize as appropriate. 4. Weight criteria using Pairwise Comparison. 5. Score alternatives against each criterion. 6. Normalize. 7. Review Results. Dr. Knowles established that the workshop would guide the stakeholders through steps 2 to 4 – deciding on the short-list of criteria and their relative weighting. Dr. Knowles asked the workshop stakeholders to introduce themselves along with which entity they were representing and what they were hoping to achieve from the workshop; the results of which are summarized in Table 4-4.

96 The Water Research Foundation Table 4-4. Workshop Stakeholders. Name Qualifications “Stakeholder Focus” Justin Mattingly WaterRF Project Manager Research Community and Potomac River Watershed rate payer. Matt Reis DC Water Director of Sustainable watershed Integrated and sustainable utility focused management watershed management Sujay Kaushal University of Maryland, Professor, Project PI Representing the Scientific Research Community Lisa Ragain Metro. Council of Governments Policy focus, translating research into policy Amelia Flannery Virginia Tech, Graduate Student, project team Environmental Science and Engineering MS student Anne Spiesman US Army Corps of Engineers Washington Drinking Water utility focus, Interested in a Aqueduct “practical tool” Greg Prelewicz Fairfax Water, Manager of Planning Drinking water utility focus, Recipient of IPR benefits Susan Glassmeyer U.S. EPA, Research Chemist “Outside” focus, Agricultural background, Research Scientist, Pam Kenel Loudon Water, Director of Water Resources Utility focus, interested in how the agricultural community can play its part Leita Bennett GHD, Associate Representing the Consulting Engineering community Bob Angelotti Upper Occoquan Service Authority, Deputy Interested in water reuse aspects of the work Executive Director of Technical Services Steve Bieber Metro. Council of Governments, Acting Water Interested in inter-community collaboration Resources Director of Programs Erik Rosenfeldt Hazen and Sawyer, Associate VP, Co-PI Representing the Consulting Engineering community

The stakeholders summarized their goals for the workshop according to this list below: • Understand each other’s perspectives. • Look for data and results to inform management, policy, decision making, and public outreach. • Develop a tool that is specific enough to be useful. • Protect source of drinking water. • Support agriculture, recreation, and fishing. • Consider affordability for ratepayers. • Develop a framework that reflects fairness and equity. • Consider other co-benefits such as flooding and air quality. • Understand the MCDA framework and consider applicability outside of this project.

Improving Water Reuse for a Healthier Potomac Watershed 97 4.2.4 Category and Criteria Development Methods and Results Dr. Knowles guided the workshop stakeholders through a series of exercises designed to identify a shortlist of salient performance criteria deemed important by the stakeholders to measure alternative performance. Biased Representation: Stakeholders are separated into groups that approximately represent their interests, such that biases in criteria selection between different types of stakeholders are reinforced. The groupings and list of criteria provided by each group are summarized in Table 4-5.

Table 4-5. Results of the Biased Representation Exercise.

Group 2: Academic Research and Group 3: Utility and Watershed Group 1: Institutional and Policy Consulting - Justin, Leita, Amelia, Management - Pam, Matt, Bob, Criterion - Lisa, Ann, Steve, Susan Sujay Greg 1 Degree of effectiveness Performance Cost 2 Consequences of implementation Cost Ease of implementation

3 Co-benefits Cost/reduction metric Does a regulatory framework exist 4 Impacts on waste balance Distribution of improvements Cost equity

5 Air emissions Aesthetics Social justice 6 Consumption of energy Recreation Economic impact 7 Incidental waste streams Local economic stimulus ‘Bang for your buck’ 8 Political palatability Ease of implementation Net benefits 9 Ease of implementation Ease of maintenance Implementable 10 Regulatory/voluntary palatability Resilience to climate change Spatial footprint

11 Number of impacted Carbon footprint stakeholders 12 Degree of uncertainty in info Energy needs, demand

13 Bio-habitat 14 Environmental benefits 15 Effectiveness certainty 16 Operability 17 Incentives 18 Mandates

Balanced Representation: Stakeholders are distributed to produced mixed groups of stakeholders with a balanced representation of views. The groups were asked to evaluate the results from the first exercise and refine the shortlist, additionally thinking of initial ways the criteria would be measured. The objective is to evaluate through discussion which criteria are truly important to all stakeholders. The groupings and list of criteria provided by each group are summarized in Table 4-6. Several of the criteria were common across all three groups. Interestingly, new criteria were introduced in the Balanced Representation exercise that had not been suggested during the Biased Representation exercise.

98 The Water Research Foundation Table 4-6. Results of the Balanced Representation Exercise. Criterion Group 1: Lisa, Ann, Steve, Group 2: Justin, Leita, Group 3: Greg, Bob, Pam, Susan Amelia, Sujay Susan, Matt 1 Cost effectiveness (life Cost/reduction (lifecycle Cost $ cycle) unit) 2 Policy/regulatory drivers Effectiveness Cost effectiveness ($/unit of removal or performance) 3 Ease of implementation Cost distribution Performance (#, acres, mg/L) (regionally/socially) 4 Resilience (climate change) Benefit distribution Implementable (H-M-L) (regionally/socially) 5 Equity Certainty Geographic distribution (H-M-L) 6 Economic impact External impacts Social impact distribution (justice) (H- (resilience?) M-L) 7 Carbon footprint Co-benefits (composite effectiveness?) 8 Energy Lifecycle – of solution (time), maintenance ($), replacement (time/$) 9 Future sustainability Uncertainty of solution/performance (probability/error bar, box whisker, end members) 10 Habitat Equity-fairness 11 Water quality 12 Green alternative 13 Ease of implementation 14 Implementation timeline 15 Would require enforcement

Improving Water Reuse for a Healthier Potomac Watershed 99 Identify similar criteria and group into categories: The combined group of stakeholders is invited to review the results of the Balanced Representation exercise, summarize similar criteria and group into broad categories. The results of this exercise are shown in Table 4-7 which shows the seven categories the stakeholders converged upon. Table 4-7. Similarity and Categorization. Criterion # Category Criterion and Method of Measurement 1 Cost effectiveness (life cycle) 2 Cost/reduction (lifecycle $$/TN) 3 Cost Cost/reduction (lifecycle $$/CEC) 4 Capital Cost $ 5 Cost effectiveness ($/unit of removal or performance) 6 Potential Impact (H-M-L) 7 Mass Removal Effectiveness Performance 8 Performance (#, acres, mg/L) 9 Water quality 10 Ease of implementation 11 Ease of implementation 12 Ease of Implementation Implementability (H-M-L) 13 Implementation timeline 14 Policy/regulatory drivers 15 Cost distribution (regionally/socially) 16 Benefit distribution (regionally/socially) 17 Equity 18 Geographic distribution (H-M-L) Equity 19 Social impact distribution (justice) (H-M-L) 20 Equity-fairness 21 Economic impact 22 Mandates 23 Longevity 24 Resilience (climate change) 25 Certainty Risk 26 External impacts (resilience) Uncertainty of solution/performance (probability/error bar, 27 box whisker, end members) 28 Habitat 29 Green alternative 30 Environment Carbon footprint 31 Energy 32 Future sustainability 33 Other Co-benefits (composite effectiveness)

100 The Water Research Foundation Elimination of Duplicates and Final Filter: The stakeholders summarize similar criteria into one representative criterion and attempt to reduce the list to the minimum number of required performance criteria. Final methods for quantitative and qualitative measurement are established. The outcomes of the criteria selection exercise are summarized in Table 4-8. The result was 19 criteria across seven categories to be weighted by the stakeholders. The basis upon which the criteria will be scored was established by the stakeholders for many of the criteria. For certain criteria, marked as TBD in Table 4-6, it was decided that the basis for scoring would be established by the project team during subsequent analysis for the alternatives.

Table 4-8. Final Outcomes of the Criteria Identification Exercise.

