“Novel Approaches to Identify Drivers of Chemical Stress in Small Rivers”

Novel approaches to identify drivers of chemical stress in small rivers

Von der Fakultät für Mathematik, Informatik und Naturwissenschaften der RWTH Aachen University zur Erlangung des akademischen Grades einer Doktorin der Naturwissenschaften genehmigte Dissertation

vorgelegt von

Liza-Marie Beckers, M.Sc.

aus Berlin

Berichter: PD. Dr. Werner Brack

Prof. Dr. Henner Hollert

Tag der mündlichen Prüfung: 15.01.2020

Diese Dissertation ist auf den Internetseiten der Universitätsbibliothek verfügbar.

„Alles geht den Bach runter.“

German proverb

iii

Eidesstattliche Erklärung

Liza-Marie Beckers

erklärt hiermit, dass diese Dissertation und die darin dargelegten Inhalte die eigenen sind und selbstständig, als Ergebnis der eigenen originären Forschung, generiert wurden.

Hiermit erkläre ich an Eides statt

1. Diese Arbeit wurde vollständig oder größtenteils in der Phase als Doktorand dieser Fakultät und Universität angefertigt;

2. Sofern irgendein Bestandteil dieser Dissertation zuvor für einen akademischen Abschluss oder eine andere Qualifikation an dieser oder einer anderen Institution verwendet wurde, wurde dies klar angezeigt;

3. Wenn immer andere eigene- oder Veröffentlichungen Dritter herangezogen wurden, wurden diese klar benannt;

4. Wenn aus anderen eigenen- oder Veröffentlichungen Dritter zitiert wurde, wurde stets die Quelle hierfür angegeben. Diese Dissertation ist vollständig meine eigene Arbeit, mit der Ausnahme solcher Zitate;

5. Alle wesentlichen Quellen von Unterstützung wurden benannt;

6. Wenn immer ein Teil dieser Dissertation auf der Zusammenarbeit mit anderen basiert, wurde von mir klar gekennzeichnet, was von anderen und was von mir selbst erarbeitet wurde;

7. Ein Teil dieser Arbeit wurde zuvor veröffentlicht und zwar in:

Beckers, L.-M.; Busch, W.; Krauss, M.; Schulze, T.; Brack, W., Characterization and risk assessment of seasonal and weather dynamics in organic pollutant mixtures from discharge of a separate sewer system. Water Research 2018, 135, 122-133.

Datum Unterschrift

Abstract The pollution of freshwater bodies with organic micropollutants poses a major threat to human health and the integrity of aquatic ecosystems. In the aquatic environment, organic micropollutants are detected in complex mixtures. Moreover, the mixture composition varies with time (e.g., by seasonal emissions of pesticides), space (e.g., downstream of wastewater treatment plants (WWTPs)) and weather conditions (e.g., due to surface runoff during rain events) leading to highly variable chemical stress in water bodies. This is a major challenge for water monitoring and management. Thus, novel approaches are needed which are able to capture mixture this complexity and variability and to identify drivers of chemical stress. These drivers include risk driving compounds, source-related fingerprints and indicator compounds which present specific exposure scenarios. The approaches developed in this dissertation were applied in the Holtemme River (Saxony-Anhalt, Germany).

In Chapter 2, a multitarget screening approach by liquid chromatography coupled to mass spectrometry in combination with a cluster analysis was applied to identify pollution patterns for seasonal emissions of WWTPs and precipitation-related emissions of a rain sewer. The identified pollution patterns were translated into risk patterns for fish, crustaceans and algae using a toxic unit approach. Acute risk for algae and crustaceans was driven by one to three individual compounds and was mainly caused by seasonal and random emissions, e.g., spills. Sublethal risk by constantly emitted pharmaceuticals posed a potential risk for all studied organisms.

In Chapter 3, spatial pollution patterns along the Holtemme River were unraveled using a nontarget screening (NTS) approach with a longitudinal cluster analysis. Three main pollution patterns were identified reflecting i) inputs from WWTPs, ii) the confluence with the Bode River and iii) diffuse and random input. The latter represented natural background and discharge of untreated wastewater via small point sources such as rain sewers. Main patterns were characterized by specific isotopologue signatures and the number of peaks in homologue series. Further subpatterns revealed source-related fingerprints of WWTP-related inputs. By structure elucidation, 25 representative compounds for these patterns were identified.

In Chapter 4, temporal pollution patterns during heavy rain events were identified using the workflow developed in Chapter 3. In six rain events, two main pollution patterns were unraveled. The two patterns represented i) pre-event emissions which were partly diluted during rain events and ii) increases of rain-related emissions via surface runoff or discharge of untreated wastewater

by combined sewer overflow. Besides a high inter-event variation in mixture composition, a common mixture representing typical rain-related pollutants was determined. From this common mixture, indicator compounds with a high intra-event intensity increase were identified by target screening and NTS. Indicator compounds, which were suitable for event-based water monitoring, included biocides and surfactants for urban surface runoff as well as piperine and chenodeoxycholic acid for emissions of untreated wastewater.

In this dissertation, approaches were developed which allow for capturing and unravelling complex mixtures of organic micropollutants. The findings of this dissertation showed that drivers of chemical stress were related to emissions of untreated wastewater, urban surface runoff and improper discharge of pesticides. This highlights the urgency for management and mitigation efforts to reduce emissions from these sources. However, the identified patterns, risk driving compounds and indicator compounds need to be tested in large scale studies to derive generally- valid organic micropollutants and strategies for water monitoring and management.

vii

Zusammenfassung Neue Ansätze zur Identifizierung von Treibern für chemischen Stress in kleinen Fließgewässern

Die Belastung von Süßwassergewässern mit organischen Mikroschadstoffen stellt ein Risiko für die menschliche Gesundheit und die Integrität aquatischer Ökosysteme dar. Organische Mikroschadstoffe werden nicht nur in komplexen Mischungen detektiert, ihre Mischungszusammensetzung variiert auch je nach Zeit (z.B. mit dem saisonalen Eintrag von Pestiziden), Ort (z.B. flussabwärts von Kläranlagen) und Witterungsbedingungen (z.B. durch den Oberflächenabfluss während Regenereignissen). Dies führt zu stark variierendem chemischen Stress im Gewässer und stellt die Entwicklung von Gewässerüberwachung und – managementstrategien vor große Herausforderungen. Es werden daher neue Methoden und Ansätze benötigt, um die Komplexität und Variabilität der Belastung mit organischen Mikroschadstoffen zu erfassen und Treiber für chemischen Stress zu identifizieren. Zu diesen Treibern zählen risikotreibende Stoffe, quellenspezifische Belastungsmuster und Indikatorsubstanzen. Die in dieser Dissertation entwickelten Methoden wurden in Fallstudien an der Holtemme (Sachsen-Anhalt) angewandt.

In Kapitel 2 wurden saisonale Belastungsmuster aus Kläranlagenabläufen sowie niederschlagsbedingte Belastungsmuster aus einem Regenkanal identifiziert. Dazu wurde eine Multi-Target Analyse mittels Flüssigkeitschromatographie gekoppelt mit Massenspektrometrie und eine Clusteranalyse durchgeführt. Die chemischen Belastungsmuster wurden in Risikomuster für Fische, Krebstiere und Algen durch Berechnung von toxischen Einheiten (toxic units) übersetzt. Dabei wurde ein akutes Risiko für Algen und Krebstiere durch ein bis drei saisonal und zufällig emittierte Mikroschadstoffe bestimmt. Subletale Effekte durch kontinuierlich eingetragene Arzneimittel stellten für alle drei Organismengruppen ein potenzielles Risiko dar.

In Kapitel 3 wurden räumliche Belastungsmuster entlang der Holtemme mittels einer Nontarget- Analyse und einer Clusteranalyse für longitudinale Datensätze ermittelt. Es wurden drei Hauptmuster bestimmt, die den Eintrag i) über Kläranlagenabläufe, ii) über den Zusammenfluss mit der Bode sowie iii) über diffusen Grundwassereintrag von natürlichen organischen Stoffen und zufälligen Eintrag von ungeklärtem Abwasser über kleinere Punktquellen, z.B. Regenwasserkanäle, repräsentierten. Die Hauptmuster wurden durch Isotopenmuster und die Anzahl von chemischen Signalen in homologen Reihen beschrieben. Weitere Untermuster identifizierten quellenspezifische Fingerabdrücke von Abwassereinträgen. Mittels

viii

Strukturauflösung wurden 25 repräsentative organische Mikroschadstoffe für die Haupt- und Untermuster identifiziert.

In Kapitel 4 wurden zeitliche Belastungsmuster während Starkregenereignissen identifiziert. Hierzu wurden die in Kapitel 3 entwickelten Methoden angewandt. Jeweils zwei Hauptmuster wurden in den sechs beprobten Starkregenereignissen entschlüsselt. Das eine Muster beschrieb Substanzen, die vor den Regenereignissen eingetragen wurden und teilweise während der Ereignisse verdünnt wurden. Das andere Muster repräsentierte Einträge von Substanzen über Oberflächenabflüsse und über den Eintrag von unbehandeltem Abwasser durch Mischwasserabschlag. Trotz hoher Variationen in der Mischungszusammensetzung zwischen Ereignissen konnte eine typische Regenmischung von organischen Mikroschadstoffen bestimmt werden. Aus dieser Mischung wurden Indikatorsubstanzen, die einen hohen Intensitätsanstieg innerhalb der Regenereignisse zeigten, mittels Target-Analyse und Nontarget-Analyse bestimmt. Geeignete Indikatorsubstanzen für die Gewässerüberwachung während Starkregenereignissen waren vor allem Biozide und Tenside, die den urbanen Oberflächeneintrag beschrieben, sowie die Stoffe Piperin und chenodeoxycholische Säure, die den Eintrag von unbehandeltem Abwasser repräsentierten.

In der vorliegenden Dissertation wurden Methoden und Ansätze für die Erfassung und Entschlüsselung komplexer Mischungen von organischen Mikroschadstoffen in Fließgewässern entwickelt. Die Ergebnisse zeigten, dass Treiber von chemischem Stress über Einträge von unbehandeltem Abwasser, urbanem Oberflächenabfluss sowie über unsachgemäße Einleitungen von Pestiziden in das Fließgewässer gelangten. Managementstrategien und Maßnahmen sind daher dringend notwendig, die diese Einleitungen reduzieren und Eintragspfade unterbinden. Die identifizierten Belastungsmuster, risikotreibenden Mikroschadstoffe und Indikatorsubstanzen sollten in großskaligen Studien überprüft werden um allgemeingültige Substanzen und Strategien für die Gewässerüberwachung und das Gewässermanagement zu bestimmen.

Table of contents Eidesstattliche Erklärung ...... iv Abstract ...... vi Zusammenfassung ...... viii Table of contents ...... x List of figures ...... xiv List of tables ...... xix Abbreviations ...... xxi Chapter 1

Introduction ...... 1 1.1 Water pollution and degradation of aquatic ecosystems ...... 1 1.2 Complex mixtures and mixture dynamics of organic micropollutants ...... 3 1.2.1 Common sources and entry pathways of organic micropollutants into lotic waters...... 3 1.2.2 Factors influencing the variability in occurrence of micropollutants in lotic waters ...... 4 1.3 Analytical methods to capture complex mixtures of organic micropollutants ...... 6 1.3.1 Chemical methods to characterize complex mixtures of organic micropollutants ...... 6 1.3.2 Multivariate statistics to identify patterns in complex mixtures of organic micropollutants ...... 8 1.3.3 Methods for risk assessment and identification of drivers of chemical stress ...... 9 1.4 The Holtemme River – a case study ...... 10 Vision and thesis’ objectives ...... 11 References ...... 14 Chapter 2

Characterization and risk assessment of seasonal and weather dynamics in organic pollutant mixtures from discharge of a separate sewer system ...... 21 Abstract ...... 22 2.1 Introduction ...... 23 2.2 Methods ...... 25 2.2.1 Study site ...... 25 2.2.2 Sampling ...... 25 2.2.3 Chemical analysis of samples ...... 26 2.2.4 Statistical analysis ...... 27

2.2.5 Calculation of toxic units ...... 28 2.3 Results and Discussion ...... 29 2.3.1 Target compounds emitted from WWTP and rain sewer ...... 29 2.3.2 Temporal patterns in pollutant mixtures ...... 30 2.3.2.1 WWTP effluent ...... 30 2.3.2.2 Precipitation-related emission patterns and emission groups in rain sewer effluent ...... 34 2.3.3 Risk estimation for temporal emissions and emission groups ...... 37 2.3.3.1 Risk patterns and risk driving compounds in WWTP effluent ...... 37 2.3.3.2 Risk patterns and risk driving compounds in rain sewer effluent ...... 40 2.3.4 Exposure in the receiving river ...... 41 2.4 Conclusions ...... 43 Acknowledgements ...... 45 References ...... 46 Chapter 3

Unraveling longitudinal pollution patterns of organic micropollutants in a small river by nontarget screening and cluster analysis ...... 51 Abstract ...... 52 3.1 Introduction ...... 53 3.2 Methods ...... 55 3.2.1 Study site ...... 55 3.2.2 Sampling ...... 56 3.2.3 Chemical analysis of samples ...... 56 3.2.4 Data processing ...... 56 3.2.5 Cluster analysis ...... 57 3.2.6 Characterization of pattern components ...... 58 3.2.7 Structure elucidation ...... 59 3.3 Results and Discussion ...... 59 3.3.1 Longitudinal peak patterns ...... 59 3.3.1.1 Main peak patterns along the river course ...... 60 3.3.1.2 Subpatterns and source-related fingerprints ...... 63 3.3.2 Characterization of pattern components ...... 65 3.3.3 Identification of ions of interest ...... 68 3.4 Conclusions ...... 73 xi

Acknowledgements ...... 74 References ...... 75 Chapter 4

Identification of precipitation-related pollution patterns and indicator compounds during heavy rain events in a small stream ...... 81 Abstract ...... 82 4.1 Introduction ...... 83 4.2 Methods ...... 85 4.2.1 Study area ...... 85 4.2.2 Sampling ...... 86 4.2.3 Data collection hydrology and physicochemical parameters ...... 86 4.2.4 Flow separation of discharge data ...... 87 4.2.5 Chemical analysis ...... 87 4.2.6 Nontarget data processing ...... 87 4.2.7 Cluster analysis ...... 88 4.2.8 Structure elucidation ...... 88 4.3 Results and Discussion ...... 89 4.3.1 Characterization of sampled heavy rain events ...... 89 4.3.2 Chemical pollution during heavy rain events...... 91 4.3.2.1 Dynamics of physicochemical parameters during heavy rain events as potential sampling triggers ...... 92 4.3.2.2 Factors influencing dynamics of pollution patterns in the receiving river...... 94 4.3.3 Identification of common peaks in “Base” and “Quick” patterns ...... 96 4.3.3.1 Common peaks of “Base” patterns – baseline mixture ...... 97 4.3.3.2 Common peaks of Quick patterns – rain-related mixture ...... 98 4.3.3.3 Agricultural pesticides in precipitation-related pollution patterns ...... 101 4.3.3.4 Identified ions of interests in the rain-related mixture ...... 102 4.4 Conclusions ...... 107 Acknowledgement ...... 109 References ...... 110 Chapter 5

Synthesis, conclusions and future research needs ...... 117 5.1 Approaches for the identification of compounds for water monitoring ...... 118 5.1.1 Approach for the identification of risk driving compounds ...... 118 xii

5.1.2 Approaches for the identification of pattern-representative and indicator compounds 119 5.2 Variability and dynamics in complex mixtures ...... 122 5.2.1 Complex mixtures from large point sources ...... 122 5.2.2 Complex mixtures from small and unknown point sources ...... 124 5.2.3 Complex mixtures from rain-related discharge ...... 125 5.3 Future water management and monitoring ...... 126 5.3.2 Future monitoring ...... 126 5.3.1 Future management ...... 128 References ...... 130 Appendix

Appendix A - Characterization and risk assessment of seasonal and weather dynamics in organic pollutant mixtures from discharge of a separate sewer system ...... S1 Appendix B - Unraveling longitudinal pollution patterns of organic micropollutants in a small river by nontarget screening and cluster analysis...... S51 Appendix C - Identification of complex pollutant patterns and representative organic micropollutants during heavy rain events in a small stream ...... S95 Acknowledgements ...... I Scientific contributions ...... II Curriculum vitae ...... III

xiii

List of figures Figure 1: Chemical status of EU river basins (EEA, 2018)...... 2 Figure 2: Overview of entry pathways of organic micropollutants into surface waters...... 3 Figure 3: Concept of thesis divided into three chapters ...... 12 Figure 4: Temporal patterns in WWTP effluent based on compound loads [mg d-1] in relation to the sampling days...... 31 Figure 5: Emission groups of compounds in WWTP effluent based on within-week variation and between-week variation...... 32 Figure 6: Temporal and weather patterns in rain sewer samples...... 34 Figure 7: Emission groups of compounds in rain sewer effluent based on loads [mg d-1] detected in each sample...... 36 Figure 8: Risk patterns of seasonal emissions from WWTP effluent...... 38

Figure 9: Seasonal variation of TUacute (left) and TUsub (right) and the contribution of individual compounds to sum TUs in WWTP effluent...... 39 Figure 10: Risk patterns of weather-related emissions from rain sewer effluent.)...... 40 Figure 11: Risk patterns of weather-related emissions from rain sewer effluent without dimethoate...... 41 Figure 12: TUs based on concentrations of main risk driving compounds in river samples in relation discharge volume in the river [m3]...... 42 Figure 13: Sampling spots at Holtemme River (Saxony-Anhalt, Germany)...... 55 Figure 14: Main spatial polllution patterns (WW, BR, DRI) identified by cluster analysis on all peaks detected by NTS...... 61 Figure 15: Spatial pollution patterns identified by cluster analysis on quantified target compounds...... 62 Figure 16: Subpatterns of main patterns (A) WW and (B) BR and (C) DRI identified by cluster analysis on all peaks included in the respective main pattern...... 63 Figure 17: Subpatterns of DRI main patterns identified by cluster analysis on all peaks with detection frequency of at least 25%...... 65 Figure 18: Plots of m/z vs. retention time of all peaks in the three main patterns (A) WW, (B) BR and (C) DRI and isotopologues assigned to isotope peaks...... 66 Figure 19: Characterization of main patterns by isotopologues and homologue series...... 67 Figure 20: Map of Holtemme River (Saxony-Anhalt, Germany)...... 85 Figure 21: Communication scheme between CSO and autosampler (TP5 Active Vacuum, MAXX® Ltd.) PLC = Programmable logic controller...... 86

xiv

Figure 22: Hydrograph of the Holtemme River at the sampling site for the year 2016...... 90 Figure 23: Chemical pollution patterns and according discharge curves A) E2905, B) E0106, C) E1306, D) 2406, E) 1306, F) E1709...... 93 Figure 24: Correlation matrix of event characteristics...... 95 Figure 25: Common peaks identified for the A) “Base” patterns and B) “Quick” patterns...... 97

Figure S 1: Output factor analysis for mixed data ...... S33 Figure S 2: Emission groups of reduced chemical data set in rain sewer effluent based on loads [mg d-1] detected in each sample...... S34 Figure S 3: Detection frequency of peaks in replicates of sample A) Holt13, B) Holt22, C) Holt38 and D) Holt17...... S59 Figure S 4: Results of cluster analysis on internal standard normalized data (left), and not- normalized data (right)...... S60 Figure S 5: Quality criteria of cluster analysis on whole data set...... S61 Figure S 6: Quality criteria of cluster analysis on WW subpatterns ...... S62 Figure S 7: Quality criteria of cluster analysis on BR subpatterns...... S63 Figure S 8: Quality criteria of cluster analysis on DRI subpatterns...... S64 Figure S 9: MS2 spectra of gabapentin-lactam in original sample and reference standard...... S80 Figure S 10: MS2 spectra of 4-methyl-7-ethylaminocoumarin in original sample and reference standard...... S80 Figure S 11: MS2 spectra of lamotrigine in original sample and reference standard...... S81 Figure S 12: MS2 spectra of methocarbamol in original sample and reference standard...... S81 Figure S 13: MS2 spectra of valsartan acid in original sample and reference standard in positive ionization...... S82 Figure S 14: MS2 spectra of valsartan acid in original sample and reference standard in negative ionization...... S82 Figure S 15: MS2 spectra of irbesartan in original sample and reference standard...... S83 Figure S 16: MS2 spectra of olmesartan in original sample and reference standard...... S83 Figure S 17: MS2 spectra of triethylene glycol monoethyl ether in original sample and reference standard...... S84 Figure S 18: MS2 spectra of n-lauroylethanolamine in original sample and reference standard...... S84 Figure S 19: MS2 spectra of lauryl betaine in original sample and reference standard...... S85 Figure S 20: MS2 spectra of lauramidopropyl betaine in original sample and reference standard...... S85 xv

Figure S 21: MS2 spectra of myristamidopropyl betaine in original sample and reference standard...... S86 Figure S 22: MS2 spectra of diethylene glycol monobutyl ether in original sample and reference standard...... S86 Figure S 23: MS2 spectra of triacetin in original sample and reference standard...... S87 Figure S 24: MS2 spectra of hexaethylene glycol in original sample and reference standard. ..S87 Figure S 25: MS2 spectra of heptaethylene glycol in original sample and reference standard. .S88 Figure S 26: MS2 spectra of octaethylene glycol in original sample and reference standard. ...S88 Figure S 27: MS2 spectra of nonaethylene glycol in original sample and reference standard. ..S89 Figure S 28: MS2 spectra of decaethylene glycol in original sample and reference standard. ..S89 Figure S 29: MS2 spectra of azelaic acid in original sample and reference standard...... S90 Figure S 30: Chromatograms and MS2 spectra of C6-SPC and C7-SPC...... S91 Figure S 31: MS2 spectra for lauryl sulfate from MassBank (top) and from original sample (bottom)...... S92 Figure S 32: pH-dependent LC retention time shift of m/z 274.2010...... S93 Figure S 33: Variation of internal standards in samples of E2905...... S99 Figure S 34: Variation of internal standards in samples of E0106...... S99 Figure S 35: Variation of internal standards in samples of E1306...... S100 Figure S 36: Variation of internal standards in samples of E2406...... S100 Figure S 37: Variation of internal standards in samples of E1307...... S101 Figure S 38: Variation of internal standards in samples of E1709...... S101 Figure S 39: Discharge at catchment gauges from 28.5.-30.5.2016...... S102 Figure S 40: Discharge at catchment gauges from 31.5.-2.6.2016...... S102 Figure S 41: Discharge at catchment gauges from 12.-14.6.2016...... S103 Figure S 42: Discharge at catchment gauges from 23. - 25.6.2016...... S103 Figure S 43: Discharge at catchment gauges from 12.-14.7.2016...... S103 Figure S 44: Discharge at catchment gauges from 16.-18.9.2016...... S104 Figure S 45: Quality criteria of cluster analysis on data set E2905...... S105 Figure S 46: Quality criteria of cluster analysis on data set E0106...... S105 Figure S 47: Quality criteria of cluster analysis on data set E1306...... S106 Figure S 48: Quality criteria of cluster analysis on data set E2406...... S106 Figure S 49: Quality criteria of cluster analysis on data set E1307...... S107 Figure S 50: Quality criteria of cluster analysis on data set E1709...... S107 Figure S 51: Physicochemical parameters during E2905...... S108 Figure S 52: Physicochemical parameters during E0106...... S108 xvi

Figure S 53: Physicochemical parameters during E1306...... S109 Figure S 54: Physicochemical parameters during E2406...... S109 Figure S 55: Physicochemical parameters during E1307...... S110 Figure S 56: Physicochemical parameters during E1709...... S110 Figure S 57: MS2 spectra of chenodeoxycholic acid in original sample and reference standard (hcd = 35)...... S111 Figure S 58: MS2 spectra of chenodeoxycholic acid in original sample and reference standard (hcd = 55)...... S111 Figure S 59: MS2 spectra of lauryl betaine in original sample and reference standard...... S112 Figure S 60: MS2 spectra of tetraacetylethylenediamine in original sample and reference standard...... S112 Figure S 61: MS2 spectra of triethylene glycol in original sample and reference standard...... S113 Figure S 62: MS2 spectra of octaethylene glycol in original sample and reference standard. . S113 Figure S 63: MS2 spectra of nonaethylene glycol in original sample and reference standard. S114 Figure S 64: MS2 spectra of decaethylene glycol in original sample and reference standard. S114 Figure S 65: MS2 spectra of undecaethylene glycol in original sample and reference standard...... S115 Figure S 66: MS2 spectra of tridecaethylene glycol in original sample and reference standard ...... S115 Figure S 67: MS2 spectra of pentaglycol ether sulfate in original sample and reference standard...... S116 Figure S 68: MS2 spectra of hexaglycol ether sulfate in original sample and reference standard...... S116 Figure S 69: MS2 spectra of heptaglycol ether sulfate in original sample and reference standard...... S117 Figure S 70: MS2 spectra of octaglycol ether sulfate in original sample and reference standard...... S117 Figure S 71: MS2 spectra of nonaglycol ether sulfate in original sample and reference standard...... S118 Figure S 72: MS2 spectra of decaglycol ether sulfate in original sample and reference standard...... S118 Figure S 73: MS2 spectra of HS2. Main fragments = m/z 89,133 and 177...... S122 Figure S 74: MS2 spectra of HS2. Main fragments = m/z 89,133 and 177...... S123 Figure S 75: MS2 spectra of HS5. Diagnostic loss of m/z 58...... S123 Figure S 76: MS2 spectra of HS6. Main fragments = m/z 59,103, 133 and 147...... S124 xvii

Figure S 77: MS2 spectra of HS8 (part 1). Main fragments = m/z 59, 88 and 115...... S124 Figure S 78: MS2 spectra of HS8 (part 2). Main fragments = m/z 59, 88 and 115...... S125 Figure S 79: MS2 spectra of HS9 (part 1). Main fragments = m/z 89, 113, 133 and 177...... S125 Figure S 80: MS2 spectra of HS9 (part 2). Main fragments = m/z 89, 113, 133 and 177...... S126 Figure S 81: MS2 spectra of HS10 (part 1). Main fragments = m/z 103, 133 and 147...... S126 Figure S 82: MS2 spectra of HS10 (part 2). Main fragments = m/z 103, 133 and 147...... S127

xviii

List of tables Table 1: Overview on factors and sources influencing occurrence of different use classes of organic micropollutants in surface water ...... 6 Table 2: Overview LC-MS/MS methods used for target screening ...... 27 Table 3: Results of structure elucidation for ions of interest ...... 71 Table 5: Characteristics of sampled heavy rain events ...... 91 Table 6: Indicator compounds for a common rain-related mixture. (Ww = wastewater, UWw = untreated wastewater, URun = urban runoff, SRun = surface runoff)...... 104

Table S1: Discharge and temperature data from the WWTP on the respective sampling days . S1 Table S2: Discharge data from the rain sewer in dry and rain weather during the sampling period ...... S2 Table S3: List of target compounds analyzed in WWTP and rain sewer samples...... S3 Table S4: LC solvent gradient for chromatographic separation of target compounds ...... S16 Table S5: Ion source parameters setting in positive and negative ionization mode ...... S16 Table S6: LC solvent gradient for analysis of LVSPE samples with LC-HRMS ...... S17 Table S7: Risk driving compounds analyzed in LVSPE river samples including information on ionization mode and MDL [ng L-1] ...... S18 Table S8: Acute effect concentrations for each compound and BQE...... S19 Table S9: Sublethal effect concentrations of target compounds ...... S23 Table S10: Loads [mg d-1] of micropollutants emitted from WWTP at different sampling days...... S35 Table S11: Loads [mg month-1] of micropollutants emitted from rain sewer at different sampling months and weather conditions...... S44 Table S12: Concentrations [ng L-1] of 14 main risk driving compounds in LVSPE samples taken during the sampling period April-November 2015 ...... S49 Table S13: Physicochemical parameters of sampling spots...... S51 Table S14: Chemicals and reagents used for LC-HRMS analysis and standards for identified compounds...... S52 Table S15: Solvent gradient for LC analysis ...... S53 Table S16: Parameters and setting of Q ExactiveTM Plus (Thermo Fisher) for fullscan experiments ...... S53 Table S17: Settings of MZmine parameters ...... S54 Table S18: Settings of parameters in R ‘nontarget’ package ...... S55 xix

Table S19: Intensities of target and prioritized unknown compounds in river samples belonging to main WW pattern ...... S65 Table S20: Intensities of target and prioritized unknown compounds in river samples belonging to main BR pattern ...... S75 Table S21: Intensities of target and prioritized unknown compounds in river samples belonging to main DRI pattern ...... S78 Table S22: Chemicals and reagents used for chemical analysis by LC-HRMS ...... S95 Table S23: Solvent Gradient elution program for chromatographic separation on LC ...... S95 Table S24: Settings of Q ExactiveTM Plus (Thermo Fisher) for fullscan experiments ...... S96 Table S25: Parameters and settings of the MZmine software ...... S96 Table S26: Parameters and setting for R ‘nontarget’ ...... S97 Table S27: Target and identified unknown organic micropollutants detected in surface water samples taken during heavy rain events ...... available under http://doi.org/10.5281/zenodo.3379087 Table S28: Identified ions of interests in rain-related mixture ...... S119

Abbreviations AChE acetylcholinesterase inhibitor AES alkyl ethoxy sulfates BR Bode River (used a pattern abbreviation for spatial pollution patterns, Chapter 3) BQE biological quality element BV between-week variation CA concentration addition CSO combined sewer overflow C47 7-diethylamino-4-methylcoumarin C47T1 7-ethylamino-4-methylcoumarin C47T2 7-amino-4-methylcoumarin DRI diffuse and random input (used a pattern abbreviation for spatial pollution patterns, Chapter 3) EC effect concentration EEA European Environmental Agency EDA effect-directed analysis EPA Environmental Protection Agency EU European Union ESA ethane sulfonic acid Eq equation (H)ESI (heated) electrospray ionization FA factor analysis FAMD factor analysis of mixed data HDX hydrogen-deuterium exchange HRMS high resolution mass spectrometry IA independent action

Ithres intensity threshold LAS linear alkylbenzene sulfonates LAU Environmental Agency for Saxony-Anhalt LC liquid chromatography LHW State Office for Flood Protection and Water Management Saxony-Anhalt LOEC lowest-observable-effect-concentrations

xxi

LVSPE large volume solid phase extraction m a.s.l. meters above sea level

th MEC95 95 percentile of the measured environmental concentration MDL method detection limits MoA mode of action MS mass spectrometry MS/MS tandem mass spectrometry msPAF multi-substance potentially affected fraction m/z mass-to-charge ratio

NO3-N nitrate nitrogen NTS nontarget screening OA oxalamic acid PEG polyethylene glycol PCA principal component analysis PCPs personal care products PLC programmable logic controller PLS-DA partial least square discrimination analysis RT retention time SPC sulfophenyl carboxylic acids SSD species sensitivity distribution TP transformation products TU toxic unit

TUacute toxic unit for acute effects (acute risk)

TUsub toxic unit for sublethal effects (sublethal risk)

TUw sum toxic units for a sampling week

TUe sum toxic units for an emission group

TUi toxic unit of individual chemical uPBT ubiquitous, persistent, bioaccumulative, toxic UN United Nations UNESCO United Nations Educational, Scientific and Cultural Organization US United States WFD Water Framework Directive

xxii

WWTP wastewater treatment plant WW wastewater (used a pattern abbreviation for spatial pollution patterns, Chapter 3) WV within-week variation QACs quaternary ammonia compounds z standard score µ mean value σ standard deviation

xxiii

Chapter 1 - Introduction

Chapter 1

Introduction

1.1 Water pollution and degradation of aquatic ecosystems Clean water is the basis for human health, the integrity of ecosystems as well as for social and economic development (UNEP, 2016, WWAP, 2015, 2019). Due to the close connection of humans to water bodies, we shaped and changed them dramatically, e.g., for the purpose of drinking water abstraction, transportation, food production, fisheries, energy generation and waste disposal. With increasing population growth, urbanization, industrialization, expansion and intensification of agriculture and impacts of climate change, evidence of freshwater quality degradation and exploitation of the Earth’s freshwater resources is observed globally (UNEP, 2016, Vörösmarty et al., 2010). Major impacts on water pollution are the introduction of pathogens, organic matter, (plastic) waste as well as inorganic and organic chemicals (Schmidt et al., 2017, UN, 2018). While the problem of metal pollution dates back to ancient history and increased dramatically with the beginning of the industrial revolution (Nriagu, 1996), organic micropollutants are only registered since the mid-1900s due their increasing use and the development of appropriate analytical tools (Richardson and Kimura, 2017). These compounds play a major role in our modern life. They are important constituents in modern medicine, personal hygiene and household activities, industrial activities and food production (Farré et al., 2008, Richardson and Ternes, 2018, Schwarzenbach et al., 2006). Many organic micropollutants are considered as “emerging pollutants” since they were only recently detected, are largely not included in water- quality regulations and their fate and potential effects in aquatic ecosystems or to humans is mostly unknown (Farré et al., 2008, Richardson and Kimura, 2017, Schwarzenbach et al., 2006).

As a response to growing awareness of water pollution, national and international authorities developed water policy instruments and implemented monitoring programs for water pollutants including organic micropollutants. Two prominent water policy instruments are the Clean Water Act in the USA established in 1972 (US EPA, 2002) and the Water Framework Directive (WFD) in the EU since 2000 (EU, 2000). The WFD is complemented by other directives focusing specifically on protection of groundwater resources (e.g., Groundwater Directive (EU, 1979)), specific emission sources (e.g., Urban Waste Water Directive (EU, 1991a)) or individual pollutants (e.g., Nitrate Directive (EU, 1991b)). However, the WFD is the first directive with the ambitious objective to achieve an overall good ecological and chemical status of European water bodies by regular 1

Chapter 1 - Introduction monitoring, assessment and mitigation measures (EU, 2000). The assessment of the chemical status is currently based on 45 “priority pollutants” which were selected as compounds posing a significant risk to aquatic ecosystems in the EU (EU, 2013). So far, the main goal of improving the status of EU surface waters initially until 2015 was not achieved. For instance, based on the assessment of the 45 priority pollutants, none of the German river basins are in a good chemical status (Figure 1) (EEA, 2018). The poor chemical status was mainly due to a few ubiquitous, persistent, bioacculumative and toxic (uPBT) pollutants such as mercury and other heavy metals and polycyclic aromatic hydrocarbons (EEA, 2018). While the selected priority pollutants are monitored for a good reason and it is important to reduce their emissions into the environment, the current approach of the WFD is not able to account for the complex chemical pollution of freshwater bodies and potential mixture effects (Brack et al., 2018). Moreover, due to changes in

Figure 1: Chemical status of EU river basins (EEA, 2018). Red color indicates river basins in which 100% of water bodies do not achieve a good status. production and consumption patterns, the list of priority pollutants needs to be regularly revised and updated. Currently, the entire WFD is under review. Besides the general call for greater linkage between chemical and ecological status assessment, one major question regarding the chemical status assessment is how to address complex pollutant mixtures of organic micropollutants in monitoring and risk assessment (Altenburger et al., 2019).

With ever more sophisticated instrumental analysis techniques, the number of detected and identified organic micropollutants will continue to increase in the future (Richardson and Ternes, 2018). Yet, we are likely just scratching at the surface of the actual pollution problem (Hollender et al., 2017). Estimations from chemical inventories indicate the potential pollution by up to 30,000 2

Chapter 1 - Introduction to 70,000 chemicals including organic micropollutants which are currently in daily use in the EU alone and may enter the aquatic ecosystem (Schwarzenbach et al., 2006). While issues of catastrophic spills and acute effects are observed in developed countries in declining frequency since the Sandoz accident in 1986 (Giger, 2009), we are facing a (ever more) complex pollution with organic micropollutants that deteriorates the chemical and ecological status of surface waters on a chronic level. This pollution is even harder to characterize and assess as the effects of the different chemical stressors are less obvious and their interactions are largely unknown.

1.2 Complex mixtures and mixture dynamics of organic micropollutants In the last decades, extensive research and monitoring efforts led not only to the detection of hundreds of organic micropollutants in aquatic environments but also revealed main pollutant sources and dynamics (Farré et al., 2008, Petrie et al., 2015, Richardson and Kimura, 2017). The following two sections provide an overview on identified sources (1.2.1) and potential factors influencing the variability in occurrence (1.2.2) of organic micropollutants in freshwater bodies, especially lotic waters such as rivers and streams.

1.2.1 Common sources and entry pathways of organic micropollutants into lotic waters There are two main pathways for organic micropollutants to enter lotic waters (Figure 2). In rural areas, diffuse sources including surface runoff and groundwater infiltration are the main entry route

Figure 2: Overview of entry pathways of organic micropollutants into surface waters. Entry pathways were adapted from (Farré et al., 2008) (PCPs = personal care products)

Chapter 1 - Introduction for organic micropollutants such as agricultural pesticides and veterinary pharmaceuticals. The input from agriculture depends on land-use type, agricultural practices; above all, proper handling of pesticides and the topographic and hydrological connection of agricultural land to the water body (Leu et al., 2004a, b, Phillips and Bode, 2004). In urban areas, large point sources like wastewater treatment plants (WWTP) discharge considerable amounts of organic micropollutants with treated wastewater. WWTPs are points of emission reflecting individual sources and activities in their catchment. Common wastewater-related compounds were identified in an EU-wide study of WWTP effluents by Loos et al. (2009). Even though, their target list of 156 compounds was far from being comprehensive, such multitarget studies strengthen the idea of deriving source-related fingerprints for these important point sources. Further point sources such as landfills (Müller et al., 2011, Roy et al., 2014), rain sewers and drainages might contribute especially during rain events to the pollution of aquatic environments (Doppler et al., 2012, Leu et al., 2004a, Wittmer et al., 2010). In the receiving water body, the pollution will be an overlay of each source’s contribution. Thus, the identification of source-related fingerprints will support the understanding and unravelling of complex mixtures in receiving water bodies (Brack et al., 2018).

1.2.2 Factors influencing the variability in occurrence of micropollutants in lotic waters Factors influencing the occurrence and concentrations of organic micropollutants in lotic water bodies are manifold and very compound-specific (Farré et al., 2008, Petrie et al., 2015). However, for general groups of organic micropollutants, certain factors can be identified which showed to influence their concentration and occurrence in lotic waters in addition to above mentioned sources (Table 1). Dynamics of wastewater-related compounds strongly depend on changes in treatment efficiencies of WWTPs. For example, heavy rain events may lead to a shorter residence time of the compounds in WWTPs and consequently inefficient treatment or even discharge of untreated wastewater via combined sewer overflow (CSO) (Launay et al., 2016, Phillips et al., 2012). Furthermore, low temperatures are expected to negatively affect the performance of the WWTP by reducing biodegradation (Vieno et al., 2005). These effects, however, depend on how pronounced the differences are between seasons at the sampling site (Conley et al., 2008, Kasprzyk-Hordern et al., 2008). Despite seasonal effects on the WWTP performance, concentration changes throughout the year and during one day may also be due to specific consumption and use patterns of chemicals, especially of pharmaceuticals and personal care products (PCPs) (Stamm et al., 2008). Seasonal as well as weekly and intra-day fluctuations are expected and partly confirmed for some pharmaceuticals, PCPs and illicit drugs, e.g., seasonal

Chapter 1 - Introduction patterns of antibiotics (Castiglioni et al., 2006) or UV stabilizers (Balmer et al., 2005), intra-day peaks of estrone (Nelson et al., 2011) and weekly patterns of illicit drugs (Thomas et al., 2012).

Diffuse pollution is influenced by stormflows as well as seasonal application for agricultural pesticides (Leu et al., 2004a). In contrast to agricultural applications, pesticide use in urban areas is not restricted to a certain season. Here, elevated concentrations were found especially during storms also after the growing season (Phillips and Bode, 2004). Due to comparably greater amounts of pesticides applied per unit area by private users (Templeton et al., 1998) and enhanced runoff from impervious surfaces (Phillips and Bode, 2004), pesticide pollution from urban areas is significant (Wittmer et al., 2010). Besides their application in gardens, pesticides are also used for roof, road and railway protection from weeds and fungi (Göbel et al., 2007, Meyer et al., 2011a, Wittmer et al., 2010).

Other important urban runoff pollutants are polycyclic aromatic and mineral oil hydrocarbons stemming from drip losses of vehicles and rubber components resulting from tire abrasion (Gasperi et al., 2014, Launay et al., 2016, Peter et al., 2018). They accumulate together with compounds from dry and wet deposition, are washed off in heavy rain events and discharged into nearby water bodies contributing to diffuse pollution (Figure 2). In urban areas, rain water is often collected in sewers connected to combined WWTPs or directly to rain storage reservoirs and water bodies resulting in point source pollution (Fig. 2) (Gasperi et al., 2014, Göbel et al., 2007).

In the receiving water body, the concentration of organic micropollutants depends largely on current flow conditions. Especially for compounds discharged from point source such as pharmaceuticals to a river, a dilution effect was observed in case of high flow rates, increasing river volume and increasing distances form the pollution source (Conley et al., 2008, Hua et al., 2006, Kasprzyk-Hordern et al., 2008). Moreover, the distance from sources affected concentrations and composition of mixtures due to elimination processes such as sorption, biodegradation and photodegradation (Vieno et al., 2005). In small rivers, dilution is generally low and thus effects of pollution sources including small point sources and diffuse pollution may have a higher contribution to the overall pollution. Consequently, a considerable risk to aquatic organisms may be observed in these small water bodies (Munz et al., 2017).

Chapter 1 - Introduction

Table 1: Overview on factors and sources influencing occurrence of different use classes of organic micropollutants in surface water

Groups according to use or Factor Source emission Pesticides Stormflows, agricultural Agricultural and application (spray drift*), developed areas, continuous runoff WWTP effluent+

Pharmaceuticals, PCPs, Heavy rain events and low WWTP effluent household and industrial temperatures affecting WWTP compound efficiency and CSO Airborne pollutants Snow melt#, stormflows Diffuse pollution Roadside pollutants Stormflows Urban areas Diffuse and point source pollution (e.g., via rain sewers or CSO) Point source pollutants Distance from source, WWTPs, rain sewers, low flow conditions in receiving drainages water body *(Schulz, 2001), +(Neumann et al., 2002), #(Meyer et al., 2011b)

Most of the studies discussed in this section focused on a few target compounds. However, in order to address real-world complex mixtures of organic micropollutants in the aquatic environment, more comprehensive screening approaches need to be developed and applied allowing for a more holistic characterization of mixture composition and variability in space, time and during extraordinary events (Altenburger et al., 2019, Brack et al., 2018).

1.3 Analytical methods to capture complex mixtures of organic micropollutants Due to developments in analytical and computational methods in the last decades, the range of analyzable organic micropollutants has been extended considerably (Hollender et al., 2017, Leendert et al., 2015, Pérez-Fernández et al., 2017). These methods allow for capturing and characterizing complex environmental pollutant mixtures.

1.3.1 Chemical methods to characterize complex mixtures of organic micropollutants In order to analyze organic micropollutants in the water phase, analytical methods involving liquid chromatography coupled to mass spectrometry (LC-MS) are preferred. In general, there are three major approaches, i.e., target screening, suspect screening and nontarget screening (NTS) (Krauss et al., 2010). Target screening is based on pre-selected compounds for which reference standards are available allowing also for quantification of detected compounds. Several hundred

Chapter 1 - Introduction compounds were analyzed simultaneously in surface waters using multitarget methods (Kuzmanović et al., 2016, Moschet et al., 2014, Munz et al., 2017, Schäfer et al., 2011). However, target screening studies are always limited to the compound selection made by researches and potentially fail to represent the actual pollutant mixture properly and bias risk assessment by missing important compounds, e.g., unknown toxicants (Altenburger et al., 2015, Malaj et al., 2014).

With the development of high resolution mass spectrometry (HRMS) and ever improving mass accuracy and resolution power, suspect and NTS became more applicable and popular (Freeling et al., 2019, Hollender et al., 2017, Richardson and Ternes, 2018, Schymanski and Williams, 2017). Suspect screening is a valuable method for organic micropollutants with known molecular formulas and structures but for which no reference standards are available (Krauss et al., 2010). Samples are screened for a list of selected masses derived from known molecular formulas. By application of smart filtering steps including exact mass, peak shape, potential ionization behavior, predicted retention times, isotopic signatures or diagnostic fragments, suspect screening is an efficient method to screen for hundreds of organic micropollutants (Moschet et al., 2013). In previous studies, suspect screening was applied to identify, e.g., pharmaceuticals, PCPs, pesticides, industrial compounds, surfactants and transformation products (Freeling et al., 2019, Gago-Ferrero et al., 2015, Hug et al., 2014, Moschet et al., 2013, Schymanski et al., 2014). Finally, NTS does not require a priori information and thus allows for an unbiased screening of environmental samples (Krauss et al., 2010). In principal, it is only limited to the boundaries of the analytical method (Altenburger et al., 2019). The greatest challenge in NTS is the huge amount of data that is generated. Thus, automated workflows and sophisticated data mining tools are required to exploit data sets generated by NTS. These workflows include three major data processing steps: data treatment, prioritization of peaks and structure elucidation. Data treatment steps involve peak picking, blank correction and isotope and adduct identification for componentization (Hollender et al., 2017). The appropriate prioritization of peaks depends on the aim of each study. Common prioritization methods are based on peak intensity, frequency of detection, specific isotopologue signatures or detection homologue series (Hollender et al., 2017). Moreover, peaks maybe prioritized by observed effects, e.g., using effect-directed analysis (EDA) (Muschket et al., 2018), or by the identification of interesting temporal trends in samples (Hollender et al., 2017, Samanipour et al., 2019)(see section 1.3.2). The most time-consuming step in NTS remains the identification of those prioritized peaks by structure elucidation. While more and more software is developed to support structure elucidation, several steps still require manual evaluation and expert knowledge (Hollender et al., 2017, Krauss et al., 2019). Besides for the identification

Chapter 1 - Introduction of toxicants by EDA (Muschket et al., 2018, Muz et al., 2017), NTS has been applied to complement target screening (Hug et al., 2014, Ruff et al., 2015), understand environmental fate of compounds (Müller et al., 2011) as well as processes during water treatment and to identify transformation products (Nürenberg et al., 2015, Schollée et al., 2015, Verkh et al., 2018). Due to its comprehensive screening, NTS allows for a more holistic characterization of complex pollutant mixtures and the identification of chemical fingerprints related to specific sources or land-use types (Altenburger et al., 2019, Brack et al., 2018, Peter et al., 2018).

1.3.2 Multivariate statistics to identify patterns in complex mixtures of organic micropollutants In order to identify underlying data structures and patterns in large data sets, multivariate statistics are a useful tool (Reimann et al., 2008a, b). Especially for the large data sets generated by NTS, the application of multivariate methods is an almost inevitable step to structure data and prioritize peaks for identification (Hollender et al., 2017). Common multivariate exploratory data analysis tools include principal component analysis (PCA), factor analysis (FA) or cluster analysis. While, essentially, all these methods aim at reducing the complexity of the data set, PCA and FA are used to reduce a large set of inter-correlated variables and visualizing the most important information by a few uncorrelated principal components or factors, respectively. The difference of the two methods lies in their approach. PCA accounts for the maximum variance of all variables and will force all variables into the common components. FA accounts for maximum inter- correlations and thus will allow for the existence of unique factors and data structures. Due to these underlying differences in the methods and several prerequisites for the data, care has to be taken when selecting one of these powerful methods (Reimann et al., 2008b). PCA was successfully applied to distinguish between different entry pathways and samples types as well as to separate parent compounds from peaks likely reflecting transformation products in wastewater (Munz et al., 2017, Schollée et al., 2015).

Cluster analysis is a popular and generally robust method to study similarities among samples or compounds, whereas more similar objects are grouped more closely together than dissimilar objects (Reimann et al., 2008a). There are various clustering methods and algorithms. The main two methods are hierarchical and partitioning clustering. Partitioning clustering methods such as k-means will subdivide the data into a pre-determined number of clusters, whereas hierarchical clustering methods group objects without subdivision. In hierarchical clustering, the number of appropriate clusters has to be determined by the user on the result of the cluster analysis (Reimann et al., 2008a). In general, both methods provide a quick and structured overview on the

Chapter 1 - Introduction data. In previous studies, several clustering methods were applied on large target and NTS data sets and supported the identification of source-related (Carpenter and Helbling, 2018) and temporal patterns (Carpenter et al., 2019, Chiaia-Hernández et al., 2017) as well as of river-basin specific pollutants (Massei et al., 2018).

Recently, first automatized workflow were developed which combine NTS and multivariate statistics making NTS applicable for routine monitoring (Hollender et al., 2017). Workflows which are able to monitor and detect peaks with sudden and intense increases indicating, e.g., spills, might be potentially used as early warning systems for water managers and drinking water suppliers (Alygizakis et al., 2019, Hollender et al., 2017, Samanipour et al., 2019).

The studies presented in this section reflect the current need to develop and share workflows that combine NTS and multivariate statistics to efficiently and quickly extract information from NTS data sets. Moreover, first studies showed the potential of linking NTS data sets and identified fingerprints to observed effects and potential risks in the aquatic environment (Hug et al., 2015, Peter et al., 2018, Zheng et al., 2012) revitalizing the idea of virtual EDA (Eide et al., 2002). So far, proper risk assessment still relies on quantified target compounds.

1.3.3 Methods for risk assessment and identification of drivers of chemical stress Aquatic ecosystems are exposed to multiple stress from chemical and non-chemical stressors (e.g., pathogens, extreme temperatures and hydromorphology changes) (Holmstrup et al., 2010). As organic micropollutants occur in complex mixtures, they might be considered as multiple stress themselves. Thus, risk assessment methods have to be applied that account for mixture toxicity and identify major risk driving compounds (Altenburger et al., 2015). There are two main concepts in order to perform an assessment for the mixture toxicity; the concept of i) concentration addition (CA) and ii) independent action (IA) (Belden et al., 2007). CA assumes that all components act with a similar mode of action (MoA), while IA assumes dissimilar action of mixture components, i.e., compounds affect different target sites independently from another. Consequently, IA requires more detailed knowledge for individual mixture components which is often not available, especially when dealing with hundreds of organic micropollutants in environmental mixtures (Backhaus and Faust, 2012). While CA outperformed IA in case of mixtures of compounds with similar MoAs, it also performed reasonably well when assessing toxicity of a mixture of compounds with similar and dissimilar MoAs (Belden et al., 2007). Thus, CA was proposed as an appropriate model for a more conservative and first precautionary risk assessment (Altenburger and Greco, 2009, Backhaus and Faust, 2012, Belden et al., 2007). Both models, CA and IA, fail in case of interactive 9

Chapter 1 - Introduction effects of mixture components (Backhaus and Faust, 2012). Up to now, there is no model which can take these potential interactions into account (Altenburger et al., 2012).

Methods for mixture risk assessment based on CA and IA models involve calculations of toxic units (TU) and the multi-substance potentially affected fraction (msPAF) approach. The msPAF model is based on species sensitivity distributions (SSD). It is able to include both CA and IA models and a diversity of species. Thus, it requires effect data for all species considered (de Zwart and Posthuma, 2005). The TU approach is rather straightforward by calculating the ratio of a compound’s environmental concentration to its (measured) effect concentration. In the WFD, biological quality elements (BQE) were defined to perform a comprehensive risk assessment. These BQEs involve organisms from different trophic levels such fish, algae and daphnia as model organisms for aquatic invertebrates (EU, 2000). Calculating sum TUs allows for the identification of the main drivers (i.e., risk driving compounds) of mixture toxicity for each BQE (Backhaus and Karlsson, 2014, Massei et al., 2018, Munz et al., 2018, Schäfer et al., 2011, Verro et al., 2009). A further consideration of MoAs in the evaluation of calculated mixture toxicity supports risk assessment by identifying risk driving MoA groups and set the focus for bioanalytical screening (Busch et al., 2016, Massei et al., 2018). The performance of all these models strongly depends on the quality of available effect data and the compounds selected for quantitative analysis in the study (Malaj et al., 2014, Moschet et al., 2014).

1.4 The Holtemme River – a case study The Holtemme River (Saxony-Anhalt, Germany) served as a case study in this dissertation. It is part of the Bode River catchment; a subcatchment of the Elbe River. The Holtemme River has a catchment size of 282 km² and a mean annual discharge of 1.31 m³s-1 (average of data from 1972- 2016) (LHW, 2019, Müller et al., 2018). The river has a length of 47 km with its source in the Harz Mountain National Park at 862 m a.s.l. and its outfall at the Bode River in the Central German Lowlands (85 m a.s.l.) (Müller et al., 2018, Wollschläger et al., 2016). After leaving the national park, it passes through the medium size towns of Wernigerode and Halberstadt and an area of intensive agriculture. Agriculture makes up 60% of the land-use in the Holtemme River catchment, followed by 30% of forests and 10% of urban areas (Wollschläger et al., 2016). Treated wastewater is emitted via two WWTPs located downstream of Wernigerode and in Halberstadt (Wollschläger et al., 2016). According to WFD, the chemical status of the Holtemme River fails to reach a good status at any point in the river due to the ubiquitous presence of mercury. The ecological status for algae and invertebrates decreases with the entrance of the river to the first urban area (LHW, 2017). Due to several weirs, dams and the channelization of the river, the 10

Chapter 1 - Introduction hydromorphology of the Holtemme River was greatly impacted (Wollschläger et al., 2016). Thus, the Holtemme River is a good example for a multiple stressed water body.

The Holtemme River was selected as an intensive research site within the network of Earth observatories called the TERrestrial ENvironmental Observatories (TERENO) (Wollschläger et al., 2016). While, TERENO focuses on terrestrial and hydrological research, it triggered a number of interdisciplinary studies and common field sampling campaigns at the Holtemme River investigating nutrients fluxes (Müller et al., 2018), spatial gradients of pathogen concentrations (Karthe et al., 2017), emission sources of anti-androgenic micropollutants (Muschket et al., 2018) as well as the identification of spatial effects of multiple stressors including organic micropollutants on the population genetic level of macroinvertebrates (Inostroza et al., 2016). This dissertation was embedded in the interdisciplinary research activities at the Holtemme River and supported the identification of chemical stress resulting from multiple emission sources in the river’s catchment. The developed approaches and methods within this dissertation were tested in case studies at the Holtemme River, but are designed to be applicable in other settings and catchments.

Vision and thesis’ objectives The complexity and variability of mixtures of organic micropollutants resulting in variable stress for aquatic organisms, is a major challenge for water managers and authorities. Thus, novel approaches are needed to address these issues and provide guidance for future water monitoring. With the development of ever more powerful analytical instruments and an increasing awareness of the enormous potential of computational methods for time-efficient data evaluation, there is a turning point from reporting and monitoring environmental concentrations of target compounds to identifying and characterizing comprehensive pollution and risk patterns in water bodies (Altenburger et al., 2019). From these patterns, risk driving compounds and indicator compounds may be determined for efficient and proper monitoring and management (Brack et al., 2018). This new line of research aims to reform monitoring towards a better safeguarding the Earth’s water resources.

Chapter 1 - Introduction

Figure 3: Concept of thesis divided into three chapters

The main aim of this thesis was to develop and apply novel approaches which are able to account for complex and variable organic pollutant mixtures and to identify drivers of chemical stress (Figure 3). These drivers include risk driving compounds contributing to mixture toxicity, but also source-related fingerprints and associated pattern-representative or indicator compounds which present specific exposure scenarios to chemical stress. The complexity of chemical pollution was addressed from three different perspectives i) temporal (Chapter 2), ii) spatial (Chapter 3) and iii) precipitation-related (Chapter 4) resulting in three major research questions.

Research question 1: How are organic micropollutants reflected in temporal emission patterns and how do these patterns translate into risk patterns and risk driving compounds for aquatic organisms?

This question was answered in Chapter 2 studying the temporal emissions over four seasons at the effluent of a WWTP. Furthermore, samples were taken at an associated rain sewer to study the effect of precipitation-related emissions as well as seasonal inputs from an urban area. In this study, a multitarget screening approach in combination with cluster analysis and risk assessment based on TUs was applied.

Research question 2: Which spatial pollution patterns can be discriminated in a small river and which source-related fingerprints can be derived?

In order to address this question, a novel open-source workflow was developed in Chapter 3. The workflow combined NTS and a cluster analysis specifically designed for longitudinal data. The 12

Chapter 1 - Introduction workflow was applied on grab samples which were taken along the course of the Holtemme River according to its flow velocity. The spatial patterns were characterized based on isotopologue signatures and detected homologue series. Subclustering revealed source-related fingerprints of WWTP emissions. Pattern-representative compounds of each main pattern were identified by structure elucidation.

Research question 3: Which precipitation-related pollution patterns result in a small river during heavy rain events and what are potential indicator compounds for event-based monitoring?

By application of the workflow developed in Chapter 3, highly time-resolved samples taken during heavy rain events at the Holtemme River were analyzed for precipitation-related pollution patterns in Chapter 4. With the support of a hydrological characterization and dynamics of physicochemical parameters, sources and dynamics of pollution patterns were identified and general considerations for event-based monitoring were drawn. A common rain-related pollutant mixture was identified from which indicator compounds for monitoring of heavy rain events were derived. Besides annotated target compounds, indicator compounds were also identified by structure elucidation of prioritized unknown peaks from the rain-related pollutant mixture.

Finally, results from the individual chapters were synthesized in Chapter 5 to draw general conclusions on drivers of chemical stress and on implications for future water monitoring and management. Furthermore, future research needs were discussed and highlighted.

This dissertation was in part based on scientific publications which were indicated at the beginning of the respective chapters and were listed under “Scientific contributions”.

Chapter 1 - Introduction

References Altenburger, R., Ait-Aissa, S., Antczak, P., Backhaus, T., Barceló, D., Seiler, T.-B., Brion, F., Busch, W., Chipman, K., de Alda, M.L., de Aragão Umbuzeiro, G., Escher, B.I., Falciani, F., Faust, M., Focks, A., Hilscherova, K., Hollender, J., Hollert, H., Jäger, F., Jahnke, A., Kortenkamp, A., Krauss, M., Lemkine, G.F., Munthe, J., Neumann, S., Schymanski, E.L., Scrimshaw, M., Segner, H., Slobodnik, J., Smedes, F., Kughathas, S., Teodorovic, I., Tindall, A.J., Tollefsen, K.E., Walz, K.-H., Williams, T.D., Van den Brink, P.J., van Gils, J., Vrana, B., Zhang, X. and Brack, W. (2015) Future water quality monitoring — Adapting tools to deal with mixtures of pollutants in water resource management. Science of The Total Environment 512– 513, 540-551. http://dx.doi.org/10.1016/j.scitotenv.2014.12.057 Altenburger, R., Brack, W., Burgess, R.M., Busch, W., Escher, B.I., Focks, A., Mark Hewitt, L., Jacobsen, B.N., de Alda, M.L., Ait-Aissa, S., Backhaus, T., Ginebreda, A., Hilscherová, K., Hollender, J., Hollert, H., Neale, P.A., Schulze, T., Schymanski, E.L., Teodorovic, I., Tindall, A.J., de Aragão Umbuzeiro, G., Vrana, B., Zonja, B. and Krauss, M. (2019) Future water quality monitoring: improving the balance between exposure and toxicity assessments of real-world pollutant mixtures. Environmental Sciences Europe 31(1), 12. 10.1186/s12302-019-0193-1 Altenburger, R. and Greco, W.R. (2009) Extrapolation concepts for dealing with multiple contamination in environmental risk assessment. Integrated Environmental Assessment and Management 5(1), 62-68. 10.1897/ieam_2008-038.1 Altenburger, R., Scholz, S., Schmitt-Jansen, M., Busch, W. and Escher, B.I. (2012) Mixture Toxicity Revisited from a Toxicogenomic Perspective. Environmental Science & Technology 46(5), 2508-2522. 10.1021/es2038036 Alygizakis, N.A., Gago-Ferrero, P., Hollender, J. and Thomaidis, N.S. (2019) Untargeted time-pattern analysis of LC-HRMS data to detect spills and compounds with high fluctuation in influent wastewater. Journal of Hazardous Materials 361, 19-29. https://doi.org/10.1016/j.jhazmat.2018.08.073 Backhaus, T. and Faust, M. (2012) Predictive Environmental Risk Assessment of Chemical Mixtures: A Conceptual Framework. Environmental Science & Technology 46(5), 2564-2573. 10.1021/es2034125 Backhaus, T. and Karlsson, M. (2014) Screening level mixture risk assessment of pharmaceuticals in STP effluents. Water Research 49, 157-165. http://doi.org/10.1016/j.watres.2013.11.005 Balmer, M.E., Buser, H.-R., Müller, M.D. and Poiger, T. (2005) Occurrence of Some Organic UV Filters in Wastewater, in Surface Waters, and in Fish from Swiss Lakes. Environmental Science & Technology 39(4), 953-962. 10.1021/es040055r Belden, J.B., Gilliom, R.J. and Lydy, M.J. (2007) How well can we predict the toxicity of pesticide mixtures to aquatic life? Integrated Environmental Assessment and Management 3(3), 364-372. 10.1002/ieam.5630030307 Brack, W., Escher, B.I., Müller, E., Schmitt-Jansen, M., Schulze, T., Slobodnik, J. and Hollert, H. (2018) Towards a holistic and solution-oriented monitoring of chemical status of European water bodies: how to support the EU strategy for a non-toxic environment? Environmental Sciences Europe 30(1), 33. 10.1186/s12302-018-0161-1 Busch, W., Schmidt, S., Kühne, R., Schulze, T., Krauss, M. and Altenburger, R. (2016) Micropollutants in European rivers: A mode of action survey to support the development of effect-based tools for water monitoring. Environmental Toxicology and Chemistry 35(8), 1887- 1899. 10.1002/etc.3460 Carpenter, C.M.G. and Helbling, D.E. (2018) Widespread Micropollutant Monitoring in the Hudson River Estuary Reveals Spatiotemporal Micropollutant Clusters and Their Sources. Environmental Science & Technology 52(11), 6187-6196. 10.1021/acs.est.8b00945

Chapter 1 - Introduction

Carpenter, C.M.G., Wong, L.Y.J., Johnson, C.A. and Helbling, D.E. (2019) Fall Creek Monitoring Station: Highly Resolved Temporal Sampling to Prioritize the Identification of Nontarget Micropollutants in a Small Stream. Environmental Science & Technology 53(1), 77-87. 10.1021/acs.est.8b05320 Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D. and Zuccato, E. (2006) Removal of Pharmaceuticals in Sewage Treatment Plants in Italy. Environmental Science & Technology 40(1), 357-363. 10.1021/es050991m Chiaia-Hernández, A.C., Günthardt, B.F., Frey, M.P. and Hollender, J. (2017) Unravelling Contaminants in the Anthropocene Using Statistical Analysis of Liquid Chromatography–High- Resolution Mass Spectrometry Nontarget Screening Data Recorded in Lake Sediments. Environmental Science & Technology 51(21), 12547-12556. 10.1021/acs.est.7b03357 Conley, J.M., Symes, S.J., Schorr, M.S. and Richards, S.M. (2008) Spatial and temporal analysis of pharmaceutical concentrations in the upper Tennessee River basin. Chemosphere 73(8), 1178-1187. http://dx.doi.org/10.1016/j.chemosphere.2008.07.062 de Zwart, D. and Posthuma, L. (2005) Complex mixture toxicity for single and multiple species: Proposed methodologies. Environmental Toxicology and Chemistry 24(10), 2665-2676. 10.1897/04-639r.1 Doppler, T., Camenzuli, L., Hirzel, G., Krauss, M., Lück, A. and Stamm, C. (2012) Spatial variability of herbicide mobilisation and transport at catchment scale: insights from a field experiment. Hydrol. Earth Syst. Sci. 16(7), 1947-1967. 10.5194/hess-16-1947-2012 EEA (2018), European Environment Agency, European waters - Assessment of status and pressures 2018. 7/2018. Eide, I., Neverdal, G., Thorvaldsen, B., Grung, B. and Kvalheim, O.M. (2002) Toxicological evaluation of complex mixtures by pattern recognition: correlating chemical fingerprints to mutagenicity. Environmental Health Perspectives 110 Suppl 6(Suppl 6), 985-988. 10.1289/ehp.02110s6985 EU (1979), European Union, Council Directive 80/68/EEC of 17 December 1979 on the protection of groundwater against pollution caused by certain dangerous substances. Off J Eur Union L 20. EU (1991a), European Union, Council Directive 91/271/EEC of 21 May 1991 concerning urban waste-water treatment. Off J Eur Union L 135. EU (1991b), European Union, Council Directive 91/676/EEC of 12 December 1991 concerning the protection of waters against pollution caused by nitrates from agricultural sources. Off J Eur Union L 327. EU (2000), European Union, Directive 2000/60/EC of the European Parliament and of the Council, establishing a framework for Community action in the field of water policy. Off J Eur Union L327:1. EU (2013), European Union, Directive 2013/39/EU of the European Parliament and the Council of 12. August 2013 amending Directives 2000/60/EC and 2008/105/EC as regards priority substances in the field of water policy. Off J Eur Union L 226/1-L226/17. Farré, M.l., Pérez, S., Kantiani, L. and Barceló, D. (2008) Fate and toxicity of emerging pollutants, their metabolites and transformation products in the aquatic environment. TrAC Trends in Analytical Chemistry 27(11), 991-1007. https://doi.org/10.1016/j.trac.2008.09.010 Freeling, F., Alygizakis, N.A., von der Ohe, P.C., Slobodnik, J., Oswald, P., Aalizadeh, R., Cirka, L., Thomaidis, N.S. and Scheurer, M. (2019) Occurrence and potential environmental risk of surfactants and their transformation products discharged by wastewater treatment plants. Sci Total Environ 681, 475-487. 10.1016/j.scitotenv.2019.04.445 Gago-Ferrero, P., Schymanski, E.L., Bletsou, A.A., Aalizadeh, R., Hollender, J. and Thomaidis, N.S. (2015) Extended Suspect and Non-Target Strategies to Characterize Emerging Polar Organic Contaminants in Raw Wastewater with LC-HRMS/MS. Environmental Science & Technology 49(20), 12333-12341. 10.1021/acs.est.5b03454

Chapter 1 - Introduction

Gasperi, J., Sebastian, C., Ruban, V., Delamain, M., Percot, S., Wiest, L., Mirande, C., Caupos, E., Demare, D., Kessoo, M.D.K., Saad, M., Schwartz, J.J., Dubois, P., Fratta, C., Wolff, H., Moilleron, R., Chebbo, G., Cren, C., Millet, M., Barraud, S. and Gromaire, M.C. (2014) Micropollutants in urban stormwater: occurrence, concentrations, and atmospheric contributions for a wide range of contaminants in three French catchments. Environmental Science and Pollution Research 21(8), 5267-5281. 10.1007/s11356-013-2396-0 Giger, W. (2009) The Rhine red, the fish dead—the 1986 Schweizerhalle disaster, a retrospect and long-term impact assessment. Environmental Science and Pollution Research 16(1), 98-111. 10.1007/s11356-009-0156-y Göbel, P., Dierkes, C. and Coldewey, W.G. (2007) Storm water runoff concentration matrix for urban areas. Journal of Contaminant Hydrology 91(1–2), 26-42. http://doi.org/10.1016/j.jconhyd.2006.08.008 Hollender, J., Schymanski, E.L., Singer, H.P. and Ferguson, P.L. (2017) Nontarget Screening with High Resolution Mass Spectrometry in the Environment: Ready to Go? Environmental Science & Technology 51(20), 11505-11512. 10.1021/acs.est.7b02184 Holmstrup, M., Bindesbøl, A.-M., Oostingh, G.J., Duschl, A., Scheil, V., Köhler, H.-R., Loureiro, S., Soares, A.M.V.M., Ferreira, A.L.G., Kienle, C., Gerhardt, A., Laskowski, R., Kramarz, P.E., Bayley, M., Svendsen, C. and Spurgeon, D.J. (2010) Interactions between effects of environmental chemicals and natural stressors: A review. Science of The Total Environment 408(18), 3746-3762. https://doi.org/10.1016/j.scitotenv.2009.10.067 Hua, W.Y., Bennett, E.R., Maio, X.-S., Metcalfe, C.D. and Letcher, R.J. (2006) Seasonality effects on pharmaceuticals and s-triazine herbicides in wastewater effluent and surface water from the Canadian side of the upper Detroit River. Environmental Toxicology and Chemistry 25(9), 2356-2365. 10.1897/05-571R.1 Hug, C., Sievers, M., Ottermanns, R., Hollert, H., Brack, W. and Krauss, M. (2015) Linking mutagenic activity to micropollutant concentrations in wastewater samples by partial least square regression and subsequent identification of variables. Chemosphere 138, 176-182. https://doi.org/10.1016/j.chemosphere.2015.05.072 Hug, C., Ulrich, N., Schulze, T., Brack, W. and Krauss, M. (2014) Identification of novel micropollutants in wastewater by a combination of suspect and nontarget screening. Environmental Pollution 184, 25-32. https://doi.org/10.1016/j.envpol.2013.07.048 Inostroza, P.A., Vera-Escalona, I., Wicht, A.-J., Krauss, M., Brack, W. and Norf, H. (2016) Anthropogenic Stressors Shape Genetic Structure: Insights from a Model Freshwater Population along a Land Use Gradient. Environmental Science & Technology 50(20), 11346-11356. 10.1021/acs.est.6b04629 Karthe, D., Lin, P.-Y. and Westphal, K. (2017) Instream coliform gradients in the Holtemme, a small headwater stream in the Elbe River Basin, Northern Germany. Frontiers of Earth Science 11(3), 544-553. 10.1007/s11707-017-0648-x Kasprzyk-Hordern, B., Dinsdale, R.M. and Guwy, A.J. (2008) The occurrence of pharmaceuticals, personal care products, endocrine disruptors and illicit drugs in surface water in South Wales, UK. Water Research 42(13), 3498-3518. http://dx.doi.org/10.1016/j.watres.2008.04.026 Krauss, M., Hug, C., Bloch, R., Schulze, T. and Brack, W. (2019) Prioritising site-specific micropollutants in surface water from LC-HRMS non-target screening data using a rarity score. Environmental Sciences Europe 31(1), 45. 10.1186/s12302-019-0231-z Krauss, M., Singer, H. and Hollender, J. (2010) LC–high resolution MS in environmental analysis: from target screening to the identification of unknowns. Analytical and Bioanalytical Chemistry 397(3), 943-951. 10.1007/s00216-010-3608-9 Kuzmanović, M., López-Doval, J.C., De Castro-Català, N., Guasch, H., Petrović, M., Muñoz, I., Ginebreda, A. and Barceló, D. (2016) Ecotoxicological risk assessment of chemical pollution in four Iberian river basins and its relationship with the aquatic macroinvertebrate

Chapter 1 - Introduction community status. Science of The Total Environment 540, 324-333. https://doi.org/10.1016/j.scitotenv.2015.06.112 Launay, M.A., Dittmer, U. and Steinmetz, H. (2016) Organic micropollutants discharged by combined sewer overflows – Characterisation of pollutant sources and stormwater-related processes. Water Research 104, 82-92. http://dx.doi.org/10.1016/j.watres.2016.07.068 Leendert, V., Van Langenhove, H. and Demeestere, K. (2015) Trends in liquid chromatography coupled to high-resolution mass spectrometry for multi-residue analysis of organic micropollutants in aquatic environments. TrAC Trends in Analytical Chemistry 67, 192- 208. https://doi.org/10.1016/j.trac.2015.01.010 Leu, C., Singer, H., Stamm, C., Müller, S.R. and Schwarzenbach, R.P. (2004a) Simultaneous Assessment of Sources, Processes, and Factors Influencing Herbicide Losses to Surface Waters in a Small Agricultural Catchment. Environmental Science & Technology 38(14), 3827-3834. 10.1021/es0499602 Leu, C., Singer, H., Stamm, C., Müller, S.R. and Schwarzenbach, R.P. (2004b) Variability of Herbicide Losses from 13 Fields to Surface Water within a Small Catchment after a Controlled Herbicide Application. Environmental Science & Technology 38(14), 3835-3841. 10.1021/es0499593 LHW (2017), State Office for Flood Protection and Water Management Saxony-Anhalt, Bericht zur Beschaffenheit der Fließgewässer und Seen in Sachsen-Anhalt 2009-2013. 1/2017. https://lhw.sachsen- anhalt.de/fileadmin/Bibliothek/Politik_und_Verwaltung/Landesbetriebe/LHW/neu_PDF/5.0_GLD/ Dokumente_GLD/Bericht_OW_2009-2013/Gewaesserbericht_2009-13.pdf LHW (2019), State Office for Flood Protection and Water Management Saxony-Anhalt, Übersicht Hochwassermeldepegel (Gauging stations for flood monitoring) (accessed on 21.8.2019). https://hochwasservorhersage.sachsen-anhalt.de/ Loos, R., Gawlik, B.M., Locoro, G., Rimaviciute, E., Contini, S. and Bidoglio, G. (2009) EU-wide survey of polar organic persistent pollutants in European river waters. Environmental Pollution 157(2), 561-568. http://dx.doi.org/10.1016/j.envpol.2008.09.020 Malaj, E., von der Ohe, P.C., Grote, M., Kühne, R., Mondy, C.P., Usseglio-Polatera, P., Brack, W. and Schäfer, R.B. (2014) Organic chemicals jeopardize the health of freshwater ecosystems on the continental scale. Proceedings of the National Academy of Sciences 111(26), 9549-9554. 10.1073/pnas.1321082111 Massei, R., Busch, W., Wolschke, H., Schinkel, L., Bitsch, M., Schulze, T., Krauss, M. and Brack, W. (2018) Screening of Pesticide and Biocide Patterns As Risk Drivers in Sediments of Major European River Mouths: Ubiquitous or River Basin-Specific Contamination? Environmental Science & Technology. 10.1021/acs.est.7b04355 Meyer, B., Pailler, J.-Y., Guignard, C., Hoffmann, L. and Krein, A. (2011a) Concentrations of dissolved herbicides and pharmaceuticals in a small river in Luxembourg. Environmental Monitoring and Assessment 180(1), 127-146. 10.1007/s10661-010-1777-9 Meyer, T., Lei, Y.D. and Wania, F. (2011b) Transport of polycyclic aromatic hydrocarbons and pesticides during snowmelt within an urban watershed. Water Research 45(3), 1147-1156. https://doi.org/10.1016/j.watres.2010.11.004 Moschet, C., Piazzoli, A., Singer, H. and Hollender, J. (2013) Alleviating the Reference Standard Dilemma Using a Systematic Exact Mass Suspect Screening Approach with Liquid Chromatography-High Resolution Mass Spectrometry. Analytical Chemistry 85(21), 10312- 10320. 10.1021/ac4021598 Moschet, C., Wittmer, I., Simovic, J., Junghans, M., Piazzoli, A., Singer, H., Stamm, C., Leu, C. and Hollender, J. (2014) How a Complete Pesticide Screening Changes the Assessment of Surface Water Quality. Environmental Science & Technology 48(10), 5423-5432. 10.1021/es500371t

Chapter 1 - Introduction

Müller, A., Schulz, W., Ruck, W.K.L. and Weber, W.H. (2011) A new approach to data evaluation in the non-target screening of organic trace substances in water analysis. Chemosphere 85(8), 1211-1219. https://doi.org/10.1016/j.chemosphere.2011.07.009 Müller, C., Musolff, A., Strachauer, U., Brauns, M., Tarasova, L., Merz, R. and Knöller, K. (2018) Tomography of anthropogenic nitrate contribution along a mesoscale river. Science of The Total Environment 615, 773-783. https://doi.org/10.1016/j.scitotenv.2017.09.297 Munz, N.A., Burdon, F.J., de Zwart, D., Junghans, M., Melo, L., Reyes, M., Schönenberger, U., Singer, H.P., Spycher, B., Hollender, J. and Stamm, C. (2017) Pesticides drive risk of micropollutants in wastewater-impacted streams during low flow conditions. Water Research 110, 366-377. http://dx.doi.org/10.1016/j.watres.2016.11.001 Munz, N.A., Fu, Q., Stamm, C. and Hollender, J. (2018) Internal Concentrations in Gammarids Reveal Increased Risk of Organic Micropollutants in Wastewater-Impacted Streams. Environmental Science & Technology 52(18), 10347-10358. 10.1021/acs.est.8b03632 Muschket, M., Di Paolo, C., Tindall, A.J., Touak, G., Phan, A., Krauss, M., Kirchner, K., Seiler, T.-B., Hollert, H. and Brack, W. (2018) Identification of Unknown Antiandrogenic Compounds in Surface Waters by Effect-Directed Analysis (EDA) Using a Parallel Fractionation Approach. Environmental Science & Technology 52(1), 288-297. 10.1021/acs.est.7b04994 Muz, M., Krauss, M., Kutsarova, S., Schulze, T. and Brack, W. (2017) Mutagenicity in Surface Waters: Synergistic Effects of Carboline Alkaloids and Aromatic Amines. Environmental Science & Technology 51(3), 1830-1839. 10.1021/acs.est.6b05468 Nelson, E.D., Do, H., Lewis, R.S. and Carr, S.A. (2011) Diurnal Variability of Pharmaceutical, Personal Care Product, Estrogen and Alkylphenol Concentrations in Effluent from a Tertiary Wastewater Treatment Facility. Environmental Science & Technology 45(4), 1228-1234. 10.1021/es102452f Neumann, M., Schulz, R., Schafer, K., Muller, W., Mannheller, W. and Liess, M. (2002) The significance of entry routes as point and non-point sources of pesticides in small streams. Water Research 36(4), 835-842. 10.1016/s0043-1354(01)00310-4 Nriagu, J.O. (1996) A History of Global Metal Pollution. Science 272(5259), 223-223. 10.1126/science.272.5259.223 Nürenberg, G., Schulz, M., Kunkel, U. and Ternes, T.A. (2015) Development and validation of a generic nontarget method based on liquid chromatography – high resolution mass spectrometry analysis for the evaluation of different wastewater treatment options. Journal of Chromatography A 1426, 77-90. https://doi.org/10.1016/j.chroma.2015.11.014 Pérez-Fernández, V., Mainero Rocca, L., Tomai, P., Fanali, S. and Gentili, A. (2017) Recent advancements and future trends in environmental analysis: Sample preparation, liquid chromatography and mass spectrometry. Analytica Chimica Acta 983, 9-41. https://doi.org/10.1016/j.aca.2017.06.029 Peter, K.T., Tian, Z., Wu, C., Lin, P., White, S., Du, B., McIntyre, J.K., Scholz, N.L. and Kolodziej, E.P. (2018) Using High-Resolution Mass Spectrometry to Identify Organic Contaminants Linked to Urban Stormwater Mortality Syndrome in Coho Salmon. Environmental Science & Technology 52(18), 10317-10327. 10.1021/acs.est.8b03287 Petrie, B., Barden, R. and Kasprzyk-Hordern, B. (2015) A review on emerging contaminants in wastewaters and the environment: Current knowledge, understudied areas and recommendations for future monitoring. Water Research 72(Supplement C), 3-27. https://doi.org/10.1016/j.watres.2014.08.053 Phillips, P.J. and Bode, R.W. (2004) Pesticides in surface water runoff in south-eastern New York State, USA: seasonal and stormflow effects on concentrations. Pest Management Science 60(6), 531-543. 10.1002/ps.879 Phillips, P.J., Chalmers, A.T., Gray, J.L., Kolpin, D.W., Foreman, W.T. and Wall, G.R. (2012) Combined Sewer Overflows: An Environmental Source of Hormones and Wastewater Micropollutants. Environmental Science & Technology 46(10), 5336-5343. 10.1021/es3001294