Qualitative (QL) or Quantitative Category Criterion (QN) Basis 1 Cost Capital Cost QN $$ 2 Performance Lifecycle Cost/reduction TN QN $$/ton TN

3 Performance Lifecycle Cost/reduction CEC QN $$/ton CEC

4 Performance Potential Impact QL H-M-L: TBD 5 Ease of Complexity of QL H-M-L: number of parties, how diffuse Implementation implementation 6 Ease of Implementation timeline to QL H-M-L: shortterm < 5, midterm 5-20, Implementation benefits longterm > 20 7 Ease of Complexity of QL H-M-L: required, supported, on Implementation Policy/regulatory drivers horizon, not on radar 8 Equity Benefit Geographic QL H-M-L: miles WC standards are met, Distribution locality, watershed, broader 9 Equity Benefit Social Distribution QL H-M-L: TBD

10 Equity Cost Geographic Distribution QL H-M-L: TBD

11 Equity Cost Social Distribution QL H-M-L: TBD 12 Equity Affordability QN TBD 13 Risk Design Lifetime QL H-M-L: shortterm < 5, midterm 5-20, longterm > 20 14 Risk Resilience QL H-M-L: TBD 15 Risk Confidence of Performance QL H-M-L: TBD

16 Environment Habitat Creation QL H-M-L: Restore, preserve, replace

17 Environment Carbon footprint/Energy QN lb CO2.e/lb TN reduced 18 Other Co-benefits QL H-M-L: reuse, avoided 19 Other Co-impacts QL H-M-L: TBD

Improving Water Reuse for a Healthier Potomac Watershed 101 4.2.5 Pairwise Comparison Methods and Results Dr. Knowles guided the workshop stakeholders through a pairwise comparison exercise, whereby each individual stakeholder compared every combination of criteria pairs against one another and relatively ranked which one they thought was most important. The input from every stakeholder was recorded. Relative importance is scored on a scale of 1-9 , where 1 means the criteria are equally important and 9 means one criterion is extremely important in comparison to the other criterion. The results are processed using Analytical Hierarchy Process (AHP) theory, which is a matrix algebra-based method of processing multiple pairwise comparisons to find the relative weight of each individual criterion within the framework of criteria. The exercise is useful for complex decision making when multiple criteria are involved in selection. Given 19 criteria, the Pairwise Comparison exercise was performed on 171 pairs, for twelve users, for a total of 2,052 datapoints. The results of the Pairwise Comparison exercise for the stakeholder group are shown in Appendix B. The weighted criteria results for each individual stakeholder are provided in Table 4-9. The following observations made regarding the data during processing: • Out of 12 users, 10 users passed the syllogism ‘consistency’ threshold (set at 0.15). Including or excluding those users (weight versus average columns below) did not significantly impact the outcomes. • There was a healthy difference of opinions about what the priorities on the project might be, as indicated by high % single standard deviation expressed against magnitude of average (26-55%). • After averaging across the group for every pairwise comparison, the consistency was excellent (0.02) suggesting the average results do a good job of reflecting the balanced group consensus. Individual stakeholder responses for each pairwise comparison were averaged to produce a representative group pairwise comparison result. The AHP weighting calculations were performed on these group averages, the results of which are shown in Figure 4-2. The top 10 criteria, accounting for 75% of total weighting are:

1 Cost/reduction TN Performance 12.9% 2 Cost/reduction CEC Performance 8.8% 3 Affordability Equity 8.6% 4 Capital Cost Cost 8.2% 5 Complexity of implementation Implementability 7.3% 6 Potential impact Performance 7.0% 7 Implementation timeline to benefits Implementability 6.3% 8 Complexity of Policy/regulation Implementability 6.0% 9 Confidence of performance Risk 5.2% 10 Geographic distribution of benefits Equity 4.6%

The project team recommends the remaining nine criteria are not significant to the selection process and that the subsequent scoring process focuses on these 10 criteria.

102 The Water Research Foundation Table 4-9. Results of the Individual Stakeholder Weighting Exercise. Std # Criterion U1 U2 U3 U4 U5 U6 U7 U8 U9 U10 U11 U12 Avg. Dev 1 Capital Cost 9% 5% 8% 11% 7% 9% 9% 11% 8% 13% 17% 10% 9.7% 35% 2 Cost/reduction (lifecycle TN/$$) 8% 8% 5% 8% 11% 16% 17% 19% 11% 11% 12% 11% 10.8% 39% 3 Cost/reduction (lifecycle CEC/$$) 12% 8% 5% 8% 10% 13% 1% 11% 9% 8% 12% 9% 9.6% 27% 4 Potential impact 8% 10% 7% 10% 9% 10% 20% 3% 10% 4% 2% 17% 7.4% 42% 5 Complexity of implementation (0-4) 7% 5% 6% 7% 2% 7% 5% 6% 5% 5% 5% 6% 5.6% 26% 6 Implementation timeline to benefits 5% 6% 5% 5% 7% 4% 3% 4% 3% 3% 2% 7% 4.5% 31% 7 Complexity of Policy/regulation 9% 4% 9% 8% 3% 9% 7% 3% 6% 3% 3% 2% 5.7% 49% 8 Geographic distribution of benefits 6% 5% 4% 4% 2% 2% 2% 6% 2% 6% 5% 6% 4.2% 39% 9 Social distribution of benefits 3% 4% 5% 3% 2% 2% 2% 4% 3% 3% 5% 3% 3.3% 31% 10 Geographic distribution of costs 5% 4% 3% 3% 2% 3% 3% 4% 3% 4% 6% 3% 3.7% 31% 11 Social distribution of costs 2% 4% 3% 3% 2% 2% 2% 5% 2% 3% 5% 1% 3.2% 37% 12 Affordability () 10% 4% 4% 6% 13% 6% 5% 7% 10% 8% 8% 3% 7.5% 39% 13 Design lifetime (5,20,100) 2% 4% 5% 3% 7% 3% 8% 1% 2% 6% 1% 6% 3.6% 55% 14 Resilience (H-M-L) 2% 5% 7% 3% 3% 4% 2% 2% 7% 2% 5% 5% 4.0% 46% 15 Confidence of performance (H-M-L) 1% 11% 7% 5% 7% 4% 5% 4% 8% 6% 2% 2% 5.7% 52% 16 Habitat creation/preservation (H-ML) 3% 3% 2% 2% 3% 3% 1% 1% 3% 4% 2% 6% 2.7% 32% 17 Carbon footprint/Energy (lb Co2/lb TN 1% 5% 4% 3% 3% 2% 6% 3% 2% 3% 2% 1% 2.8% 45% etc.) 18 Co-benefits (H-M-L) 5% 2% 7% 3% 5% 2% 1% 2% 3% 3% 3% 1% 3.4% 49% 19 Co-impacts (externalities) 1% 3% 5% 4% 2% 1% 1% 1% 2% 3% 3% 1% 2.7% 46% CONSISTENCY RATIO 0.14 0.09 0.09 0.11 0.07 0.07 0.22 0.11 0.09 0.07 0.08 0.27

Improving Water Reuse for a Healthier Potomac Watershed 103

Figure 4-2. Results of the AHP Weighting Exercise for Each Criterion and Summary by Category.

104 The Water Research Foundation 4.3 Defining the Metrics for Each Alternative To compare the four alternatives for managing nutrients and emerging contaminants within the MCDA framework, each alternative must be assigned a “score” for each of the top 10 criteria. This score can be based upon quantitative or qualitative measures, depending on appropriate comparison metrics. For example, the five of the top six categories selected by the panel are all “quantifiable” according to data collected within this study (Cost/reduction TN, Cost/reduction CEC, Capital Cost, Potential Impact), while the others would be categorized as qualitative (Affordability, Complexity of implementation, Implementation timeline to benefits, Complexity of Policy/regulation, Confidence of performance, Geographic distribution of benefits). 4.3.1 Quantitative Metrics – Performance and Cost 4.3.1.1 Defining Baseline Performance In order to contextualize the performance and cost quantitative calculations, a short summary of the status and remaining load reduction requirements associated with the Chesapeake Bay Total Maximum Daily Load (TMDL) is provided. The U.S. EPA established the Chesapeake Bay TMDL in December 2010 in order to reduce nutrient and sediment loads into the Bay. Billed as a comprehensive “pollution diet” for the Bay, the TMDL includes accountability to regional jurisdiction with the aim of restoring clean water in the Chesapeake Bay and the region’s streams, creeks, and rivers. The TMDL encompassing a 64,000-square-mile watershed, consisting of numerous rivers and stream, and includes the entirety of the Potomac River watershed. The TMDL identifies the necessary pollution reductions from major sources of nitrogen, phosphorus and sediment across the Bay jurisdictions and sets pollution limits necessary to meet water quality standards. Specifically, the TMDL set Bay watershed limits of 185.9 million pounds of nitrogen, 12.5 million pounds of phosphorus and 6.45 billion pounds of sediment per year. This equates to a 25% reduction in nitrogen, 24% reduction in phosphorus and 20% reduction in sediment. The pollution limits were further divided by jurisdiction and major river basin and the TMDL is designed to ensure that all pollution control measures needed to fully restore the Bay and its tidal rivers are in place by 2025. As part of the TMDL process, jurisdictions were required to develop Watershed Implementation Plans (WIPs), in three Phases. Currently, jurisdictions are operating under Phase III WIPs, with plans being implemented to achieve 2025 Bay restoration goals. With the submission of WIP III, extensive accounting was performed by jurisdictions in order to understand pollutant load reductions which have been made since enaction of the Bay TMDL, and compare those to reductions required under the TMDL. Table 4-10 summarizes this information for the States of Virginia and Maryland, the two states which border the Potomac for the majority of the River and contain the vast majority of the watershed.