Chapter 1 - Introduction

Reimann, C., Filzmoser, P., Garret, R.G. and Dutter, R. (2008a) Statistical Data Analysis Explained: Applied Environmental Statistics with R, pp. 233-247, John Wiley&Sons Ltd., Chichester, UK. Reimann, C., Filzmoser, P., Garret, R.G. and Dutter, R. (2008b) Statistical Data Analysis Explained: Applied Environmental Statistics with R, pp. 211-232, John Wiley&Sons Ltd., Chichester, UK. Richardson, S.D. and Kimura, S.Y. (2017) Emerging environmental contaminants: Challenges facing our next generation and potential engineering solutions. Environmental Technology & Innovation 8, 40-56. https://doi.org/10.1016/j.eti.2017.04.002 Richardson, S.D. and Ternes, T.A. (2018) Water Analysis: Emerging Contaminants and Current Issues. Analytical Chemistry 90(1), 398-428. 10.1021/acs.analchem.7b04577 Roy, J.W., Van Stempvoort, D.R. and Bickerton, G. (2014) Artificial sweeteners as potential tracers of municipal landfill leachate. Environmental Pollution 184, 89-93. https://doi.org/10.1016/j.envpol.2013.08.021 Ruff, M., Mueller, M.S., Loos, M. and Singer, H.P. (2015) Quantitative target and systematic non-target analysis of polar organic micro-pollutants along the river Rhine using high- resolution mass-spectrometry – Identification of unknown sources and compounds. Water Research 87(Supplement C), 145-154. https://doi.org/10.1016/j.watres.2015.09.017 Samanipour, S., Kaserzon, S., Vijayasarathy, S., Jiang, H., Choi, P., Reid, M.J., Mueller, J.F. and Thomas, K.V. (2019) Machine learning combined with non-targeted LC-HRMS analysis for a risk warning system of chemical hazards in drinking water: A proof of concept. Talanta 195, 426-432. https://doi.org/10.1016/j.talanta.2018.11.039 Schäfer, R.B., von der Ohe, P.C., Kühne, R., Schüürmann, G. and Liess, M. (2011) Occurrence and Toxicity of 331 Organic Pollutants in Large Rivers of North Germany over a Decade (1994 to 2004). Environmental Science & Technology 45(14), 6167-6174. 10.1021/es2013006 Schmidt, C., Krauth, T. and Wagner, S. (2017) Export of Plastic Debris by Rivers into the Sea. Environmental Science & Technology 51(21), 12246-12253. 10.1021/acs.est.7b02368 Schollée, J.E., Schymanski, E.L., Avak, S.E., Loos, M. and Hollender, J. (2015) Prioritizing Unknown Transformation Products from Biologically-Treated Wastewater Using High- Resolution Mass Spectrometry, Multivariate Statistics, and Metabolic Logic. Analytical Chemistry 87(24), 12121-12129. 10.1021/acs.analchem.5b02905 Schulz, R. (2001) Comparison of spray drift- and runoff-related input of azinphos-methyl and endosulfan from fruit orchards into the Lourens River, South Africa. Chemosphere 45(4), 543-551. https://doi.org/10.1016/S0045-6535(00)00601-9 Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B., Johnson, C.A., von Gunten, U. and Wehrli, B. (2006) The Challenge of Micropollutants in Aquatic Systems. Science 313(5790), 1072-1077. 10.1126/science.1127291 Schymanski, E.L., Singer, H.P., Longrée, P., Loos, M., Ruff, M., Stravs, M.A., Ripollés Vidal, C. and Hollender, J. (2014) Strategies to Characterize Polar Organic Contamination in Wastewater: Exploring the Capability of High Resolution Mass Spectrometry. Environmental Science & Technology 48(3), 1811-1818. 10.1021/es4044374 Schymanski, E.L. and Williams, A.J. (2017) Open Science for Identifying “Known Unknown” Chemicals. Environmental Science & Technology 51(10), 5357-5359. 10.1021/acs.est.7b01908 Stamm, C., Alder, A.C., Fenner, K., Hollender, J., Krauss, M., McArdell, C.S., Ort, C. and Schneider, M.K. (2008) Spatial and Temporal Patterns of Pharmaceuticals in the Aquatic Environment: A Review. Geography Compass 2(3), 920-955. 10.1111/j.1749-8198.2008.00104.x Templeton, S., Zilberman, D. and Yoo, S.J. (1998) Peer Reviewed: An Economic Perspective on Outdoor Residential Pesticide Use. Environmental Science & Technology 32(17), 416A-423A. 10.1021/es983726s

Chapter 1 - Introduction

Thomas, K.V., Bijlsma, L., Castiglioni, S., Covaci, A., Emke, E., Grabic, R., Hernández, F., Karolak, S., Kasprzyk-Hordern, B., Lindberg, R.H., Lopez de Alda, M., Meierjohann, A., Ort, C., Pico, Y., Quintana, J.B., Reid, M., Rieckermann, J., Terzic, S., van Nuijs, A.L.N. and de Voogt, P. (2012) Comparing illicit drug use in 19 European cities through sewage analysis. Science of The Total Environment 432, 432-439. https://doi.org/10.1016/j.scitotenv.2012.06.069 UN (2018), United Nations, Sustainable Development Goal 6: Synthesis Report 2018 on Water and Sanitation. New York, USA. www.unwater.org/app/uploads/2018/07/SDG6_SR2018_web_v5.pdf UNEP (2016), United Nations Environment Programme, A Sanpshot of the World's Water Quality: Towards a global assessment. Nairobi, Kenya. http://wedocs.unep.org/bitstream/handle/20.500.11822/19524/UNEP_WWQA_report_03052016. pdf?sequence=1&isAllowed=y US EPA (2002), United States Environmental Protection Agency, Clean Water Act. 33 U.S.C. §1251 et seq. (1972). Verkh, Y., Rozman, M. and Petrovic, M. (2018) A non-targeted high-resolution mass spectrometry data analysis of dissolved organic matter in wastewater treatment. Chemosphere 200, 397-404. https://doi.org/10.1016/j.chemosphere.2018.02.095 Verro, R., Finizio, A., Otto, S. and Vighi, M. (2009) Predicting Pesticide Environmental Risk in Intensive Agricultural Areas. II: Screening Level Risk Assessment of Complex Mixtures in Surface Waters. Environmental Science & Technology 43(2), 530-537. 10.1021/es801858h Vieno, N.M., Tuhkanen, T. and Kronberg, L. (2005) Seasonal Variation in the Occurrence of Pharmaceuticals in Effluents from a Sewage Treatment Plant and in the Recipient Water. Environmental Science & Technology 39(21), 8220-8226. 10.1021/es051124k Vörösmarty, C.J., McIntyre, P.B., Gessner, M.O., Dudgeon, D., Prusevich, A., Green, P., Glidden, S., Bunn, S.E., Sullivan, C.A., Liermann, C.R. and Davies, P.M. (2010) Global threats to human water security and river biodiversity. Nature 467, 555. 10.1038/nature09440 Wittmer, I.K., Bader, H.P., Scheidegger, R., Singer, H., Lück, A., Hanke, I., Carlsson, C. and Stamm, C. (2010) Significance of urban and agricultural land use for biocide and pesticide dynamics in surface waters. Water Research 44(9), 2850-2862. http://dx.doi.org/10.1016/j.watres.2010.01.030 Wollschläger, U., Attinger, S., Borchardt, D., Brauns, M., Cuntz, M., Dietrich, P., Fleckenstein, J.H., Friese, K., Friesen, J., Harpke, A., Hildebrandt, A., Jäckel, G., Kamjunke, N., Knöller, K., Kögler, S., Kolditz, O., Krieg, R., Kumar, R., Lausch, A., Liess, M., Marx, A., Merz, R., Mueller, C., Musolff, A., Norf, H., Oswald, S.E., Rebmann, C., Reinstorf, F., Rode, M., Rink, K., Rinke, K., Samaniego, L., Vieweg, M., Vogel, H.-J., Weitere, M., Werban, U., Zink, M. and Zacharias, S. (2016) The Bode hydrological observatory: a platform for integrated, interdisciplinary hydro-ecological research within the TERENO Harz/Central German Lowland Observatory. Environmental Earth Sciences 76(1), 29. 10.1007/s12665-016-6327-5 WWAP (2015), UNESCO World Water Assessment Programme, The United Nations world water development report 2015: water for a sustainable world. 978-92-3-100099-7. Paris, France. WWAP (2019), UNESCO World Water Assessment Programme, The United Nations World Water Development Report 2019: Leaving No One Behind. 978-92-3-100309-7. Paris, France. Zheng, W., Wang, X., Tian, D., Zhang, H., Tian, W., Andersen, M.E., Zheng, Y., Sun, X., Jiang, S., Cao, Z., He, G. and Qu, W. (2012) Pollution Trees: Identifying Similarities among Complex Pollutant Mixtures in Water and Correlating Them to Mutagenicity. Environmental Science & Technology 46(13), 7274-7282. 10.1021/es300728q

Chapter 2 – Temporal chemical and risk patterns

Chapter 2

Characterization and risk assessment of seasonal and weather dynamics in organic pollutant mixtures from discharge of a separate sewer system

Declaration of authors’ contributions:

Liza-Marie Beckers organized and performed the sampling campaign, sample preparation, chemical analysis, chemical data analyses and toxic unit calculation, statistical evaluation and prepared the manuscript. Wibke Busch provided effect concentrations needed for the toxic unit calculations and supervised the risk assessment. Martin Krauss supervised and supported the chemical analysis and evaluation of chemical data of the samples. Martin Krauss and Tobias Schulze gave advice on the sampling campaign and the statistical evaluation of the chemical data. Werner Brack advised the risk assessment and the interpretation of chemical and risk patterns.

All co-authors read and commented on the manuscript.

This chapter is based on the journal article

Characterization and risk assessment of seasonal and weather dynamics in organic pollutant mixtures from discharge of a separate sewer system Liza-Marie Beckers, Wibke Busch, Martin Krauss, Tobias Schulze, Werner Brack Water Research 2018, Volume 135, pages 122-133

Chapter 2 – Temporal chemical and risk patterns

Abstract Sites of wastewater discharge are hotspots for pollution of freshwater bodies with organic micropollutants and are often associated with adverse effects to aquatic organisms. The assessment, monitoring and managment of these hotspots is challenged by variations in the pollutant mixture composition due to season, weather conditions and random spills. In this study, temporal exposure patterns in organic micropollutant mixtures from wastewater discharge were unraveled and respective acute and sublethal risks for aquatic organisms were analyzed. Samples were taken from two components of a separate sewer system i) a WWTP and ii) a rain sewer of a medium size town as well as from the receiving river in different seasons. Rain sewer samples were separately collected for rain and dry - weather conditions. In total, 149 compounds were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS). By considering the pollution dynamics in the point sources, the complexity of pollutant mixtures was reduced by k- means clustering to a few emission groups representing temporal and weather-related pollution patterns. From these groups, BQE - specific risk patterns were derived. In most cases, one main risk driving emission group and one to three individual risk driving compounds were identified for each BQE. While acute risk for fish was quite low, algae were exposed to seasonally emitted herbicides (terbuthylazine, spiroxamine) and crustaceans to randomly spilled insecticides (diazinon, dimethoate). Sublethal risks for all BQE were strongly influenced by constantly emitted pollutants, above all, pharmaceuticals. Variability of risks in the river was mainly driven by water discharge of the river rather than by season or peak events. Overall, the studied WWTP represented the major pollution source with a specific emission of agricultural compounds. However, the investigated rain sewer showed to be a constant pollution source with untreated wastewater due to faulty connections and was an important entry route for high loads of insecticides and biocides due to spills or incorrect disposal. By considering these pollution and risk dynamics, monitoring strategies may be optimized with a special focus on times of low flow conditions in the river, rain events and seasonally emitted risk drivers.

Chapter 2 – Temporal chemical and risk patterns

2.1 Introduction

Synthetic organic chemicals are essential for our modern lifestyle. They are used as pharmaceuticals, pesticides, and biocides, as well as in all types of industrial production and as components of everyday consumer products (Farré et al., 2008, Richardson and Ternes, 2018). As organic micropollutants they are ubiquitous in the aquatic environment and may be detected in great numbers and in complex and variable mixture compositions (Busch et al., 2016, Loos et al., 2009). Wastewater including urban stormwater is an important source for these pollutants potentially affecting aquatic organisms (Inostroza et al., 2016, König et al., 2017, Munz et al., 2017, Münze et al., 2017). Consequently, a comprehensive characterization, assessment, monitoring and management of wastewater effluents and receiving waters is crucial. This requires considering variations in the mixture composition due to season, weather conditions and random spills (Petrie et al., 2015, Wittmer et al., 2010).

In small rivers, WWTP effluents but also rain sewers may significantly contribute to the overall discharge and pollution load (Munz et al., 2017, Wittmer et al., 2010). In separate sewer systems, stormwater is discharged directly to rain storage reservoirs and receiving water bodies with less or no treatment. Consequently, polycyclic aromatic hydrocarbons from atmospheric deposition or traffic as well as pesticides and biocides used for urban pest control in private or public settings may be washed off from surfaces at rain events and enter surface waters via rain sewers (Gasperi et al., 2014, Wittmer et al., 2010).

While only limited data are available for rain sewers, pollutants and pollutant mixtures emitted from WWTPs were extensively studied (e.g. Kostich et al., 2014, Loos et al., 2013, Munz et al., 2017). Loos et al. (2013) found 80% of their 156 target compounds in the effluents of 90 WWTPs across Europe. Munz et al. (2017) detected 50% of 57 compounds at eight time points in the effluents of 24 Swiss WWTPs. These studies suggest the occurrence of common baseline pollution patterns. In addition, the review by Petrie et al. (2015) summarizes knowledge on diurnal, weekly, seasonal and precipitation-related patterns of individual compounds or compound classes entering or emitted from WWTPs due to changing WWTP performance or consumption patterns. Season- specific emission was observed for the insect repellent DEET (Nelson et al., 2011), pharmaceuticals such as antibiotics and nonsteroidal anti-inflammatory drugs (e.g. Castiglioni et al., 2006, Golovko et al., 2014, Vieno et al., 2005) and pesticides (Neumann et al., 2002).

It may be hypothesized that the complex contamination in receiving rivers reflects an overlay of constant baseline emission of municipal wastewater components via WWTPs and seasonal and

Chapter 2 – Temporal chemical and risk patterns event-driven pollution patterns from WWTPs, rain sewers, agricultural runoff and other sources. Temporal pollution patterns and seasonal concentration peaks may result in temporal risk and survival patterns of aquatic organisms (Ashauer et al., 2007). In order to enhance the efficiency and explanatory power of monitoring of organic micropollutants, priority mixtures and compounds may be identified on the basis of pollution and effect patterns (Altenburger et al., 2015). Simple indicators for risk by toxic chemicals are toxic units (TU), which are the ratio of a measured environmental concentration of a compound to its respective effect or lethal concentration. Sum TUs characterize mixture effects. To get a quick, but holistic overview on the risk, the WFD defined BQE as representative organisms groups for the aquatic community for monitoring and status assessment (EEA, 2008).

The objective of the present study was to unravel temporal exposure and risk patterns of mixtures of organic micropollutants in wastewater discharge from a separate sewer system including a municipal WWTP and an associated rain sewer. In detail, four major questions were addressed: 1) How can pollutants from wastewater discharge be grouped? 2) Can these groups be used to derive and discriminate risk patterns for different BQE? 3) Are there individual dominating drivers of risk or are the risks driven by the complex mixtures? 4) To what extent do WWTP and rain sewer effluent contribute to BQE-specific risks in the river? Finally, this study contributed to the discussion on future monitoring strategies and management of pollution hotspots.

Chapter 2 – Temporal chemical and risk patterns

2.2 Methods

2.2.1 Study site

The study was performed at the River Holtemme (Saxony Anhalt, Germany) at a conventional WWTP (80,000 population equivalents), which receives municipal and industrial wastewater from the municipality of Wernigerode and nine villages in an area of intensive agriculture. The WWTP’s catchment is connected by 180 km of sewers. For biological wastewater treatment, the WWTP uses activated sludge technology in its aeration tank with the addition of ferric salt for phosphorus removal. The WWTP operates in a separate sewer system. Stormwater is drained via rain sewers. Consequently, the WWTP is not strongly affected by rainfalls and has a hydraulic retention time of about 72 hours. Discharge and temperature data of the effluent during the sampling periods are shown in the appendix (Table S1). The studied rain sewer discharges approximately 6 km upstream of the WWTP effluent into the river. It serves the old town and a residential area of Wernigerode with private houses, gardens, roads and commercial areas. The discharge data of the rain sewer and precipitation data are presented in Table S2. Discharge data at the nearest river gauge were retrieved from the public database of the (LHW, 2017).

2.2.2 Sampling

From the WWTP effluent, eight daily composite samples were taken during one week in May, July/August and October 2015, and February 2016, respectively. Each composite sample was compiled of subsamples taken every two minutes. In total, 32 samples (250 ml each) were provided by the WWTP operators.

Monthly composite samples from a rain sewer from April 2015 to April 2016 were provided by the Environmental Agency for Saxony-Anhalt (LAU). Under dry conditions, volume-proportional samples were collected by an auto-sampler equipped with an ultrasonic probe. Samples of 150 ml were collected after 16 m³ of water passed the sampling point. Rain samples were taken discharge-proportionally every two minutes if discharge rates exceeded 30 L s-1. The samples were taken continuously over one month and immediately stored in separate five-liter containers placed in an onsite freezer connected to the auto-sampler. For each weather condition and month, one-liter samples were provided by the LAU. The samples from May and June 2015 could not be analyzed due to transport damage.

Monthly composite samples were also taken from the river approximately 1.2 km downstream of the WWTP effluent. In hourly intervals, subsamples of 500 ml were automatically collected using 25

Chapter 2 – Temporal chemical and risk patterns an on-site large volume solid phase extraction (LVSPE) device described by Schulze et al. (2017). Since the monitoring station was not operable in the winter months due to possible frost damage, LVSPE samples could only be taken from April to November 2015.

All samples were stored at -20°C until analysis.

2.2.3 Chemical analysis of samples

In total, 149 compounds were selected for target analysis of water samples by liquid chromatography-tandem mass spectrometry (LC-MS/MS). The selection was based on previous knowledge of the occurrence of compounds in river water, general consumption habits as well as specific industrial activities in the study area. Finally, the list included pharmaceuticals, pesticides, biocides, household and industrial compounds, artificial sweeteners and important transformation products (TP) (Table S3). Chemical analysis was performed on a 1260 Infinity LC system (Agilent) coupled to a QTrap 6500 MS/MS (ABSciex) with an IonDrive Turbo V electrospray ion source system. From each sample, an aliquot of 1 mL was taken for chemical analysis. For the WWTP samples, a 1:10 dilution was additionally prepared. To avoid any losses due to sample enrichment, the 1 mL aliquots were prepared for direct large volume injection (100 µL). To each aliquot, 10 µL of a 2 mM ammonium formate buffer, 25 µL of methanol and 25 µL of an internal standard mixture containing 40 isotope- labelled compounds (40 ng mL-1) were added. Chromatographic separation was performed by gradient elution on a 1260 Infinity LC system (Agilent) with a Kinetex C18 column (50 x 3 mm x 2.6 µm particle size, Phenomenex) guarded by a pre-column (5 x 3 mm) of the same type and an in-line filter (0.2 µm). A QTrap 6500 MS/MS (ABSciex) with an IonDrive Turbo V electrospray ion source was used for detection in scheduled multiple reaction monitoring mode. The target compounds were addressed with three LC-MS methods according to ionization behavior (Table 2). Details on LC-MS/MS settings are presented in Tables S4 and S5. Individual compound settings are provided in Table S3. For quantification, matrix-matched calibration standards were prepared with water from a pristine stream (Wormsgraben, Harz Mountains, Germany). The calibration included ten standards covering a range from 1 to 1000 ng L-1. The method detection limits (MDL) ranged from 0.3 to 50 ng L-1 (Table S3). For each sample batch, quality control samples were prepared accordingly using selected analyzed river water samples spiked with 10, 50 and 200 ng L-1, respectively. The compound concentrations in WWTP and rain sewer samples were quantified with internal standards using the MultiQuantTM Software (Sciex). For the quality control samples, a deviation of

Chapter 2 – Temporal chemical and risk patterns

±20% from the expected value was allowed. Details on the preparation and analysis of LVSPE samples are provided in the supporting information (Appendix A2.3).

Table 2: Overview LC-MS/MS methods used for target screening

Method LC mobile phase Flow rate Ionization Number of [µL min-1] mode compounds

1 A: Water + 0.1% formic acid 0.40 ESI positive 52 B: MeOH + 0.1% formic acid

2 A: Water + 2 mM NH4 formate 0.40 ESI positive 62 B: MeOH + 2 mM NH4 formate

3 A: Water + 1 mM NH4F + 0.35 ESI negative 35

1 mM NH4 formate B: MeOH

2.2.4 Statistical analysis

In order to account for different discharge across the sampling periods, the concentration data were converted into loads using discharge data provided by the operators of the WWTP and the LAU.

Statistical analysis was performed in R (R Core Team, 2017). Data from the WWTP and the rain sewer were analyzed separately due to different system and sampling settings. Prior to statistical analysis, the data was transformed by the natural logarithm, scaled and centered to reduce skewness and ensure equal variance of all variables. Values below MDL were treated as zeroes. Due to several non-detects in the data sets, the ‘glog’ function from the package ‘FitAR’ was used (McLeod and Zhang, 2008). Temporal patterns of all detected compounds and sampling days or months were displayed by a heatmap (function ‘heatmap.2’, R package ‘gplots’ (Warnes et al., 2016)). K-means clustering was applied as an exploratory data analysis tool to identify emission groups among all organic micropollutants in the WWTP and rain sewer effluent. The factoextra package was used for k-means clustering and the preparation of graphs (Kassambara and Mundt, 2016). The number of clusters was determined by the elbow method, which calculates the within sum of squares for different numbers of k clusters (“wss” method, ‘fviz_nbclust’ function). The clustering of WWTP compounds was based on the between-week (BV) and within-week (WV) variation of a compound’s load calculated according to equation (Eq.) (1) and (2). From the WV values of each week, an average WV was calculated for each compound. The R package FactoMineR was used for analysis of mixed data (FAMD) (Le et al., 2008).

Chapter 2 – Temporal chemical and risk patterns

Eq. (1) Calculation of between-week variation (BV):

휎(푀푒푎푛푠 푝푒푟 푤푒푒푘 ) 퐵푉 = ∗ 100% µ(푀푒푎푛푠 푝푒푟 푤푒푒푘 )

Eq. (2) Calculation of within-week variation (WV):

σ = standard deviation

휎 (퐿표푎푑푠 푎푙푙 푠푎푚푝푙푒푠 푝푒푟 푤푒푒푘) (∑ ) 4 푤푒푒푘푠 µ(퐿표푎푑푠 푎푙푙 푠푎푚푝푙푒푠 푝푒푟 푤푒푒푘) 푊푉 = ∗ 100% 4

µ = mean value

2.2.5 Calculation of toxic units

Toxic units were calculated for acute toxicity (TUacute) and sublethal risk (TUsub) such as effects on growth, reproduction and behavior. The 95th percentile of the measured environmental

-1 concentration (MEC95) [mg L ] of each compound was calculated for a) each sampling week of the WWTP effluent, b) for each emission group of the WWTP effluent and the rain sewer effluent and c) for each weather condition in the rain sewer samples. The MEC95 was divided by acute effect concentrations (ECs) and lowest-observable-effect-concentrations (LOEC), respectively, for fish, green algae and daphnia as a representative for crustaceans. The ECs were either based on the 5th percentile of the measured acute ECs retrieved from the US Environmental Protection Agency’s ECOTOX database (according to Busch et al. (2016)) or on predicted ECs (read-across or ECOSAR) in case of missing effect data (Table S8). Details on selected LOECs retrieved from the ECOTOX database are provided in Table S9. Risk driving compounds were determined by calculating the contribution of the individual chemical’s toxicity (TUi) to the sum of all individual toxicity values of a sampling week (TUw) or an emission group (TUe) (Eq.3). Furthermore, the contribution of an emission group to sum toxicity of all emission groups (Eq.4) was assessed.

Eq. 3: Contribution of individual compound (i)

푇푈푖 %i = ∗100% 푇푈푤/푒

Eq. 4: Contribution of emission group (e)

푇푈푒 %e = ∗100% ∑푇푈푒

Chapter 2 – Temporal chemical and risk patterns

2.3 Results and Discussion

2.3.1 Target compounds emitted from WWTP and rain sewer

In total, 89 out of 149 target organic micropollutants were detected in the WWTP effluent (Table S10). In general, concentrations of common target compounds (e.g., carbamazepine, tramadol, benzotriazole, terbuthylazine) were in the same range as concentrations detected in Europe-wide surveys of WWTP effluents (Loos et al., 2013, Munz et al., 2017). Since the WWTP is part of a separate sewer system, slightly lower concentrations were observed for insecticides that are used as outdoor biocides and for the turf herbicide mecoprop than by Munz et al. (2017). Those compounds were likely emitted via rain sewers during runoff and did not pass the WWTP. On the other hand, high concentrations of agricultural pesticides in May, for example boscalid and MCPA (max = 962 ng L-1 and 17,836 ng L-1, respectively) were in contrast to previous studies and suggested a strong influence of the surrounding agricultural activities on the WWTP emissions. Here, stormwater runoff could not explain this input but rather incorrect disposal and cleaning practices. Even though seasonal pesticide peaks were reported before in WWTP effluent (Munz et al., 2017, Neumann et al., 2002, Wittmer et al., 2010), it was not expected that this WWTP contributed directly and to a great extent to the pesticide emissions. In this study, the WWTP was also an important entry route for fungicides such as propiconazole and epoxiconazole. Pharmaceutical industry in the investigated catchment was potentially the source for high concentrations of amitriyptyline and pipamperone. The concentrations of pipamperone were similar to those in a WWTP investigated by Van De Steene et al. (2010), which received wastewater from a chemo pharmaceutical plant. In other WWTPs not connected to pharmaceutical industry, the concentrations of pipamperone were at least two magnitudes lower (Van De Steene et al., 2010).

In the rain sewer samples, 67 of 149 target compounds were detected (Table S11). The samples shared 55 compounds with the WWTP effluent, but at lower concentrations. The presence of pharmaceuticals and - most importantly - the artificial sweetener cyclamate indicated the discharge of untreated wastewater from the rain sewer. Cyclamate is almost completely removed in WWTPs, thus it is a suitable tracer for raw wastewater (Buerge et al., 2009). According to the operators of the local wastewater system, the connection rate was about 98%, but some households were still erroneously connected to the rain sewer system. The rain sewer was also characterized by the emission of biocides and urban pesticides as described by Wittmer et al. (2010).

Chapter 2 – Temporal chemical and risk patterns

By including both point sources in the study, use and emission patterns of common wastewater pollutants from households as well as source-specific pollutants from agriculture and private and urban pest control were identified.

2.3.2 Temporal patterns in pollutant mixtures In the following sections, temporal patterns in pollutant mixtures were discussed. Detailed concentration data for each compound and sample can be found in Tables S10 and S11.

2.3.2.1 WWTP effluent

Temporal emission patterns The discharge volume of the WWTP fluctuated only by 14% over the whole sampling period. The temperature in the biological treatment compartment (i.e., aeration tank) showed a minimum temperature of 11.1°C in February and a maximum temperature of 19.4°C in July/August, with mean temperatures of 15.9°C in May and of 17.7°C in October (Table S1). However, clear seasonal differences in the pollution load and composition were detected (Figure 4). In general, the lowest loads were emitted in July. Since the respective influent samples were not analyzed, no conclusions could be made on the removal efficiency of the WWTP. Rather overall patterns resulting from variations in treatment efficiencies and input were determined. However, low pollutant loads in summer months were also detected in previous studies and were explained by increased biodegradation in the WWTP (Castiglioni et al., 2006, Munz et al., 2017, Vieno et al., 2005). In contrast to other studies, the highest loads were observed in May, especially for pharmaceuticals. Usually higher pharmaceutical concentrations and loads were observed in winter due to decreased biodegradation and increased consumption (Castiglioni et al., 2006, Golovko et al., 2014, Vieno et al., 2005). As also the most diverse pollutant mixture was detected in May, the higher loads possibly resulted from a higher burden on the microbial community thus reducing the efficiency of biodegradation in the system (Onesios et al., 2009). Seasonal differences were driven by month-specific clusters of target compounds (Figure 4). For example, high loads of herbicides were detected in May, fungicide loads increased in October and February showed a specific emission of a few pharmaceuticals.

Chapter 2 – Temporal chemical and risk patterns

Figure 4: Temporal patterns in WWTP effluent based on compound loads [mg d-1] in relation to the sampling days. Data was log-transformed, scaled and centered.

Chapter 2 – Temporal chemical and risk patterns

Emission groups in the pollutant mixture The target compounds were separated into four emission groups (Figure 5) based on the variation of each compound load within a week (Eq. 2) and among the weeks (Eq. 1). In general, the four groups were distinguished into two “constant” (i.e., group 2 and group 4), and two “seasonal” emission groups (i.e., group 1 and 3).

Figure 5: Emission groups of compounds in WWTP effluent based on within-week variation and between- week variation. Ellipses represent 95% confidence interval. Full compound names and details are given in Table S10.

Compounds observed at constant levels were assigned to group 2. The lowest WV and BV was 17% and 14%, respectively, for lidocaine. Considering these minimum variations, this group was neither affected by seasonal nor by great weekly fluctuations and thus reflected baseline emission. Most members of this group were pharmaceuticals for long-term treatments (e.g., carbamazepine, beta-blockers) or daily used drugs (e.g., diclofenac, lidocaine). For these compounds, findings of this study were in agreement with previous studies (Castiglioni et al., 2006, Golovko et al., 2014, Munz et al., 2017, Vieno et al., 2005). For antibiotics like sulfamethoxazole and trimethoprim, a

Chapter 2 – Temporal chemical and risk patterns rather constant emission was detected in agreement with a study by Marx et al. (2015), indicating a joint prescription of these antibiotics in Germany throughout the year. In contrast, seasonal peaks for sulfamethoxazole have been detected by Castiglioni et al. (2006) and Golovko et al. (2014). Finally, steady emission was observed for industrial compounds with wide and constant areas of application, e.g., benzotriazole, and for the legacy herbicide fenuron and the biocides carbendazim and fipronil. Those biocides are applied as a fungicide in outdoor-paints and as an insecticide in fly traps, respectively (BAuA, 2017).

In contrast to group 2, group 4 compounds were characterized by a higher WV. This variation was observed for high-consumption compounds (e.g., propiconazole and saccharin) or might have resulted from variations in the production process, for example, of the pharmaceuticals pipamerone and melperone, from which medicines are formulated batch-like in a local plant. Furthermore, low concentrations around the MDL led to higher variation for the herbicide metabolites and low-use fungicides.

Seasonal compounds were characterized by a high BV and were covered in groups 1 and 3. Group 3 included micropollutants, which were i) constantly emitted within each sampling week, however at significantly different loads among the weeks or ii) were only found in one of the weeks. Most agricultural pesticides were among the seasonal compounds with peak application in May. Thus, group 3 compounds were considered as “season-specific”. Five pharmaceuticals were assigned to this group. They had seasonal peaks in February (i.e., sertraline and nitrendipin) and in October (i.e., loperamide), respectively. The occurrence of antihistamine cetirizine (peak in May) and the anti-coughing agent ambroxol (peak in February) indicated season-dependent consumption. Interestingly, the artificial sweetener acesulfame was assigned to this group with a clear peak in the May week. Since acesulfame is rather persistent during wastewater treatment processes (Buerge et al., 2009), an increased influent load was the only explanation. However, to my knowledge season-dependent removal efficiencies of acesulfame and other sweeteners have not been studied yet.

Group 1 contained seasonal compounds with high WV resulting from four phenomena. 1) In case of the agricultural pesticide pethoxamide, this was due to a short application peak in May indicating that the effluent directly reflected the agricultural activities of the surrounding area. 2) For other compounds, apparent seasonal emission resulted from non-systematic consumption of privately used chemicals or accidental spills (e.g., dimethoate and diazinon) or 3) from a few individual data points around the MDL (e.g., the legacy pesticide TP deisopropylatrazine). 4) Additionally, compounds such as 2-napthtalenesulfonic acid showed a large general variation most likely

Chapter 2 – Temporal chemical and risk patterns because of its wide area of application. Due to these diverse reasons for the higher WV, this group was called “seasonal-random”.

Overlaps of the clusters’ confidence intervals may be reduced by repeated and extended sampling to clarify group membership. Furthermore, the intervals indicated outliers, e.g., p- toluenesulfonamide (TSA) and mecoprop (MCP), which did not fit to any cluster.

2.3.2.2 Precipitation-related emission patterns and emission groups in rain sewer effluent In the rain sewer samples, the most prominent pattern was driven by rain- and dry-weather conditions (Figure 6). Based on the compounds’ load profile, dry weather samples were clearly separated from rain weather samples. These dynamics were in agreement with Wittmer et al. (2010).

Figure 6: Temporal and weather patterns in rain sewer samples (D = dry; R = rain). Compound loads were log- transformed, scaled and centered prior to clustering.

Chapter 2 – Temporal chemical and risk patterns

The compounds were assigned to two emission groups representing micropollutants either related to wastewater (i.e., faulty connections) or surface runoff (Figure 7). The validity of the cluster analysis was confirmed by FAMD and k-means clustering with a reduced dataset (Figures S1 and S2).

In general, pharmaceuticals dominated the chemical profile under dry weather conditions indicating wastewater as a source (group 1). Pesticides and biocides correlated with wet weather conditions (group 2). The legacy pesticide atrazine (banned since 1991) and pesticide metabolites were detected only in dry-weather samples due to their continuous low concentrations, which were diluted below the MDL during rain events. The input of these compounds occurs likely via infiltrating groundwater as atrazine is still often present after these years in aquifers and the vadose zone in Germany and released in low concentrations into surface waters (Vonberg et al., 2014).

Surface runoff (group 2) was characterized by biocides, fungicides, insecticides and urban pesticides. Pollution with these compounds from urban areas is considerable due to comparably larger amounts of herbicides applied by urban users than by farmers (Blanchoud et al., 2004). Furthermore, larger fractions of precipitation enter receiving waters as runoff from paved areas with impermeable materials than from areas with permeable soil. Biocides and fungicides are often applied on facades and other outdoor surfaces. Herbicides like MCPA and mecoprop are commonly applied on turf for weed control (Phillips and Bode, 2004, Wittmer et al., 2010). The runoff of terbuthylazine was unexpected as there is no agricultural area connected to the rain sewer.

Chapter 2 – Temporal chemical and risk patterns

Figure 7: Emission groups of compounds in rain sewer effluent based on loads [mg d-1] detected in each sample. Data was standardized prior to PCA and clustering; zeros treated by glog transformation. Ellipses represent 95% confidence interval. Full compound names and details are given in Table S11.

Outliers of both groups showed a random and precipitation-independent emission behavior (e.g., propiconazole) or had concentrations close or below the MDL (e.g., 2-octyl-4-isothiazolin). Deisopropylatrazine and N-acetyl-4-aminoantipyrine showed correlation with dry-weather discharge but were outliers due to many non-detects.

In contrast to patterns in the WWTP effluent, clear seasonal dynamics could not be confirmed in rain sewer discharge. In general, the pharmaceuticals found were associated with long-term treatment and were therefore constantly emitted. Distinct high loads were only observed in October (Figure 6), which possibly resulted from maintenance works in the sewer system. For rain-related emissions, clear seasonal patterns were missing due to the lack of strong rain events (discharge > 30 L s-1) in most fall and winter months. Still, privately used pesticides are not as systematically and efficiently applied as agricultural pesticides (Templeton et al., 1998). Inter- seasonal presence of these compounds was observed in streams due to an extended application period and continuous runoff from urban areas (Phillips and Bode, 2004). Moreover, some non- 36

Chapter 2 – Temporal chemical and risk patterns agricultural pesticides have a broad field of application as biocides e.g., in facade protection. Single peaks might be further due to spills and incorrect disposal (Phillips and Bode, 2004, Wittmer et al., 2010). This probably holds true for the high peak of dimethoate in the November dry-weather sample. The rain sewer system is quite sensitive to discharge variations and immediately reflects the activities in its catchment.

2.3.3 Risk estimation for temporal emissions and emission groups

In order to understand temporal dynamics of risks for fish, crustaceans and algae, acute (TUacute) and sublethal risks (TUsub) were associated with emission groups of chemicals as identified in chapter 3.2. The pie charts describe the contribution of each emission group (Eq. 4) and the main drivers of each group (Eq. 3) for BQE (Figures 8 - 11).

2.3.3.1 Risk patterns and risk driving compounds in WWTP effluent Acute and sublethal risk from wastewater discharge of the WWTP was assessed per season and emission group. Concerning the sum toxicity of each sampling week, similar seasonal patterns for acute and for sublethal risk were observed for all BQE (Figure 8A). Only sublethal risk for fish did not show any strong temporal dependency. In general, acute toxic risk for fish was lower compared to risks for other BQE and was driven by the seasonally emitted compounds; metolachlor and MCPA (Figure 8B). The potential toxic effect of these two herbicides was mainly explained by their high maximum effluent concentrations (i.e., 849 ng L-1 and 17,836 ng L-1, respectively). The exceedence of TUacute by TUsub by up to three orders of magnitude indicated the underestimation of risk for fish if only acute toxicity is considered (Figure 8A). Due to physiological similarities to mammals, fish are potentially more susceptible to pharmaceuticals, which are not designed to exhibit acute toxicity but to be biologically active chemicals (Corcoran et al., 2010). In fact, constantly emitted pharmaceuticals explained 90% of the TUsub for fish (Figure 8E). The main driver was citalopram (65%), which was demonstrated to alter the swimming behavior of fish (Olsén et al., 2014). This might be important for predator-prey relations. Antidepressants were reported to affect a number of physiological functions (Corcoran et al., 2010). For example, the second driver, amitriptyline (19%), decreased the body length of fish embryos (Yang et al., 2014). Yet, long-term consequences of exposure to these compounds in wildlife and on population level are largely unknown.

Chapter 2 – Temporal chemical and risk patterns

For daphnia, TUacute and TUsub were driven by a diazinon peak in October (Figure 8C,F). However, in agreement with Munz et al. (2017), diazinon also dominated the risk pattern in the absence of peak concentrations (Figure 9). Diazinon is no longer approved for plant protection, but is still registered as an insecticide in pet collars against fleas. The application in private households might explain the rather random emissions (Wittmer et al., 2010). Due to the long hydraulic retention time of 72 hours, the diazinon concentrations detected during a few days in October likely resulted from a single spill.

The second driver for TUacute was fipronil (10%), which was constantly emitted at low Figure 8: Risk patterns of seasonal emissions from concentrations (Figure 8C). Like diazinon, WWTP effluent. (A) Sum TUacute and TUsub for seasonal emission from WWTP effluent (based on weekly MEC95 fipronil is registered for veterinary and indoor of each compound). Contributions of WWTP emission groups and individual risk drivers to sum TUacute and pest control. Both are neuroactive compounds TUsub are shown for fish (B,E), daphnia (C,F) and algae (D,G) (based on MEC95 of each compound in respective targeting different receptors (i.e., diazinon is an emission group). (NFA= N-Formyl-4-aminoantipyrine; FIP = Fipronil; DFC = Diclofenac). acetylcholinesterase (AChE) inhibitor and fipronil disrupts GABA receptors) (Busch et al., 2016). Due to their high acute toxicity to non-target insects, they were banned as plant protection products. However, their continuous application in private households still poses an issue for aquatic ecosystems.

The temporal risk for algae was explained by the seasonal emission of agricultural pesticides (Figure 8D,G). The herbicide terbuthylazine (51%) and the fungicide spiroxamine (21%) explained most of the TUacute (Figure 8D). Both drivers were season-specific for May, while spiroxamine showed a short peak application in the May week and was thus assigned to group 1. Terbutylazine is a potent photosynthesis inhibitor, while spiroxamine inhibits the sterol biosynthesis (Busch et al., 2016). A similar emission pattern was observed for MCPA, which was the main driver for TUsub (Figure 8G).

In addition to seasonal sublethal risk drivers, daphnia and algae were constantly exposed to sublethal concentrations of diclofenac (Figure 8F,G). Diclofenac may affect growth in daphnia and cell multiplication in algae (Dietrich et al., 2010, Lawrence et al., 2012). The anti-inflammatory 38

Chapter 2 – Temporal chemical and risk patterns agent was previously identified as a main risk driver in environmental mixtures (Busch et al., 2016, Munz et al., 2017) and was associated with risk to fish and mammals (Corcoran et al., 2010).

When looking at the risk in the individual sampling weeks, the importance of seasonal risk drivers

Figure 9: Seasonal variation of TUacute (left) and TUsub (right) and the contribution of individual compounds to sum TUs in WWTP effluent.TUs based on seasonal MEC95 concentrations of all detected target compounds. becomes very clear. Lower total risk was often accompanied by a decreasing contribution of the main risk drivers and increasing importance of other compounds leading to a more complex risk pattern and highlighting the need for further investigations of mixture effects (Figure 9).