Improving Water Reuse for a Healthier Potomac Watershed 105 Table 4-10. Potomac River Estimates of Percent Reduction to Achieve 2025 TMDL. Data estimated from Chesapeake Bay Foundation (2019). Maryland Virginia 2018 Load 2025 Goal 2018 Load 2025 Goal (106 lb/yr) (106 lb/yr) Required % (106 lb/yr) (106 lb/yr) Required % Sector TN / TP TN / TP Reduction TN / TP TN / TP Reduction Agriculture NP 22.1 / 0.63 18.9 / 0.53 14% / 16% 20 / 1.5 12 / 0.9 40% / 40% Urban/Suburb NP 9.2 / 0.68 7.5 / 0.48 18% / 29% 10.9 / 1.35 10.3 / 1.2 5.5% / 11% Point WWTP 10 / 0.48 11.5 / 0.79 Achieved 12.5 / 1.0 16 / 12 Achieved Total 52 / 4.0 48 / 3.6 7.7% / 10% 58.1 / 6.5 56 / 6.2 3.6% / 4.6%

Table 4-10 indicates that point source reduction goals (2025) have been achieved in Virginia and Maryland, with the most significant reductions still required in agriculture (Virginia) and urban/suburban runoff (Maryland). For the purposes of this analysis, an average TOTAL reduction for the Potomac was assumed to be equivalent to load-weighted average reduction requirements for Maryland and Virginia, 5.5% TN reduction and 7.3% TP reduction. 4.3.1.2 Quantitative Performance Metrics Efficiency of Reduction – CECs Data from Tables 2-6 and 2-8 was analyzed to understand the efficiency of CEC reduction by BMPs in each sector. Non-point source performance for CECs was determined by averaging reductions of measure estrogens and pesticides. Point source calculations for impact of Advanced Water Treatment was performed by comparing the flow-averaged concentrations from the UOSA facility as compared to the flow-averaged concentrations associated with the Blue Plains WWTP and the Seneca WWTP. Point source data for advanced nutrient control of CECs at advanced nutrient control WWTPs was not available in the collected data, as all sampled facilities were advanced nutrient control facilities. Previous data has shown mixed results for CEC control from the nutrient control process at the Blue Plains WWTP, as have previous studies. A recent compilation of literature found in Tetra Tech (2019), indicated mixed performance with some studies reporting greater than 90% removal of some CECs with advanced nutrient control, while others showing “despite the substantial improvements in removal of BOD, TSS, nutrients, and other regulated water quality parameters, there seemed to be no significant increase in reduction of CEC concentrations as a result of facility upgrades”. This data was compiled to develop an estimate for CEC removal with advanced nutrient treatment found at Potomac Watershed facilities. Table 4-11. CEC Reduction Performance Summary. CEC Removal Performance Agriculture Urban BMPs Point – WRF Point - AWT EDCs 89% 56% 63% Pesticides 28% 44% 67% All CECs 46% 43% 25%* 65% *Estimated from Compilation of recent literature studies.

Potential for Impact - Sector-Specific BMP Implementation Extensive work was performed within this study to estimate concentrations and loads of nutrients and several emerging contaminants to the Potomac Watershed, summarized in Table 2-7. Additionally, Year 2 work was utilized to assess impacts of BMPs on potentially reducing loads of nutrients and CECs for agricultural and urban stormwater BMPs, as well as advanced nutrient removal in wastewater treatment and even advanced wastewater treatment. The results are summarized in Table 4-12. In order to assess impact of BMPs on nutrient and CEC loads, Table 4-12 was developed utilizing BMP performance data in an effort to summarize reduction in TN as a function of BMP adoption throughout the watershed, per sector. In order to provide a realistic assessment, the performance evaluation was

106 The Water Research Foundation performed as a function of 10%, 20%, and 50% BMP adoption throughout the watershed, per individual category (agriculture, urban, point source water reclamation and point source advanced water treatment. In order to simplify comparison, any value impacting total load by less an 0.5% was neglected from the analysis. The values in bold are associated with TN and TP reductions greater than the goal reduction values. Again, the analysis indicates that implementing BMPs throughout the watershed to target agriculture inputs provides the most significant reductions in nutrients, and CECs. Table 4-12. Total Watershed Load Reductions Associated With Sector-Specific BMP Implementation. 10% Watershed BMP 20% Watershed BMP 50% Watershed BMP Implementation Implementation Implementation Constituent Agr Urb WRF AWT Agr Urb WRF AWT Agr. Urb. WRF AWT Water 0.8% 2.0% TDN 3.0% 6.1% 0.6% 0.7% 0.7% 15.7% 1.5% 1.8% 1.8% SRP 3.8% 0.7% 0.8% 0.8% 7.6% 1.5% 0.9% 0.9% 19.0% 3.7% 2.3% 2.3% DOC 1.2% 0.8% 2.4% 1.6% 0.2% 5.9% 4.1% 0.6% 1.1% E1 4.1% 1.3% 8.2% 2.5% 20.5% 6.2% Atrazine 6.9% 13.9% 0.9% 34.7% 2.1% Metolachlor 3.6% 0.9% 7.2% 1.7% 18.0% 4.4% 1.0% Simazine 2.4% 4.8% 12.1% 0.7% Prometon 0.9% 1.9% 4.7% Notes: Highlighted cells (yellow): Greater than target reduction of TN and TP. Assumptions: 1. Nonpoint source removal of TDN based on results of this study, WRF and AWT removal based on assumed regulation of 1 mg/L TDN (from current 3 mg/L average) 2. Nonpoint source removal of SRP performance based, WRF and AWT based on assumed regulation of 50% additional reduction 3. Nonpoint source removal of CECs based on observations. Point source (WRF) removal based on comparison of advanced nutrient control (Blue Plains) to Seneca. Point source (AWT) based on comparison of AWT for reuse (UOSA) vs. average of WRF.

4.3.1.3 Quantitative Cost Metrics Cost metrics were developed to understand impacts of implementing nutrient and CEC reduction strategies. The cost metrics identified as of interest in the workshop evaluation include incremental costs of nitrogen removal ($/lb N), costs of removing contaminants of emerging concern ($/lb CEC), along with an aggregate Capital Cost of implementation. For informational purposes, the cost of phosphorous removal ($/lb-P) was also included. Incremental Cost Metrics Incremental costs were found through literature review. Costs of non-point source BMP implementation for nutrient control in the Chesapeake Bay watershed have been compiled regularly for estimating expenditures. These compilations show that implementation costs can vary widely, so a range of incremental costs are provided from several sources. Costs for BMPs at point sources (advanced nutrient control at Water Reclamation Facilities (WRF), and nutrient control through Advanced Water Treatment (AWT) Technologies) were compiled on a treated flow basis. These numbers have been normalized to $/lb-N with the assumption that implementation in the Potomac would be targeted to remove TN from 3 mg/L to 1 mg/L. Additionally, unit costs for CEC co-management were calculated by calculating N removal costs for each strategy and applying to back-calculate an incremental cost of management for a range of CECs. For simplicity, $/lb-CEC removed values are shown for atrazine (atz), which shows the lowest incremental

Improving Water Reuse for a Healthier Potomac Watershed 107 cost removal of CECs evaluated along with estrone (E1), which showed the highest incremental cost removal of CECs evaluated. Estimates of BMP implementation are summarized in Table 4-13.

Table 4-13. Incremental Costs of BMP Implementation to Address TN, TP, and CECs. Sector $/lb-N $/gpd (for N) $/lb-P $/lb-CEC5 Agriculture 50 (26 – 90 in MD)1 N/A 1,000 (VA, up to 1,400)2 $0.4M (atz) – 100 (5 – 600 in VA)2 $140M (E1) Urban 633 (375 – 1222 in MD)1 N/A 10,000 (VA, Up to $8.1M (atz) - 300 (5 – 1000 in VA)2 80,000)2 $562M (E1) WRF 1,1254 7 (0.2 – 15)3 $283M (atz) – $136B (E1) AWT 2,4104 15 (7 – 22)3 $215M (atz) - $301B (E1) Notes: 1. Range presented in Environmental Finance Center, University of Maryland (2015). 2. Range presented in Environmental Finance Center, University of Maryland (2017). 3. Range presented U.S. EPA (2015). 4. Calculated assuming upgrades required to achieve 2mg/L TN removal (3mg/L  1 mg/L) 5. Calculated based upon $ required spent per annual lb-TN spent for each technology

Aggregate Cost Metrics Total Capital Cost required to achieve Chesapeake Bay TMDL limits for TN by 2025 were calculated according to sector contributions. As described in Table 4-10, achieving the TN TMDL requires a total TN removal contribution of 7.7% in Maryland and 3.6% in Virginia. To describe the Potomac, a combined VA and MD TN percent reduction was calculated at 5.5% TN load reduction, or 4.5 x 106 lbs N/yr in the Potomac (calculated as 5.5% of the Total TN load observed in Table 2-7). With 4.5 x106 lb-N/yr serving as the base load, calculations were performed to describe the percent removal from the observed load (Table 2-7) categorized by sector, summarized in Table 4-14. The data clearly indicate that removing the required TN load through urban runoff control or point source control present significantly greater demands for controlling N loadings through either urban nonpoint or point source control when compared to agriculture nonpoint source control. Further, cost information provided in Table 4-13 can be applied to suggest a sector-specific cost of compliance, assuming required load reductions are applied individually per sector. The cost information further illustrates the economic advantages of requiring a significant portion of the removal requirements be achieved through agriculture based TN reductions. Table 4-14. Sector-Specific Aggregate Costs Required for Removing TN to Achieve 2025 TMDL. 2018 TN Load Sector Specific Removal to Sector Specific Cost to Achieve Sector (106 lb-N) Achieve 4.5M lb-N Removal 4.5M lb-N Removal Agriculture NP 65.3 6.8% $225 M Urban NP 12.1 37.2% $1.13 B Point (WRF or AWT) 4.5 100% $3.3 B - $7.1 B

Recognizing challenges associated with relying solely on one sector to achieve all nutrient load reductions, Table 4-15 presents results from an optimization exercise, where cost of compliance was compared across three scenarios. The scenarios considered are an agriculture only reduction scenario (analogous to that described in Table 4-14), an equal distribution of reduction across non-point sources (agriculture and urban stormwater), and an equal distribution of reduction across non-point and point sources. The analysis indicates that even though the per sector load reduction requirements are less, the cost of compliance increases significantly when the compliance responsibility is shared.

108 The Water Research Foundation Table 4-15. Aggregate Costs of “Shared” TN Reduction to Achieve 2025 TMDL. Required % Removal Aggregate Cost of Sector Specific Cost to Achieve Scenario from Each Sector Compliance 4.5M lb-N Removal Agriculture only 18.2% $225 M $225 M Combined NP 16.5% 37.2% $306 M All Combined 14.9% 100% $767 M – $1.3 B

4.3.2 Qualitative Metrics In addition to the four quantitative metrics (incremental costs for Nitrogen and CECs, aggregate capital cost, and performance), six metrics which are semi-quantitative or qualitative in nature were identified. These metrics need to be defined by the user on a 1-5 scale, and include affordability, complexity of Implementation, Timeline to benefits (subsequent to implementation), complexity of policy/regulation, confidence of performance, and geographic distribution of benefits. Each individual user or stakeholder will determine their scoring for each qualitative metric, and the following information is provided as information regarding each metric. Affordability Capital costs and incremental costs are provided as separate quantitative criteria addressing cost and performance, respectively, but affordability is included as an equity benefit, used to describe differences in impacts to a single user. On the surface, affordability seems skewed to favor of large, publicly funded ventures, such as POTWs and urban infrastructure and unfavorable for agriculture BMPs, which often require cost share and partnerships with individual farms. However, several state and federal cost-share and grant/loan programs exist for each alternative, which are described below: • Agriculture BMPs – Programs exist in both Maryland and Virginia to specifically fund agricultural BMPs, providing farm-level aid for improvements which reduce nonpoint pollutant loads to the Chesapeake Bay. Grant programs exist to enhance affordability for individual landowners. These programs include:

o Maryland: Cost share and subsidy programs provide most of the assistance for agriculture BMP implementation in Maryland. The programs are implemented as partnerships between the state and federal agencies, and while they are not directly tied to nutrient emissions, many do incentivize implementation of key water quality improvement practices. For the entire state of Maryland, these programs are expected to provide more than $730M to subsidize agricultural improvements, significantly improving affordability of these measures.

o Virginia: Virginia’s Water Quality Improvement Fund was established to provide funding for nonpoint projects. The program is administered by the Virginia Department of Environmental Quality and the Department of Conservation and Recreation with the purpose of providing “Water Quality Improvement Grants to local governments, soil and water conservation districts, institutions of higher education and individuals for point and nonpoint source pollution prevention, reduction and control programs.” Within this program, the Virginia Natural Resources Commitment Fund supports installation of agricultural BMPs and associated technical assistance, stipulating 55% for matching grants within or partially within the Chesapeake Bay Watershed. Since implementation in 2010, funding for the program has varied widely from year- to-year, but has averaged ~$30M annually. Projecting this funding level represents more than $450M over the 15 year TMDL implementation period, with nearly $250M (55%) targeted to matching grants within the Bay Watershed.

Improving Water Reuse for a Healthier Potomac Watershed 109 • Stormwater BMPs – Regulatory programs have been developed to force municipalities to fund Urban BMP implementation, with some fund available through select grant programs.

o Maryland – Financing stormwater management has traditionally been the responsibility of local governments. In most communities, stormwater management projects are funded from the general fund, and must compete with other funding priorities. More recently, communities have developed service fees and stormwater enterprise funds. While still paid by local taxpayers, these funds provide dedicated funding for stormwater management projects. IN 2012, House Bill 987 was passed which required localities develop mandatory fee-based stormwater financing and revenue programs within urban communities across the state, subject to MS4 requirements.

o Virginia – The Virginia Stormwater Local Assistance Fund (SLAF) was created in 2013, providing matching grants to local governments to plan, design, and implement cost-effective BMPs that reduce pollutant loads from stormwater runoff. In its existence, the Fund has been funded at varying levels, ranging between $5M in 2016 to $35M in 2014. Funded projects are tied to affordable BMPs, with a threshold of $50,000 or less to treat each pound of Total Phosphorous, a representative pollutant. In addition, the Virginia Clean Water Revolving Loan Fund provides low-interest loans to support water quality improvements projects throughout the state. • Point Source Treatment – POTWs can receive low interest loans, but ultimately finance improvements through customer rates.

o Virginia – The Commonwealth of Virginia provides grant and funding opportunities for localities to construct wastewater treatment facilities, through the Virginia Clean Water Revolving Loan Fund (low interest loans for WWTP upgrades and innovative solutions), The Virginia Quality Improvement Fund (finances nutrient reduction strategies at publicly owned wastewater treatment facilities primarily within the Chesapeake Bay Watershed), and the Green Project Reserve (projects that utilize green or “soft path” practices to complement hard/gray infrastructure, adapt practices that reduce environmental footprints for water and wastewater treatment, provide energy conservation, and help utilities provide sustainable solutions) Complexity of Implementation • Agriculture – Nonpoint BMPs in the agriculture sector can range from relatively simple measures to implement with high nutrient reductions, including practices such as livestock exclusion, grass buffers, effective manure management) to more complicated measures (e.g., upland precision intensive rotational grazing, wetland restoration, tree planting, land retirement). Generally Non- complex projects • Urban – Nonpoint BMPs in the urban stormwater sector can be significantly more complicated than agriculture BMPs, specifically due to proximity to large population centers, which introduce complications such as more landowner approval for larger projects, managing impacts to the population (i.e., utilities, traffic, urban infrastructure, etc.). In addition, managing the intended use of large scale stormwater management projects for long-term performance can be challenging. Moderate complexities associated with urban infrastructure and retrofits. • Point Source – These are the most complex implementation strategies, requiring advancement of technology in centralized wastewater treatment facilities. Nutrient reduction processes and advanced water treatment technologies are often complex and expensive to design, build, and operate and require trained professionals to ensure they are meeting performance. Very complex implementation of advanced technology.