Chapter 2 – Temporal chemical and risk patterns

2.3.3.2 Risk patterns and risk driving compounds in rain sewer effluent The chemical profiles in the rain sewer were mainly driven by the weather conditions but most compounds were present in both conditions. Therefore, TUs were calculated for dry and rain conditions based on the MEC95 of each detected compound. Likewise, the risk contribution of the two emission groups was evaluated for both weather conditions (Figure 10). Higher TUs were observed for fish during dry weather. The TUacute in the rain sewer was similar to TUacute in May in the WWTP effluent. The TUacute for algae was slightly higher during rain events. For daphnia, again an extraordinary peak was observed (TUacute of 2.3). This peak was due to a high concentration (5161 ng L-1) of the insecticide dimethoate in the November sample. Like the risk driving compounds for daphnia in WWTP effluent, dimethoate is a neuroactive compound inhibiting the AChE like diazinon (Busch et al., 2016). It is approved for private use as an insecticide in gardens. The extraordinary peak in the dry sample in November likely resulted from incorrect disposal after the application period leading to acute risk at the effluent. This peak strongly influenced the TUacute patterns for fish and daphnia. Dimethoate always dominated the acute toxicity for daphnia (Figure 10C), while fipronil strongly contributed to TUacute in rain events and in the absence of the spill event (Figure 11C). For fish, dimethoate explained

59% of the TUacute in dry weather. In the absence of the dimethoate spill, the TUacute decreased by 50% for fish and by two orders of magnitude for daphnia, for which fipronil became the main diver in dry samples (Figure 11A,C). For fish, Figure 10: Risk patterns of weather-related carbendazim and losartan contributed now to the emissions from rain sewer effluent. (A) Sum TUacute and TUsub for dry and rain emission from rain sewer TUacute in dry weather. Dimethoate slightly effluent (based on MEC95 of each compound in each weather condition). Contributions of weather conditions affected the TUsub of fish (Figure 10E). Hence, in and individual risk drivers to sum TUacute and TUsub are shown for fish (B,E), daphnia (C,F) and algae (D,G) the absence of the spill, citalopram and (based on MEC95 of each compound in each weather condition). (ATP = Amitriptyline; CBZ = Carbamazepine; amitriptyline increased in their contribution CP = Citalopram; DCF = Diclofenac; DIU = Diuron; (Figure 11D). The TU pattern for daphnia did DMT= Dimethoate; DNP = 2,4-Dinitrophenol; FIP = sub Fipronil; FPP= Fenpropimorph; TBA= Terbuthylazine, not change. Carbendazim is classified as TBY= Terbutryn).

Chapter 2 – Temporal chemical and risk patterns

hazardous for human and environmental health (ECHA, 2017).

For algae, the herbicides diuron and terbuthlyazine were important acute risk drivers under both weather conditions (Figure 10D). Elevated concentrations of diuron and terbuthlyazine were observed during runoff events. However, they were still present in low concentrations during dry discharge (Figure 6). Similarly to carbendazim, diuron is used in biocides in outdoor paints (BAuA, 2017). Often both compounds are used as constituents of the same product and are consequently washed off together. However, diuron specifically inhibits photosynthesis (Busch et al., 2016). Thus, algae are especially affected by it.

Overall, TUsub were lower than in WWTP effluent Figure 11: Risk patterns of weather-related but were dominated by the same pharmaceuticals emissions from rain sewer effluent without dimethoate. (A) Comparison of TUacute and TUsub in (Figures 8 and 10). Additionally, TUsub for daphnia presence and absence of dimethoate spill in November dry weather sample as well as contributions of risk was driven by an exceptionally high metoprolol drivers to these TUs for fish (B,D) and daphnia (C,E) without dimethoate spill. (NFA= N-Formyl-4- concentration in the September rain sample (3837 aminoantipyrine, NAA = N-Acetyl-4-aminoantipyrine). -1 ng L ). For algae, TUsub was similar in rain and dry conditions with diuron as main driver (Figure 10G).

Four out of seven risk drivers in the rain sewer effluent were biocides. Similar to the patterns in the WWTP effluent, randomly used and discarded insecticides posed high risk to aquatic organisms. The assessment of both point sources implied that acute and sublethal risk was driven by mainly one emission group. In all cases, one to three dominating risk driving compounds could be identified.

2.3.4 Exposure in the receiving river In order to assess the contribution of the studied WWTP and rain sewer to the risk in the receiving river, monthly composite LVSPE samples were analyzed taken from the river downstream of these two point sources. The risk assessment was based on the main risk driving compounds identified in chapter 2.3.3 (Table S12). In general, the TUs increased from May with a peak in July and

Chapter 2 – Temporal chemical and risk patterns

August (Figure 12). The discrepancy between the expected risk pattern based on the identified emission groups (section 2.3.3.1) and the high summer risk can be explained by the low discharge volume in the river and thus lower dilution from June onwards (Figure 12) (Ankley et al., 2007).

This effect can be clearly seen for TUsub for fish, which was mostly driven by constantly emitted pharmaceuticals, i.e., similar river discharge resulted in similar TUsub (Figure 12b). A season- specific risk pattern due to seasonal application could still be deduced for algae. Higher dilution in spring and input from diffuse sources during summer rains contributed to elevated risk even after the main application period (Neumann et al., 2002). Thus, these seasonally emitted pollutants require monitoring in higher temporal resolution during the respective peak times and rain events in the following months. Generally, rain discharge calls for event-based monitoring. Short peak emissions via the WWTP effluent and the rain sewer were averaged out in monthly composite samples and the rain sewer effluent was strongly diluted in the river. For crustaceans, average risk in the river was dominated by constantly emitted fipronil. Only, the dimethoate spill in November was intense enough to be observed in the monthly composite river samples and contributed to the monthly average TUacute. Concerning the contribution of point sources to the discharge of the river, the rain sewer played a minor role (0.01 - 2.3%). Still, adverse acute and sublethal effects to organisms might be observed at the vicinity of the effluent. Moreover, similar patterns and contributions may be assumed for other rain sewers discharging into the river and contributing to the pollution and risk patterns upstream of the WWTP. The studied WWTP contributed up to 40% to the river discharge under low flow conditions and was the main contributor to the pollution with organic micropollutants in the river. The final pollution and risk pattern in the river was mainly driven by the total river discharge (Ankley et al., 2007).

Figure 12: TUs based on concentrations of main risk driving compounds in river samples in relation discharge volume in the river [m3].

Chapter 2 – Temporal chemical and risk patterns

2.4 Conclusions Clustering WWTP effluents according to temporal pollution patterns revealed a clear seasonality in the emission reflected by compounds such as agricultural pesticides in spring or fungicides in fall. The concentrations of other compounds such as many pharmaceuticals were quite constant. In the rain sewer, the discrimination between rain and dry discharge dominated contamination patterns, which were driven by surface runoff and illicit connections. In addition to constantly and seasonally emitted pollutants, there were randomly emitted non-agricultural pesticide and biocide peaks in both WWTP effluent and rain sewer calling for management. These patterns may be used for hypothesis testing in other catchments with similar and different wastewater systems (e.g., concerning CSO structures, land use and hydraulic retention time characteristics).

Chemical emission patterns could be directly translated to BQE-specific risk patterns. While fish were potentially most affected by sublethal effects of constantly emitted chemicals (i.e., baseline emission), algae were exposed to seasonal risks. Random emission of insecticides and biocides from private and urban sources were a strong potential threat for crustaceans.

High risks were typically related to one or three risk drivers, while there may be more contributing chemicals at lower risks. This particularly applied for algae and crustaceans, which were under high risk during pesticide applications and spills. Variability of acute and sublethal risks in the river was mainly driven by water discharge of the river rather than by season or peak events. Due to the lack of sublethal effect data for several compounds and endpoints, the assessment provided only a rough estimate on the potential risks to aquatic organisms. However, this estimate already highlighted the importance of sublethal effects on growth, behavior and reproduction in risk assessments and especially underlined the major role of constantly emitted pharmaceuticals in these effects for all BQE. Furthermore, identified risk driving compounds and risk patterns should be tested in bioassays to confirm their effects and investigate mixture effects on the BQE.

Despite the occurrence of occasionally high loads of insecticides and biocides in the rain sewer, the WWTP was still the more important pollution source with a considerable emission of agricultural pesticides and compounds used and produced by local industries. The risk posed by non-agricultural pesticides and biocides may require rethinking of approval and stormwater management, while input of agricultural pesticides may be reduced by awareness-rising of professional users.

Chapter 2 – Temporal chemical and risk patterns

By considering pollution dynamics due to temporal and precipitation-related influences, water authorities may optimize monitoring strategies suggesting high frequency monitoring during main emission times of pesticides and focusing on low water discharge seasons. Furthermore, event- based monitoring will support the identification of peak contamination triggered by rain events but will fail to capture random spills. Event sampling might be especially important for catchments with WWTPs strongly affected by rainfall, e.g., combined sewer systems.

Chapter 2 – Temporal chemical and risk patterns

Acknowledgements

This study was supported by SOLUTIONS project that is funded by the European Union Seventh Framework Programme (FP7-ENV-2013-two-stage Collaborative project) under grant agreement number 603437. The authors thank the WWTP operators and the Landesamt für Umweltschutz for providing samples and discharge data. Special thanks to Jörg Ahlheim for taking LVSPE samples as well as Margit Petre, Hubert Schupke and Cindy Weidauer for preparing and analyzing LVSPE samples. Mick Wu is acknowledged for advice on statistical methods. JChem for Office was used for structure property prediction and calculation, JChem for Office 6.2.1.1020, 2014, ChemAxon (http://www.chemaxon.com).

Chapter 2 – Temporal chemical and risk patterns

References Altenburger, R., Ait-Aissa, S., Antczak, P., Backhaus, T., Barceló, D., Seiler, T.-B., Brion, F., Busch, W., Chipman, K., de Alda, M.L., de Aragão Umbuzeiro, G., Escher, B.I., Falciani, F., Faust, M., Focks, A., Hilscherova, K., Hollender, J., Hollert, H., Jäger, F., Jahnke, A., Kortenkamp, A., Krauss, M., Lemkine, G.F., Munthe, J., Neumann, S., Schymanski, E.L., Scrimshaw, M., Segner, H., Slobodnik, J., Smedes, F., Kughathas, S., Teodorovic, I., Tindall, A.J., Tollefsen, K.E., Walz, K.-H., Williams, T.D., Van den Brink, P.J., van Gils, J., Vrana, B., Zhang, X. and Brack, W. (2015) Future water quality monitoring — Adapting tools to deal with mixtures of pollutants in water resource management. Science of The Total Environment 512– 513, 540-551. http://dx.doi.org/10.1016/j.scitotenv.2014.12.057 Ankley, G.T., Brooks, B.W., Huggett, D.B., Sumpter and John, P. (2007) Repeating History: Pharmaceuticals in the Environment. Environmental Science & Technology 41(24), 8211-8217. 10.1021/es072658j Ashauer, R., Boxall, A.B.A. and Brown, C.D. (2007) New Ecotoxicological Model To Simulate Survival of Aquatic Invertebrates after Exposure to Fluctuating and Sequential Pulses of Pesticides. Environmental Science & Technology 41(4), 1480-1486. 10.1021/es061727b BAuA (2017), Federal Institute for Occupational Safety and Health, eBIOMELD (accessed on 29.05.2017). https://www.biozid-meldeverordnung.de/offen/ Blanchoud, H., Farrugia, F. and Mouchel, J.M. (2004) Pesticide uses and transfers in urbanised catchments. Chemosphere 55(6), 905-913. https://doi.org/10.1016/j.chemosphere.2003.11.061 Buerge, I.J., Buser, H.-R., Kahle, M., Müller, M.D. and Poiger, T. (2009) Ubiquitous Occurrence of the Artificial Sweetener Acesulfame in the Aquatic Environment: An Ideal Chemical Marker of Domestic Wastewater in Groundwater. Environmental Science & Technology 43(12), 4381-4385. 10.1021/es900126x Busch, W., Schmidt, S., Kühne, R., Schulze, T., Krauss, M. and Altenburger, R. (2016) Micropollutants in European rivers: A mode of action survey to support the development of effect-based tools for water monitoring. Environmental Toxicology and Chemistry 35(8), 1887- 1899. 10.1002/etc.3460 Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D. and Zuccato, E. (2006) Removal of Pharmaceuticals in Sewage Treatment Plants in Italy. Environmental Science & Technology 40(1), 357-363. 10.1021/es050991m Corcoran, J., Winter, M.J. and Tyler, C.R. (2010) Pharmaceuticals in the aquatic environment: A critical review of the evidence for health effects in fish. Critical Reviews in Toxicology 40(4), 287-304. 10.3109/10408440903373590 Dietrich, S., Ploessl, F., Bracher, F. and Laforsch, C. (2010) Single and combined toxicity of pharmaceuticals at environmentally relevant concentrations in Daphnia magna – A multigenerational study. Chemosphere 79(1), 60-66. https://doi.org/10.1016/j.chemosphere.2009.12.069 ECHA (2017), European Chemicals Agency, Substance information - Carbendazim (accessed on 25.07.2017). https://echa.europa.eu/substance-information/- /substanceinfo/100.031.108?_disssubsinfo_WAR_disssubsinfoportlet_backURL=https%3A%2F %2Fecha.europa.eu%2Fde%2Fsearch-for- chemicals%3Fp_p_id%3Ddisssimplesearch_WAR_disssearchportlet%26p_p_lifecycle%3D0%2 6p_p_state%3Dnormal%26p_p_mode%3Dview%26p_p_col_id%3Dcolumn- 1%26p_p_col_count%3D1%26_disssimplesearch_WAR_disssearchportlet_sessionCriteriaId%3 DdissSimpleSearchSessionParam101401500990999267 EEA (2008), European Environment Agency, Water Framework Directive and monitoring (accessed on 03.08.2017). https://www.eea.europa.eu/themes/water/status-and- monitoring/water-framework-directive-and-monitoring

Chapter 2 – Temporal chemical and risk patterns

Farré, M.l., Pérez, S., Kantiani, L. and Barceló, D. (2008) Fate and toxicity of emerging pollutants, their metabolites and transformation products in the aquatic environment. TrAC Trends in Analytical Chemistry 27(11), 991-1007. https://doi.org/10.1016/j.trac.2008.09.010 Gasperi, J., Sebastian, C., Ruban, V., Delamain, M., Percot, S., Wiest, L., Mirande, C., Caupos, E., Demare, D., Kessoo, M.D.K., Saad, M., Schwartz, J.J., Dubois, P., Fratta, C., Wolff, H., Moilleron, R., Chebbo, G., Cren, C., Millet, M., Barraud, S. and Gromaire, M.C. (2014) Micropollutants in urban stormwater: occurrence, concentrations, and atmospheric contributions for a wide range of contaminants in three French catchments. Environmental Science and Pollution Research 21(8), 5267-5281. 10.1007/s11356-013-2396-0 Golovko, O., Kumar, V., Fedorova, G., Randak, T. and Grabic, R. (2014) Seasonal changes in antibiotics, antidepressants/psychiatric drugs, antihistamines and lipid regulators in a wastewater treatment plant. Chemosphere 111, 418-426. https://doi.org/10.1016/j.chemosphere.2014.03.132 Inostroza, P.A., Vera-Escalona, I., Wicht, A.-J., Krauss, M., Brack, W. and Norf, H. (2016) Anthropogenic Stressors Shape Genetic Structure: Insights from a Model Freshwater Population along a Land Use Gradient. Environmental Science & Technology 50(20), 11346-11356. 10.1021/acs.est.6b04629 Kassambara, A. and Mundt, F. (2016). factoextra: Extract and Visualize the Results of Multivariate Data Analyses. R package version 1.0.3. König, M., Escher, B.I., Neale, P.A., Krauss, M., Hilscherová, K., Novák, J., Teodorović, I., Schulze, T., Seidensticker, S., Kamal Hashmi, M.A., Ahlheim, J. and Brack, W. (2017) Impact of untreated wastewater on a major European river evaluated with a combination of in vitro bioassays and chemical analysis. Environmental Pollution 220(Part B), 1220-1230. https://doi.org/10.1016/j.envpol.2016.11.011 Kostich, M.S., Batt, A.L. and Lazorchak, J.M. (2014) Concentrations of prioritized pharmaceuticals in effluents from 50 large wastewater treatment plants in the US and implications for risk estimation. Environmental Pollution 184, 354-359. http://doi.org/10.1016/j.envpol.2013.09.013 Lawrence, J.R., Zhu, B., Swerhone, G.D.W., Roy, J., Tumber, V., Waiser, M.J., Topp, E. and Korber, D.R. (2012) Molecular and microscopic assessment of the effects of caffeine, acetaminophen, diclofenac, and their mixtures on river biofilm communities. Environmental Toxicology and Chemistry 31(3), 508-517. 10.1002/etc.1723 Le, S., Josse, J. and Husson, F. (2008) FactoMineR: An R Package for Multivariate Analysis. Journal of Statistical Software 25(1), 1-18. LHW (2017), State Office for Flood Protection and Water Management Saxony-Anhalt, Datenportal Gewässerkundlicher Landesdienst Sachsen-Anhalt (Database State Waterways Service) (accessed on 12.7.2017). http://gldweb.dhi-wasy.com/gld-portal/ Loos, R., Carvalho, R., António, D.C., Comero, S., Locoro, G., Tavazzi, S., Paracchini, B., Ghiani, M., Lettieri, T., Blaha, L., Jarosova, B., Voorspoels, S., Servaes, K., Haglund, P., Fick, J., Lindberg, R.H., Schwesig, D. and Gawlik, B.M. (2013) EU-wide monitoring survey on emerging polar organic contaminants in wastewater treatment plant effluents. Water Research 47(17), 6475-6487. http://dx.doi.org/10.1016/j.watres.2013.08.024 Loos, R., Gawlik, B.M., Locoro, G., Rimaviciute, E., Contini, S. and Bidoglio, G. (2009) EU-wide survey of polar organic persistent pollutants in European river waters. Environmental Pollution 157(2), 561-568. http://dx.doi.org/10.1016/j.envpol.2008.09.020 Marx, C., Mühlbauer, V., Schubert, S., Oertel, R., Ahnert, M., Krebs, P. and Kuehn, V. (2015) Representative input load of antibiotics to WWTPs: Predictive accuracy and determination of a required sampling quantity. Water Research 76, 19-32. http://dx.doi.org/10.1016/j.watres.2015.02.049 McLeod, A.I. and Zhang, Y. (2008) Improved Subset Autoregression: With R Package. Journal of Statistical Software 28(2).

Chapter 2 – Temporal chemical and risk patterns

Munz, N.A., Burdon, F.J., de Zwart, D., Junghans, M., Melo, L., Reyes, M., Schönenberger, U., Singer, H.P., Spycher, B., Hollender, J. and Stamm, C. (2017) Pesticides drive risk of micropollutants in wastewater-impacted streams during low flow conditions. Water Research 110, 366-377. http://dx.doi.org/10.1016/j.watres.2016.11.001 Münze, R., Hannemann, C., Orlinskiy, P., Gunold, R., Paschke, A., Foit, K., Becker, J., Kaske, O., Paulsson, E., Peterson, M., Jernstedt, H., Kreuger, J., Schüürmann, G. and Liess, M. (2017) Pesticides from wastewater treatment plant effluents affect invertebrate communities. Science of The Total Environment 599, 387-399. http://dx.doi.org/10.1016/j.scitotenv.2017.03.008 Nelson, E.D., Do, H., Lewis, R.S. and Carr, S.A. (2011) Diurnal Variability of Pharmaceutical, Personal Care Product, Estrogen and Alkylphenol Concentrations in Effluent from a Tertiary Wastewater Treatment Facility. Environmental Science & Technology 45(4), 1228-1234. 10.1021/es102452f Neumann, M., Schulz, R., Schafer, K., Muller, W., Mannheller, W. and Liess, M. (2002) The significance of entry routes as point and non-point sources of pesticides in small streams. Water Research 36(4), 835-842. 10.1016/s0043-1354(01)00310-4 Olsén, K.H., Ask, K., Olsén, H., Porsch-Hällström, I. and Hallgren, S. (2014) Effects of the SSRI citalopram on behaviours connected to stress and reproduction in Endler guppy, Poecilia wingei. Aquatic Toxicology 148(Supplement C), 113-121. https://doi.org/10.1016/j.aquatox.2013.12.032 Onesios, K.M., Yu, J.T. and Bouwer, E.J. (2009) Biodegradation and removal of pharmaceuticals and personal care products in treatment systems: a review. Biodegradation 20(4), 441-466. 10.1007/s10532-008-9237-8 Petrie, B., Barden, R. and Kasprzyk-Hordern, B. (2015) A review on emerging contaminants in wastewaters and the environment: Current knowledge, understudied areas and recommendations for future monitoring. Water Research 72(Supplement C), 3-27. https://doi.org/10.1016/j.watres.2014.08.053 Phillips, P.J. and Bode, R.W. (2004) Pesticides in surface water runoff in south‐eastern New York State, USA: seasonal and stormflow effects on concentrations. Pest Management Science 60(6), 531-543. 10.1002/ps.879 R Core Team (2017). R: A language and environment for statistical computing. v3.4.3. R Foundation for Statistical Computing Richardson, S.D. and Ternes, T.A. (2018) Water Analysis: Emerging Contaminants and Current Issues. Analytical Chemistry 90(1), 398-428. 10.1021/acs.analchem.7b04577 Templeton, S., Zilberman, D. and Yoo, S.J. (1998) Peer Reviewed: An Economic Perspective on Outdoor Residential Pesticide Use. Environmental Science & Technology 32(17), 416A-423A. 10.1021/es983726s Van De Steene, J.C., Stove, C.P. and Lambert, W.E. (2010) A field study on 8 pharmaceuticals and 1 pesticide in Belgium: Removal rates in waste water treatment plants and occurrence in surface water. Science of The Total Environment 408(16), 3448-3453. http://dx.doi.org/10.1016/j.scitotenv.2010.04.037 Vieno, N.M., Tuhkanen, T. and Kronberg, L. (2005) Seasonal Variation in the Occurrence of Pharmaceuticals in Effluents from a Sewage Treatment Plant and in the Recipient Water. Environmental Science & Technology 39(21), 8220-8226. 10.1021/es051124k Vonberg, D., Vanderborght, J., Cremer, N., Pütz, T., Herbst, M. and Vereecken, H. (2014) 20 years of long-term atrazine monitoring in a shallow aquifer in western Germany. Water Research 50, 294-306. https://doi.org/10.1016/j.watres.2013.10.032 Warnes, G.B., Bolker, B., Bonebakker, L., Gentleman, R., Liaw, W.H.A., Lumley, T., Maechler, M., Magnusson, A., Moeller, S., Schwartz, M. and Venables, B. (2016) gplots: Various R Programming Tools for Plotting Data. R package version 3.0.1. Wittmer, I.K., Bader, H.P., Scheidegger, R., Singer, H., Lück, A., Hanke, I., Carlsson, C. and Stamm, C. (2010) Significance of urban and agricultural land use for biocide and pesticide 48

Chapter 2 – Temporal chemical and risk patterns dynamics in surface waters. Water Research 44(9), 2850-2862. http://dx.doi.org/10.1016/j.watres.2010.01.030 Yang, M., Qiu, W., Chen, J., Zhan, J., Pan, C., Lei, X. and Wu, M. (2014) Growth inhibition and coordinated physiological regulation of zebrafish (Danio rerio) embryos upon sublethal exposure to antidepressant amitriptyline. Aquatic Toxicology 151(Supplement C), 68- 76. https://doi.org/10.1016/j.aquatox.2013.12.029

Chapter 3 – Spatial chemical patterns

Chapter 3

Unraveling longitudinal pollution patterns of organic micropollutants in a small river by nontarget screening and cluster analysis

Declaration of authors’ contributions:

Liza-Marie Beckers organized and performed the sampling campaign, sample preparation, chemical analysis, chemical data analyses and structure elucidation, statistical evaluation and prepared the manuscript. Werner Brack supported the interpretation of pollutant patterns and source-related fingerprints. Janek Paul Dann gave advice on the evaluation and interpretation of the results from pH- dependent LC-retention experiments. Martin Krauss supported the sampling campaign, supervised chemical analysis and nontarget screening including all structure elucidation methods and gave advice on multivariate statistics. Erik Müller gave advice on multivariate statistics and provided R scripts. Tobias Schulze supported the sampling campaign and gave advice on multivariate statistics.

All co-authors read and commented on the manuscript.

This chapter is based on the journal article

Unraveling longitudinal pollution patterns of organic micropollutants in a small river by nontarget screening and cluster analysis Liza-Marie Beckers, Werner Brack, Janek Paul Dann, Martin Krauss, Erik Müller, Tobias Schulze Science of the Total Environment (under peer review)

Chapter 3 – Spatial chemical patterns

Abstract The pollution of aquatic ecosystems with complex and largely unknown mixtures of organic micropollutants challenges water monitoring strategies. In this study, an open-source workflow based on NTS was implemented to unravel longitudinal pollution patterns of organic micropollutants along a river course. The 47 km long Holtemme River (Saxony-Anhalt, Germany), a tributary of the Bode River, was used as a case study. Sixteen grab samples were taken along the river and analyzed by LC-HRMS. A cluster analysis specifically designed for longitudinal data sets was applied to identify spatial pollutant patterns and prioritize peaks for compound identification. Three main pollution patterns were identified representing pollutants entering i) from WWTPs, ii) at the confluence with the Bode River and iii) from diffuse and random inputs via small point sources and groundwater input. By further subclustering of the main patterns, source-related fingerprints were revealed. The main patterns were characterized by specific isotopologue signatures and the number of peaks in homologue series indicating major (pollution) sources. Furthermore, 25 out of 38 representative compounds for the patterns were identified by structure elucidation. The workflow represents an important contribution to the ongoing attempts to understand, monitor, prioritize and manage complex environmental mixtures and may be applied to other settings.

Chapter 3 – Spatial chemical patterns

3.1 Introduction Aquatic ecosystems are contaminated with a complex and largely unknown mixture of organic micropollutants emitted from a number of pollution sources (Richardson and Kimura, 2017). Furthermore, compounds stemming from natural processes are added to this mixture. While hundreds of compounds were already identified and detected in freshwater bodies by target screening, the quantity and diversity of detected target compounds and their occurrence in complex and variable mixtures still poses a major challenge for monitoring and water management (Altenburger et al., 2015, Brack et al., 2018b, Busch et al., 2016). Moreover, the analysis of pre- selected target compounds can only provide a snapshot of the actual pollution and strongly influences risk assessment (Hollender et al., 2017, Malaj et al., 2014, Moschet et al., 2014). Thus, novel approaches are needed to characterize these mixtures, link them to sources and prioritize yet unknown organic micropollutants for identification in order to allow for efficient mitigation of pollutant emissions (Altenburger et al., 2015).

NTS by liquid chromatography coupled to high-resolution mass spectrometry (LC-HRMS) provides an unbiased approach for capturing this complexity. It was used as a tool to understand transport of organic micropollutants as well as processes in drinking water (Müller et al., 2011) and wastewater treatment (Nürenberg et al., 2015) such as formation of transformation products (Schollée et al., 2015) and degradation of dissolved organic matter (Verkh et al., 2018). It was further applied to complement target screening (Hug et al., 2014, Ruff et al., 2015, Schymanski et al., 2014b) and used in effect-directed analysis to identify unknown toxicants (Muschket et al., 2018, Muz et al., 2017).

NTS generates a huge amount of data (e.g., up to millions of peaks in a set of 360 samples before data treatment (Carpenter et al., 2019)) and thus, the application of multivariate statistics becomes inevitable. Using exploratory data analysis tools, the complexity of the data set can be reduced and data structures may be unravelled (Carpenter et al., 2019, Hollender et al., 2017, Schollée et al., 2015). For example, time-trend analysis was used recently to detect temporal changes of individual peaks at the influent of a WWTP (Alygizakis et al., 2019). This is a valid approach for extracting individual compounds with potentially interesting trends. However, in order to draw more general conclusions on mixture dynamics, cluster analysis proofed to be a valuable and time- efficient tool (Carpenter et al., 2019, Chiaia-Hernández et al., 2017). By means of clustering techniques, similarities among organic micropollutants within complex mixtures were identified and sorted into distinct spatial and temporal chemical or ecotoxicological patterns (Carpenter and Helbling, 2018, Carpenter et al., 2019, Chiaia-Hernández et al., 2017, Peter et al., 2018, Zheng et

Chapter 3 – Spatial chemical patterns al., 2012). These patterns may reflect source-related or effect-related fingerprints (Brack et al., 2018b, Carpenter and Helbling, 2018, Peter et al., 2018, Zheng et al., 2012) and can be used as a prioritization tool for the identification of individual peaks as pattern-representatives (Carpenter et al., 2019, Chiaia-Hernández et al., 2017). In a longitudinal setting, the advantages of time-series analysis and the reduction potential of cluster analysis can be combined to identify groups of variables with similar longitudinal behaviour. Genolini et al. (2015) developed a partitioning cluster analysis for longitudinal data (‘kml’) originally designed for epidemiological data. Here, each variable’s course is seen as a trajectory and similar trajectories are clustered together. This approach is potentially faster than a two-step procedure as applied by Chiaia-Hernández et al. (2017) or a stepwise comparison of spatial samples (Ruff et al., 2015). With the application of NTS in combination with this novel clustering technique, it was hypothesized that continuous longitudinal pollution patterns resulting from diffuse and point sources can be distinguished in a small river.

The objective of this study was to demonstrate this novel open-source workflow on a data set generated by NTS of water samples taken according to the flow velocity along a river course. Using a multi pollution source catchment as a case study, the study investigated I) whether the new approach allows for the discrimination of specific pollution patterns including patterns of natural background and source-related fingerprints? II) Whether the patterns can be generally characterized based on isotopologue signatures and homologue series? III) What are representative compounds for these patterns? Thus, the identification of source-specific pollutant patterns supported peak prioritization for the identification of unknown but representative compounds of these patterns.

Chapter 3 – Spatial chemical patterns

3.2 Methods

3.2.1 Study site

The Holtemme River (Saxony-Anhalt, Germany) was chosen as a case study (Figure 13). The Holtemme River is part of the Bode subcatchment of the Elbe basin. From its source in the national park of the Harz Mountains to its confluence with the Bode River, it spans over 47 km passing two medium size towns and an area of intensive agriculture. Both towns have a WWTP with activated sludge treatment, which discharge into the river. The catchment of the first WWTP (WWTP I) covers an urban and rural area of 300 km² with about 50,000 inhabitants and an industrial input of about 15,000 population equivalents. It operates in a separate sewer system with a connection rate of 98% and has a hydraulic retention time of about 72 hours. The second WWTP (WWTP II) covers a mostly urban area of 143 km² with about 36,800 inhabitants connected to the WWTP. The input from industry contributes with approximately 5,400 population equivalents. WWTP II operates in a combined sewer system with a connection rate of 99.4%. The hydraulic retention time varies from about 54 hours in dry weather to 15 hours during rain events. On the sampling day, the daily discharge of the WWTPs was 10042 m³ and 6617 m³, respectively. The WWTP effluents can be considered as the largest tributaries of the Holtemme River contributing with about 34% and 23% to the river’s discharge on the sampling day, respectively.

Figure 13: Sampling spots at Holtemme River (Saxony-Anhalt, Germany).

Chapter 3 – Spatial chemical patterns

3.2.2 Sampling

Grab samples of 500 mL each were collected along the river at 16 spots (Figure 13). Aliquots of 1 mL of each sample were taken for chemical analysis. The time of sampling was adjusted to the river’s flow velocity to sample the same water package at each spot. Details on the sampling spots including information on physicochemical properties of the samples are shown in the appendix (Table S13).

3.2.3 Chemical analysis of samples

Samples were prepared for direct large volume injection (100 µL). For sample preparation, 10 µL of a 2 mM ammonium formate buffer, 25 µL of methanol and 25 µL of an internal standard mixture containing 40 isotope-labelled compounds (40 ng mL-1) were added to 1 mL of sample. Chemical analysis was performed on an UltiMate 3000 LC system (Thermo Scientific) coupled to a quadrupole-Orbitrap MS (Q Exactive Plus, Thermo Scientific) with a heated electrospray ionization (HESI) source. Details on chemicals and reagents and on LC and MS settings are provided in the appendix, sections B2.1 and B2.2. At the beginning and at the end of each batch, calibration

-1 standards were run at four levels (1, 10, 100 and 1000 ng L ). Solvent blanks (95% H2O/ 5% methanol) were analyzed at least after every sixth sample accounting for background contamination.

3.2.4 Data processing

Raw data from the LC-HRMS analysis were converted into .mzML format (centroid mode) by ProteoWizard v3.0.18265 (Chambers et al., 2012). Peak lists were generated using the software MZmine v2.32 (Pluskal et al., 2010), settings are given in the appendix (Table S17). Repeatability of the chemical analysis and peak picking was checked by injecting replicates of selected samples (Appendix B3.1). The peak lists were exported to Microsoft Excel® for blank correction. Blank correction was performed according to Eq. 5 on peak lists generated by MZmine. Signals below that threshold in the samples were removed. Furthermore, an intensity cut-off at peak heights below 5000 in negative mode and 50000 in positive mode was included to remove noise added by gap filling. For annotated target compounds, calibration standards were considered for logical increase in peak heights.

Chapter 3 – Spatial chemical patterns

Eq. 5: Calculation of intensity threshold (Ithres)

Ithres = µ(IBlk) + 2*σ(IBlk)

µ(IBlk) = mean of peak intensities in blanks; σ(IBlk) = standard deviation of peak intensities in blanks

Prior to cluster analysis, isotope peaks identified by the R package ‘nontarget’ v1.9 (Loos and Singer, 2017, R Core Team, 2017) (Appendix B2.4) were removed and cleaned peak lists were merged. Adduct peaks were not removed at this stage, several false positives were detected when examining target compounds. If a target compound was detected in both ionization modes, the one showing higher intensities was kept. Some typically detected target compounds were missed during peak detection by MZmine due to poor peak shapes were added by analysis on a LC- MS/MS system (QTrap 6500 MS/MS, ABSciex) and manual integration with MultiQuant Software (Sciex). Details on the LC-MS/MS method are described elsewhere (Beckers et al., 2018) and in section 2.2.3. For the added target compounds, the intensity cutoff limit was not an exclusion criterion as they were manually integrated and were analyzed with a full calibration curve ranging from 1 to 1000 ng L-1 (Beckers et al., 2018). In total, seven compounds were added. These compounds included the wastewater marker compounds acesulfame and saccharin (Buerge et al., 2009) as well as the pharmaceuticals pipamperone, diphenhydramine, ofloxacin, ciprofloxacin and metoprolol acid, which were detected as important wastewater compounds in a previous study (Beckers et al., 2018).

3.2.5 Cluster analysis

Cluster analysis was performed on componentized peak lists of the 16 river samples. The statistical evaluation of the data was based on peak heights normalized by internal standards to account for matrix effects (Appendix B3.2, Figure S4). Prior to cluster analysis, the normalized peak heights were scaled to unit variance (i.e., z-score scaling) according to Eq. 6. Scaling ensures that all variables spread over the same range, i.e., all variables have equal variances. Non detects (i.e., zeros) were not removed from the data set and treated by scaling as no type of pollution pattern should be excluded.

Chapter 3 – Spatial chemical patterns

Eq. 6: Scaling to unit variance

푥 − µ 푧 = 휎 z = standard score, µ = mean, σ = standard deviation

Cluster analysis was performed in R using the R package ‘kml’ to unravel longitudinal patterns and clusters of peaks along a river course (Genolini et al., 2015, R Core Team, 2017). The cluster analysis in ‘kml’ was customized by using the distance function ‘diss.CORT’ from the R package ‘TSclust’ (Montero and Vilar, 2014). This function compares trajectories based on the change in direction and rate at each spot. Thus, this distance function fitted better to the spatial data set and helped to mitigate the assumption of spherical data by Euclidean distance used in the k-means algorithm. The script for the cluster analysis is provided in the appendix, section B2.5. The final number of clusters was chosen according to a consensus score of the incorporated quality criteria (Appendix B3.3). The analysis was performed on the entire data set as well as on the resulting clusters to identify potential subpatterns masked by main patterns. The ‘kml’ package provided probabilities of individuals belonging to the different clusters. However, these probabilities should be seen as indications rather than absolute values as they depend on normal distribution of each peak’s data which does not apply for single detects.

3.2.6 Characterization of pattern components

The R ‘nontarget’ package was used for the characterization of the peaks in the different patterns by identifying isotopologue signatures, adducts and homologue series (Appendix B2.4) (Loos and Singer, 2017). The analysis was based on the most representative sample of each pattern (section 3.3.2). The most representative sample of each pattern was the sample in which maximum intensities of peaks in the respective pattern were observed. The sample name consists of the abbreviation “Holt” for Holtemme River and a number corresponding to the river kilometer. For the wastewater pattern, this included samples Holt17 and Holt31 corresponding to the sampling spots downstream of each of the WWTPs. Samples Holt9 and Holt26 were analyzed as representatives for the diffuse and random input pattern and Holt42 for the Bode River pattern. Information on isotopologues and homologues series was merged with information on cluster assignment and displayed in scatter plots (R packages ‘ggplot2’ (Wickham, 2016) and ‘ggpubr’ (Kassambara, 2018)).

Chapter 3 – Spatial chemical patterns

3.2.7 Structure elucidation

Peaks were selected for structure elucidation by intensity. The top high-intensity peaks were selected in the representative samples from the common wastewater pattern (top = 10), the subpattern specific for WWTP I (top = 10) and the WWTP II (top = 5) as well as the subpattern specific for the Bode River input (top = 10) and the diffuse and random input pattern (top = 10). Chemical formulas were generated with the QualBrowser in XCalibur (Thermo Scientific). Calculated formulas were tested for plausibility regarding the isotopic pattern in the QualBrower and submitted for a probable formula query in ChemSpider (Royal Society of Chemistry, 2015) and CompTox (US EPA, 2019) database. Further information for structure elucidation was obtained by re-analyzing samples again in data-dependent MS², hydrogen-deuterium exchange (HDX) and pH-dependent chromatography experiments according to Muz et al. (2017). Fragment lists from respective MS² spectra were submitted to MetFrag v2.3 (Ruttkies et al., 2016) (web tool) to obtain candidate lists. HDX experiments provided information on exchangeable hydrogens in a molecule (Ruttkies et al., 2019), while pH-dependent chromatography supported the identification of probable pKa values (Dann et al., 2016). Experimentally determined pKa value ranges were compared to calculated acidic and basic pKa values by JChem for Office (Excel). Spectral similarity was checked for candidates in MassBank (Horai et al., 2010) and CFM-ID (Allen et al., 2014). Details on the complete workflow for structure elucidation are provided in the appendix, section B2.6. Finally, the level of identification for each structure was reported according to confidence levels introduced by Schymanski et al. (2014a).

3.3 Results and Discussion In the data set, 14,235 peaks were extracted in negative and 50,446 peaks in positive mode. After blank correction and removal of isotope peaks, the final list contained 23,485 peaks including 141 annotated target compounds. The performance of replicate analyses is presented in the appendix, section B3.1.

3.3.1 Longitudinal peak patterns

Cluster analysis is an exploratory data analysis tool which reduced the data set to three main patterns. The plausibility of the patterns was checked by annotated target compounds and quantified target compounds (Figure 15) as well as knowledge on potential sources at the Holtemme River. Due to the nature of partitioning cluster analysis, every variable (i.e., every peak) needs to be assigned to one of the clusters. This might be problematic for variables in the 59

Chapter 3 – Spatial chemical patterns overlapping region of clusters. Thus, the main pattern did not reflect each peak’s intensity course. In order to “clean-up” the main pattern and identify finer structures and source-related fingerprints in the data set, a second subclustering of the main patterns was performed (section 3.3.1.2). The probability of belonging to the assigned cluster and the intensities in river samples is presented for target and prioritized unknown compounds in Tables S19-21.