110 The Water Research Foundation Implementation – Timeline to Benefits There has been significant research into determining the timeline to benefits of nutrient reduction strategies on ecological health of the Chesapeake Bay. As significant portions of the current nutrient loads within and arriving in the Bay every year can be attributed to “legacy” deposits of nutrients in the sediments of Bay tributaries, it is expected to be many years for ecological impacts to be felt. Ancillary benefits of many of these improvements, however, may be felt much sooner, specifically quality of life benefits and infrastructure hardening associated with urban stormwater management improvements (i.e., more greenspace and less safety concerns/storm damage). Additionally, advance water treatment strategies can provide for potable reuse (as in the Occoquan system), potentially providing for water source reliability for regional drinking water purveyors.

Complexity of Policy/Regulation • Agriculture – Developing policy or regulations to govern agriculture emissions of pollutants has proven very difficult. With potentially broad impacts to the regional economy (>300,000 jobs in Virginia are tied to agriculture), regulation has been strongly opposed and pollutant load reductions rely heavily on subsidies and voluntary action. Virginia does not have individual permits for animal feeding operations. In recent years, several legislative proposals that would have required mandatory agricultural BMPs in Virginia have failed to gain support. In Maryland, the state’s Nutrient Management Planning program requires all farmers grossing $2,500 a year or mor or livestock producers with 8,000 pounds or more of live animal weight to follow nutrient management plants (NMPs) when fertilizing crops and managing animal manure. In addition, discharges from large animal feeding operations are regulated through Clean Water Act NPDES permits, targeting nutrient load reductions. • Stormwater – Urban stormwater is regulated in both Maryland and Virginia, particularly through each state’s Municipal Separate Storm Sewer Systems (MS4), which regulates nutrient loads that are the result of existing urban development (pre-1994). In Maryland, the 10 largest subdivisions (nine counties and the City of Baltimore) are Phase I MS4 communities, and they account for more than 75% of the state’s total urban nutrient load to the Chesapeake Bay. Regulations in Maryland are requiring in MS4 permit renewals implementation of plans to address waste load allocations, and to treat runoff from impervious surfaces. In Virginia, more than half of the nitrogen and phosphorous loads stem from unregulated sources, namely areas that are not subject to MS4 discharge permits. • Point Sources – Point Source pollution through WWTPs has been heavily regulated in the Bay Watershed, with mandatory limits reductions regularly imposed on POTWs with both Virginia and Maryland. In the Potomac, most utilities are achieving less than 3.0 mg/L TN and 0.3 mg/L TP effluent discharge, nearing the limit of feasibility, and POTWs have led the way toward nutrient load reductions in the region. From a regulatory perspective, imposing limits on POTWs is relatively easy to do. An exception to this rule lies within potable reuse, which is currently regulated on a case-by- case basis. Confidence of Performance • Agriculture – In Clary et al. (2017), the WE&RF published a Final Report entitled Agricultural Best Management Practices Database: Version 2.0 Data Summary. In the report, the authors indicate that “For the most part, the initial findings of the AgBMPDB Version 2.0 align with expectations for BMP performance as presented in the literature”, particularly as they related to nutrient management practices, no-till and conservation tillage practices, and cover crops. However, the authors also state that “challenges of effectively analyzing agricultural research data are evident due to the number of variables that combine to determine pollutant loading and BMP performance at a given site”. The limited confidence in accurately assessing BMP performance data for agriculture BMPs was also observed in this study, evidenced by wide variability in comparing co-located BMP

Improving Water Reuse for a Healthier Potomac Watershed 111 performance, and led to limited confidence in understanding BMP performance. Additionally, longterm confidence in agriculture BMP performance is tied to maintenance, which often must be performed by individual landowners, leading to a high potential for reduced performance in the future. This leads to low confidence of performance for ag-based BMPs. • Urban – Performance of Urban Stormwater BMPs has been studied extensively, with compilations of performance data readily available in the International Stormwater BMP Database (http://www.bmpdatabase.org/). Extensive summaries of the data have been published, the most recent being the 2016 International Stormwater BMP Database 2016 Summary Statistics (Clary et al. 2017b). The report compiles extensive performance data for nutrient removal, sediment load limitation, as well as reduction of bacteria and metals. Available since 1996, the database has been successful in its overall purpose, “to provide scientifically sound information to improve the design, selection and performance of BMPs.” “As of the most recent release in November 2016, the BMP Database contains data sets from nearly 650 BMP studies through the U.S. and several other countries”. This extensive dataset provides a greater degree of confidence in assessing urban BMP performance as compared to agriculture based BMPs. In addition, as these efforts are often publicly funded projects, maintenance of the BMPs is often rolled into municipal operating budgets. And while funding of project O&M may suffer periodically, the accountability involved is significantly more manageable than if their continued success was destined by the whims of individual landowners. This leads to moderate-to-high confidence of performance for Urban BMPs. • Point Source – Point source nutrient reduction strategies at water reclamation facilities have been studied for decades and are very well understood. In addition, long-term performance in guaranteed through regulatory constructs that force POTWs to operate under measured permitted limitations, proving their performance on annual bases. This leads high confidence of performance for Point Source BMPs. Geographic Distribution of Benefits • Agriculture – As shown in Figure 1-3, the agriculture land use in the Potomac River Watershed is primarily located in the Upper Potomac, in the Shenandoah Valley and North into Maryland. Benefits of agriculture BMPs therefore, provide ecological benefits to much of the impacted Potomac, plus public health benefits to drinking water utilities located primarily in the downstream DC Metro area. Therefore, agriculture BMPs are expected to have high geographic distribution of benefits concerning both ecological and human health. • Urban – The DC Metro area is the primary high-density urban center in the Potomac River Watershed, with a large impact of both urban and suburban development. The urban center is located at the Fall line, with relatively limited urban or agriculture development below the fall. However, the entirety of the Bay ecosystem lies below the urban centers as well. Therefore, urban BMPs are expected to have moderate geographic distribution of benefits concerning ecological health, but limited public health benefits • Point Source – Most opportunities for impacting wastewater lie in the urban centers. However, the Washington, D.C. metropolitan area has a long history of reusing reclaimed water in an urban water cycle. In addition, compared to the runoff driven sectors, the treated wastewater flows are modest. Therefore, point source (advanced nutrient control) provides limited geographic distribution of benefits concerning ecological health, but high public health benefits.

112 The Water Research Foundation 4.4 Scoring Alternatives To complete the HazenConverge Multi-criteria Decision Support Exercise, raw scores were developed for each of the criteria stated above. As described earlier, categories were developed through a stakeholder workshop process, and direct pairwise comparison was undertaken to develop criteria weighting. Scoring for the criteria were developed above, and summarized in Table 4-16. Following Table 4-16 are two screen captures from the HazenConverge Tool (Figures 4-3 and 4-4), displaying a summary of the criteria and score inputs to the Tool, respectively.