3.3.1.1 Main peak patterns along the river course

According to the score of the quality criteria (Figure S5), three main patterns were unraveled in the river data set by cluster analysis (Figure 14). The wastewater (“WW”) pattern included 9,811 peaks (about 42% of the data set) with two distinct peaks at the sampling spots downstream of the WWTPs and low or no signals in the headwater. Also, most of the target compounds (n = 100, Table S19) belonging mostly to the group of pharmaceuticals, industrial compounds and pesticides were assigned to this group. The Bode River (“BR”) pattern (7,776 peaks or 33% of all peaks) showed a distinct and sudden increase in peak intensity at the last sampling spot in the river, which coincides with the confluence with the Bode River. As there are no major tributaries in the Holtemme River between sampling sites 40 and 42, those peaks likely originated from the Bode River. Target compounds of BR pattern included mostly industrial compounds and industrially used biocides (i.e., isothiazolinones, Table S20). The diffuse and random input (“DRI”) pattern (5,910 peaks) was a mirror image of the WW pattern and included about 25% of all peaks. It showed higher intensities in the headwater regions with a decrease downstream of the WWTP effluent sites potentially due to dilution (section 3.2.1). Thus, the peaks represented by this pattern were not associated with WWTP effluents. The few target compounds that were assigned to this pattern were mainly pesticide metabolites as well as the legacy pesticide atrazine and artificial sweeteners (Table S21). The presence of the artificial sweeteners cyclamate and saccharin suggested the input of untreated wastewater as they are largely degraded during the wastewater treatment process (Buerge et al., 2009). A previous study identified a rain sewer as a small point source for untreated wastewater and random spills in this headwater region due to faulty or illicit connections in the sewer network (Beckers et al., 2018). Also the occurrence of pesticides and their metabolites might be explained by the input from rain sewers and other drainages as well as from groundwater (Doppler et al., 2012, Kolpin et al., 2000, Reemtsma et al., 2013). During this sampling campaign, the total discharge was solely generated by base flow mainly generated by groundwater as well as by contributions from tributaries (including WWTP effluents) (Müller et al., 2018). Furthermore, this pattern also contained many unidentified peaks which showed consistently high intensities over the whole river course. They likely represented natural

Chapter 3 – Spatial chemical patterns background compounds. Thus, this pattern summarized both diffuse and random input of organic micropollutants and natural compounds.

Figure 14: Main spatial polllution patterns (WW, BR, DRI) identified by cluster analysis on all peaks detected by NTS.

In future studies, the stability of these patterns should be tested for temporal variations due to changing flow conditions (i.e., effects of dilution) and seasonal influences (Beckers et al., 2018) (e.g., pesticide applications in spring or changes in industrial production). Especially the dynamics in the DRI pattern, which summarized diffuse background input as well as small point sources, might have a higher variability in time due to random discharges (Beckers et al., 2018).

Chapter 3 – Spatial chemical patterns

The kml analysis was tested and validated with a data set containing a subset quantified target compounds according to Beckers et al. (2018). Two main patterns could be unraveled (Figure 15). The two patterns resembled the WW and DRI pattern. The BR pattern was not detected as it contained mostly surfactants and industrial compounds which were not part of the smaller target compound list. These results confirmed the suitability of kml analysis for large data sets, on the one hand, and supported the use of NTS for the comprehensive detection of dynamics in complex mixtures, on the other hand.

Figure 15: Spatial pollution patterns identified by cluster analysis on quantified target compounds.

Chapter 3 – Spatial chemical patterns

3.3.1.2 Subpatterns and source-related fingerprints

Based on the score of the quality criteria (Figure S6), cluster analysis of the WW pattern revealed four subpatterns (Figure 16A). The majority of peaks were assigned to subpattern WW1, which represented peaks associated with both WWTPs. Subpatterns WW2 and WW3 included peaks which were more associated with either one of the WWTPs, i.e., peaks which solely or mainly originated from one of the WWTPs. Specific input from WWTP I included fungicides, the antibiotics roxithromycin and azithromycin, as well as coumarin derivatives (Table S19). The latter were previously identified as the main drivers for anti-androgenic activity at this sampling site (Muschket et al., 2018). Several pharmaceuticals (e.g., acetaminophen and ketoprofen) were associated to a larger extent with WWTP II even though they were emitted from both WWTPs. The relatively higher input from WWTP II might be explained by shorter hydraulic residence times and thus less efficient treatment of WWTP II (section 3.2.1). The subpatterns WW1, WW2 and WW3 clearly assigned peaks to their sources. Thus, they may be seen as source-related fingerprints, whereas the WW1 subpattern is a fingerprint for common wastewater compounds with lower variability and the WW2 and WW3 subpatterns are fingerprints for wastewater-related compounds Figure 16: Subpatterns of main patterns (A) WW and (B) BR and (C) DRI identified by cluster analysis on with more variable discharges or specific all peaks included in the respective main pattern. sources in the WWTPs’ catchments. Many of the compounds in these patterns were among frequently detected compounds at European

Chapter 3 – Spatial chemical patterns

WWTPs including the sweetener acesulfame, pharmaceuticals (e.g., carbamazepine, citalopram, diclofenac and sulfamethoxazole), pesticides (e.g., MCPA) and corrosion inhibitors such as benzotriazoles (Loos et al., 2013, Munz et al., 2017). Subpattern WW4 contained compounds which were predominant at the first sampling spot (Figure 16A), and showed only small intensity increases downstream of both WWTPs. Already in the headwater region, there is some anthropogenic influence due to a small battery factory and a hotel prior to sampling spot 3. Both treat their wastewater in septic tanks and discharge rain water to the Holtemme River.

Likewise, subclustering of the BR pattern (Figure 16B and S7) revealed subpatterns of peaks that also occurred at the spots downstream of the WWTPs (i.e., BR2, BR4). However, the sampling spot with highest peak intensities was still the river mouth for all sub-patterns (i.e., BR1-4).

Subclustering of the DRI pattern was more challenging and represented preliminary results on the pattern dynamics (Figure 16C). The subpatterns indicated a few sampling sites with elevated intensities in the urban regions (i.e., sampling spot 9, 11, 15 and 26) (Figure 16C). These subpatterns contained compounds related to wastewater input likely reflecting inputs from small point sources such as rain sewers. The high variation of some peaks among sampling spots is likely due to very random and inconsistent inputs from these sources as they directly reflect activities in their catchment (Beckers et al., 2018). Thus, the cluster analysis, especially with detailed subclustering has the potential to detect even smaller point sources. The effects of single detects on the identified clusters was tested for the DRI pattern as 30% of peaks were detected at only 25% of the sampling sites. To evaluate the effect a detection threshold of 25% was set and subclustering was performed once again. While small point sources were lost, the four subpatterns remained (Figure 17). This confirmed the robustness of partitioning analysis which is hardly disturbed by single detections. By repeated sampling, the origin of peaks in DRI patterns may become more defined and background may be better separated from input of small point sources.

Chapter 3 – Spatial chemical patterns

Figure 17: Subpatterns of DRI main patterns identified by cluster analysis on all peaks with detection frequency of at least 25%.

3.3.2 Characterization of pattern components

The main patterns were investigated for characteristic mass-to-charge ratio (m/z) and retention time (RT) distributions as well as for the number of peaks with specific isotopologue signatures and homologue series (Figures 18 and 19). Halogenated compounds are interesting from an ecotoxicological point of view as they are good indicators for anthropogenic compounds (Ruff et al., 2015). Sulfur-containing compounds especially in combination with homologue series hint towards surfactants (Alygizakis et al., 2019, Gago-Ferrero et al., 2015).

By plotting m/z values against RT of the pattern components, distinct differences between the DRI pattern and the two other patterns (i.e., WW and BR) were identified (Figure 18). The DRI pattern was dominated by compounds eluting at or close to the column dead time with high intensities (i.e., RT < 1 min). A lot of potentially halogenated and sulfur-containing compounds were among these peaks (Figures 18C and 19 A,B). A closer inspection of these peaks in the DRI pattern revealed that no molecular formulas with halogens could be fitted to the observed m/z. This was often due to missing 13C peaks or poor peak shapes and resolution. For a better identification of these compounds and increased performance in peak picking (Appendix B3.1), an improved chromatographic separation of highly hydrophilic compounds on a more polar stationary phase would be required.

Chapter 3 – Spatial chemical patterns

Also the patterns WW and BR included such early eluting peaks with this isotopologue signature. However, in these patterns more halogenated and sulfur-containing compounds were detected with higher RTs (Figure 18A, B).

Figure 18: Plots of m/z vs. retention time of all peaks in the three main patterns (A) WW, (B) BR and (C) DRI and isotopologues assigned to isotope peaks.

The number of peaks assigned as part of a homologue series was evaluated per pattern (Figure 19C). The number of homologue peaks increased downstream of the effluent of the two WWTPs and almost doubled with the confluence with the Bode River. Similarly, a lot of sulfur-containing compounds were related to the WW and BR pattern (Figures 18 A,B and 19A). Dissolved organic matter originating from wastewater has a distinctly high content of sulfur-containing species in comparison to dissolved organic matter from pristine waters (Greenwood et al., 2012). The investigation of changes in dissolved organic matter homologue series during wastewater treatment showed that especially compounds with CH2-series are not readily degradable during treatment (Verkh et al., 2018). In previous studies, several (sulfur-containing) homologue series were identified as surfactants in wastewater (Alygizakis et al., 2019, Gago-Ferrero et al., 2015, Peter et al., 2018, Schymanski et al., 2014b). Thus, the presence of these compounds in the WW and BR pattern supported the urban and industrial contributions indicated by target compounds

Chapter 3 – Spatial chemical patterns

(section 3.3.1.1). Follow-up studies in the Bode River should reveal where this high contribution of compounds in homologue series (potentially surfactants) originate from. Some of these ions of interests were identified (section 3.3.3).

A consistently low number of peaks in a homologue series were related to the DRI pattern (Figure 19C). Most of these homologue series (>90%) showed a mass increment of 14 m/z representing a CH2 group. This group is commonly seen in anthropogenic homologue series but was also discovered in homologue series of natural compounds such as humic and fulvic acids (Stenson et al., 2002). Besides low numbers of sulfur-containing compounds detected in the DRI pattern, the homologues series in this pattern likely reflect natural background. The results suggested that natural compounds make up a considerable part in the pollutant mixtures detected along the river. Further efforts are necessary to study these compounds, especially, because they play a role in the overall ecosystem health (Pignatello and Xing, 1996) and in water treatment (Neale et al., 2012).

Figure 19: Characterization of main patterns by isotopologues and homologue series. Contributions of patterns to number of potentially halogenated (A) and sulfur-containing (B) compounds as well as to peaks involved in a homologue series (C) for sampling spots representative for the main patterns along the Holtemme River.

Chapter 3 – Spatial chemical patterns

3.3.3 Identification of ions of interest

In addition to target compounds, representative ions of interests were identified to different levels of confidence (Schymanski et al., 2014a). The identification focused on high intensity peaks in the common WW pattern (WW1) as well as the two WWTP-specific patterns (WW2 and WW3) and the DRI and BR pattern. The results are summarized in Tables 3 and S19-21. Based on determined molecular formulas, plausible candidate structures were selected using MS2 spectra, pKa values (indicated pH-dependent retention times) and number of exchangeable hydrogens. Finally, commercial relevance was considered as an indicator for the probability to occur in a wastewater-impacted river. The MS2 spectra of the compounds in the original sample and the respective reference standards are presented in the appendix, section B3.4.2.

In the WW patterns, several pharmaceuticals (lamotrigine, methocarbamol, irbesartan and olmesartan) and some pharmaceutical transformation products (gabapentin-lactam and valsartan acid) could be confirmed by reference standards. The peak of lamotrigine was correctly identified by the R ‘nontarget’ package as ion with chlorine isotopes. Lamotrigine was assigned to the WW3 subpattern and showed a distinct peak at WWTP I (Table S19). The intensity was reduced to 30% of the original peak over the course of the river. WWTP I had a specific input of some pharmaceuticals such as the antidepressant pipamperone (Table S19). This might be explained by the presence of a pharmaceutical manufacturer connected to the WWTP as there is no difference in hospital size or specialization among the two towns at the Holtemme River. Lamotrigine is a ubiquitous pharmaceutical previously detected, e.g., in the Rhine River, in Swiss WWTP effluents and a US estuary (Carpenter and Helbling, 2018, Munz et al., 2017, Muz et al., 2017, Ruff et al., 2015). The other identified pharmaceuticals showed similar intensities at both WWTP effluent sites (Table S19). Methocarbamol is a muscle relaxant and irbesartan, olmesartan and valsartan (the latter detected as its transformation product valsartan acid) are used for treatment of hypertension. The high intensity and detections in other studies can be explained by high consumption volumes of these widely used pharmaceuticals (Carpenter and Helbling, 2018, Munz et al., 2017). Irbesartan was detected in 100% of WWTP effluents in EU-wide study (Loos et al., 2013). Gabapentin-lactam is a human metabolite of the anticonvulsant gabapentin and is more stable under environmental conditions than the parent compound (Henning et al., 2018). Gabapentin was part of the target list and has been assigned to the WW2 subpattern showing a 50% higher intensity in the effluent of WWTP II than in the effluent of WWTP I, while the intensity of gabapentin-lactam was similar in both WWTP effluents. The lower gabapentin to gabapentin- lactam ratio in the effluent of WWTP I might be explained by a more efficient treatment of WWTP I. 68

Chapter 3 – Spatial chemical patterns

Furthermore, 4-methyl-7-ethylaminocoumarin was identified by a reference standard as specific to WWTP I (Table S19). Coumarin derivatives were identified as ecotoxicologically relevant compounds specifically emitted from this WWTP (Muschket et al., 2018). 4-Methyl-7- ethylaminocoumarin is the transformation product of 4-methyl-7-diethylaminocoumarin. Like the parent compound, it has an anti-androgenic effect. However it is less potent than its parent compound (Muschket et al., 2018). The sulfophenyl carboxylic acids (SPC) C6-SPC and C7-SPC were tentatively identified at confidence level 2b. Their identification matched the isotopologue and homologue patterns revealed in section 3.3.2 as representatives of a sulfur-containing homologue series. SPCs are main degradation products of linear alkylbenzene sulfonates (LAS) and have been detected in the aquatic environment and WWTP effluents (Lara-Martín et al., 2011). No records were available in MassBank spectral library for C6-SPC or C7-SPC. However, diagnostic fragments (183.0123 m/z and 197.0279 m/z) and ionization were matched to previous studies (Appendix B3.4.3, Figure S30) (Gonsior et al., 2011, Lara-Martín et al., 2011). Moreover, the mass increment 14 m/z suggested a CH2 - homologue series.

Seven out of 21 ions of interest were identified at level 4 in the WW pattern. By application of the pH-dependent LC retention method, two of these peaks with the same molecular formula behind the m/z 274.2010 were separated (Appendix B3.4.4, Figure S32). Even though, the two compounds could not be fully identified, one peak must belong to a carboxylic acid and the other to a compound with a basic functional group with a basic pKa between 2.6 and 6.4, e.g., primary, secondary, tertiary aromatic amines or triazine derivates. The limits of proper pKa calculation were exemplified for irbesartan, olmesartan and 4-methyl-7-ethylaminocoumarin. Here, the calculated pKa did not correspond to the structures suggested by the pH-dependent LC retention (Table 3).

Thus, care that has to been taken in the evaluation of calculated pKa values. Only for two ions in the WW pattern, no unequivocal molecular formula could be determined.

The BR pattern was dominated by peaks which were predominantly showing ammonium adducts

+ + + [M+NH4] but also the [M+H] and [M+Na] adducts. Five of these peaks were identified (level 1) as polyethylene glycols (PEGs) with the general molecular formula C2nH4n+2On+1. They are usually detected as these adducts (Alygizakis et al., 2019, Lara-Martín et al., 2011, Peter et al., 2018). PEGs have a broad field of application in industrial and household products and may enter via rain sewers during surface runoff (Peter et al., 2018) as well as via treated (Schymanski et al., 2014b) and untreated (Gago-Ferrero et al., 2015) wastewater input. PEGs were also observed at other sampling sites at the Holtemme River, e.g., in urban regions and at the weir (Table S20), but not as dominant as at the confluence with the Bode River. Moreover, the intensities dropped downstream of the WWTP effluents suggesting a removal by WWTPs in agreement with other 69

Chapter 3 – Spatial chemical patterns studies (Freeling et al., 2019). The results coincided with the overall patterns revealed by isotopologue signatures and homologue series detection (section 3.3.2) which suggested a specific contribution of Bode River to the Holtemme River, e.g., by untreated wastewater or a specific point source. Moreover, other surfactants and industrial compounds were identified at this spot including triacetin, diethylene glycol monobutyl ether and azelaic acid (level 1). Triacetin was identified in surface waters and groundwater (Schwarzbauer and Ricking, 2010, Sorensen et al., 2015) and was previously linked to specific industrial effluents and proposed as an indicator for the production of paper and inks (Botalova et al., 2011). However, it has a broad range of other industrial applications as a food additive, plasticizer and in pharmaceutical products suggesting a variety of sources. Azelaic acid was intensively studied in and associated with airborne organic particulate matter as a photochemical oxidation product of unsaturated fatty acids (Hyder et al., 2012, Wang et al., 2002). In this study, azelaic acid was only detected at the sampling spot at the river mouth (Table S20) which contradicts an input from of atmospheric deposition. However, it is also used in PCPs (DrugBank, 2019), which might explain its local occurrence in the Holtemme River. Again, these specifically high occurrences in the BR pattern call for further in-depth investigations on sources in the Bode River and dynamics at this particular sampling spot.

In the DRI pattern, five out of eight ions of interest could be identified to level 1 as constituents of cocamidopropylbetaine as well as n-lauroylethanolamine and triethylene glycol monomethyl ether. Cocamidopropylbetaine and n-lauroylethanolamine are surfactants mainly used in PCPs (ECHA, 2019a, b). These compounds were not related to the input of treated wastewater, as they were not detected downstream of WWTP effluents, but were specifically high in the urban area upstream of WWTP I (Table S21) suggesting input of untreated wastewater via rain sewers (Beckers et al., 2018). They were clustered together with the target compound lauryl diethanolamide in the DRI pattern (Table S21). In absence of a reference standard, lauryl sulfate was tentatively identified at level 2a (Appendix B3.4.3, Figure S31). It was previously identified in untreated wastewater (Alygizakis et al., 2019). Triethylene glycol monomethyl ether and lauryl sulfate were related to point source pollution at a sampling spot close to a rain sewer and at sampling spot Holt36, which is at a weir (Table S21, Figure 13). The site-specific detection of these compounds might suggest an input of raw wastewater and surface runoff via rain sewers, their quick removal from the water phase and a remobilization in the weir area from deposited sediments, respectively.

Chapter 3 – Spatial chemical patterns

Table 3: Results of structure elucidation for ions of interest

Peak identifiers Candidate identifiers Number of exchangeable Results of pH-dependent Level of Sample # hydrogens LC retention confidence information m/z Mono- RT Adduct Formula Name From HDX For acidic pKa basic pKa Pattern isotopic [min] experiment candidate mass structure

exp calc exp calc

+ 154.1226 153.1153 8.5 [M+H] C9H15NO Gabapentin-lactam 1 1 >10 14.9 <2.6 1 WW1

+ 204.1019 203.0943 10.0 [M+H] C12H13NO2 4 -Methyl-7- 1 1 >10 <2.6 4.3 1 WW3 ethylaminocoumarin + 223.1440 222.1364 9.2 [M+H] C12H18N2O2 1 >10 <2.6 4 WW1

+ 249.1847 248.1771 11.2 [M+H] C16H24O2 0 >10 <2.6 4 WW1

+ 256.0149 255.0073 5.2 [M+H] C9H7Cl2N5 Lamotrigine 4 4 2.6 5.9 1 WW3 < pKa <6.4 + 269.1283 268.1207 11.6 [M+H] C16H16N2O2 2 >10 <2.6 4 WW3

+ 327.1337 326.1261 11.9 [M+H] C18H18N2O4 2 <6.4 4 WW3

+ 212.2008 211.1932 11.9 [M+H] C13H25NO 1 >10 <2.6 4 WW1

+ 214.1437 213.1361 8.1 [M+H] C11H19NO3 nr <6.4 4 WW2

+ 242.1021 241.0945 7.4 [M+H] C11H15NO5 Methocarbamol nr 3 >10 13.6 <2.6 -3.4 1 WW1

+ 267.0871 266.0799 9.3 [M+H] C14H10N4O2 2 2 <6.4 3.5 1

- Valsartan acid WW1 265.0729 266.0801 9.3 [M-H] C14H10N4O2 2 2 <6.4 3.5 1

+ 274.2010 273.1934 7.7 [M+H] C14H27NO4 <6.4 4 WW2

+ 274.2010 273.1934 7.7 [M+H] C14H27NO4 nr 2.6< 4 WW2 pKa <6.4

354.0716 353.0640 8.9 [M+H]+ not 1 >10 <2.6 5 WW1 determined + 429.2397 428.2321 11.2 [M+H] C25H28N6O Irbesartan 1 1 6.4< 4.3 1 WW1 pKa <10 443.2188 442.2112 10.2 [M+H]+ not 1 6.4< 5 WW2 determined pKa <10 + 447.2136 446.206 9.0 [M+H] C24H26N6O3 Olmesartan nr 3 6.4< 5.0 1 WW2 pKa <10

Chapter 3 – Spatial chemical patterns

Peak identifiers Candidate identifiers Number of exchangeable Results of pH-dependent Level of Sample hydrogens LC retention confidence# information m/z Mono- RT Adduct Formula Name From HDX For acidic pKa basic pKa Pattern isotopic [min] experiment candidate mass structure exp calc exp calc

- 271.0649 272.0725 9.4 [M-H] C12H16O5S C6-SPC 2 2 <6.4 nc 2b WW2

- 285.0806 286.0882 10.6 [M-H] C13H18O5S 2 2 <6.4 3.9 2b C7-SPC WW2 304.1208 286.0875 11.7 [M+NH4] C13H18O5S 2 2 <6.4 -1.8 2b + + 165.1121 164.1048 1.8 [M+H] C7H16O4 Triethylene glycol nr 1 nr 15.1 nr -2.7 1 DRI monomethyl ether + 244.2270 243.2197 13.3 [M+H] C14H29NO2 N- 2 2 >10 15.5 <2.6 -1.3 1 DRI lauroylethanolamine + 272.2581 271.2508 12.5 [M+H] C16H33NO2 Lauryl betaine 0 0 >10 2.6 <2.6 nc 1 DRI

+ 343.2953 342.2880 12.7 [M+H] C19H38N2O3 Lauramidopropyl 1 1 >10 15.9 <2.6 -1.1 1 DRI betaine + 345.2745 344.2672 12.6 [M+H] C19H32N6 3 6.4< 4 DRI pKa <10 + 371.3265 370.3192 13.7 [M+H] C21H42N2O3 Myristamidopropyl 2 2 >10 15.9 <2.6 -1.1 1 DRI betaine - 265.1482 266.1558 18.5 [M-H] C12H26O4S Lauryl sulfate 1 1 <6.4 -1.5 2a DRI

- 293.1762 294.1838 12.0 [M-H] C17H26O4 2 <6.4 4 DRI

+ 155.1542 154.1466 1.4 [M+H] C9H18N2 nr nr nr 4 BR1

+ 163.1328 162.1252 7.9 [M+H] C8H18O3 Diethylene glycol nr 1 nr 15.1 nr -2.7 1 BR1 monobutyl ether 236.1127 218.0790 7.2 [M+NH4] C9H14O6 Triacetin 0 0 nr nc nr -6.5 1 BR1 + 300.2014 282.1678 5.7 [M+NH4] C12H26O7 Hexaethylene glycol nr 2 >10 14.8 <2.6 -2.7 1 BR1 + 344.2276 326.1941 6.6 [M+NH4] C14H30O8 Heptaethylene nr 2 >10 14.8 <2.6 -2.7 1 BR1 + glycol 388.2539 370.2202 7.1 [M+NH4] C16H34O9 Octaethylene glycol 2 2 >10 14.8 <2.6 -2.7 1 BR1 + 432.2802 414.2465 7.5 [M+NH4] C18H38O10 Nonaethylene glycol 2 2 >10 14.8 <2.6 -2.7 1 BR1 + 476.3065 458.2727 7.9 [M+NH4] C20H42O11 Decaethylene glycol 2 2 >10 14.8 <2.6 -2.7 1 BR1 + - 187.0979 188.1055 8.8 [M-H] C9H16O4 Azelaic acid 2 2 nr 4.1 nr nc 1 BR1 #level of confidence according to Schymanski et al. (2014), nr = no results obtained from experiments, nc= not calculable by JChem

Chapter 3 – Spatial chemical patterns

3.4 Conclusions

The analytical power of NTS is continuously increasing and the volume of NTS data produced is increasing exponentially. However, the availability of concepts and tools to structure and exploit these huge data sets is lacking behind. The present study demonstrated how innovative analytical workflows integrating multivariate statistical approaches emerging from different areas of research help identifying pollution patterns and source-related fingerprints in the highly complex pollutant mixtures. To my knowledge, this is the first study to apply a longitudinal cluster analysis on a nontarget data set efficiently separating peaks originating from different sources. The identified patterns suggested a high abundance of natural background in environmental chemical mixtures which could be separated from clear anthropogenic inputs, but request further investigation. The cluster analysis was robust enough to identify main pollution patterns despite many single detects in the data set. By means of isotopologue fingerprints and homologue series as well as detected target and identified nontarget compounds, the patterns were related to inputs from WWTPs, specific pollutants at the river’s mouth and point pollution of untreated wastewater. The exchange of identified patterns in environmental mixtures and source-related fingerprints is encouraged among researchers to test their validity in other water bodies and point sources and allow for their complementation. The proposed workflow is extendable to and should be tested in other settings (e.g. larger rivers, river stretches) to quickly identify pollution hotspots or pathways or identifying temporal dynamics. Thus, the approach presented here is an important building block in the ongoing attempts to understand, monitor, prioritize and manage complex environmental mixture (Brack et al., 2018a).

Chapter 3 – Spatial chemical patterns

Acknowledgements

This study was supported by SOLUTIONS project funded by the European Union Seventh Framework Programme (FP7-ENV-2013-two-stage Collaborative project) under grant agreement number 603437. The authors thank the WWTP operators for providing effluent samples, discharge data and information on the WWTP catchments. The authors further acknowledge Christin Müller (UFZ) for providing the map of the study site and Andreas Musolff (UFZ) for calculating flow velocities of the river. A free academic license of JChem for Office (Excel) was used for structure based property calculation, JChem for Office 6.2.1, 2014, ChemAxon (http://www.chemaxon.com). The QExactive Plus LC-HRMS used is part of the major infrastructure initiative CITEPro (Chemicals in the Terrestrial Environment Profiler) funded by the Helmholtz Association.

Chapter 3 – Spatial chemical patterns

References Allen, F., Pon, A., Wilson, M., Greiner, R. and Wishart, D. (2014) CFM-ID: a web server for annotation, spectrum prediction and metabolite identification from tandem mass spectra. Nucleic acids research 42(Web Server issue), W94-W99. 10.1093/nar/gku436 Altenburger, R., Ait-Aissa, S., Antczak, P., Backhaus, T., Barceló, D., Seiler, T.-B., Brion, F., Busch, W., Chipman, K., de Alda, M.L., de Aragão Umbuzeiro, G., Escher, B.I., Falciani, F., Faust, M., Focks, A., Hilscherova, K., Hollender, J., Hollert, H., Jäger, F., Jahnke, A., Kortenkamp, A., Krauss, M., Lemkine, G.F., Munthe, J., Neumann, S., Schymanski, E.L., Scrimshaw, M., Segner, H., Slobodnik, J., Smedes, F., Kughathas, S., Teodorovic, I., Tindall, A.J., Tollefsen, K.E., Walz, K.-H., Williams, T.D., Van den Brink, P.J., van Gils, J., Vrana, B., Zhang, X. and Brack, W. (2015) Future water quality monitoring — Adapting tools to deal with mixtures of pollutants in water resource management. Science of The Total Environment 512– 513, 540-551. http://dx.doi.org/10.1016/j.scitotenv.2014.12.057 Alygizakis, N.A., Gago-Ferrero, P., Hollender, J. and Thomaidis, N.S. (2019) Untargeted time-pattern analysis of LC-HRMS data to detect spills and compounds with high fluctuation in influent wastewater. Journal of Hazardous Materials 361, 19-29. https://doi.org/10.1016/j.jhazmat.2018.08.073 Beckers, L.-M., Busch, W., Krauss, M., Schulze, T. and Brack, W. (2018) Characterization and risk assessment of seasonal and weather dynamics in organic pollutant mixtures from discharge of a separate sewer system. Water Research 135, 122-133. https://doi.org/10.1016/j.watres.2018.02.002 Botalova, O., Schwarzbauer, J. and Sandouk, N.a. (2011) Identification and chemical characterization of specific organic indicators in the effluents from chemical production sites. Water Research 45(12), 3653-3664. https://doi.org/10.1016/j.watres.2011.04.012 Brack, W., Escher, B., Müller, E., Schmitt-Jansen, M., Schulze, T., Slobodnik, J. and Hollert, H. (2018a) Towards a holistic and solution-oriented monitoring of chemical status of European water bodies: how to support the EU strategy for a non-toxic environment? Environmental Sciences Europe DOI: 10.1186/s12302-018-0161-1. Brack, W., Escher, B.I., Müller, E., Schmitt-Jansen, M., Schulze, T., Slobodnik, J. and Hollert, H. (2018b) Towards a holistic and solution-oriented monitoring of chemical status of European water bodies: how to support the EU strategy for a non-toxic environment? Environmental Sciences Europe 30(1), 33. 10.1186/s12302-018-0161-1 Buerge, I.J., Buser, H.-R., Kahle, M., Müller, M.D. and Poiger, T. (2009) Ubiquitous Occurrence of the Artificial Sweetener Acesulfame in the Aquatic Environment: An Ideal Chemical Marker of Domestic Wastewater in Groundwater. Environmental Science & Technology 43(12), 4381-4385. 10.1021/es900126x Busch, W., Schmidt, S., Kühne, R., Schulze, T., Krauss, M. and Altenburger, R. (2016) Micropollutants in European rivers: A mode of action survey to support the development of effect-based tools for water monitoring. Environmental Toxicology and Chemistry 35(8), 1887- 1899. 10.1002/etc.3460 Carpenter, C.M.G. and Helbling, D.E. (2018) Widespread Micropollutant Monitoring in the Hudson River Estuary Reveals Spatiotemporal Micropollutant Clusters and Their Sources. Environmental Science & Technology 52(11), 6187-6196. 10.1021/acs.est.8b00945 Carpenter, C.M.G., Wong, L.Y.J., Johnson, C.A. and Helbling, D.E. (2019) Fall Creek Monitoring Station: Highly Resolved Temporal Sampling to Prioritize the Identification of Nontarget Micropollutants in a Small Stream. Environmental Science & Technology 53(1), 77-87. 10.1021/acs.est.8b05320 Chambers, M.C., Maclean, B., Burke, R., Amodei, D., Ruderman, D.L., Neumann, S., Gatto, L., Fischer, B., Pratt, B., Egertson, J., Hoff, K., Kessner, D., Tasman, N., Shulman, N., Frewen, B., Baker, T.A., Brusniak, M.-Y., Paulse, C., Creasy, D., Flashner, L., Kani, K., Moulding, C., Seymour, S.L., Nuwaysir, L.M., Lefebvre, B., Kuhlmann, F., Roark, J., Rainer, P., 75

Chapter 3 – Spatial chemical patterns

Detlev, S., Hemenway, T., Huhmer, A., Langridge, J., Connolly, B., Chadick, T., Holly, K., Eckels, J., Deutsch, E.W., Moritz, R.L., Katz, J.E., Agus, D.B., MacCoss, M., Tabb, D.L. and Mallick, P. (2012) A cross-platform toolkit for mass spectrometry and proteomics. Nature Biotechnology 30, 918. https://doi.org/10.1038/nbt.2377 Chiaia-Hernández, A.C., Günthardt, B.F., Frey, M.P. and Hollender, J. (2017) Unravelling Contaminants in the Anthropocene Using Statistical Analysis of Liquid Chromatography–High- Resolution Mass Spectrometry Nontarget Screening Data Recorded in Lake Sediments. Environmental Science & Technology 51(21), 12547-12556. 10.1021/acs.est.7b03357 Dann, J.P., Brack, W. and Krauss, M. (2016). pH-Dependent Retention Time Measurement for the Identification of Unknown Substances with LC-HRMS. (30.05.2016).NonTarget2016, Ascona, Switzerland. 10.5281/zenodo.3267256 Doppler, T., Camenzuli, L., Hirzel, G., Krauss, M., Lück, A. and Stamm, C. (2012) Spatial variability of herbicide mobilisation and transport at catchment scale: insights from a field experiment. Hydrol. Earth Syst. Sci. 16(7), 1947-1967. 10.5194/hess-16-1947-2012 DrugBank (2019), Azelaic acid (accessed on 24.04.2019). https://www.drugbank.ca/drugs/DB00548 ECHA (2019a), 1-Propanaminium, 3-amino-N-(carboxymethyl)-N,N-dimethyl-, N-coco acyl derivs., hydroxides, inner salts (accessed on 23.04.2019). https://echa.europa.eu/de/substance-information/-/substanceinfo/100.057.308 ECHA (2019b), N-(2-hydroxyethyl)dodecanamide (accessed on 23.04.2019). https://echa.europa.eu/de/substance-information/-/substanceinfo/100.005.055 Freeling, F., Alygizakis, N.A., von der Ohe, P.C., Slobodnik, J., Oswald, P., Aalizadeh, R., Cirka, L., Thomaidis, N.S. and Scheurer, M. (2019) Occurrence and potential environmental risk of surfactants and their transformation products discharged by wastewater treatment plants. Sci Total Environ 681, 475-487. 10.1016/j.scitotenv.2019.04.445 Gago-Ferrero, P., Schymanski, E.L., Bletsou, A.A., Aalizadeh, R., Hollender, J. and Thomaidis, N.S. (2015) Extended Suspect and Non-Target Strategies to Characterize Emerging Polar Organic Contaminants in Raw Wastewater with LC-HRMS/MS. Environmental Science & Technology 49(20), 12333-12341. 10.1021/acs.est.5b03454 Genolini, C., Alacoque, X., Sentenac, M. and Arnaud, C. (2015) kml and kml3d: R Packages to Cluster Longitudinal Data. Journal of Statistical Software 65(4), 34. 10.18637/jss.v065.i04 Gonsior, M., Zwartjes, M., Cooper, W.J., Song, W., Ishida, K.P., Tseng, L.Y., Jeung, M.K., Rosso, D., Hertkorn, N. and Schmitt-Kopplin, P. (2011) Molecular characterization of effluent organic matter identified by ultrahigh resolution mass spectrometry. Water Research 45(9), 2943-2953. https://doi.org/10.1016/j.watres.2011.03.016 Greenwood, P.F., Berwick, L.J. and Croué, J.P. (2012) Molecular characterisation of the dissolved organic matter of wastewater effluents by MSSV pyrolysis GC–MS and search for source markers. Chemosphere 87(5), 504-512. https://doi.org/10.1016/j.chemosphere.2011.12.051 Henning, N., Kunkel, U., Wick, A. and Ternes, T.A. (2018) Biotransformation of gabapentin in surface water matrices under different redox conditions and the occurrence of one major TP in the aquatic environment. Water Research 137, 290-300. https://doi.org/10.1016/j.watres.2018.01.027 Hollender, J., Schymanski, E.L., Singer, H.P. and Ferguson, P.L. (2017) Nontarget Screening with High Resolution Mass Spectrometry in the Environment: Ready to Go? Environmental Science & Technology 51(20), 11505-11512. 10.1021/acs.est.7b02184 Horai, H., Arita, M., Kanaya, S., Nihei, Y., Ikeda, T., Suwa, K., Ojima, Y., Tanaka, K., Tanaka, S., Aoshima, K., Oda, Y., Kakazu, Y., Kusano, M., Tohge, T., Matsuda, F., Sawada, Y., Hirai, M.Y., Nakanishi, H., Ikeda, K., Akimoto, N., Maoka, T., Takahashi, H., Ara, T., Sakurai, N., Suzuki, H., Shibata, D., Neumann, S., Iida, T., Tanaka, K., Funatsu, K., Matsuura, F., Soga, T.,

Chapter 3 – Spatial chemical patterns

Taguchi, R., Saito, K. and Nishioka, T. (2010) MassBank: a public repository for sharing mass spectral data for life sciences. Journal of Mass Spectrometry 45(7), 703-714. 10.1002/jms.1777 Hug, C., Ulrich, N., Schulze, T., Brack, W. and Krauss, M. (2014) Identification of novel micropollutants in wastewater by a combination of suspect and nontarget screening. Environmental Pollution 184(Supplement C), 25-32. https://doi.org/10.1016/j.envpol.2013.07.048 Hyder, M., Genberg, J., Sandahl, M., Swietlicki, E. and Jönsson, J.Å. (2012) Yearly trend of dicarboxylic acids in organic aerosols from south of Sweden and source attribution. Atmospheric Environment 57, 197-204. https://doi.org/10.1016/j.atmosenv.2012.04.027 Kassambara, A. (2018). ggpubr: 'ggplot2' Based Publication Ready Plots. v0.2. Kolpin, D.W., Thurman, E.M. and Linhart, S.M. (2000) Finding minimal herbicide concentrations in ground water? Try looking for their degradates. Science of The Total Environment 248(2), 115-122. https://doi.org/10.1016/S0048-9697(99)00535-5 Lara-Martín, P.A., González-Mazo, E. and Brownawell, B.J. (2011) Multi-residue method for the analysis of synthetic surfactants and their degradation metabolites in aquatic systems by liquid chromatography–time-of-flight-mass spectrometry. Journal of Chromatography A 1218(30), 4799-4807. https://doi.org/10.1016/j.chroma.2011.02.031 Loos, M. and Singer, H. (2017) Nontargeted homologue series extraction from hyphenated high resolution mass spectrometry data. Journal of Cheminformatics 9(1), 12. 10.1186/s13321-017-0197-z Loos, R., Carvalho, R., António, D.C., Comero, S., Locoro, G., Tavazzi, S., Paracchini, B., Ghiani, M., Lettieri, T., Blaha, L., Jarosova, B., Voorspoels, S., Servaes, K., Haglund, P., Fick, J., Lindberg, R.H., Schwesig, D. and Gawlik, B.M. (2013) EU-wide monitoring survey on emerging polar organic contaminants in wastewater treatment plant effluents. Water Research 47(17), 6475-6487. http://dx.doi.org/10.1016/j.watres.2013.08.024 Malaj, E., von der Ohe, P.C., Grote, M., Kühne, R., Mondy, C.P., Usseglio-Polatera, P., Brack, W. and Schäfer, R.B. (2014) Organic chemicals jeopardize the health of freshwater ecosystems on the continental scale. Proceedings of the National Academy of Sciences 111(26), 9549-9554. 10.1073/pnas.1321082111 Montero, P. and Vilar, J.A. (2014) TSclust: An R Package for Time Series Clustering. Journal of Statistical Software 62(1), 43. 10.18637/jss.v062.i01 Moschet, C., Wittmer, I., Simovic, J., Junghans, M., Piazzoli, A., Singer, H., Stamm, C., Leu, C. and Hollender, J. (2014) How a Complete Pesticide Screening Changes the Assessment of Surface Water Quality. Environmental Science & Technology 48(10), 5423-5432. 10.1021/es500371t Müller, A., Schulz, W., Ruck, W.K.L. and Weber, W.H. (2011) A new approach to data evaluation in the non-target screening of organic trace substances in water analysis. Chemosphere 85(8), 1211-1219. https://doi.org/10.1016/j.chemosphere.2011.07.009 Müller, C., Musolff, A., Strachauer, U., Brauns, M., Tarasova, L., Merz, R. and Knöller, K. (2018) Tomography of anthropogenic nitrate contribution along a mesoscale river. Science of The Total Environment 615, 773-783. https://doi.org/10.1016/j.scitotenv.2017.09.297 Munz, N.A., Burdon, F.J., de Zwart, D., Junghans, M., Melo, L., Reyes, M., Schönenberger, U., Singer, H.P., Spycher, B., Hollender, J. and Stamm, C. (2017) Pesticides drive risk of micropollutants in wastewater-impacted streams during low flow conditions. Water Research 110, 366-377. http://dx.doi.org/10.1016/j.watres.2016.11.001 Muschket, M., Di Paolo, C., Tindall, A.J., Touak, G., Phan, A., Krauss, M., Kirchner, K., Seiler, T.-B., Hollert, H. and Brack, W. (2018) Identification of Unknown Antiandrogenic Compounds in Surface Waters by Effect-Directed Analysis (EDA) Using a Parallel Fractionation Approach. Environmental Science & Technology 52(1), 288-297. 10.1021/acs.est.7b04994 Muz, M., Krauss, M., Kutsarova, S., Schulze, T. and Brack, W. (2017) Mutagenicity in Surface Waters: Synergistic Effects of Carboline Alkaloids and Aromatic Amines. Environmental Science & Technology 51(3), 1830-1839. 10.1021/acs.est.6b05468

Chapter 3 – Spatial chemical patterns

Neale, P.A., Antony, A., Bartkow, M., Farre, M., Heitz, A., Kristiana, I., Tang, J.Y.M. and Escher, B.I. (2012) Bioanalytical assessment of the formation of disinfection by-products in a drinking water treatment plant. Environmental Science & Technology 46(18), 10317–10325. 10.1016/j.watres.2011.05.038 Nürenberg, G., Schulz, M., Kunkel, U. and Ternes, T.A. (2015) Development and validation of a generic nontarget method based on liquid chromatography – high resolution mass spectrometry analysis for the evaluation of different wastewater treatment options. Journal of Chromatography A 1426, 77-90. https://doi.org/10.1016/j.chroma.2015.11.014 Peter, K.T., Tian, Z., Wu, C., Lin, P., White, S., Du, B., McIntyre, J.K., Scholz, N.L. and Kolodziej, E.P. (2018) Using High-Resolution Mass Spectrometry to Identify Organic Contaminants Linked to Urban Stormwater Mortality Syndrome in Coho Salmon. Environmental Science & Technology 52(18), 10317-10327. 10.1021/acs.est.8b03287 Pignatello, J.J. and Xing, B. (1996) Mechanisms of Slow Sorption of Organic Chemicals to Natural Particles. Environmental Science & Technology 30(1), 1-11. 10.1021/es940683g Pluskal, T., Castillo, S., Villar-Briones, A. and Orešič, M. (2010) MZmine 2: Modular framework for processing, visualizing, and analyzing mass spectrometry-based molecular profile data. BMC Bioinformatics 11, 395-395. 10.1186/1471-2105-11-395 R Core Team (2017). R: A language and environment for statistical computing. v3.4.3. R Foundation for Statistical Computing Reemtsma, T., Alder, L. and Banasiak, U. (2013) Emerging pesticide metabolites in groundwater and surface water as determined by the application of a multimethod for 150 pesticide metabolites. Water Research 47(15), 5535-5545. https://doi.org/10.1016/j.watres.2013.06.031 Richardson, S.D. and Kimura, S.Y. (2017) Emerging environmental contaminants: Challenges facing our next generation and potential engineering solutions. Environmental Technology & Innovation 8, 40-56. https://doi.org/10.1016/j.eti.2017.04.002 Royal Society of Chemistry (2015) ChemSpider. http://www.chemspider.com/ Ruff, M., Mueller, M.S., Loos, M. and Singer, H.P. (2015) Quantitative target and systematic non-target analysis of polar organic micro-pollutants along the river Rhine using high- resolution mass-spectrometry – Identification of unknown sources and compounds. Water Research 87(Supplement C), 145-154. https://doi.org/10.1016/j.watres.2015.09.017 Ruttkies, C., Schymanski, E.L., Strehmel, N., Hollender, J., Neumann, S., Williams, A.J. and Krauss, M. (2019) Supporting non-target identification by adding hydrogen deuterium exchange MS/MS capabilities to MetFrag. Analytical and Bioanalytical Chemistry. 10.1007/s00216-019-01885-0 Ruttkies, C., Schymanski, E.L., Wolf, S., Hollender, J. and Neumann, S. (2016) MetFrag relaunched: incorporating strategies beyond in silico fragmentation. Journal of Cheminformatics 8, 3-3. 10.1186/s13321-016-0115-9 Schollée, J.E., Schymanski, E.L., Avak, S.E., Loos, M. and Hollender, J. (2015) Prioritizing Unknown Transformation Products from Biologically-Treated Wastewater Using High- Resolution Mass Spectrometry, Multivariate Statistics, and Metabolic Logic. Analytical Chemistry 87(24), 12121-12129. 10.1021/acs.analchem.5b02905 Schwarzbauer, J. and Ricking, M. (2010) Non-target screening analysis of river water as compound-related base for monitoring measures. Environmental Science and Pollution Research 17(4), 934-947. 10.1007/s11356-009-0269-3 Schymanski, E.L., Jeon, J., Gulde, R., Fenner, K., Ruff, M., Singer, H.P. and Hollender, J. (2014a) Identifying Small Molecules via High Resolution Mass Spectrometry: Communicating Confidence. Environmental Science & Technology 48(4), 2097-2098. 10.1021/es5002105 Schymanski, E.L., Singer, H.P., Longrée, P., Loos, M., Ruff, M., Stravs, M.A., Ripollés Vidal, C. and Hollender, J. (2014b) Strategies to Characterize Polar Organic Contamination in Wastewater: Exploring the Capability of High Resolution Mass Spectrometry. Environmental Science & Technology 48(3), 1811-1818. 10.1021/es4044374 78

Chapter 3 – Spatial chemical patterns

Sorensen, J.P.R., Lapworth, D.J., Nkhuwa, D.C.W., Stuart, M.E., Gooddy, D.C., Bell, R.A., Chirwa, M., Kabika, J., Liemisa, M., Chibesa, M. and Pedley, S. (2015) Emerging contaminants in urban groundwater sources in Africa. Water Research 72, 51-63. https://doi.org/10.1016/j.watres.2014.08.002 Stenson, A.C., Landing, W.M., Marshall, A.G. and Cooper, W.T. (2002) Ionization and Fragmentation of Humic Substances in Electrospray Ionization Fourier Transform-Ion Cyclotron Resonance Mass Spectrometry. Analytical Chemistry 74(17), 4397-4409. 10.1021/ac020019f US EPA (2019) United States Environmental Protection Agency, CompTox Chemicals Dashboard. https://comptox.epa.gov/dashboard Verkh, Y., Rozman, M. and Petrovic, M. (2018) A non-targeted high-resolution mass spectrometry data analysis of dissolved organic matter in wastewater treatment. Chemosphere 200, 397-404. https://doi.org/10.1016/j.chemosphere.2018.02.095 Wang, G., Niu, S., Liu, C. and Wang, L. (2002) Identification of dicarboxylic acids and aldehydes of PM10 and PM2.5 aerosols in Nanjing, China. Atmospheric Environment 36(12), 1941-1950. https://doi.org/10.1016/S1352-2310(02)00180-2 Wickham, H. (2016). ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York Zheng, W., Wang, X., Tian, D., Zhang, H., Tian, W., Andersen, M.E., Zheng, Y., Sun, X., Jiang, S., Cao, Z., He, G. and Qu, W. (2012) Pollution Trees: Identifying Similarities among Complex Pollutant Mixtures in Water and Correlating Them to Mutagenicity. Environmental Science & Technology 46(13), 7274-7282. 10.1021/es300728q

Chapter 4 – Precipitation-related chemical patterns

Chapter 4

Identification of precipitation-related pollution patterns and indicator compounds during heavy rain events in a small stream

This chapter is currently prepared for submission to an international peer-reviewed journal.

Chapter 4 – Precipitation-related chemical patterns

Abstract Heavy rain events potentially lead to an extraordinary exposure of aquatic organisms to complex and variable mixtures of organic micropollutants via the input of untreated wastewater and surface runoff. Thus, comprehensive screening approaches are necessary to capture and fully characterize the impact of such rain events on overall river pollution. In this study, precipitation- related pollution patterns of six rain events in a small river were unraveled by NTS and cluster analysis. The events were triggered by the overflow of a combined sewer system and sampled in 30-min composite samples over eight hours to detect highly time-resolved patterns in the receiving river. Two main pollution patterns were identified in each event representing i) pre-event emissions which were partially diluted and ii) the increase of rain-related pollutants during heavy rain events. The duration of the rain-related exposure depended on the duration of CSO and the amount of pre-event rain. Longest rain-related exposure up to more than four hours was observed for intense rain events after periods of drought. Even though, the mixture composition between rain events was highly variable (including between 216 and 336 detected target compounds), a typical rain- related mixture and 54 indicator compounds for water monitoring of heavy rain events were identified. These compounds showed a high intra-event intensity fold change or were solely introduced during rain events. Reliable indicator compounds for event-based monitoring of rain- related input included, among others, biocides and surfactants for urban surface runoff as well as piperine and chenodeoxycholic acid for emissions of untreated wastewater. The latter two compounds were reported in surface water for the first time. The high contribution of surfactants to the rain-related mixtures was highlighted by NTS. The study contributed to the understanding of pollutant patterns, dynamics and sources during heavy rain events and derived common indicator compounds suitable for event-based monitoring.

Chapter 4 – Precipitation-related chemical patterns

4.1 Introduction Sudden and intense rain events pose major challenges for water managers and risk to the aquatic environment. Heavy rain events potentially lead to an extraordinary exposure of aquatic organisms to organic micropollutants via the input of untreated wastewater and surface runoff (Kayhanian et al., 2008, Launay et al., 2016). Consequently, these events have been in the focus of research. One major issue is that the composition of stormwater pollution is very complex and variable (e.g.,Launay et al., 2016, Peter et al., 2018). Thus, a comprehensive approach is needed to capture and reduce this complexity and support water monitoring and management strategies accounting for event-related mixture complexity and dynamics (Altenburger et al., 2019, Brack et al., 2018).

Major point sources for rain-related discharge are CSO and rain sewers. In the past, they were studied intensively regarding inputs of suspended solids, pathogens, metals and selected organic micropollutants (e.g.,Gasperi et al., 2012, Launay et al., 2016, Madoux-Humery et al., 2013, Métadier and Bertrand-Krajewski, 2012, Phillips et al., 2012, Wittmer et al., 2010). Discharge form CSO itself integrates several emission sources (e.g., urban surface runoff and wastewater). Depending on individual chemical properties and the contribution of rain water to the total discharge, CSO can lead to peak concentrations of certain pollutants and to dilution of others (Launay et al., 2016). For example, elevated concentrations during CSO were observed for compounds which are largely removed during wastewater treatment, e.g., hormones (Phillips et al., 2012). Similarly, pollutants from diffuse sources like pesticides and biocides as well as surfactants and rubber additives are predominately emitted via rain-related surface runoff (Beckers et al., 2018, Launay et al., 2016, Meyer et al., 2011, Peter et al., 2018, Phillips and Bode, 2004, Wittmer et al., 2010). The event-specific emissions depend on a number of factors such as land-use cover, sewer infrastructure and meteorological (e.g., antecedent dry period) and hydrological conditions (e.g., soil moisture and flow conditions) as well as duration and onset of a rain event and daily or seasonal use of chemicals (Krein et al., 2013, Launay et al., 2016, Madoux- Humery et al., 2013, Métadier and Bertrand-Krajewski, 2012, Phillips and Bode, 2004, Wittmer et al., 2010).

Thus, routine event-based monitoring in rivers has to account for all these sources and variations and, consequently, reliable indicator compounds for heavy rain events are needed. In several studies, indicator compounds have been proposed for emission sources and processes (Jekel et al., 2015) and wastewater (Dickenson et al., 2011, Nödler et al., 2016). However, these studies relied on already pre-selected target compounds. In order to obtain a comprehensive assessment

Chapter 4 – Precipitation-related chemical patterns and characterization mixtures of organic micopollutants in heavy rain events, NTS has to be performed.

NTS offers an unbiased approach to fully characterize complex environmental mixtures (Altenburger et al., 2019, Brack et al., 2018, Hollender et al., 2017). In order to identify patterns in large data sets obtained from NTS, major data treatment steps including multivariate statistics are inevitable. In combination with multivariate exploratory data analysis tools, such as time trend or cluster analysis, extraordinary temporal peaks and trends can be determined in a data set and selected for further compound identification (e.g., Alygizakis et al., 2019, Carpenter et al., 2019, Hollender et al., 2017, Samanipour et al., 2019). Recently, Peter et al. (2018) applied a NTS approach to link micropollutants in stormwater to observed risks in a receiving water body and derive a chemical signature for the pollutant mixture inducing these effects. It has been proposed and shown that complex pollution patterns in surface waters result from a combination of source- related patterns (or such signatures) (Beckers et al., submitted, Brack et al., 2018, Peter et al., 2018), and the same is potentially true for precipitation-related patterns.

Thus, it may be hypothesized that with the application of NTS and cluster analysis, typical pollution patterns for heavy rain events can be revealed independent from inter-event variations of individual pollutants. From these general patterns, indicator compounds suitable for monitoring heavy rain events can be derived.

The Holtemme River with its multi-source catchment was used as a case study. With the support of a hydrological assessment and physicochemical parameters, the study aimed at i) the identification of precipitation-related pollutant patterns, their sources and correlation with hydrological and physicochemcial parameters, ii) the identification of a common precipitation- related pollutant mixture and iii) the prioritization and identification of common rain event pollutants (i.e., indicator compounds). The study supported the development of strategies for event-based monitoring. This was the first study using a nontarget approach to unravel time-resolved pollution patterns of organic micropollutants during heavy rain events in a small river.

Chapter 4 – Precipitation-related chemical patterns

4.2 Methods 4.2.1 Study area The study was performed at the Holtemme River (Saxony-Anhalt, Germany) (Figure 20). The Holtemme River is 47 km long and has it source in the Harz Mountains. It flows from the national park through the medium-sized towns of Wernigerode and Halberstadt and an area with intensive agriculture. There are two WWTPs in the catchment. WWTP I serves the town of Wernigerode and surrounding villages and operates in a separate sewer system. WWTP II operates in a combined sewer system. It covers the mostly urban area of Halberstadt (143 m²). During heavy rain events, stormwater and untreated wastewater bypass WWTP II and are discharged into the Holtemme River via the main CSO structure. The studied CSO is in front of WWTP II and is an emergency measure in case of overload of pumps and bar screens. It is usually only activated during sudden and intense increases of influent. The sampling site was approximately 9 km downstream of the CSO structure. The area between the CSO and the sampling site was mostly used for agricultural purposes (Figure 20).

Figure 20: Map of Holtemme River (Saxony-Anhalt, Germany).

Chapter 4 – Precipitation-related chemical patterns

4.2.2 Sampling The CSO was used as trigger for heavy and sudden rain events that potentially lead to increased surface runoff from urban and rural areas besides discharge of untreated wastewater. In case of a CSO event, the WWTP received a signal from an ultrasonic senor at the CSO structure. The signal was transmitted to the automatic sampler installed 9 km downstream of the CSO structure. The communication scheme between the sensor at the CSO, the WWTP and the automatic sampler is displayed in Figure 21. During an event, 16 30-min composite samples were taken by the automatic sampler equipped with a cooling unit (TP5 C Active Vacuum, MAXX® Ltd.). The sampler was equipped with a telemetry module MT-021 v.1.51 (Inventia Ltd.). Likewise, a telemetry module was connected to a programmable logic controller (PLC) system at the WWTP to transmit the CSO signal to the module at the sampler. The sampling started immediately with the beginning of the CSO and lasted for 8 hours. Samples of 200 ml each were taken in 10-min intervals. The samples were stored in glass bottles at 4°C. Samples were collected latest two days after the event.

Figure 21: Communication scheme between CSO and autosampler (TP5 Active Vacuum, MAXX® Ltd.) PLC = Programmable logic controller. 4.2.3 Data collection hydrology and physicochemical parameters Discharge data of the gauging stations in the Holtemme River (daily means, hourly data and 15- min interval data) were downloaded from the data portal of the State Office for flood protection and water management of Saxony-Anhalt (LHW, 2019). Precipitation data (hourly data) from the station in the village of Sargstedt were downloaded from the archive of the German Weather Service (DWD) (DWD, 2019). Information on CSO discharge amounts was provided by the municipal utilities. This data was only available in hourly and for one event in 15-min intervals.

General physicochemcial parameters such as electric conductivity, NO3-N and turbidity were routinely analyzed at the sampling station by the UFZ in 15-min intervals. Correlations among event characteristics were identified and visualized using the R package ‘corrplot’ (Wei and Simko, 2017). The correlations were tested using different correlation coefficient (i.e., “pearson”, “kendall” and “spearman”).

Chapter 4 – Precipitation-related chemical patterns

4.2.4 Flow separation of discharge data Total discharge represents a mixture of base flow and quick flow. Quick flow integrates surface runoff and lateral flow close to the surface (i.e., interflow) (Bosch et al., 2017). Base flow separation was mathematically performed using the Eckardt (Q90/Q50) method. This method showed the best performance as it was very close to the average base flow determined by all methods. Details on the method are provided elsewhere (Tarasova et al., 2018). The method was applied on discharge data (15-min intervals) of the gauging station at the sampling site of the entire year 2016.

4.2.5 Chemical analysis From the 30-min composite samples, aliquots of 1 mL were taken for chemical analysis by LC- HRMS. Samples were prepared for large volume direct injection (100 µL) by adding 10 µL of a 2 mM ammonium formate buffer, 25 µL of methanol and 25 µL of an internal standard mixture containing 40 isotope-labelled compounds (40 ng mL-1). Chromatographic separation was performed on a Kinetex 2.6 μm EVO C18 (50x2.1 mm) column using an UltiMate 3000 LC system (Thermo Scientific). The LC system was coupled to a quadrupole-Orbitrap MS (Q Exactive Plus, Thermo Scientific) with a HESI source. Details on parameters and settings of the instruments are presented in the appendix, C1.2. For quality control, solvent blanks (95% LCMS grade water/ 5% LCMS grade methanol) were run after every batch of six samples. Furthermore, calibration standards at levels 1, 10, 100 and 1000 ng L-1 containing a mix of all target compounds and internal standards were analyzed at the beginning and end of each batch. The matrix effect was evaluated by plotting the intensity change of the internal standards across all samples (Appendix C2.1).

4.2.6 Nontarget data processing Raw files obtained from LC-HRMS analysis were converted into the mzML format (centroid mode) using Proteowizard v3.0.18265 (Chambers et al., 2012). Peak picking, alignment, target annotation and gap filling were performed by the software MZmine v2.32 (Pluskal et al., 2010). The target list contained 411 compounds for positive ionization mode and 83 compounds for negative ionization mode. Their correct annotation was checked for a plausible increase in peak height along the calibration standards. The peaklists were subjected to blank correction and isotope clean-up using an R workflow based on the package ‘nontarget’ (Loos, 2016). Peaks below the calculated blank threshold (Eq. 7) were removed. Moreover, peaks with intensities below 5000 were removed to reduce noise. Isotopes were identified using the R ‘nontarget’ package. Settings are described in the appendix, C1.4. Identified isotope peaks were removed from the final peaklist. 87

Chapter 4 – Precipitation-related chemical patterns

Eq. 7: Calculation of intensity threshold (Ithres)

Ithres = µ(IBlk) + 2*σ(IBlk)

µ(IBlk) = mean of peak intensities in blanks; σ(IBlk) = standard deviation of peak intensities in blanks

4.2.7 Cluster analysis Prior to cluster analysis, peak intensities were scaled and centered (Eq. 8). This ensured that all remaining peaks had equal variances.

Eq. 8: Scaling to unit variance

푥 − µ 푧 = 휎 z = standard score, µ = mean, σ = standard deviation

Pollution patterns were identified for each event using the R package ‘kml’ (Genolini et al., 2015). Kml cluster analysis is based on the partitioning cluster analysis algorithm and was designed to unravel longitudinal patterns in epidemiological data sets. The method has been applied and validated in a previous study by Beckers et al. (submitted). The function ‘diss.Cort’ from the R package ‘TSclust’ was used as a distance function (Montero and Vilar, 2014) to fit the temporal data set. Settings of the kml analysis are provided in the appendix, section C1.5. For the identification of common peaks in rain-related mixtures, peaklists of the respective “Quick” patterns from each event were meta-aligned in MZmine using the .mzTab format (Appendix C1.3). Common peaks were filtered in Microsoft Excel and visualized by a Venn diagram using the R package ‘venn’ (Dusa, 2018). The peaks of common the rain event mixture were defined by their m/z and RT and were displayed in scatter plots using R ‘nontarget’ (Loos, 2016). Homologue series were identified in this mixture by R ‘nontarget’ and highlighted in the scatter plots (Loos and Singer, 2017).

4.2.8 Structure elucidation The identification of ions of interests focused on the top 30 high-intensity peaks in ESI positive mode (i.e., ranked by mean maximum intensity among events) in the rain-related mixture (4.3.3.4) and members of homologue series which were also detected in the small rain event E2406. Molecular formulas were determined using QualBrowser from Xcalibur (ThermoScientific) and were submitted to a query in Comptox database (US EPA, 2019). In case of no hits, another query

Chapter 4 – Precipitation-related chemical patterns was submitted to ChemSpider database (Royal Society of Chemistry, 2015). Data-dependent MS² spectra were obtained for the ions of interests for which potential molecular formulas could be generated. Experimental spectra of 67 ions of interest were submitted to MetFrag v2.3 (web tool) to obtain a candidate list using ChemSpider database (Ruttkies et al., 2016). Further details on structure elucidation are described in the appendix (B2.6.1, B2.6.2). For homologue series, candidates were checked for structures possibly occurring in series with the identified m/z shift. The level of confidence of identification for each compound was reported according to Schymanski et al. (2014a).

4.3 Results and Discussion The results and discussions are structured into three sections. In the first section, the sampled events are briefly characterized. In the second section, chemical pollution patterns are discussed in the context of discharge patterns and physicochemical parameters in order to understand the occurrence of pollution patterns and derive general advice for sampling of heavy rain events in small streams. Finally, common target compounds in the “Base” and “Quick” patterns and prioritized unknown peaks among events in the “Quick” are identified and indicator compounds for rain-related emissions are discussed.

4.3.1 Characterization of sampled heavy rain events Six heavy rain events associated with CSO were successfully sampled from May to September 2016. The main CSO structure immediately before the entrance of the main channel into the WWTP was used as a trigger for heavy rain events. This overflow is activated in case of a sudden and intense increase of inflow, which exceeds the capacity of pumps and bar screens of the WWTP. Each CSO event was an immediate response to a sudden and short peak in precipitation, which was usually reflected by a peak in river discharge and an increased contribution of quick flow (Figure 22). Thus, the CSO was an appropriate and reliable trigger for sudden and intense rain events. The main characteristics of the events are listed in Table 4. Variable amounts of raw wastewater and stormwater were discharged from the CSO into Holtemme River. The rain event E2406 was the shortest event with exceptionally low CSO amounts (i.e., 30 m³). Since there was just rain prior to the CSO, it may have been a short wave of backwater from WWTP resulting from a short-term overload of the system. Thus, E2406 did not represent a typical CSO event. However, it is still is an example of a rain event potentially leading to surface runoff and influencing the chemical exposure in the receiving water body.

Chapter 4 – Precipitation-related chemical patterns

Figure 22: Hydrograph of the Holtemme River at the sampling site for the year 2016. Sampled events are marked. Precipitation data was obtained from the DWD station in Sargstedt, river discharge data from LHW gauge Nienhagen.

The events were categorized into “local” and “catchment” events. This characterization was based on comparing discharge levels at the two gauging stations within the Holtemme River (Figures 20 and S39-44). If a discharge peak was detected on the day of the event at the upstream gauge, the event was considered as catchment rain event, likely affecting WWTP I and rain sewers in upstream urban areas. E2905, E0106, E2406 and E1307 were catchment events; while the other two events represented local heavy rain events.

Chapter 4 – Precipitation-related chemical patterns

Table 4: Characteristics of sampled heavy rain events

Parameter Events Abbreviation E2905 E0106 E1306 E2406 E1307 E1709 Date of event 29.05.2016 01.06.2016 13.06.2016 24.06.2016 13.07.2016 17.09.2016 Start of sampling 04:35 am 12:19 pm 04:04 pm 07:18 pm 09:56 pm 01:31 am End of sampling 12:35 pm 08:19 pm 12:04 am 03:18 am 05:56 am 09:31 am Start of CSO 04:34 am 12:18 pm 04:03 pm 07:17 pm 09:55 pm 01:30 am End of CSO 06:00 am 02:00 pm 8:00 pm 09:00 pm 01:00 am 06:00 am Duration of CSO 01:30 01:40 04:30 01:40 03:00 04:30 [hh:min] Amount of CSO 550 1,500 55,000 30 6,480 37,080 discharge [m³] Amount of 3.1 3.3 9.4 none 3.6 16.0 precipitation during event [mm] Amount of 6.7 8 6.3 12.1 32.7 8.4 precipitation immediately prior event [mm] Max rain intensity 4.8 2 6.2 8.6 32.4 5.2 prior to event [mm h-1] Antecedent rain 14.1 11.6 0.4 10.9 1 none (<7 d) [mm] Scale of event Catchment Catchment Local Catchment Catchment Local Duration of 2 3 4 2 3.5 4 “Quick” pattern [h] Start quick flow 5.5 5.5 1 6 1.5 2.5 [h after sampling started]

4.3.2 Chemical pollution during heavy rain events After blank correction and removal of isotope peaks, the final peaklists of each rain event contained between 16,092 peaks (E2406) and 29,225 peaks (E1306). By means of kml cluster analysis, main pollution patterns during heavy rain events were identified. In each event, the complex environmental mixtures could be summarized into two or three pollution patterns (Figure 23A-E and S45-50). Target compounds associated with each of the patterns in each event are presented in Table S27. Roughly, one pattern increased with or shortly after the discharge peak (i.e., quick flow) and will further be called pattern “Quick”. The other pattern was either decreased or remained stable after the discharge peak and will further be called pattern “Base”. The two general patterns represented i) pre-event emissions partly diluted during rain events (“Base” pattern) and ii) rain-related emissions via surface runoff and discharges of untreated wastewater (“Quick” pattern). The two patterns agreed with patterns identified in previous studies from point sources and river systems (e.g., Krein et al., 2013, Launay et al., 2016, Meyer et al., 2011, Phillips et al., 2012, Wittmer et al., 2010). In order to get a better understanding of the sources and

Chapter 4 – Precipitation-related chemical patterns dependencies of these patterns and draw general conclusions for event-based monitoring, the pollution patterns were compared with dynamics of physicochemical parameters (section 4.3.2.1) and event characteristics were tested for correlation (section 4.3.2.2).

4.3.2.1 Dynamics of physicochemical parameters during heavy rain events as potential sampling triggers The dynamics of physicochemical parameters during rain events are displayed in the appendix, Figures S52-57. Independent from any (hydrological) differences among the events, changes in discharge marked a turning point in pollution dynamics (Figure 23A-E). However, the peak of the “Quick” pattern was slightly delayed compared to discharge peak in several events (Figure 23 B,C,E,F). The delay between discharge peak and pollutant maximum may have resulted from discharges and runoff of pollution sources from distances further away from the sampling site or by soil retention (Phillips and Bode, 2004; Meyer et al., 2011). This phenomenon is described as the kinematic wave effect (Meyer et al., 2011). When the sampling occurs closer to the runoff source, the pollution peak might even arrive earlier than discharge peak (i.e., first flush) (Meyer et al., 2011, Reoyo-Prats et al., 2018). In this case, sampling triggered by discharge would miss peak concentrations.

Besides discharge, changes in turbidity indicated changes in pollution dynamics. Increases of turbidity with quick flow suggested resuspension of sewer sediments, runoff and mobilization of organic matter by erosion of soil and river banks as well as destabilization of biofilms (Jeanneau et al., 2015, Passerat et al., 2011). In fact, turbidity was a better marker for the main “Quick” pattern (“Quick 2”) during the event E0106. During this event, catchment rain caused several lower peaks in discharge, while turbidity only peaked with the beginning of the main “Quick” pattern. Moreover, changes in discharge were sometimes less pronounced in this small stream (e.g., during E2406), while turbidity increased at least by a factor of 37. Lowest maximum turbidity values were observed in E2406 (i.e., 10.7 NTU, Figures 23D and S54). This value might be a trigger for sampling CSO and runoff events, but the threshold value needs to be established for every river individually. During E1709, the highest turbidity values were measured (max. 391 NTU) over the entire study period (Figures 23E and S56). Due to the long antecedent dry period, there was a long accumulation phase of organic matter. In general, the “Quick” pattern lasted longer than peaks in discharge or turbidity (Figure 23A-E). Thus, sampling should last longer than peaks in physicochemical parameters and discharge to capture full event exposure.

Chapter 4 – Precipitation-related chemical patterns

Electric conductivity and NO3-N concentrations are useful indicators for dilution of river water (i.e., groundwater and treated wastewater) by nutrient-poor rain water (Meyer et al., 2011).

Figure 23: Chemical pollution patterns and according discharge curves A) E2905, B) E0106, C) E1306, D) 2406, E) 1306, F) E1709. Gray background displays scaled peak height of individual peaks in each sample.

Chapter 4 – Precipitation-related chemical patterns

During E2905, E0106 and E1306 the quick flow peak was associated with a drop in electric conductivity and NO3-N concentration. This dilution effect was negligible during the low increase in quick flow during E2406. During E1307, the contribution of rain water only occurred after the second discharge peak and was not observed at all during E1709. During E1709, NO3-N correlated with increase in turbidity suggesting direct surface runoff of fertilizer and runoff from agricultural land (Meyer et al., 2011, Müller et al., 2018). Together with the high and sharp turbidity peak, E1709 represented a typical flash flood event. Moreover, the electric conductivity remained constantly high (i.e., 1063 µS cm-1) over the entire sampling time (Figure S57). These high values were in agreement with values observed in (raw) wastewater during dry weather conditions (Passerat et al., 2011). They are also typical for the Holtemme River as both WWTPs are the main tributaries during low flow conditions (Beckers et al., submitted).

4.3.2.2 Factors influencing dynamics of pollution patterns in the receiving river The relationships between event characteristics and pollution patterns in the receiving river (Table 4) were investigated to determine factors which influence the dynamics and occurrence of precipitation-related pollution patterns in the receiving river (Figure 24). Due to the low number of samples, these correlations represented preliminary results but indicated potential relationships. The correlation analysis revealed factors that influence the duration of the “Quick” pattern and the time of the turning point in pollution patterns in the river.

The duration of the “Quick” pattern increased with precipitation during the CSO event and thus with the duration and amount of CSO. Consequently, the “Quick” pattern reflected discharges from the CSO, urban runoff as well as discharges from other smaller point sources, i.e., rain sewers. As the sampling site was close to the river mouth, the different point sources cannot be fully separated anymore (Wittmer et al., 2010). The peak of the “Quick” pattern lasted up to more than four hours. Due to mixed land-use in the catchment and considerable distance to the urban area, high and sharp pollutant peaks (i.e., first flush) were not observed. Rather, the detected runoff and wastewater pollutant peaks were elongated due to distance, retention by soil and integration of several diffuse and point sources (Krein et al., 2013, Meyer et al., 2011). Only during the event E0106, a short peak prior to the main “Quick” pattern was observed (“Quick 1”, Figure 23C). This peak could not be correlated to the main CSO discharge and likely represented a temporary overflow of a point source such as a storm sewer closer to the sampling site. The target compounds associated with this pattern did not reflect specific type of source, i.e., mixture of wastewater-related and runoff compounds (Table S27B). During the intense catchment event E2905, the second “Quick” pattern, i.e., “Quick 2”, likely represented rain-related discharge of sources further upstream of WWTP II, e.g., WWTP I and rain sewers (Figure 23A). This hypothesis 94

Chapter 4 – Precipitation-related chemical patterns was supported by an increase of coumarin derivates with the “Quick 2” pattern. Coumarin derivates are specifically discharged by WWTP I (Beckers et al., submitted, Muschket et al., 2018).

The duration of “Quick” pattern was longer and the pattern turning point occurred earlier in case of less pre-event rain. Pre-event rain leads to more hydrophilic soil conditions and, thus, may have increased subsurface flow, which is slower than immediate and direct surface runoff (Krein et al., 2013, Meyer et al., 2011). The later turning point in pollution patterns and shorter duration of the main “Quick” patterns were pronounced in the events E2905 and E2406 (Figure 23A, D). Thus, longer exposure to pollutants in the “Quick” pattern might be expected during intense rain events with no pre-event rain such as during E1306, E1307 and E1709. However, the duration of CSO discharge is the dominant factor as observed for E0106.

Figure 24: Correlation matrix of event characteristics.

The occurrence of pollution peaks resulting from heavy rain events depended on several factors that are partially well known but difficult to predict for each event (e.g., pre-event rain or duration of CSO). The limitations of certain potential triggers such as discharge or turbidity as well as factors influencing the occurrence of pollution patterns in the receiving river were discussed above (section 4.3.2.1 and 4.3.2.2) and need to be considered in the design of routine event-based 95

Chapter 4 – Precipitation-related chemical patterns monitoring strategies in other rivers. For water managers relying on a few composite samples, a comprehensive overview of approaches for obtaining representative composite samples is provided by Ort et al. (2010).

4.3.3 Identification of common peaks in “Base” and “Quick” patterns The mixtures of organic micropollutants in the sampled events were quite diverse and contained a number of compounds from different sources. Between 216 and 336 target compounds were annotated in the different events (Table S27). According to previous quantitative target studies, the event mean concentrations of organic micropollutants can vary by at least one order of magnitude between rain event discharges (Krein et al., 2013, Launay et al., 2016). Thus, the mixture composition may change, especially when looking at concentrations of individual compounds. Consequently, each event represented a unique exposure of the environment. However, the general pollution patterns in this study were observed to be quite similar among different events. Thus, common peaks in these general patterns were identified regardless of all variations discussed here and in previous studies. By the application of unbiased NTS, these mixtures were more comprehensively described by including also yet unknown peaks and compounds.

Common peaks were identified for the “Base” and “Quick” patterns (Figure 25). Peaks of the small rain event E2406 were not used as an exclusion criterion as only small amounts (i.e., 30 m²) of untreated wastewater was discharged and the contribution of quick flow to the overall discharge was rather low. The distribution of common peaks defined by m/z and RT for the “Base” and “Quick” patterns, respectively, was displayed in Figure 25. The lists of common peaks were analyzed for the presence of homologue series by R ‘nontarget’ and if detected highlighted in the scatterplots. No homologue series were identified in the common peaks of the “Base” patterns.

Chapter 4 – Precipitation-related chemical patterns

Figure 25: Common peaks identified for the A) “Base” patterns and B) “Quick” patterns. “Quick1” and “Quick2” patterns of E2905 and E0106 were merged into one peaklist, respectively. Common peaks are displayed in retention time vs. mass-to-charge ratio (m/z) scatterplots. Homologue series detected in the “Quick” pattern are represented in colored lines.

4.3.3.1 Common peaks of “Base” patterns – baseline mixture The chemical mixture of common “Base” peaks represented peaks that were constantly present in the river regardless of intra-day or seasonal variations. It contained 705 peaks, which predominately eluted close to the dead time of the column (Figure 25A). Chromatographic separation on more hydrophilic stationary phases would be necessary to clearly separate these early eluting compounds. Investigations in a previous study linked these peaks to potentially natural compounds (Beckers et al., submitted). Moreover, some of these compounds might represent metal complexes as often no 13C peak was observed. However, further LC-MS experiments are necessary to unravel the origin of these omnipresent peaks.

Only six target compounds were identified in the baseline mixture including the pharmaceutical fluconazole and gabapentin-lactam, the industrial compound acridine and the coumarin derivate 7-amino-4-methylcoumarin (C47T2) as well as the pesticide metabolites dimethachlor oxalamic acid (OA) and metazachlor ethane sulfonic acid (ESA). Except for the latter three, these 97

Chapter 4 – Precipitation-related chemical patterns compounds represent widely used and highly consumed chemicals emitted from WWTPs. Gabapentin-lactam is the human metabolite of the anticonvulsant gabapentin. It has a higher stability against biotic degradation and is more persistent in the aquatic environment as compared to its parent compound, which is largely removed during activated sludge removal (Henning et al., 2018). While usually gabapentin belonged to “Quick” pattern, gabapentin-lactam was diluted with increasing discharge (“Base” pattern) (e.g., Table S27E). Thus, the ratio of parent compound to transformation product might be a good proxy for the contribution of untreated wastewater. This was also proposed for valsartan and its transformation product valsartan acid by Nödler et al. (2016).

Coumarin compounds are specifically emitted from WWTP I which is largely unaffected by rain events (Beckers et al., submitted, Muschket et al., 2018). C47T2 is a transformation product of 7- ethylamino-4-methylcoumarin (C47T1) which again is formed from 7-diethylamino-4- methylcoumarin (C47) during wastewater treatment (Muschket et al., 2018). During E2905, C47 and C47T1 were part of the “Quick 2” pattern potentially reflecting a temporary overflow of WWTP I due to a large catchment rain event (Figure S39). Since there was no increase in intensity observed for C47T2, it may be hypothesized that the shorter hydraulic retention times and consequently less efficient treatment or even a temporary discharge of untreated wastewater of WWTP I during the event E2905 led to a reduced formation of C47T2. However, the transformation kinetics of these compounds need to be studied more in-depth to confirm this hypothesis.

Season-independent input of diffuse sources was reflected by the pesticide metabolites dimethachlor OA and metazachlor ESA. In general, OA and ESA pesticide metabolites are commonly detected in groundwater (Kolpin et al., 2000). Thus, their presence in the baseline mixture indicated the continuous contribution of micropollutants via groundwater input during the entire study period. This contribution is likely diluted during rain events during which the input from surface runoff is dominating.

4.3.3.2 Common peaks of Quick patterns – rain-related mixture The common rain-related chemical mixture comprised 2,209 peaks including 115 homologue series mostly belonging to C2H4O groups (i.e., m/z shift of 44) and C3H6O groups (i.e., m/z shift of

58) as well as CH2 series (m/z shift of 14) (Figure 25B). The rain-related mixture reflected inputs from untreated wastewater and urban surface runoff. Indicator compounds for rain-related input were summarized in Table 5. Indicator compounds were checked to represent the course of the main “Quick” patterns and were ranked based on their intra-event intensity fold changes. Compounds with a high intra-event change are potentially good indicators for monitoring heavy 98

Chapter 4 – Precipitation-related chemical patterns rain events. However, compounds with lower intra-event fold change were also included on this list in order to provide a more comprehensive list of potential indicator compounds which should be investigated in other rivers. Furthermore, the list included indicator compounds for different sources and use classes. In some cases, related compounds (e.g., transformation products) were listed as alternative compounds for monitoring. These alternative compounds represented the same source but showed lower intra-event changes.

In total, 58 target compounds were annotated in the common rain-related mixture. The increase of the artificial sweeteners saccharin and cyclamate after the peak discharge indicated an input of untreated wastewater (Buerge et al., 2009) or insufficiently treated wastewater due to shorter hydraulic retention times. Cyclamate and saccharin are established indicators for untreated wastewater and showed intra-event intensity fold changes of up to 48 and 23, respectively. The artificial sweetener acesulfame indicated a general increased contribution of wastewater (Buerge et al., 2009). Thus, dilution by rain water was circumvented for some compounds by the wastewater contributions from others sources. Sources other than CSO are both WWTPs and may also include pipe leakages (Musolff et al., 2010), leachates from landfill, manure (Jmaiff Blackstock et al., 2019) or septic tanks (Lapointe et al., 2017). These latter sources might be not well connected to the main river during dry weather conditions.

Other wastewater-related compounds were cotinine, the insect repellent icaridin, the food constituents genistein and piperine, the pharmaceuticals n-acteyl-4-aminoantipyrine (metabolite of metamizol), metformin, mycophenolic acid, ketoprofen, valsartan, pravastatin as well as clopidogrel and compounds used in PCPs. All of these compounds are consumed in high volumes and thus are potentially good wastewater indicators. High WWTP removal efficiencies were reported e.g., for ketoprofen and valsartan (Kasprzyk-Hordern et al., 2009), cotinine (Buerge et al., 2008) and genistein (Bacaloni et al., 2005). This supported the findings by Phillips et al. (2012) that CSO and overflow of other related point sources is an important source for compounds with a high WWTP removal.

This is the first time that piperine was reported in surface waters. Due to its ubiquitous usage as the main constituent of pepper and high intensities, it seemed to be a reliable marker –similar to artificial sweeteners or cotinine- for untreated wastewater. In fact, the observed intensity increase of piperine during an event was even higher than for the artificial sweeteners (i.e., factor of 93, Table 5). However, more data should be gathered on piperine in the aquatic environment to support this finding and understand its fate.