Improving Water Reuse for a Healthier Potomac Watershed 113 Table 4-16. Summary of Criteria Scores, With Justification. Criterion Approach to Defining Agriculture Urban Point - WRF Point - AWT Cost/reduction TN Quantitative – Literature Review $50 / lb-N $633 / lb-N $1,125 / lb-N $2410 / lb-N

CEC Reduction Efficiency Quantitative – Performance Data 46% 43% 25%* 65% (*lit. review estimate)

Affordability Qualitative – Availability of support Grant funding Funded through Funded through POTW rate payers. funding available operating budgets. Some grant funding available

Complexity of Qualitative – Literature Assessment Non-complex tech Moderate complex Complex implementation of advanced Implementation options urban retrofits. technology

Capital Cost Quantitative – Agg. Cost to required $225 M $306 M $767 M $1.3B 2025 TN TMDL Potential for Impact Quantitative – % of 2018 observed 6.8% 37.2% 100% load (this study) required for 2025 TN TMDL Implementation timeline Qualitative - Literature Assessment Ecological and Human Health Benefits Likely to take 10 – 15 years after implementation for all to benefits approaches. Complexity of Qualitative – Regulatory Review Complex, limited Modestly complex– MS4 Not complex – robust No consistent reuse Policy/Regulation current structure programs currently exist. structure exists regulatory infrastructure Confidence of Qualitative – Literature Assessment Nutrients – Low Nutrients – High Nutrients – High Nutrients – High performance confidence confidence confidence confidence CECs – low CECs – low confidence CECs – low CECs – high confidence confidence confidence Geographic distribution of Qualitative – Ecological / Human Ecological – high Ecological – mod. Ecological – low Ecological – low. benefits Health Benefits Assessment Human – mod. Human - low Human - low Human - high

114 The Water Research Foundation

Figure 4-3. Screen Capture Describing Criteria Development and Weighting Input Into the HazenConverge Tool.

Improving Water Reuse for a Healthier Potomac Watershed 115

Figure 4-4. Screen Capture Describing Raw Score Ranges and Scores Input Into the HazenConverge Tool.

116 The Water Research Foundation Figure 4-5 summarizes the results for the weighted, criteria driven analysis. With each alternative receiving a potential maximum of 100 points, there was a clear difference in total scores for the four alternatives, with agriculture “winning” with 71 points, urban BMPs and Water Reclamation clustered at 53 and 44 points, respectively, and advanced water treatment well behind collecting only 32 points. The Figure also displays relative criteria category weightings with Cost and Implementability holding the most weight with 28% and 26%, respectively, and performance and equity (of cost and benefits) clustered at 21% and 18% impact.

Figure 4-5. Results Summary Panel, With a Breakdown of Weighted Criteria Scores. Agr = Agriculture, Urb = Urban, WRF = Water Reclamation Facility, AWT = Advanced Water Treatment

Improving Water Reuse for a Healthier Potomac Watershed 117 The results summarized in Figure 4-5 indicate that the Stakeholder Panel’s choice in scoring criteria and importance of each reflected an inherent concern about achieving equity in achieving the Bay TMDL goals. With cost metrics driving the evaluation, explicitly focused on reducing nutrients, and equity of costs/benefits, the results clearly reflect the stakeholder panel’s collective opinion that agriculture BMPs provide the most opportunity for impacting human and ecological health, as the least cost and most affordable option. As an example, for the agriculture sector, these two categories accounted for 44 of the 71 points scored, while for the advanced water treatment option, the sum accounted for only 11 points (the entire alternative achieved only 32 points). While developing a regulatory or policy framework for requiring implementation of BMPs within the Agriculture sector was recognized to be a challenge compared to other alternatives, this criteria was weighted low enough in the opinion of the stakeholders that it did not significanty affect the analysis. To provide for additonal analyisis Figures 4-6, 4-7, and 4-8 provide graphical depictions of the same results, as unweighted scores according to category (Figure 4-6), weighted scores according to criteria category (Figure 4-7), and unweighted scores according to criteria category (Figure 4-8).

Figure 4-6. Alternative Scoring Arranged According to Unweighted Criteria.

118 The Water Research Foundation

Figure 4-7. Alternative Scoring Arranged According to Weighted Category.

Figure 4-8. Alternative Scoring Arranged According to Unweighted Category.

Improving Water Reuse for a Healthier Potomac Watershed 119 4.5 Conclusions of the Multi-Criteria Cost/Benefit Assessment A multi-criteria decision analysis was performed to evaluate four alternatives for co-managing nutrients and contaminants of emerging concern. In order to define and rank the importance of criteria for the analysis, a stakeholder workshop was held which resulted in identification and prioritization of 10 specific criteria for analysis, grouped within five categories including cost, performance, equity, implementability, and risk. Scores were developed for each category, with quantitative scores developed for four criteria (cost/reduction N, CEC removal Efficiency, Capital Cost, and Potential Impact), and qualitative scores developed for six criteria (Affordability, Complexity of Implementation, Implementation Timeline to Benefits, Complexity of Policy/Regulation, Confidence of Performance, and Geographic Distribution of Benefits). When weighted scores were developed for each alternative, co- management of nutrients and CECs through implementation of Agriculture BMPs was the highest ranked options, collecting 71 total points (of 100). The next highest score was achieved by Urban BMPs (53 points), followed by point source management at Water Reclamation Facilities (44 points), and point source management through Advanced Water Treatment of Reclaimed Water (32 points).

120 The Water Research Foundation CHAPTER 5 Conclusions

Conclusions were developed based on the analysis performed herein, corresponding with specific tasks of the project, as follows: • Hot Spot and Hot Moment Analysis. • Contaminant Source Allocation. • Impact of Best Management Practices. • Impact of Planned Indirect Potable Reuse. • Multi-Criteria Decision Analysis.

5.1 Hot Spot and Hot Moment Analysis An intensive, quarterly sampling campaign of thirty-one primary tributaries of the Potomac River were sampled during 2016-2017 across seasonal flow regimes. Results of the campaign were used to develop “hot-spot” and “hot-moment” analyses for impacts of sub-watersheds throughout the region. • Hotspots of bulk estrogenic activity (measures with the bioluminescent Yeast Estrogen Screen (BLYES) were found primarily in sub-watersheds dominated by pasture/hay type land uses. The “hot moment” of BLYES was during the spring high-flow period. The BLYES was not significantly correlated with estrogenic chemical compounds (exclusively estrone and estrone-3S). • Herbicides were dominated by atrazine and metolachlor and were far higher in three sub- watersheds than the others. At most sites, concentrations of atrazine and metolachlor were positively correlated with BLYES, and their hot moment was also during the spring high-flow period. • Total dissolved nitrogen (TDN) hotspots occurred in sub-watersheds with larger percentages of cropland land use and TDN concentration was highly correlated with watersheds dominated by cropland. Nitrate isotopic tracers also suggested that soil nitrogen (N) (derived from fertilizers) was the primary N source within the Potomac River. The hot moment for elevated N concentration was during the summer low-flow period. • Concentrations of soluble reactive phosphorus (SRP) increased with increasing degree of agricultural land use, and it was also high in watersheds containing a large percentage of wetlands. TDN was correlated with three common herbicides, while SRP was positively correlated with BLYES. • Hotspots of dissolved organic carbon (DOC) were located in urban areas and DOC concentration was not correlated with BLYES or herbicides. 5.2 Contaminant Source Allocation Results from the monthly sampling campaign in Year 1 were utilized to estimate average annual loads for point sources (POTWs), agriculture (cropland and pasture), and urban. Table 2-7 summarizes the percent of total annual load to the Potomac watershed according to land use, for point sources, agriculture, urban, and forested land use. The analysis shows evidence of the outsized impact of both agriculture and urban non-point sources for EDCs and herbicides, compared to water input to the watershed. The results further suggest that herbicides and TDN, and BLYES and SRP may have common sources. Thus, BLYES and nutrients may be co-managed by applying best management practices for phosphorous (P).