Chapter 4 – Precipitation-related chemical patterns

Compounds used in PCPs included quaternary ammonia compounds (QACs), betaines, UV filters and triclosan. QACs (e.g., benzyldimethyldodecylammonium) are cationic surfactants largely removed during wastewater treatment by sludge adsorption and biodegradation (Clara et al., 2007). QACs were associated with the development of multi-resistant bacteria and with risks to aquatic organisms, especially algae (Zhang et al., 2015). In Germany, they were banned for use as biocides in 2016 (BAuA (2019)). Betaines were associated with inputs of untreated wastewater to the Holtemme River in a previous study (Beckers et al., submitted). Also UV filters (e.g., benzophenone-3) are not completely removed by WWTPs, but their removal rates vary a lot (Kasprzyk-Hordern et al., 2009, Phillips et al., 2012, Wu et al., 2018). Their association with the “Quick” pattern indicated an input of UV filters with insufficiently treated wastewater. Benzophenones were previously identified as endocrine disrupting compounds. A steady exposure with pulses during rain events to these compounds may cause chronic effects, especially in fish (Tsui et al., 2014). Triclosan was commonly detected in treated wastewater and CSO, whereas lower or comparable concentrations were observed during stormflows in CSO discharge (Launay et al., 2016, Phillips et al., 2012). In this study, triclosan was clearly associated with the “Quick” pattern and in fact, it was often only detected with the incoming runoff wave. Mobilization of suspended matter in sewers might have contributed to pollution load during rain events (Hajj- Mohamad et al., 2014, Launay et al., 2016).

Furthermore, several plastic additives and some industrial compounds showed a strong increase with the “Quick” patterns including bisphenol S and tri(butoxyethyl)phosphate, perfluorooctanesulfonic acid, melamine and the surfactant N,N-dimethyldodecylamine-N-oxide. Higher concentrations of the flame retardant tri(butoxyethyl)phosphate were observed in urban streams during stormflows than during dry weather, even though comparable levels were detected in the WWTP effluents and CSO (Launay et al., 2016, Phillips and Chalmers, 2009). For tri(butoxyethyl)phosphate, additional input from road runoff was proposed by Phillips et al. (2012) which explained the higher concentrations during rain events. Melamine compounds were identified as components of runoff water related to salmon coho mortality (Peter et al., 2018). This calls for a further risk assessment of the pollutant mixtures identified in this study.

The common rain-related mixture also contained benzothiazole rubber additives (i.e., 2-(4- morpholinyl)benzothiazole, 2-benzothiazolsulfonic acid, 2-(methylthio)benzothiazole and N- cyclohexyl-2-benzothiazolamine and N,N’-diemethyl-p-phenyldiamine) and urban biocides (i.e., carbendazim and cybutryne (Irgarol)). Rubber additives were identified as tire-derived markers (Kumata et al., 2002). They potentially reflected surface runoff from urban areas which is in agreement with previous studies (Launay et al., 2016, Wittmer et al., 2010). The contribution of 100

Chapter 4 – Precipitation-related chemical patterns urban areas and surfaces like facades to pollution with biocides and privately used pesticides was shown to be substantial (Wittmer et al., 2010). The increased exposure to biocides via surface runoff potentially increases risk for aquatic organisms as these compounds are highly potent. For example, carbendazim was identified as one of the main risk drivers for fish in urban storm water and is potentially harmful to humans and other aquatic organisms (Beckers et al., 2018, ECHA, 2017, Phillips and Bode, 2004). Carbendazim was often detected in stormwater in other catchments and point sources during heavy rain events (Gasperi et al., 2014, Wittmer et al., 2010, Launay et al., 2016). Still, it is registered for use, while the use of cybutryne as an antifouling biocide is prohibited in the EU since January 2017 (UBA, 2016). As this study was conducted in 2016, new monitoring data should reveal if cybutryne emissions are declining in the aquatic environment.

Further compounds reflecting surface runoff were fungicides and growth regulators (sometimes used in combination) such as myclobutanil and daminozide. Besides agricultural application, they are also used for ornamental plants and fruits, which could explain their steady input over the entire study period potentially from parks or garden centers. None of these pesticides are allowed for use in households and private gardens (BVL, 2019). Myclobutanil has been detected in an urban stream by Philips and Bode (2004) but not in a non-urban stream supporting the findings of this study.

The peaks contained in the “Quick” pattern of E2406 were not used as an exclusion criterion for the identification of the common rain-related mixtures as it represented a small rain event with almost no CSO. Still, 35 target compounds of the common rain-related mixture were also detected in the “Quick” pattern of E2406 (Table 5). Moreover, ten homologue series were observed in E2406, which were prioritized for structure elucidation. Thus, even without considerable CSO discharge, rain events can lead to substantial changes in pollution in the receiving water body.

4.3.3.3 Agricultural pesticides in precipitation-related pollution patterns In this study, several agricultural pesticides were detected in the patterns of the different events (Table S27). While they did not form a separate cluster as a result of different runoff dynamics, their number and association with “Base” and “Quick” patters changed between events. In total, the number of pesticides dropped from 107 in early June to 55 in September indicating a dominance of seasonal input versus input during rain events. For example, the broadly applied agricultural pesticides MCPA and boscalid were only observed in the “Quick 2” pattern of E2905 and the “Base” pattern of E0106. As the “Quick 2” pattern likely resulted from an overflow at WWTP I, their input was not related to direct runoff during heavy rain events. Rather considerable

101

Chapter 4 – Precipitation-related chemical patterns seasonal input of these two compounds was observed via WWTP I and related to improper practices and faulty connections in the sewer system (Beckers et al., 2018).

Prosulfocarb was the only agricultural pesticide in the common rain-related mixture. This herbicide is registered for agricultural use only (BVL, 2019). A continuous wash off during rain events throughout the entire study period might have resulted from the long mean transition time of water in the catchment, which was estimated to be approximately 20 months (Lutz et al., 2018). However, pesticide input depends on many factors, which can hardly be generalized or foreseen, e.g., topographic and hydrological connection between the agricultural area and the river as well as weather conditions prior and post application (Krein et al., 2013, Leu et al., 2004, Phillips and Bode, 2004). Thus, pesticides are not always affected by rain-related runoff and their concentration peaks might occur later in the receiving river via slower subsurface flows (Phillips and Bode, 2004), which were possibly missed during the 8 hour sampling campaign of this study. Carpenter et al. (2019) identified various concentration-discharge relationships for different pesticides and transformation products. Consequently, agricultural pesticides may not be reliable indicator compounds for the monitoring heavy rain events.

4.3.3.4 Identified ions of interests in the rain-related mixture Structure elucidation efforts focused on common peaks in the “Quick” patterns, i.e., the rain- related mixture, in order to derive more potential indicator compounds for event-based monitoring. Even though a quite comprehensive target screening was already performed, further compounds, which showed a high intra-event intensity fold change and were represented in the “Quick” pattern of each event, were revealed by NTS. Compounds identified to level 1 were added to Table 5 as indicator compounds for a common rain-related mixture. Ions of interests identified to a lower confidence level are presented in Table S28. Three individual compounds were identified to level 1, i.e., the bile acid chenodeoxycholic acid as well as lauryl betaine and tetraacetylethylenediamine. Chenodeoxycholic acid is one of the main bile acids excreted by humans and is also commercially synthesized for therapeutic purposes. So far, it was never reported in surface waters. It showed the highest intra-event intensity fold changes of all indicator compounds discussed here. Thus, it might be a valuable and ubiquitous marker for untreated wastewater independent from local consumption patterns. Tetraacetylethylenediamine is used in laundary detergents and in paper production as a bleach activator (PubChem, 2019). Lauryl betaine is another constituent of cocamidopropylbetaine. Other constituents were previously identified in the Holtemme River and were part of the target list of this study (Beckers et al., submitted). Three compounds were identified to level 4 including two fatty acids.

102

Chapter 4 – Precipitation-related chemical patterns

Furthermore, several surfactants as members of homologue series were identified by structure elucidation efforts. Surfactants commonly occur in homologue series and are usually detected as

[M+NH4]+ and [M+Na]+ adducts as in this study (Lara-Martín et al., 2011, Schymanski et al., 2015). Not for all surfactants commercial standards could be obtained and thus, they were identified to lower confidence levels (Table S28). Two surfactant series were confirmed by a reference standard (level 1), i.e., alkyl ethoxy sulfates (AES) and PEGs. AES were previously detected in raw wastewater (Alygizakis et al., 2019, Gago-Ferrero et al., 2015, Schymanski et al.,

2014b). PEGs with the sum formula C2nH4n+2On+1 were among the top-ranked compounds regarding maximum intensities; however they were not detected during the small rain event E2406 suggesting a more specific rain-related discharge. Several (different) PEG-series were detected by Peter et al. (2018) in stormwater runoff and raw wastewater (Gago-Ferrero et al., 2015, Lara-

Martín et al., 2011). Other tentative (level 3) surfactants included CH3(CH2)9(CH2CH2O)nOH, Bis-

PEG-acids COOHCH2(CH2OCH2)nCH2COOH and COOHCH2O(CH2CH2O)n-2CH2COOH as well as m-PEG-acid COOH(CH2CH2O)nCH3. These groups were characterized by diagnostic fragments (Appendix C2.5.2 and C2.5.3).

In general, surfactants clearly followed the “Quick” pattern indicating their increased introduction with untreated wastewater and surface runoff. While some of the surfactant compounds were present prior to the discharge peak, their extreme intra-event fold change in intensity (Table 5) makes them valuable markers for rain-related input (Corada-Fernández et al., 2017). In general, surfactants are generally well removed by WWTPs (Clara et al., 2007, Freeling et al., 2019, Lara- Martín et al., 2011). This is supported by the findings of a previous spatial study conducted under dry weather conditions. During dry weather conditions, the number of peaks in homologues series dropped downstream of WWTP effluents (Beckers et al., submitted). Due to their ubiquitous use, their specific sources cannot be separated. However, surfactants make up a substantial amount of peaks in a rain-related mixture (Figure 25B). Their occurrence in complex surfactant mixtures (including transformation products) poses a potential threat to aquatic organisms (Freeling et al., 2019, Peter et al., 2018).

103

Chapter 4 – Precipitation-related chemical patterns

Table 5: Indicator compounds for a common rain-related mixture. (Ww = wastewater, UWw = untreated wastewater, URun = urban runoff, SRun = surface runoff).

Compound name Exact Intra- Detected Source Use class Alternative Other mass event in E2406 compound studies fold change* Target compounds Triclosan 287.9512 NA No UWw Biocide Philips et al., 2012, Launay et al., 2016 Bisphenol S 250.0300 6-183 Yes UWw Plastic additive N,N-Dimethyldodecyl- 229.2406 3-160 Yes UWw Surfactant N,N-Dimethyldecyl- amine-N-oxide amine Piperine 285.1365 4-93 Yes UWw Food ingredient Lauryl diethanolamide 287.2460 4-63 No UWw PCP Lauramidopropyl- betaine Cyclamate 179.0616 7-48 Yes UWw Food ingredient Benzyldimethyl- 304.2999 2-28 No UWw PCP Benzyldimethyl- dodecylammonium hexadecylammonium Metformin 129.1014 4-23 Yes UWw Pharmaceutical Saccharin 182.9990 3-21 Yes UWw Industrial compound Mycophenolic acid 320.1260 NA-19 Yes UWw Pharmaceutical Cotinine 176.0950 3-14 Yes UWw Semiluxury food Beckers et al., 2018 Metoprolol acid 267.1471 2-14 No UWw Pharmaceutical Cybutryne (Irgarol) 253.1361 3-13 Yes URun Biocide Terbutryn Beckers et al., 2018 (Terbutryn) Tri(butoxyethyl)- 398.2433 2-13 Yes UWw Plastic additive Triphenylphosphate Philips and phosphate Chalmers, 2009 Perfluorooctanesulfonic 499.9375 4-12 Yes UWw Industrial compound acid Valsartan 435.2270 3-12 Yes UWw Pharmaceutical Losartan Nödler et al., 2016 Triethylcitrate 276.1209 3-11 No UWw Plastic additive N-cyclohexyl-2- 232.1034 2-10 Yes URun Rubber additive 2-Benzothiazolesulfonic Launay et benzothiazole-amine acid, al., 2016

104

Chapter 4 – Precipitation-related chemical patterns

Compound name Exact Intra- Detected Source Use class Alternative Other mass event in E2406 compound studies fold change* 2- (Methylthio)benzothiazole Icaridin 229.1678 2-10 Yes UWw PCP Carbendazim 191.0695 2-9 Yes URun Biocide Launay et al., 2016, Wittmer et al., 2010, Gasperi et al., 2014, Beckers et al., 2018 Ketoprofen 254.0943 2-9 Yes UWw Pharmaceutical Naproxen N-acetyl-4-aminoantipyrine 245.1164 2-8 Yes UWw Pharmaceutical Acesulfame 162.9939 2-7 No Ww Food ingredient Launay et al., 2016 Genistein 270.0528 3-8 Yes UWw Food ingredient König et al. 2017 Melamine 126.0654 1.5-6 Yes UWw Industrial compound N,N’-Diemethyl-p- 136.1000 2-5 Yes URun Rubber additive phenyldiamine Prosulfocarb 251.1344 2-5 No SRun Pesticide Benzophenone-3 228.0786 3-4 Yes Ww UV filter Benzophenone-4 Philips et al., 2012 m-Xylene-4- 186.0351 2-4 Yes UWw Industrial compound sulfonic-acid Myclobutanil 288.1142 2-4 Yes URun Pesticide Phillips and Bode, 2004 Quinoline N-oxide 145.0528 1.3-4 No UWw Industrial compound 1-Ethyl-3-methyl- 111.0917 2-3 Yes UWw Industrial compound imidazolium 4-(Dimethylamino)- 122.0844 2-3 Yes UWw Industrial compound pyridine Daminozide 160.0848 2-3 Yes URun Pesticide (plant growth regulator) Clopidogrel 321.0590 2-3 Yes UWw Pharmaceutical Pravastatin 424.2461 2-3 Yes UWw Pharmaceutical

105

Chapter 4 – Precipitation-related chemical patterns

Compound name Exact Intra- Detected Source Use class Alternative Other mass event in E2406 compound studies fold change*

Nontarget peaks (level 1) Chenodeoxycholic acid 392.293 NA- 2166 Yes UWw Human bile acid Lauryl betaine 271.251 3-17 No UWw PCP Tetraacetylethylenediamine 228.1110 11-731 Yes UWw Industrial compound PEG-3 150.0892 3-9 No UWw/SRun Surfactant PEG-8 370.22 12-58 No UWw/SRun Surfactant PEG-9 414.2463 12-47 No UWw/SRun Surfactant PEG-10 458.2725 14-35 No UWw/SRun Surfactant PEG-11 502.2989 14-40 No UWw/SRun Surfactant PEG-13 590.3512 13-32 No UWw/SRun Surfactant Pentaglycol ether sulfate 406.1514 NA-174 Yes UWw/SRun Surfactant Hexaglycol ether sulfate 450.1778 NA-135 Yes UWw/SRun Surfactant Heptagylcol ether sulfate 494.204 NA-101 Yes UWw/SRun Surfactant Octaglycol ether sulfate 538.2304 NA-258 Yes UWw/SRun Surfactant Nonaglycol ether sulfate 582.2555 NA-57 Yes UWw/SRun Surfactant Decaglycol ether sulfate 626.2817 NA Yes UWw/SRun Surfactant * Intensity fold change of NA indicates that intensity minimum of zero during an event

106

Chapter 4 – Precipitation-related chemical patterns

4.4 Conclusions In this study, a longitudinal cluster analysis was successfully applied to NTS data sets obtained from time-resolved sampling of heavy rain events in a small river. The analysis showed to be rough and robust enough to reduce the complexity of the pollutant mixture in the river to two or three distinct patterns. Rain-related input dramatically changed the pollution dynamics in a small stream, i.e., by dilution of the pre-event emissions (“Base” pattern) and input of rain-related emissions (“Quick” pattern). The “Quick” pattern summarized the emissions of a number of potential emission sources, whose input occurred simultaneously at the sampling point leading to a complex rain-related exposure. No separate pattern was observed for diffuse runoff, e.g., of agricultural pesticides in these heavy rain events.

While the CSO was a reliable trigger for the occurrence of heavy rain events, the change of pollution patterns was usually reflected by a peak in turbidity and an increase in discharge. Event- based monitoring in rivers has to account for several factors determining the occurrence of pollution patterns. Most importantly, the “Quick” pattern may last longer than peaks in physicochemical parameters and discharge. Especially during summer events with more direct surface runoff and intense rain events, exposure to rain-related pollutants lasted more than four hours. Moreover, the kinematic wave effect might lead to a delay between discharge and pollution peak. Ideally, triggers and trigger values should be adapted to each river system.

Despite great inter-event variations of individual compounds, a typical rain-related mixture was obtained. This list of target indicator compounds was extended by structure elucidation efforts of prioritized nontarget peaks in rain-related mixture. Most indicator compounds identified here were in agreement with dynamics observed during heavy rain events at point sources and reflect urban surface runoff (e.g., biocides and surfactants) and untreated wastewater (e.g., artificial sweeteners). Due to the overlay of several pollution sources in the river, some differences in compounds dynamics were observed compared to point sources studies. Reliable compounds for event-based monitoring include biocides and surfactants for urban surface runoff as well as piperine and chenodeoxycholic acid for untreated wastewater. Piperine and chenodeoxycholic acid are reported in surface water for the first time. These compounds showed high intra-event fold changes or were solely introduced with the “Quick” pattern. Moreover, they also reflected changes in smaller rain events.

In follow-up studies, the proposed indicator compounds should be tested for suitability as indicators for risk to aquatic organisms (Brack et al., 2018), especially since these events 107

Chapter 4 – Precipitation-related chemical patterns represent a re-occurring pulse exposure on top of constant emissions. Considerable risk is expected with regard to urban runoff such as biocides (Beckers et al., 2018) and surfactants (Freeling et al., 2019, Peter et al., 2018).

108

Chapter 4 – Precipitation-related chemical patterns

Acknowledgement The authors thank Jörg Ahlheim (UFZ) and Patrick Schütze (UFZ) for supporting the technical development of the sampling campaign. Patrick Fink (UFZ) and Florian Zander (UFZ) are acknowledged for providing data on physicochemical parameters at the sampling station. The support of the operators of the WWTP and the municipalities is greatly acknowledged. This study was supported by SOLUTIONS project funded by the European Union Seventh Framework Programme (FP7-ENV-2013-two-stage Collaborative project) under grant agreement number 603437. The QExactive Plus LC-HRMS used is part of the major infrastructure initiative CITEPro (Chemicals in the Terrestrial Environment Profiler) funded by the Helmholtz Association.

109

Chapter 4 – Precipitation-related chemical patterns

References Altenburger, R., Brack, W., Burgess, R.M., Busch, W., Escher, B.I., Focks, A., Mark Hewitt, L., Jacobsen, B.N., de Alda, M.L., Ait-Aissa, S., Backhaus, T., Ginebreda, A., Hilscherová, K., Hollender, J., Hollert, H., Neale, P.A., Schulze, T., Schymanski, E.L., Teodorovic, I., Tindall, A.J., de Aragão Umbuzeiro, G., Vrana, B., Zonja, B. and Krauss, M. (2019) Future water quality monitoring: improving the balance between exposure and toxicity assessments of real-world pollutant mixtures. Environmental Sciences Europe 31(1), 12. 10.1186/s12302-019-0193-1 Alygizakis, N.A., Gago-Ferrero, P., Hollender, J. and Thomaidis, N.S. (2019) Untargeted time-pattern analysis of LC-HRMS data to detect spills and compounds with high fluctuation in influent wastewater. Journal of Hazardous Materials 361, 19-29. https://doi.org/10.1016/j.jhazmat.2018.08.073 Bacaloni, A., Cavaliere, C., Faberi, A., Foglia, P., Samperi, R. and Laganà, A. (2005) Determination of isoflavones and coumestrol in river water and domestic wastewater sewage treatment plants. Analytica Chimica Acta 531(2), 229-237. https://doi.org/10.1016/j.aca.2004.10.037 BAuA (2019) Federal Institute for Occupational Safety and Health, eBIOMELD (accessed on 24.06.2019). https://www.biozid-meldeverordnung.de/offen/ Beckers, L.-M., Busch, W., Krauss, M., Schulze, T. and Brack, W. (2018) Characterization and risk assessment of seasonal and weather dynamics in organic pollutant mixtures from discharge of a separate sewer system. Water Research 135, 122-133. https://doi.org/10.1016/j.watres.2018.02.002 Beckers, L.M., Brack, W., Dann, J.P., Krauss, M., Müller, E. and Schulze, T. (submitted) Unraveling longitudinal pollution patterns of organic micropollutants in a small river by nontarget screening and cluster analysis. Bosch, D.D., Arnold, J.G., Allen, P.G., Lim, K.-J. and Park, Y.S. (2017) Temporal variations in baseflow for the Little River experimental watershed in South Georgia, USA. Journal of Hydrology: Regional Studies 10, 110-121. https://doi.org/10.1016/j.ejrh.2017.02.002 Brack, W., Escher, B.I., Müller, E., Schmitt-Jansen, M., Schulze, T., Slobodnik, J. and Hollert, H. (2018) Towards a holistic and solution-oriented monitoring of chemical status of European water bodies: how to support the EU strategy for a non-toxic environment? Environmental Sciences Europe 30(1), 33. 10.1186/s12302-018-0161-1 Buerge, I.J., Buser, H.-R., Kahle, M., Müller, M.D. and Poiger, T. (2009) Ubiquitous Occurrence of the Artificial Sweetener Acesulfame in the Aquatic Environment: An Ideal Chemical Marker of Domestic Wastewater in Groundwater. Environmental Science & Technology 43(12), 4381-4385. 10.1021/es900126x Buerge, I.J., Kahle, M., Buser, H.-R., Müller, M.D. and Poiger, T. (2008) Nicotine Derivatives in Wastewater and Surface Waters: Application as Chemical Markers for Domestic Wastewater. Environmental Science & Technology 42(17), 6354-6360. 10.1021/es800455q BVL (2019), Verzeichnis zugelassener Pflanzenschutzmittel (List of registered plant protection products) (accessed on 30.6.2019). https://apps2.bvl.bund.de/psm/jsp/index.jsp Carpenter, C.M.G., Wong, L.Y.J., Johnson, C.A. and Helbling, D.E. (2019) Fall Creek Monitoring Station: Highly Resolved Temporal Sampling to Prioritize the Identification of Nontarget Micropollutants in a Small Stream. Environmental Science & Technology 53(1), 77-87. 10.1021/acs.est.8b05320 Chambers, M.C., Maclean, B., Burke, R., Amodei, D., Ruderman, D.L., Neumann, S., Gatto, L., Fischer, B., Pratt, B., Egertson, J., Hoff, K., Kessner, D., Tasman, N., Shulman, N., Frewen, B., Baker, T.A., Brusniak, M.-Y., Paulse, C., Creasy, D., Flashner, L., Kani, K., Moulding, C., Seymour, S.L., Nuwaysir, L.M., Lefebvre, B., Kuhlmann, F., Roark, J., Rainer, P., Detlev, S., Hemenway, T., Huhmer, A., Langridge, J., Connolly, B., Chadick, T., Holly, K., Eckels, J., Deutsch, E.W., Moritz, R.L., Katz, J.E., Agus, D.B., MacCoss, M., Tabb, D.L. and

110

Chapter 4 – Precipitation-related chemical patterns

Mallick, P. (2012) A cross-platform toolkit for mass spectrometry and proteomics. Nature Biotechnology 30, 918. https://doi.org/10.1038/nbt.2377 Clara, M., Scharf, S., Scheffknecht, C. and Gans, O. (2007) Occurrence of selected surfactants in untreated and treated sewage. Water Research 41(19), 4339-4348. https://doi.org/10.1016/j.watres.2007.06.027 Corada-Fernández, C., Candela, L., Torres-Fuentes, N., Pintado-Herrera, M.G., Paniw, M. and González-Mazo, E. (2017) Effects of extreme rainfall events on the distribution of selected emerging contaminants in surface and groundwater: The Guadalete River basin (SW, Spain). Science of The Total Environment 605-606, 770-783. https://doi.org/10.1016/j.scitotenv.2017.06.049 Dickenson, E.R.V., Snyder, S.A., Sedlak, D.L. and Drewes, J.E. (2011) Indicator compounds for assessment of wastewater effluent contributions to flow and water quality. Water Research 45(3), 1199-1212. https://doi.org/10.1016/j.watres.2010.11.012 Dusa, A. (2018). venn: Draw Venn Diagrams. v1.7. DWD (2019), Climate Data Center (CDC) (accessed on 27.05.19). ftp://ftp- cdc.dwd.de/pub/CDC/observations_germany/climate/ ECHA (2017), European Chemicals Agency, Substance information - Carbendazim (accessed on 25.07.2017). https://echa.europa.eu/substance-information/- /substanceinfo/100.031.108?_disssubsinfo_WAR_disssubsinfoportlet_backURL=https%3A%2F %2Fecha.europa.eu%2Fde%2Fsearch-for- chemicals%3Fp_p_id%3Ddisssimplesearch_WAR_disssearchportlet%26p_p_lifecycle%3D0%2 6p_p_state%3Dnormal%26p_p_mode%3Dview%26p_p_col_id%3Dcolumn- 1%26p_p_col_count%3D1%26_disssimplesearch_WAR_disssearchportlet_sessionCriteriaId%3 DdissSimpleSearchSessionParam101401500990999267 Freeling, F., Alygizakis, N.A., von der Ohe, P.C., Slobodnik, J., Oswald, P., Aalizadeh, R., Cirka, L., Thomaidis, N.S. and Scheurer, M. (2019) Occurrence and potential environmental risk of surfactants and their transformation products discharged by wastewater treatment plants. Sci Total Environ 681, 475-487. 10.1016/j.scitotenv.2019.04.445 Gago-Ferrero, P., Schymanski, E.L., Bletsou, A.A., Aalizadeh, R., Hollender, J. and Thomaidis, N.S. (2015) Extended Suspect and Non-Target Strategies to Characterize Emerging Polar Organic Contaminants in Raw Wastewater with LC-HRMS/MS. Environmental Science & Technology 49(20), 12333-12341. 10.1021/acs.est.5b03454 Gasperi, J., Zgheib, S., Cladière, M., Rocher, V., Moilleron, R. and Chebbo, G. (2012) Priority pollutants in urban stormwater: Part 2 – Case of combined sewers. Water Research 46(20), 6693-6703. http://dx.doi.org/10.1016/j.watres.2011.09.041 Genolini, C., Alacoque, X., Sentenac, M. and Arnaud, C. (2015) kml and kml3d: R Packages to Cluster Longitudinal Data. Journal of Statistical Software 65(4), 34. 10.18637/jss.v065.i04 Hajj-Mohamad, M., Aboulfadl, K., Darwano, H., Madoux-Humery, A.S., Guérineau, H., Sauvé, S., Prévost, M. and Dorner, S. (2014) Wastewater micropollutants as tracers of sewage contamination: analysis of combined sewer overflow and stream sediments. Environmental Science: Processes & Impacts 16(10), 2442-2450. 10.1039/C4EM00314D Henning, N., Kunkel, U., Wick, A. and Ternes, T.A. (2018) Biotransformation of gabapentin in surface water matrices under different redox conditions and the occurrence of one major TP in the aquatic environment. Water Research 137, 290-300. https://doi.org/10.1016/j.watres.2018.01.027 Hollender, J., Schymanski, E.L., Singer, H.P. and Ferguson, P.L. (2017) Nontarget Screening with High Resolution Mass Spectrometry in the Environment: Ready to Go? Environmental Science & Technology 51(20), 11505-11512. 10.1021/acs.est.7b02184 Jeanneau, L., Denis, M., Pierson-Wickmann, A.C., Gruau, G., Lambert, T. and Petitjean, P. (2015) Sources of dissolved organic matter during storm and inter-storm conditions in a

111

Chapter 4 – Precipitation-related chemical patterns lowland headwater catchment: constraints from high-frequency molecular data. Biogeosciences 12(14), 4333-4343. 10.5194/bg-12-4333-2015 Jekel, M., Dott, W., Bergmann, A., Dünnbier, U., Gnirß, R., Haist-Gulde, B., Hamscher, G., Letzel, M., Licha, T., Lyko, S., Miehe, U., Sacher, F., Scheurer, M., Schmidt, C.K., Reemtsma, T. and Ruhl, A.S. (2015) Selection of organic process and source indicator substances for the anthropogenically influenced water cycle. Chemosphere 125, 155-167. https://doi.org/10.1016/j.chemosphere.2014.12.025 Jmaiff Blackstock, L.K., Wawryk, N.J.P., Jiang, P., Hrudey, S.E. and Li, X.-F. (2019) Recent applications and critical evaluation of using artificial sweeteners to assess wastewater impact. Current Opinion in Environmental Science & Health 7, 26-33. https://doi.org/10.1016/j.coesh.2018.09.002 Kasprzyk-Hordern, B., Dinsdale, R.M. and Guwy, A.J. (2009) The removal of pharmaceuticals, personal care products, endocrine disruptors and illicit drugs during wastewater treatment and its impact on the quality of receiving waters. Water Research 43(2), 363-380. https://doi.org/10.1016/j.watres.2008.10.047 Kayhanian, M., Stransky, C., Bay, S., Lau, S.L. and Stenstrom, M.K. (2008) Toxicity of urban highway runoff with respect to storm duration. Science of The Total Environment 389(2), 386-406. https://doi.org/10.1016/j.scitotenv.2007.08.052 Kolpin, D.W., Thurman, E.M. and Linhart, S.M. (2000) Finding minimal herbicide concentrations in ground water? Try looking for their degradates. Science of The Total Environment 248(2), 115-122. https://doi.org/10.1016/S0048-9697(99)00535-5 Krein, A., Keßler, S., Meyer, B., Pailler, J.Y., Guignard, C. and Hoffmann, L. (2013) Concentrations and loads of dissolved xenobiotics and hormones in two small river catchments of different land use in Luxembourg. Hydrological Processes 27(2), 284-296. 10.1002/hyp.9209 Kumata, H., Yamada, J., Masuda, K., Takada, H., Sato, Y., Sakurai, T. and Fujiwara, K. (2002) Benzothiazolamines as Tire-Derived Molecular Markers: Sorptive Behavior in Street Runoff and Application to Source Apportioning. Environmental Science & Technology 36(4), 702-708. 10.1021/es0155229 Lapointe, B.E., Herren, L.W. and Paule, A.L. (2017) Septic systems contribute to nutrient pollution and harmful algal blooms in the St. Lucie Estuary, Southeast Florida, USA. Harmful Algae 70, 1-22. https://doi.org/10.1016/j.hal.2017.09.005 Lara-Martín, P.A., González-Mazo, E. and Brownawell, B.J. (2011) Multi-residue method for the analysis of synthetic surfactants and their degradation metabolites in aquatic systems by liquid chromatography–time-of-flight-mass spectrometry. Journal of Chromatography A 1218(30), 4799-4807. https://doi.org/10.1016/j.chroma.2011.02.031 Launay, M.A., Dittmer, U. and Steinmetz, H. (2016) Organic micropollutants discharged by combined sewer overflows – Characterisation of pollutant sources and stormwater-related processes. Water Research 104, 82-92. http://dx.doi.org/10.1016/j.watres.2016.07.068 Leu, C., Singer, H., Stamm, C., Müller, S.R. and Schwarzenbach, R.P. (2004) Simultaneous Assessment of Sources, Processes, and Factors Influencing Herbicide Losses to Surface Waters in a Small Agricultural Catchment. Environmental Science & Technology 38(14), 3827-3834. 10.1021/es0499602 LHW (2019), Datenportal Gewässerkundlicher Landesdienst Sachsen-Anhalt (Database State Waterways Service) (accessed on 28.01.2019). http://gldweb.dhi-wasy.com/gld-portal/ Loos, M. (2016). nontarget: Detecting Isotope, Adduct and Homologue Relations in LC- MS Data. v1.9. Loos, M. and Singer, H. (2017) Nontargeted homologue series extraction from hyphenated high resolution mass spectrometry data. Journal of Cheminformatics 9(1), 12. 10.1186/s13321-017-0197-z Lutz, S.R., Krieg, R., Müller, C., Zink, M., Knöller, K., Samaniego, L. and Merz, R. (2018) Spatial Patterns of Water Age: Using Young Water Fractions to Improve the Characterization of

112

Chapter 4 – Precipitation-related chemical patterns

Transit Times in Contrasting Catchments. Water Resources Research 54(7), 4767-4784. 10.1029/2017wr022216 Madoux-Humery, A.-S., Dorner, S., Sauvé, S., Aboulfadl, K., Galarneau, M., Servais, P. and Prévost, M. (2013) Temporal variability of combined sewer overflow contaminants: Evaluation of wastewater micropollutants as tracers of fecal contamination. Water Research 47(13), 4370-4382. http://dx.doi.org/10.1016/j.watres.2013.04.030 Métadier, M. and Bertrand-Krajewski, J.L. (2012) The use of long-term on-line turbidity measurements for the calculation of urban stormwater pollutant concentrations, loads, pollutographs and intra-event fluxes. Water Research 46(20), 6836-6856. http://dx.doi.org/10.1016/j.watres.2011.12.030 Meyer, B., Pailler, J.-Y., Guignard, C., Hoffmann, L. and Krein, A. (2011) Concentrations of dissolved herbicides and pharmaceuticals in a small river in Luxembourg. Environmental Monitoring and Assessment 180(1), 127-146. 10.1007/s10661-010-1777-9 Montero, P. and Vilar, J.A. (2014) TSclust: An R Package for Time Series Clustering. Journal of Statistical Software 62(1), 43. 10.18637/jss.v062.i01 Müller, C., Musolff, A., Strachauer, U., Brauns, M., Tarasova, L., Merz, R. and Knöller, K. (2018) Tomography of anthropogenic nitrate contribution along a mesoscale river. Science of The Total Environment 615, 773-783. https://doi.org/10.1016/j.scitotenv.2017.09.297 Muschket, M., Di Paolo, C., Tindall, A.J., Touak, G., Phan, A., Krauss, M., Kirchner, K., Seiler, T.-B., Hollert, H. and Brack, W. (2018) Identification of Unknown Antiandrogenic Compounds in Surface Waters by Effect-Directed Analysis (EDA) Using a Parallel Fractionation Approach. Environmental Science & Technology 52(1), 288-297. 10.1021/acs.est.7b04994 Musolff, A., Leschik, S., Reinstorf, F., Strauch, G. and Schirmer, M. (2010) Micropollutant Loads in the Urban Water Cycle. Environmental Science & Technology 44(13), 4877-4883. 10.1021/es903823a Nödler, K., Tsakiri, M., Aloupi, M., Gatidou, G., Stasinakis, A.S. and Licha, T. (2016) Evaluation of polar organic micropollutants as indicators for wastewater-related coastal water quality impairment. Environmental Pollution 211, 282-290. https://doi.org/10.1016/j.envpol.2016.01.014 Ort, C., Lawrence, M.G., Rieckermann, J. and Joss, A. (2010) Sampling for Pharmaceuticals and Personal Care Products (PPCPs) and Illicit Drugs in Wastewater Systems: Are Your Conclusions Valid? A Critical Review. Environmental Science & Technology 44(16), 6024-6035. 10.1021/es100779n Passerat, J., Ouattara, N.K., Mouchel, J.-M., Vincent, R. and Servais, P. (2011) Impact of an intense combined sewer overflow event on the microbiological water quality of the Seine River. Water Research 45(2), 893-903. https://doi.org/10.1016/j.watres.2010.09.024 Peter, K.T., Tian, Z., Wu, C., Lin, P., White, S., Du, B., McIntyre, J.K., Scholz, N.L. and Kolodziej, E.P. (2018) Using High-Resolution Mass Spectrometry to Identify Organic Contaminants Linked to Urban Stormwater Mortality Syndrome in Coho Salmon. Environmental Science & Technology 52(18), 10317-10327. 10.1021/acs.est.8b03287 Phillips, P. and Chalmers, A. (2009) Wastewater Effluent, Combined Sewer Overflows, and Other Sources of Organic Compounds to Lake Champlain1. JAWRA Journal of the American Water Resources Association 45(1), 45-57. 10.1111/j.1752-1688.2008.00288.x Phillips, P.J. and Bode, R.W. (2004) Pesticides in surface water runoff in south-eastern New York State, USA: seasonal and stormflow effects on concentrations. Pest Management Science 60(6), 531-543. 10.1002/ps.879 Phillips, P.J., Chalmers, A.T., Gray, J.L., Kolpin, D.W., Foreman, W.T. and Wall, G.R. (2012) Combined Sewer Overflows: An Environmental Source of Hormones and Wastewater Micropollutants. Environmental Science & Technology 46(10), 5336-5343. 10.1021/es3001294 Pluskal, T., Castillo, S., Villar-Briones, A. and Orešič, M. (2010) MZmine 2: Modular framework for processing, visualizing, and analyzing mass spectrometry-based molecular profile data. BMC Bioinformatics 11, 395-395. 10.1186/1471-2105-11-395 113

Chapter 4 – Precipitation-related chemical patterns

PubChem (2019), Tetraacetylethylenediamine (accessed on 10.8.2019). https://pubchem.ncbi.nlm.nih.gov/compound/Tetraacetylethylenediamine Reoyo-Prats, B., Aubert, D., Sellier, A., Roig, B. and Palacios, C. (2018) Dynamics and sources of pharmaceutically active compounds in a coastal Mediterranean river during heavy rains. Environmental Science and Pollution Research 25(7), 6107-6121. 10.1007/s11356-017- 0880-7 Royal Society of Chemistry (2015) ChemSpider. http://www.chemspider.com/ Ruttkies, C., Schymanski, E.L., Wolf, S., Hollender, J. and Neumann, S. (2016) MetFrag relaunched: incorporating strategies beyond in silico fragmentation. Journal of Cheminformatics 8, 3-3. 10.1186/s13321-016-0115-9 Samanipour, S., Kaserzon, S., Vijayasarathy, S., Jiang, H., Choi, P., Reid, M.J., Mueller, J.F. and Thomas, K.V. (2019) Machine learning combined with non-targeted LC-HRMS analysis for a risk warning system of chemical hazards in drinking water: A proof of concept. Talanta 195, 426-432. https://doi.org/10.1016/j.talanta.2018.11.039 Schymanski, E.L., Jeon, J., Gulde, R., Fenner, K., Ruff, M., Singer, H.P. and Hollender, J. (2014a) Identifying Small Molecules via High Resolution Mass Spectrometry: Communicating Confidence. Environmental Science & Technology 48(4), 2097-2098. 10.1021/es5002105 Schymanski, E.L., Singer, H.P., Longrée, P., Loos, M., Ruff, M., Stravs, M.A., Ripollés Vidal, C. and Hollender, J. (2014b) Strategies to Characterize Polar Organic Contamination in Wastewater: Exploring the Capability of High Resolution Mass Spectrometry. Environmental Science & Technology 48(3), 1811-1818. 10.1021/es4044374 Schymanski, E.L., Singer, H.P., Slobodnik, J., Ipolyi, I.M., Oswald, P., Krauss, M., Schulze, T., Haglund, P., Letzel, T., Grosse, S., Thomaidis, N.S., Bletsou, A., Zwiener, C., Ibáñez, M., Portolés, T., de Boer, R., Reid, M.J., Onghena, M., Kunkel, U., Schulz, W., Guillon, A., Noyon, N., Leroy, G., Bados, P., Bogialli, S., Stipaničev, D., Rostkowski, P. and Hollender, J. (2015) Non-target screening with high-resolution mass spectrometry: critical review using a collaborative trial on water analysis. Analytical and Bioanalytical Chemistry 407(21), 6237-6255. 10.1007/s00216-015-8681-7 Tarasova, L., Basso, S., Zink, M. and Merz, R. (2018) Exploring Controls on Rainfall- Runoff Events: 1. Time Series-Based Event Separation and Temporal Dynamics of Event Runoff Response in Germany. Water Resources Research 54(10), 7711-7732. 10.1029/2018wr022587 Tsui, M.M.P., Leung, H.W., Lam, P.K.S. and Murphy, M.B. (2014) Seasonal occurrence, removal efficiencies and preliminary risk assessment of multiple classes of organic UV filters in wastewater treatment plants. Water Research 53, 58-67. https://doi.org/10.1016/j.watres.2014.01.014 UBA (2016), Umweltbundesamt, EU Commission declines to authorise antifouling ingredient (accessed on 26.8.2019). https://www.umweltbundesamt.de/en/press/pressinformation/eu-commission-declines-to- authorise-antifouling US EPA (2019) United States Environmental Protection Agency, CompTox Chemicals Dashboard. https://comptox.epa.gov/dashboard Wei, T. and Simko, V. (2017). R package "corrplot": Visualization of a Correlation Matrix. v0.84. Wittmer, I.K., Bader, H.P., Scheidegger, R., Singer, H., Lück, A., Hanke, I., Carlsson, C. and Stamm, C. (2010) Significance of urban and agricultural land use for biocide and pesticide dynamics in surface waters. Water Research 44(9), 2850-2862. http://dx.doi.org/10.1016/j.watres.2010.01.030 Wu, M.H., Li, J., Xu, G., Ma, L.D., Li, J.J., Li, J.S. and Tang, L. (2018) Pollution patterns and underlying relationships of benzophenone-type UV-filters in wastewater treatment plants and their receiving surface water. Ecotoxicol Environ Saf 152, 98-103. 10.1016/j.ecoenv.2018.01.036

114

Chapter 4 – Precipitation-related chemical patterns

Zhang, C., Cui, F., Zeng, G.-m., Jiang, M., Yang, Z.-z., Yu, Z.-g., Zhu, M.-y. and Shen, L.- q. (2015) Quaternary ammonium compounds (QACs): A review on occurrence, fate and toxicity in the environment. Science of The Total Environment 518-519, 352-362. https://doi.org/10.1016/j.scitotenv.2015.03.007

115

116

Chapter 5 – Synthesis, conclusions and future perspective

Chapter 5

Synthesis, conclusions and future research needs The main objective of this thesis was to identify drivers of chemical stress by developing and applying novel approaches which are able to account for complex and variable mixtures of organic micropollutants (Figure 3). This dissertation investigated temporal (Chapter 2), spatial (Chapter 3) and precipitation-related (Chapter 4) variability of pollutant mixtures in a small river. Three major research questions were addressed in this dissertation.