Improving Water Reuse for a Healthier Potomac Watershed 121 5.3 Impacts of Best Management Practices Samples from paired watershed sites, chosen to represent similar land uses (agriculture non-point; urban non-point; conventional, enhanced nutrient, and advanced water treatment point source) were analyzed to correlate implementation of BMPs with reductions in nutrients, EDCs, and pesticides. The following conclusions were drawn from the analysis. • Nonpoint source BMPs showed mixed success in reducing nutrients. Specifically: o TDN was effectively removed with agriculture BMPs tested (38%) o SRP effectively removed with agriculture BMPs tested (51%) o TDN not effectively removed with urban BMPs tested (<20%) o SRP effectively removed with urban BMPs tested (45%) • Nonpoint source BMPs showed significant reductions in EDCs and pesticides. o EDCs were removed with agriculture BMPs tested (89%) and urban BMPs tested (56%). o Pesticides were less effectively removed with agriculture BMPs tested (28%) and urban BMPs tested (44%). • Point source BMPs (Enhanced Nutrient Removal and Advanced Water Treatment) significantly reduced TDN (66%) and SRP (>50%). • Point source BMPs (Water Reclamation Facilities and Advanced Water Treatment) significantly reduced EDCs (Estrogens not observed from WRF discharges) and pesticides (67% AWT). The evaluation was furthered to ascertain the impact of implementing sector specific BMPs on loads to the Potomac Watershed. An analysis of the impact of 10%, 20%, and 50% BMP implementation per sector on the overall loads to the Potomac Watershed was performed, with agriculture BMP implementation providing by far the most potential for nutrient and EDC/pesticide reductions, as shown in Table 4-12. 5.4 Impact of Planned and Unplanned Indirect Potable Reuse The Occoquan system is a planned indirect potable reuse system, located fully within the Potomac Watershed. The Occoquan reservoir serves as the Water Supply for the Griffith Water Treatment Plant, owned and operated by Fairfax Water. The Upper Occoquan Service Authority (UOSA) is an advanced WRF located in the Occoquan Watershed, discharging to the Bull Run, a direct tributary of the Occoquan Reservoir. The UOSA WRF discharges a significant portion of the flow into the reservoir on a daily basis. Levels of nutrients, EDCs, and pesticides were compared from UOSA effluent to the Bull Run background concentrations, to assess impacts of UOSA effluent on Occoquan system water quality. The following conclusions were drawn. • UOSA discharge contained significantly higher levels of TDN and SRP than background levels in the Bull Run. However, SRP did not significantly impact concentrations within the Bull Run. Nitrate (a component of TDN) is purposefully discharged from UOSA into the Occoquan Reservoir in winter in order to prevent anaerobic conditions in the reservoir and improve water quality in the reservoir. • The analysis of EDCs and pesticides in the UOSA discharge yielded the following:

o No estrogens detected in effluent and even zenoestrogens like nonylphenol and bisphenol-A were at lower levels in UOSA effluent than background Bull Run o Pesticides were present in UOSA effluent, at levels greater than and impacting Bull Run. All levels detected were below any U.S. EPA maximum contaminant limit, acute level benchmarks, or health-based screening levels available. o UOSA contributed significantly to the Occoquan mass detected for several pesticides, including several neonicoinoid insecticides and triazine herbicides.

122 The Water Research Foundation o Other compounds, including bisphenol-A, fipronis, chlothianidin present int the Occoquan suggest sources other than UOSA effluent. o Only three SOC loads to the reservoir can be attributed to UOSA discharge (acetamiprid, atrazine, and Tribufos), under average flow and rainfall conditions. To further address the question of impact of planned vs. unplanned indirect potable reuse, the raw water to the Griffith WTP and the Corbalis WTP were monitoring. The Corbalis WTP served as the unplanned IPR test case as the source of water for this facility is the Potomac River. The analysis yielded the following conclusions. • There was very little difference in nutrients and in situ water quality parameters between the two facilities. • Both the unplanned (Corbalis WTP) and planned (Griffith WTP) intakes were similarly impacted from EDCs and pesticides, in terms of number of contaminants detected. • Average concentrations of observed contaminants were calculated. Generally, o Corbalis WTP had higher average concentrations of industrial compounds (4,-nonlyphenole and 4-tert-octylphenol); insecticides (acetamiprid, dichlorvos, and fenchlorphos); and one triazine herbicide (simazine). o Griffith WTP had higher concentrations of two triazin herbicides (atrazine and prometon); neonicotinoids (chlothianidin, dinotefuran, and imidacloprid); bisphenol-A, and metolachlor. • All levels detected at the WTP intakes were below any U.S. EPA maximum contaminant limit, acute level benchmarks, or health-based screening levels available. 5.5 Multi-Criteria Decision Analysis A multi-criteria decision analysis was performed to evaluate four alternatives for co-managing nutrients and contaminants of emerging concern, including implementation of Non-point source (Agriculture, Urban Stormwater) BMPs, increased nutrient control at WRFs, and Advanced Water Treatment for Planned Indirect Potable Reuse. When weighted scores were developed for each alternative, informed by a convened stakeholder workshop of academic, utility, consulting, and NGO representatives; co- management of nutrients and CECs through implementation of Agriculture BMPs was the highest ranked option, collecting 71 total points (of 100). The next highest score was achieved by Urban BMPs (53 points), followed by point source management at Water Reclamation Facilities (44 points), and point source management through Advanced Water Treatment of Reclaimed Water (32 points). With cost metrics driving the evaluation, explicitly focused on reducing nutrients, and equity of costs/benefits, the results clearly reflect the stakeholder panel’s collective opinion that agriculture BMPs provide the most opportunity for impacting human and ecological health, as the least cost and most affordable option.

Improving Water Reuse for a Healthier Potomac Watershed 123

124 The Water Research Foundation APPENDIX A

Presentations from Multi-Criteria Decision Framework Workshop

Improving Water Reuse for a Healthier Potomac Watershed 125

126 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 127

128 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 129

130 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 131

132 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 133

134 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 135

136 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 137

138 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 139

140 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 141

142 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 143

144 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 145

146 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 147

148 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 149

150 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 151

152 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 153

154 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 155

156 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 157

158 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 159

160 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 161

162 The Water Research Foundation

Improving Water Reuse for a Healthier Potomac Watershed 163

164 The Water Research Foundation APPENDIX B

Results of the Pairwise Comparison Exercise

Improving Water Reuse for a Healthier Potomac Watershed 165

166 The Water Research Foundation References

Álvarez-Salgado, X.A. and E.J. Miller. 1998. Simultaneous determination of dissolved organic carbon and total dissolved nitrogen in seawater by high temperature catalytic oxidation: conditions for precise shipboard measurements. Marine Chemistry, 62(3-4): 325-333. Ator, S.W., J.D. Blomquist, J.W. Brakebill, J.M. Denis, M.J. Ferrari, C.V. Miller, and H. Zappia. 1998. Water Quality in the Potomac River Basin, Maryland, Pennsylvania, Virginia, West Virginia, and the District of Columbia, 1992-96, U.S. Geological Survey Circular 1166, pp 38. Batt, A.L. and D.S. Aga. 2005. Simultaneous analysis of multiple classes of antibiotics by ion trap LC/MS/MS for assessing surface water and groundwater contamination. Anal Chem., 77(9): 2940-7. Bishop, P.L., W.D. Hively, J.R. Stedinger, M.R. Rafferty, J.L. Lojpersberger, and J.A. Bloomfield. 2005. Multivariate analysis of paired watershed data to evaluate agricultural best management practice effects on stream water phosphorus. J. Environ. Qual., 34 (3): 1087-1101. Blazer, V.S., L.R. Iwanowicz, H. Henderson, P.M. Mazik, J.A. Jenkins, D.A. Alvarez, and J.A. Young. 2012. Reproductive endocrine disruption in smallmouth bass (Micropterus dolomieu) in the Potomac River basin: spatial and temporal comparisons of biological effects. Environ Monit Assess., 184(7): 4309-34. Casciotti, K.L., D.M. Sigman, M.G. Hastings, J.K. Böhlke, and A. Hilkert. 2002. Measurement of the oxygen isotopic composition of nitrate in seawater and freshwater using the denitrifier method. Anal Chem., 74(19): 4905-12. Chesapeake Bay Foundation. 2019. “2019 Chesapeake Bay State of the Blueprint: Pennsylvania, Maryland, and Virginia” Accessed July 20, 2020. https://www.cbf.org/document-library/clean-water- blueprint/2019-state-of-the-blueprint.pdf. Ciparis, S., L.R. Iwanowicz, and J.R. Voshell. 2012. Effects of watershed densities of animal feeding operations on nutrient concentrations and estrogenic activity in agricultural streams. Sci Total Environ., 414: 268-276. Clary J., Jones J., Leisenring M., Strecker E., Koryto K. 2017a. Final Report: Agricultural Best management Practices Database (AgBMPDB) Version 2.0 Data Summary. Water Environment & Reuse Foundation, WE&RF Project: SIWM14T16/15T16. Clary J., Jones J., Leisenring M., Strecker E., Koryto K. 2017b. Final Report: International Stormwater BMP Database 2016 Summary Statistics. Water Environment & Reuse Foundation, WE&RF ISBN: 978-1- 94124-285-8. Dernoeden, P.H, J.E. Kaminski, and S.J. McDonald. 2005. Maryland 2004 smooth crabgrass control studies in turf. Proceedings Northeastern Weed Science Society 59: 97-98. Duan, S.W. and S.S. Kaushal. 2013. Warming increases carbon and nutrient fluxes from sediments in streams across land use. Biogeosciences, 10: 1193-1207. Duan, S.W., Y.X. He, S.S. Kaushal, T.S. Bianchi, N.D. Ward, and L.D. Guo. 2017. Impact of Wetland Decline on Decreasing Dissolved Organic Carbon Concentrations along the Mississippi River Continuum. Front. Mar. Sci. doi:10.3389/fmars.2016.00280. https://www.osti.gov/servlets/purl/1347840. Duan, S.W., K. Delaney-Newcomb, S.S. Kaushal, S.E.G. Findlay, and K.T. Belt. 2014. Potential effects of leaf litter on water quality in urban watersheds. Biogeochemistry, 121: 61-80.

Improving Water Reuse for a Healthier Potomac Watershed 167 Dutta, S.K., S.P. Inamdar, J. Tso, and D.S. Aga. 2012. Concentrations of free and conjugated estrogens at different landscape positions in an agricultural watershed receiving poultry litter. Water, Air, & Soil Pollution, 223: 2821-2836. Environmental Finance Center, University of Maryland. 2015. Maryland’s Chesapeake Bay Restoration Financing Strategy Final Report Prepared for the Maryland Department of Natural Resources and the Maryland Department of the Environment. Accessed 8/17/2020 at https://arch.umd.edu/sites/default/files/docs/publications/md_chesa_resotration_financing_2015_0.pdf. Environmental Finance Center, University of Maryland. 2017. Strategies for Financing Chesapeake Bay Restoration in Virginia. Prepared for the Chesapeake Bay Program Office on behalf of the Commonwealth of Virginia. Accessed 8/17/2020 at https://www.chesapeakebay.net/documents/Strategies_for_Financing_Ches_Bay_Restoration_in_VA_F INAL_9.26.17.pdf Golterman, H.L. 2001. Phosphate release from anoxic sediments or `What did Mortimer really write?' Hydrobiologia, 450: 99-106. Gordon, L.I., J.C. Jennings, A.A. Rosse, and J.M. Krest. 1993. A suggested protocol for continuous flow automated analysis of seawater nutrients (phosphate, nitrate, nitrite and silicic acid) in the WOCE Hydrographic program and the Joint Global Ocean Fluxes Study. WOCE Hydrographic Program Office, Methods Manual WHPO 91-1, 55 p. Hanselman, T.A., D.A. Graetz, and A.C. Wilkie. 2003. Manure-borne estrogens as potential environmental contaminants: a review. Environ Sci Technol., 37(24): 5471-8. Hayes, T., K. Haston, M.T., A. Hoang, C. Haeffele, and A. Vonk. 2003. Atrazine-induced hermaphroditism at 0.1 ppb in American leopard frogs (Rana pipiens): laboratory and field evidence. Environ Health Perspect., 111(4): 568-575. Hayes, T.B., L.L. Anderson, V.R. Beasley, S.R. deSolla, T. Iguchi, H. Ingraham, P. Kestemont, J. Kniewald, Z. Kniewald, V.S. Langlois, E.H. Luque, K.A. McCoy, M. Muñoz-de-Toro, T. Oka, C.A. Oliveira, F. Orton, S. Ruby, M. Suzawa, L.E. Tavera-Mendoza, V.L. Trudeau, A.B. Victor-Costa, and E. Willingham. 2011. Demasculinization and feminization of male gonads by atrazine: Consistent effects across vertebrate classes. Journal of Steroid Biochemistry and Molecular Biology, 127:64-73, doi:10.1016/j.jsbmb.2011.03.015. He, P. and D.S. Aga. 2019. Comparison of GC-MS/MS and LC-MS/MS for the analysis of hormones and pesticides in surface waters: Advantages and pitfalls. Analytical Methods, 11, DOI: 10.1039/C8AY02774A. Kaushal, S.S., P.M. Groffman, L.E. Band, E.M. Elliott, C.A. Shields, and C. Kendall. 2011. Tracking nonpoint source nitrogen pollution in human-impacted watersheds. Environ Sci Technol, 45(19): 8225-8232. Kaushal, S.S., K. Delaney-Newcomb, S.E.G. Findlay, T.A. Newcomer, S. Duan, M.J. Pennino, G.M. Sivirichi, A.M. Sides-Raley, M.R. Walbridge, and K.T. Belt. 2014. Longitudinal patterns in carbon and nitrogen fluxes and stream metabolism along an urban watershed continuum. Biogeochemistry, 121: 23-44. Kendall, C. and J.J. McDonnell (editors). 1998. Isotope Tracers in Catchment Hydrology (1st Edition). Amsterdam in the Netherlands: Elsevier Science. O’Neil, M.J., ed. 2006. The Merck Index. 14th ed. Whitehouse Station: Merck & Co., Inc. Pennino, M.J., S.S. Kaushal, S.N. Murthy, J.D. Blomquist, J.C. Cornwell, and L.A. Harris. 2016. Sources and transformations of anthropogenic nitrogen along an urban river – estuarine continuum. Biogeosciences, 13(22): 6211-6228.

168 The Water Research Foundation Sanseverino, J., R.K. Gupta, A.C. Layton, S.S. Patterson, S.A. Ripp, L. Saidak, M.L. Simpson, T.W. Schultz, and G.S. Sayler. 2005. Use of Saccharomyces cerevisiae BLYES expressing bacterial bioluminescence for rapid, sensitive detection of estrogenic compounds. Appl Environ Microbiol. 71(8): 4455-60. Salste, L., P. Leskinen, M. Virta, and L. Kronberg. 2007. Determination of estrogens and estrogenic activity in wastewater effluent by chemical analysis and the bioluminescent yeast assay. Sci Total Environ., 378: 343-351. Shanks, K. and A. Soehl. 2008. “Riparian protection improves in-stream habitat and water quality.” EMDE 3 (7), http://mde.maryland.gov/programs/ResearchCenter/eMDE/Pages/vol3no7/ success.aspx. Tetra Tech Center for Ecological Sciences. (2019). Assessing Benefits of Wastewater Treatment Plant Nutrient Control Upgrades on Toxic Contaminants. Prepared for: Toxics Work Group, Chesapeake Bay Trust. Tso, J. and D.S. Aga. 2010. A systematic investigation to optimize simultaneous extraction and liquid chromatography tandem mass spectrometry analysis of estrogens and their conjugated metabolites in milk. J Chromatogr A 1217: 4784-4795. Tso, J., S. Dutta, S. Inamdar, and D.S. Aga. 2011. Simultaneous analysis of free and conjugated estrogens, sulfonamides, and tetracyclines in runoff water and soils using solid‐phase extraction and liquid chromatography‐tandem mass spectrometry. J. Agric. Food Chem. 59: 2213-2222. United States Environmental Protection Agency (U.S. EPA). 2015. A Compilation of Cost Data Associated with the Impacts and Control of Nutrient Pollution. U.S. EPA Office of Water. EPA 820-F-15-096. Accessed 8/17/2020 at https://www.epa.gov/sites/production/files/2015-04/documents/nutrient-economics- report-2015.pdf. United States Environmental Protection Agency (U.S. EPA). 2019. “4-Nonylphenol.” Accessed August 6, 2019. https://iaspub.epa.gov/tdb/pages/contaminant/contaminantOverview.do?contaminantId=10220. USGS. 2019. “Streamstats.” Accessed August 11, 2018. https://streamstats.usgs.gov/ss/. Vanderford, B.J. and S.A. Snyder. 2006. Analysis of Pharmaceuticals in Water by Isotope Dilution Liquid Chromatography/Tandem Mass Spectrometry. Environ. Sci. Technol., 40(23): 7312-7320.

Improving Water Reuse for a Healthier Potomac Watershed 169

170 The Water Research Foundation

advancing the science of water®

1199 North Fairfax Street, Suite 900 6666 West Quincy Avenue Alexandria, VA 22314-1445 Denver, CO 80235-3098