Research question 1 (Chapter 2): How are organic micropollutants reflected in temporal emission patterns and how do these patterns translate into risk patterns and risk driving compounds for aquatic organisms?

Emissions from a WWTP effluent were divided into four temporal emissions patterns reflecting constantly and seasonally emitted organic micropollutants as well as their two subgroups containing pollutants with more random variations. Emissions of organic micropollutants from a rain sewer were separated into two emission groups. Pollutants were separated into an emission group resulting from illicit sewer connections representing compounds associated with untreated wastewater as well as a group representing surface runoff pollutants. High acute or sublethal risk for BQE was mostly dominated by one risk driving emission group and one to three individual risk driving compounds.

Research question 2 (Chapter 3): Which spatial pollution patterns can be discriminated in a small river and which source-related fingerprints can be derived?

Three main spatial pollutant patterns were discriminated in the Holtemme River representing input from i) WWTPs, ii) the Bode River at the confluence and iii) diffuse and small point sources. By subclustering of the main WWTP pattern, three source-related fingerprints were determined for common wastewater compounds and specific emissions for each WWTP. Moreover, representative compounds were identified for each pattern.

Research question 3 (Chapter 4): Which precipitation-related pollution patterns result in a small river during heavy rain events and what are potential indicator compounds for event-based monitoring?

117

Chapter 5 – Synthesis, conclusions and future perspective

During heavy rain events, two main precipitation-related pollutant patterns were identified in the Holtemme River. The patterns represented i) pre-event emissions which were partially diluted and ii) pollutants increasing with rain-related discharge during the rain event. 54 potential indicator compounds for event-based monitoring could be derived by target compound annotation and NTS. The potentially most suitable indicator compounds were characterized by a high intra-event intensity increase or were solely introduced by rain-related discharge.

In the following sections, the approaches used in this dissertation are discussed (5.1) and results of the individual chapters are synthesized (5.2) in the context of current research and water monitoring as well as future research needs. The synthesis of results from Chapters 2-4, which focused on different aspects of pollutant mixture complexity and variability, allows for drawing more general conclusions for future water management and monitoring (5.3). Within section 5.3 further needs for interdisciplinary research and stakeholder communication are addressed.

5.1 Approaches for the identification of compounds for water monitoring Appropriate organic micropollutants for monitoring efforts such as under the WFD have to be selected under consideration of risk assessment and comprehensive screening (Altenburger et al., 2019, Brack et al., 2018). In this thesis, two approaches were applied: i) multitarget screening to identify temporal emission patterns which were translated into risk patterns (Chapter 2) and ii) NTS in combination with cluster analysis for a more holistic characterization of complex mixtures and identification of spatial and precipitation-related pollution patterns (Chapters 3 and 4). The first approach led to the identification of risk driving compounds, while the second approach identified source-related fingerprints, pattern-representative and indicator compounds.

5.1.1 Approach for the identification of risk driving compounds In order to assess the potential impact of exposure of aquatic organisms to variable mixtures of organic micropollutants and identify risk driving compounds, emission patterns of pollutants were linked to effect patterns. In Chapter 2, temporal and precipitation-related pollutant patterns were translated into BQE-specific risk patterns by using a TU approach. This approach allowed for identifying risk driving compounds and directly characterizing them according to their emission dynamics, which are valuable information for developing monitoring and management strategies. Results showed that, the risk was explained by one to three individual risk driving compounds in case of high acute risk, which was in agreement with other studies (Munz et al., 2017, Schäfer et al., 2011). High acute risk was typically only observed for algae and crustaceans due to seasonal 118

Chapter 5 – Synthesis, conclusions and future perspective emissions of pesticides or rather random spills of insecticides. In case of insecticide emissions, also small point sources represented a considerable entry route. For fish, acute risk was rather low. However, there is potential threat for fish by sublethal effects such as effects on growth, reproduction and behavior caused by constantly emitted compounds. This constant exposure played a potentially major role for all BQE. Results in Chapter 2 showed that out of the main risk driving compounds determined for acute and sublethal risk for algae, crustaceans and fish, only diuron and terbutryn are currently considered as priority pollutants under the WFD (EU, 2013). This mismatch was also observed by Moschet et al. (2014) and Schäfer et al. (2011). Besides the fact that the most risk driving pesticides are currently not on the WFD priority list, the list also does not reflect risk drivers of sublethal effects for fish.

Future research Risk assessment depends strongly on available effect data, which in turn depends on the performance and applicability of toxicity tests. In this dissertation, available effect data originated from single-substance tests. These tests cannot account for interactive mixture effects. Moreover, they are usually designed to study acute, short-term effects leading to a lack of chronic effect data to estimate effects of long-term exposure (Ankley et al., 2007). Thus, true chronic as well as mixture effects remain largely unknown. Also the risk assessment in this study was affected by the quality of available effect data. Still, a simple risk estimation based on TUs is already more advanced and informative than the current chemical status assessment under the WFD (Altenburger et al., 2019, Brack et al., 2018). In general, an even better link between chemical exposure and effects, i.e., between chemical and ecological status assessment, would require the investigation of long-term chronic exposure of constantly emitted pollutants as well as seasonal emissions and event-related pulses under consideration of other (non-chemical) stressors. This is a daunting task and is a major challenge for future research.

5.1.2 Approaches for the identification of pattern-representative and indicator compounds While quantitative multitarget screening methods are valuable for identifying risk driving compounds, there is still a high chance of missing ecotoxicologically relevant compounds. For comprehensive characterization of complex pollutant mixtures, NTS has to be applied (Altenburger et al., 2019). While spatial screening studies will help to understand the extent and dimensions of pollution with organic micropollutants, they will also contribute to shed light on where these pollutants are coming from and reveal source-related fingerprints (Brack et al., 2018). A novel and open-source workflow combing NTS and longitudinal cluster analysis was developed

119

Chapter 5 – Synthesis, conclusions and future perspective in Chapter 3. The workflow was applied in Chapter 3 for the identification of spatial pollution patterns and source-related fingerprints following a discrete water package along the river course of the Holtemme River. The cluster analysis efficiently reduced the individual spatial courses of 23,485 peaks to the three main patterns reflecting i) inputs from WWTPs (“WW” pattern), ii) input from the Bode River and iii) diffuse and random input. A general challenge of partitioning cluster analysis algorithms is to determine the number of clusters (Reimann et al., 2008) with the aim to achieve the lowest number of clusters possible which sufficiently explain the trends of the highest amounts of peaks possible. In this dissertation, annotated target compounds supported the decision on the appropriate number of clusters and the interpretation of patterns. Furthermore, and a quantified subset of target compounds was used for validation of the workflow. However, when dealing with large data sets of some ten thousand peaks, false cluster assignment of peaks cannot be ruled out. By further subclustering of main patterns more detailed data structures were revealed and cluster assignment of individual peaks was improved. The cluster analysis showed to be robust against peaks with low detection frequencies. While sufficient subclusters revealed small point sources based on single detects, main patterns were not influenced by those peaks. In order to derive clear source-related fingerprints, variations of source-related peaks have to be high enough to be separated from trends of other peaks and natural background. The cluster analysis would be substantially improved if peaks associated with natural background could be sorted out in a clean-up step prior to cluster analysis. Results showed that source interpretation, especially of the diffuse and random pattern, suffered from a high amount of peaks potentially reflecting natural background besides anthropogenic input from diffuse sources and small point sources. Therefore, typical fingerprints or characteristics of natural background would need to be established in future studies to improve the characterization of anthropogenic pollution in aquatic ecosystems.

In Chapter 4, the workflow developed in Chapter 3 was used for the identification of temporal pollution dynamics during heavy rain events. In each event, two or three main pollution patterns were identified representing i) dilution of pre-event peaks by increasing flow (“Base” pattern) and ii) rain-related emission via surface runoff and discharges of untreated wastewater (“Quick” pattern). While the pollution patterns among different rain events were quite similar, the mixture composition varied a lot. The identification of the variability of mixture composition required further data analysis steps. By comparing mixture components in the “Quick” patterns from different events, a common rain-related mixture and potential indicator compounds were derived suitable for monitoring such events.

120

Chapter 5 – Synthesis, conclusions and future perspective

In this dissertation, the workflow was successfully applied to study spatial and temporal patterns in a small river. However, its applicability should be tested in larger rivers with much higher dilution. First results from a study in the Danube River showed that variations of pollutants in the Danube River were too small, i.e., the impact of point sources was too low in comparisons to natural variations for patterns to be detected (Hashmi, 2019). On the other hand, spatial trends in the Rhine River could be revealed for target compounds (Ruff et al., 2015). Thus, there is substantial merit in conducting further analysis using the workflow developed in this dissertation in large rivers.

Future research

Data mining of NTS data sets Data mining becomes an essential task in analytical chemistry as new instruments provide huge amounts of data per sample requiring processing by automated workflows. The field of computational mass spectrometry may be well established in proteomics and metabolomics but has just recently found its way into environmental chemistry (Hollender et al., 2017). With the newly available computational tools and approaches, major challenges emerged that have to be faced in order to develop workflows that are robust, reproducible and applicable for routine monitoring, e.g., as shown at the Rhine monitoring station (Hollender et al., 2017). In the last couple of years, several open-source workflows for data treatment and pattern identification were developed (Alygizakis et al., 2019, Carpenter et al., 2019, Samanipour et al., 2019a). In this thesis, an open-source workflow was developed towards the identification and quick and comprehensive characterization of complex mixtures of organic micropollutants. However, the identification of unknown compounds remains a time-consuming step and requires an array of structure elucidation methods and expert knowledge (Hollender et al., 2017). The further optimization and automatization of this step should be the focus of future studies. Workflows have to be further developed to allow for automatized annotation of sum formulas and linkages to databases as well as componentization and clustering of common features. Moreover, NTS studies have to be linked with effects and risk assessment. While current real-time monitoring NTS approaches might be capable of detecting peaks with abnormal temporal variations, e.g., indicating spills, the combination with effect-based tools might lead to a more comprehensive early warning system and an understanding of cause-effect relationships. Recent studies also promoted the concept of virtual EDA, which might speed up and support the identification of toxicants (Altenburger et al., 2019, Hug et al., 2015).

Quality control and data sharing Important issues in the development of NTS tools are measures for quality control. Currently, there is a discussion about reproducibility of NTS (Hites and Jobst, 2018, Samanipour et al., 2019b). 121

Chapter 5 – Synthesis, conclusions and future perspective

Even before data evaluation by multivariate statistics, data treatment is an essential and differently approached topic by individual research groups, e.g. concerning sample preparation and blank correction (Samanipour et al., 2019b). Quality control and assurance as well as their communication need to be more streamlined and similarly reported as in target screening studies (Hites and Jobst, 2018, Samanipour et al., 2019b). A first step in this direction has been made concerning the results of NTS and structure elucidation by Schymanski et al. (2014). As in this dissertation, the defined confidence levels are broadly applied by the NTS community for the reporting of the identification level of unknown compounds. Similar, guidelines and standards would be helpful for data treatment, e.g., blank correction or replicate analysis. With a concept for quality control, data sharing should also be brought forward (Schymanski and Williams, 2017). Data and method sharing, e.g., by repositories of raw data files or sorted peaklists and developed workflows, will allow for a more efficient exploitation of information contained in NTS data sets by retrospective screening (Alygizakis et al., 2018) and more reproducible and transparent results (Samanipour et al., 2019b). Furthermore, it will path the way for large meta-studies and comprehensive assessments of chemical stress in the environment (Altenburger et al., 2019, Alygizakis et al., 2018).

5.2 Variability and dynamics in complex mixtures The variability in complex pollutant mixtures was investigated in Chapters 2-4 looking at different aspects of dynamics and sources. In the following sections 5.2.1 to 5.2.3, findings for the variability in complex pollutant mixtures were synthesized and conclusions were drawn for future research.

5.2.1 Complex mixtures from large point sources In Chapter 2, temporal emission and risk patterns in the effluent of one WWTP operating in a separate sewer system were studied. In Chapter 3, source-related fingerprints of the two WWTPs at the Holtemme River were identified. In the following, the results obtained from these two studies on WWTP emissions were linked to derive general conclusions on WWTP emissions.

WWTPs are major and well-studied point sources of freshwater pollution with organic micropollutants (Kostich et al., 2014, Loos et al., 2013). When assessing temporal emissions from WWTPs, previous studies claimed significant seasonal changes for different organic micropollutants (Castiglioni et al., 2006, Vieno et al., 2005). In several studies, however, the “intrinsic” variability within a season due to influent changes, variations in treatment efficiency and flow was not reported. In this dissertation, these issues were considered in order to detect true changes in seasonal emissions. As presented in Chapter 2, organic micropollutants emitted from 122

Chapter 5 – Synthesis, conclusions and future perspective a WWTP were clustered into two temporal emission groups representing constant and seasonally emissions as well as two groups of randomly emitted compounds. Already constantly emitted compounds showed a minimum weekly variation of 17%. Seasonally and randomly emitted compounds varied up to 135% within one week and 275% among seasons. With the identification of seasonally emitted agricultural pesticides, the influence of catchment-specific activities on WWTP effluent was highlighted. The studied WWTP in Chapter 2 operated in a separate sewer system. Theoretically, only municipal and industrial wastewater entered the WWTP, while rain- related surface runoff results in diffuse runoff or was discharged via rain sewers. The input of agricultural pesticides in considerable loads suggested improper handling and practices by users and has also been observed in other studies (Neumann et al., 2002, Wittmer et al., 2010). Constantly emitted compounds represented baseline pollution from the WWTP. These baseline pollutants can be linked to source-related fingerprints identified in Chapter 3. In Chapter 3, longitudinal cluster analysis along the Holtemme River revealed three main spatial pollutant patterns, one of which reflected emissions from the two WWTPs at the Holtemme River. By further subclustering, a wastewater pattern representing compounds equally emitted from both WWTPs was identified. 70% of the constantly emitted compounds identified in Chapter 2 were part of the common wastewater pattern (“WW1”) identified in Chapter 3. These compounds mainly belonged to the group of pharmaceuticals (e.g., carbamazepine, citalopram, diclofenac) as well as included the corrosion inhibitor benzotriazole and the biocide fipronil which were all widely detected in European WWTPs (Loos et al., 2013, Munz et al., 2017). Even though a comprehensive comparison between the patterns identified in Chapters 2 and 3 was not possible due to different screening approaches, this relationship among constantly and commonly emitted compounds strengthened the proposal of source-related fingerprints which should be further tested for suitability in large-scale studies. A mismatch possibly due to commonly but seasonally emitted compounds highlighted the importance to consider temporal and spatial variation in order to derive suitable compounds for monitoring. Moreover, specific fingerprints for each WWTP were identified in Chapter 3. These fingerprints included compounds and peaks which were more specific to one of the two WWTPs. This specific discharge likely resulted from catchment-specific activities (e.g., emissions of coumarin derivates (Muschket et al., 2018)) or variations in treatment technology and, consequently, different removal efficiencies. Thus, when testing these fingerprints identified in Chapter 3 in large-scale studies, prioritization might be given to common wastewater compounds (WW1) followed by the WWTP-specific patterns (WW2 and WW3). Moreover, the identified fingerprints should be tested for temporal variations.

123

Chapter 5 – Synthesis, conclusions and future perspective

5.2.2 Complex mixtures from small and unknown point sources In Chapter 2, emissions and related risk for BQE at a rain sewer were investigated. In Chapter 3, the influence of small point sources and the confluence with the Bode River on the overall pollution in the Holtemme River was revealed. Here, two patterns described i) diffuse and random input and ii) specific contributions from the Bode River. In the following, the results on complex pollutant mixtures from small and unknown point sources identified in Chapters 2 and 3 are synthesized to assess the impact of such sources on the overall river pollution.

In practice, sewer networks are always prone to error due to historical faulty connections, pipe leakage (Musolff et al., 2010) or improper discharges (Wittmer et al., 2010). Unintended discharge of untreated municipal wastewater was observed via a rain sewer at the Holtemme River (Chapters 2). While less and lower concentrated organic micropollutants were detected in the rain sewer pollutant mixture compared to WWTP emissions, there was a steady and continuous discharge of wastewater markers (e.g., pharmaceuticals and artificial sweeteners) as well as urban pesticides and biocides. Results of Chapter 2 showed that the emitted loads from one rain sewer were negligible compared to emissions from the associated WWTP. However, when looking at the overall pollution in the Holtemme River (Chapter 3), the emissions of small point sources such as rain sewers contributed – among peaks from natural and diffuse sources - to the formation of one main spatial pollution pattern (i.e., the DRI pattern) which summarized 23% of the detected peaks. Hence, these small point sources have a considerable influence on the overall pollution pattern in the receiving water body. Moreover, by longitudinal cluster analysis, compounds that were specific to emissions from these small point sources were revealed including artificial sweeteners and surfactants used in PCPs (Chapter 3). Random discharges of micropollutants resulted in several single detects in Chapters 2 and 3. For instance, the random and illicit input of the insecticide dimethoate led to an acute risk for crustaceans in the vicinity of the rain sewer and contributed to the overall increased risk for crustaceans during this month in the downstream river (Chapter 2).

With the application of longitudinal cluster analysis, a specific and high contribution of the Bode River to the pollution of the Holtemme River at the confluence of both rivers was identified (Chapter 3). Bode-specific pollutants included industrial compounds as well as surfactants. To this point, the source of these pollutants in the Bode River remains unknown. However, an industrial point source with insufficient wastewater treatment or emission of extraordinary high loads is suspected as surfactants are generally well removed in conventional WWTPs (Freeling et al., 2019). The complex pollutant mixtures identified in the DRI and BR pattern were

124

Chapter 5 – Synthesis, conclusions and future perspective substantially different from the wastewater patterns and would have been missed by target screening alone. The analysis performed in Chapter 3 proved its applicability to separate different source-related patterns from complex mixtures in a small river and to identify pattern- representative compounds.

5.2.3 Complex mixtures from rain-related discharge In Chapter 2, risks resulting from rain-related surface runoff via a rain sewer were investigated. In Chapter 4, pollution patterns during heavy rain events leading to CSO were unravelled in the Holtemme River. In the following, rain-related pollutant mixtures identified from a rain sewer and in the receiving river after CSO were discussed to derive general conclusions on pollution dynamics and mixture composition of rain-related discharge.

Six heavy rain events associated with CSO were investigated in Chapter 4. Heavy rain events led to an exchange of pollution patterns in the river by i) partial dilution of peaks in the “Base” pattern and ii) an increase of peaks in the “Quick” pattern. The pollutant mixture in the individual rain events varied strongly in number and composition. Between 216 and 336 target compounds were detected in the precipitation-related pollutant patterns. These results highlighted findings of Chapter 2 and emphasized the importance of considering temporal variations in complex pollutant mixtures and seasonal emissions, above all, of agricultural pesticides. In fact, emissions of agricultural pesticides seemed to be more influenced by seasonal application than by rain-related runoff. In general, hydrological flow conditions and precipitation prior and during a rain event affected the pollution patterns in the river. Nevertheless, common peaks for all “Quick” patterns were identified. These peaks included wastewater compounds that are easily degradable in WWTPs and urban surface runoff. By extended target screening and NTS, 54 indicator compounds for monitoring emissions during heavy rain events were determined. Two indicator compounds for untreated wastewater were identified for the first time in surface water. These compounds included the pepper constituent piperine and the human bile acid chenodeoxycholic acid. Both showed an intra-event intensity fold changes up 93% and 2166%, respectively. Chenodeoxycholic acid is a potentially very robust indicator for wastewater emissions as it is independent from variable consumption patterns. As also shown in Chapter 2, urban emissions were generally less affected by seasons but more by weather conditions. Urban emissions clearly shaped pollution and risk patterns resulting in peaks of urban biocides and privately used pesticides as well as surfactants (Chapter 2 and 4). Biocides and pesticides used in urban areas dominated the calculated risk for aquatic organisms emitted from the rain sewer (Chapter 2). But also surfactants originating from road runoff were recently linked to acute lethal effects for fish

125

Chapter 5 – Synthesis, conclusions and future perspective

(Peter et al., 2018). These findings revealed neglects in recent water management strategies as the considerable contribution of urban emissions calls for management and mitigation efforts to reduce emissions of highly potent micropollutants from urban areas.

The integral discussion of the results from Chapter 2, 3 and 4 underlined the presence of ubiquitous micropollutants as well as specific compounds in temporal, spatial or precipitation- related mixtures. In general, temporal variation had a major influence on in the overall pollution but also rain-related runoff and river flow conditions shaped chemical stress in the receiving water body. Studies accounting for all these variations would result in an enormous amount of samples. Also the separate studies in this thesis either focused on high temporal resolution at a few sampling spots (Chapters 2 and 4) or at a high spatial resolution at many sampling spots (Chapter 3). In future studies, a stronger link with modelling approaches might allow for assessing large scale river pollution with organic micropollutants and filling data gaps by using extrapolations (Brack et al., 2017, Lindim et al., 2016).

5.3 Future water management and monitoring The results of this dissertation clearly showed that drivers of chemical stress such as risk driving compounds, source-related fingerprints and indicator compounds are linked to discharge of treated and untreated wastewater, as well as unintended or illicit discharge of highly potent urban contaminants. Thus, the improvement of treatment efficiencies and the reclamation of safe, treated wastewater is important to manage the water quality and water balance, mitigate chemical stress and ultimately reduce threats to human and environment health (Eggen et al., 2014, WWAP, 2017).

5.3.2 Future monitoring Based on the findings of this dissertation, general trends resulting in possible monitoring strategies were deduced. Potential compounds for monitoring were derived from the identification of i) main risk driving compounds (Chapter 2), ii) pattern-representative compounds (Chapter 3) and iii) indicator compounds for heavy rain events (Chapter 4).

Water monitoring schemes should be designed and implemented that take into account low flow conditions, risks of constantly emitted compounds as well as heavy rain events and associated urban runoffs and increased contributions of untreated wastewater. Sampling devices could be connected to gauging stations that record discharge and water levels in real-time. WWTP effluents as well as confluences with other major streams were clear hotspots and should be in the focus

126

Chapter 5 – Synthesis, conclusions and future perspective for monitoring. More advanced monitoring would involve effect-based tools. Effect-based tools are able to monitor effects of complex mixtures and avoid missing important toxicants or pre-selecting compounds for monitoring (Brack et al., 2017). For effect-based methods, appropriate trigger values have to be defined indicating a threshold for acceptable risk in monitored water bodies (Escher et al., 2018). These tools would also allow for monitoring random spills of hazardous compounds.

In this dissertation, seasonal emissions of agricultural pesticides and major emissions of those compounds from point sources were observed calling for mitigation efforts at these point sources. In contrast, inputs of agricultural pesticides via diffuse sources and during heavy rain events were not detected within the studies of this thesis. This could be due to the study area in which point sources seemed to have larger impact than diffuse sources. Currently, a German-wide project (“Kleingewässer Monitoring”) conducts a pesticide monitoring in small streams focusing on smaller diffuse input during rain events in the main application period of those compounds (UFZ, 2018a). Results will show if pesticides are considerably introduced into surface water during rain events and how they will affect aquatic organisms. Other challenges of future water monitoring are addressed in European research projects such SOLUTIONS (UFZ, 2018b), MARS (University of Duisburg-Essen, 2018) and networks, e.g., NORMAN (NORMAN, 2019). Currently, the results from the SOLUTIONS project are published in policy briefs and research articles and will shape future water monitoring including the WFD (Brack et al., 2019).

Interdisciplinary research for future monitoring campaigns Solving real-world environmental problems requires interdisciplinary approaches that transcend conventional academic boundaries (Fox and Rohlich, 1968, Nature, 2015). By looking through an interdisciplinary lens, multiple perspectives from different water-related research fields are integrated in order to set the scope of the problem, develop a project design, select tools, conduct sampling and interpret and communicate the results and outcomes (Brown et al., 2015). However, the task of combining all these disciplines from the start of a project and understanding their importance in the investigated issue is not easy and straightforward.

In future monitoring campaigns, sampling could be improved by a closer link to hydrology, e.g., to account for flow variations (Ort et al., 2010). This requires more equipment but will benefit the collection of flow-representative composite samples. In this dissertation, the flow variations were considered by calculating loads (Chapter 2), sampling according to flow velocities (Chapter 3) and comparing pollutant patterns to discharge patterns (Chapter 4). Evaluation of hydrological data strongly supported pattern interpretation during heavy rain events (Chapter 4), while

127

Chapter 5 – Synthesis, conclusions and future perspective chemical analysis and toxicology were closely linked via risk assessment methods (Chapter 2). More effort has to be made to effectively combine chemistry and (community) ecology (Altenburger et al., 2019) and thus derive cause-effect relationships. Due to different points of view, interdisciplinary experiments, field studies and assessments are quite challenging (Brown et al., 2015). Moreover, long-term ecological studies in the field are very demanding and complex (Kuebbing et al., 2018). Current approaches which are trying to bridge the gap between chemical exposure and observed effects in the aquatic environment, like the SPEAR (SPEcies At Risk) index (Liess and Ohe, 2005) or ecology-directed analysis, need to be further developed and applied in more studies to understand complex stressor relationships in aquatic ecosystems and design even more effective and comprehensive monitoring strategies (Altenburger et al., 2019).

5.3.1 Future management Treated wastewater contains an unknown number organic micropollutants and TPs (Schollée et al., 2015). As shown in Chapters 2 and 3, WWTPs are hotspots of pollution and risks to aquatic organisms. Upgrading WWTPs is one of the hot topics in research and water management (Eggen et al., 2014). First results from a Swiss case study look promising as risk was reduced for aquatic organisms downstream of a WWTP effluent (Munz et al., 2018). However different technologies have to be carefully evaluated, especially for the production of new TPs (Bourgin et al., 2018). Urban pollution via small point sources is another point source that has to be managed. Here, especially biocides as well as urban pesticides call for management. Urban (unintentional) emissions of biocides clearly shaped risk and pollution patterns in Chapters 2, 3 and 4. Moreover surfactants, which potentially cause adverse effects in the environment (Freeling et al., 2019, Peter et al., 2018), were identified as compounds resulting from emissions of untreated wastewater and surface runoff. Some of these emissions can be avoided or at least reduced by further water treatment steps and solutions have to be developed for all emission sources (Eggen et al., 2014).

Managing pollution in the Holtemme River – Stakeholder communication The Holtemme River is a prime example of a multiple stressed freshwater body (Wollschläger et al., 2016). As shown in Chapters 2, 3 and 4, aquatic organisms were constantly exposed to a complex mixture of organic micropollutants due to the low dilution in the Holtemme River and major contributions from two WWTPs. In low flow conditions or during heavy rain events leading to CSO, the stream resembled more a pipe for wastewater than a river. Interactions with stakeholders and sharing information with the public, raises awareness and increases understanding of pollution with organic micropollutants. Within the frame of this thesis, a first

128

Chapter 5 – Synthesis, conclusions and future perspective stakeholder workshop was conducted at the Holtemme River, on January 18th 2017, involving researchers from different research disciplines, water managers and environmental authorities, WWTP operators and representatives of local farmers. While no detailed management options were discussed during the workshop, it was used as an opportunity to build a network and inform stakeholders about first research results. The workshop also provided an opportunity for stakeholders to share their expert knowledge on different systems and local conditions and therefore contribute to an iterative evaluation of scientific findings. Moreover, the workshop increased the visibility and awareness of water research and the problems it tackles for the local stakeholders. Especially, the issue of seasonal introduction of agricultural pesticides particularly due to improper cleaning practices of spraying equipment was discussed with representatives from the local farmers and water managers. Stakeholder engagement and workshops shall be continued beyond this dissertation to communicate results and raise awareness on issues with complex pollution of organic micropollutants.

129

Chapter 5 – Synthesis, conclusions and future perspective

References Altenburger, R., Brack, W., Burgess, R.M., Busch, W., Escher, B.I., Focks, A., Mark Hewitt, L., Jacobsen, B.N., de Alda, M.L., Ait-Aissa, S., Backhaus, T., Ginebreda, A., Hilscherová, K., Hollender, J., Hollert, H., Neale, P.A., Schulze, T., Schymanski, E.L., Teodorovic, I., Tindall, A.J., de Aragão Umbuzeiro, G., Vrana, B., Zonja, B. and Krauss, M. (2019) Future water quality monitoring: improving the balance between exposure and toxicity assessments of real-world pollutant mixtures. Environmental Sciences Europe 31(1), 12. 10.1186/s12302-019-0193-1 Alygizakis, N.A., Gago-Ferrero, P., Hollender, J. and Thomaidis, N.S. (2019) Untargeted time-pattern analysis of LC-HRMS data to detect spills and compounds with high fluctuation in influent wastewater. Journal of Hazardous Materials 361, 19-29. https://doi.org/10.1016/j.jhazmat.2018.08.073 Alygizakis, N.A., Samanipour, S., Hollender, J., Ibáñez, M., Kaserzon, S., Kokkali, V., van Leerdam, J.A., Mueller, J.F., Pijnappels, M., Reid, M.J., Schymanski, E.L., Slobodnik, J., Thomaidis, N.S. and Thomas, K.V. (2018) Exploring the Potential of a Global Emerging Contaminant Early Warning Network through the Use of Retrospective Suspect Screening with High-Resolution Mass Spectrometry. Environmental Science & Technology 52(9), 5135-5144. 10.1021/acs.est.8b00365 Ankley, G.T., Brooks, B.W., Huggett, D.B., Sumpter and John, P. (2007) Repeating History: Pharmaceuticals in the Environment. Environmental Science & Technology 41(24), 8211-8217. 10.1021/es072658j Bourgin, M., Beck, B., Boehler, M., Borowska, E., Fleiner, J., Salhi, E., Teichler, R., von Gunten, U., Siegrist, H. and McArdell, C.S. (2018) Evaluation of a full-scale wastewater treatment plant upgraded with ozonation and biological post-treatments: Abatement of micropollutants, formation of transformation products and oxidation by-products. Water Research 129, 486-498. https://doi.org/10.1016/j.watres.2017.10.036 Brack, W., Ait-Aissa, S., Altenburger, R., Cousins, I., Dulio, V., Escher, B., Focks, A., Ginebreda, A., Hering, D., Hilscherová, K., Hollender, J., Hollert, H., Kortenkamp, A., de Alda, M.L., Posthuma, L., Schymanski, E., Segner, H. and Slobodnik, J. (2019) Let us empower the WFD to prevent risks of chemical pollution in European rivers and lakes. Environmental Sciences Europe 31(1), 47. 10.1186/s12302-019-0228-7 Brack, W., Dulio, V., Ågerstrand, M., Allan, I., Altenburger, R., Brinkmann, M., Bunke, D., Burgess, R.M., Cousins, I., Escher, B.I., Hernández, F.J., Hewitt, L.M., Hilscherová, K., Hollender, J., Hollert, H., Kase, R., Klauer, B., Lindim, C., Herráez, D.L., Miège, C., Munthe, J., O'Toole, S., Posthuma, L., Rüdel, H., Schäfer, R.B., Sengl, M., Smedes, F., van de Meent, D., van den Brink, P.J., van Gils, J., van Wezel, A.P., Vethaak, A.D., Vermeirssen, E., von der Ohe, P.C. and Vrana, B. (2017) Towards the review of the European Union Water Framework Directive: Recommendations for more efficient assessment and management of chemical contamination in European surface water resources. Science of The Total Environment 576(Supplement C), 720-737. https://doi.org/10.1016/j.scitotenv.2016.10.104 Brack, W., Escher, B.I., Müller, E., Schmitt-Jansen, M., Schulze, T., Slobodnik, J. and Hollert, H. (2018) Towards a holistic and solution-oriented monitoring of chemical status of European water bodies: how to support the EU strategy for a non-toxic environment? Environmental Sciences Europe 30(1), 33. 10.1186/s12302-018-0161-1 Brown, R.R., Deletic, A. and Wong, T.H.F. (2015) Interdisciplinarity: How to catalyse collaboration. Nature 525(315-317). 10.1038/525315a Carpenter, C.M.G., Wong, L.Y.J., Johnson, C.A. and Helbling, D.E. (2019) Fall Creek Monitoring Station: Highly Resolved Temporal Sampling to Prioritize the Identification of Nontarget Micropollutants in a Small Stream. Environmental Science & Technology 53(1), 77-87. 10.1021/acs.est.8b05320

130

Chapter 5 – Synthesis, conclusions and future perspective

Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D. and Zuccato, E. (2006) Removal of Pharmaceuticals in Sewage Treatment Plants in Italy. Environmental Science & Technology 40(1), 357-363. 10.1021/es050991m Eggen, R.I.L., Hollender, J., Joss, A., Schärer, M. and Stamm, C. (2014) Reducing the Discharge of Micropollutants in the Aquatic Environment: The Benefits of Upgrading Wastewater Treatment Plants. Environmental Science & Technology 48(14), 7683-7689. 10.1021/es500907n Escher, B.I., Aїt-Aїssa, S., Behnisch, P.A., Brack, W., Brion, F., Brouwer, A., Buchinger, S., Crawford, S.E., Du Pasquier, D., Hamers, T., Hettwer, K., Hilscherová, K., Hollert, H., Kase, R., Kienle, C., Tindall, A.J., Tuerk, J., van der Oost, R., Vermeirssen, E. and Neale, P.A. (2018) Effect-based trigger values for in vitro and in vivo bioassays performed on surface water extracts supporting the environmental quality standards (EQS) of the European Water Framework Directive. Science of The Total Environment 628-629, 748-765. https://doi.org/10.1016/j.scitotenv.2018.01.340 EU (2013), European Union, Directive 2013/39/EU of the European Parliament and the Council of 12. August 2013 amending Directives 2000/60/EC and 2008/105/EC as regards priority substances in the field of water policy. Off J Eur Union L 226/1-L226/17. Fox, I.K. and Rohlich, G.A. (1968) Interdisciplinary Research in Environmental Sciences. Journal (Water Pollution Control Federation) 40(3), 395-398. Freeling, F., Alygizakis, N.A., von der Ohe, P.C., Slobodnik, J., Oswald, P., Aalizadeh, R., Cirka, L., Thomaidis, N.S. and Scheurer, M. (2019) Occurrence and potential environmental risk of surfactants and their transformation products discharged by wastewater treatment plants. Sci Total Environ 681, 475-487. 10.1016/j.scitotenv.2019.04.445 Hashmi, M.A.K. (2019) Chemical and bioanalytical screenings for the unravelling of micropollutants from river Danube, RWTH Aachen University, Aachen. Hites, R.A. and Jobst, K.J. (2018) Is Nontargeted Screening Reproducible? Environmental Science & Technology 52(21), 11975-11976. 10.1021/acs.est.8b05671 Hollender, J., Schymanski, E.L., Singer, H.P. and Ferguson, P.L. (2017) Nontarget Screening with High Resolution Mass Spectrometry in the Environment: Ready to Go? Environmental Science & Technology 51(20), 11505-11512. 10.1021/acs.est.7b02184 Hug, C., Sievers, M., Ottermanns, R., Hollert, H., Brack, W. and Krauss, M. (2015) Linking mutagenic activity to micropollutant concentrations in wastewater samples by partial least square regression and subsequent identification of variables. Chemosphere 138, 176-182. https://doi.org/10.1016/j.chemosphere.2015.05.072 Kostich, M.S., Batt, A.L. and Lazorchak, J.M. (2014) Concentrations of prioritized pharmaceuticals in effluents from 50 large wastewater treatment plants in the US and implications for risk estimation. Environmental Pollution 184, 354-359. http://doi.org/10.1016/j.envpol.2013.09.013 Kuebbing, S.E., Reimer, A.P., Rosenthal, S.A., Feinberg, G., Leiserowitz, A., Lau, J.A. and Bradford, M.A. (2018) Long-term research in ecology and evolution: a survey of challenges and opportunities. Ecological Monographs 88(2), 245-258. 10.1002/ecm.1289 Liess, M. and Ohe, P.C.V.D. (2005) Analyzing effects of pesticides on invertebrate communities in streams. Environmental Toxicology and Chemistry 24(4), 954-965. 10.1897/03- 652.1 Lindim, C., van Gils, J. and Cousins, I.T. (2016) A large-scale model for simulating the fate & transport of organic contaminants in river basins. Chemosphere 144, 803-810. https://doi.org/10.1016/j.chemosphere.2015.09.051 Loos, R., Carvalho, R., António, D.C., Comero, S., Locoro, G., Tavazzi, S., Paracchini, B., Ghiani, M., Lettieri, T., Blaha, L., Jarosova, B., Voorspoels, S., Servaes, K., Haglund, P., Fick, J., Lindberg, R.H., Schwesig, D. and Gawlik, B.M. (2013) EU-wide monitoring survey on emerging polar organic contaminants in wastewater treatment plant effluents. Water Research 47(17), 6475-6487. http://dx.doi.org/10.1016/j.watres.2013.08.024

131

Chapter 5 – Synthesis, conclusions and future perspective

Moschet, C., Wittmer, I., Simovic, J., Junghans, M., Piazzoli, A., Singer, H., Stamm, C., Leu, C. and Hollender, J. (2014) How a Complete Pesticide Screening Changes the Assessment of Surface Water Quality. Environmental Science & Technology 48(10), 5423-5432. 10.1021/es500371t Munz, N.A., Burdon, F.J., de Zwart, D., Junghans, M., Melo, L., Reyes, M., Schönenberger, U., Singer, H.P., Spycher, B., Hollender, J. and Stamm, C. (2017) Pesticides drive risk of micropollutants in wastewater-impacted streams during low flow conditions. Water Research 110, 366-377. http://dx.doi.org/10.1016/j.watres.2016.11.001 Munz, N.A., Fu, Q., Stamm, C. and Hollender, J. (2018) Internal Concentrations in Gammarids Reveal Increased Risk of Organic Micropollutants in Wastewater-Impacted Streams. Environmental Science & Technology 52(18), 10347-10358. 10.1021/acs.est.8b03632 Muschket, M., Di Paolo, C., Tindall, A.J., Touak, G., Phan, A., Krauss, M., Kirchner, K., Seiler, T.-B., Hollert, H. and Brack, W. (2018) Identification of Unknown Antiandrogenic Compounds in Surface Waters by Effect-Directed Analysis (EDA) Using a Parallel Fractionation Approach. Environmental Science & Technology 52(1), 288-297. 10.1021/acs.est.7b04994 Musolff, A., Leschik, S., Reinstorf, F., Strauch, G. and Schirmer, M. (2010) Micropollutant Loads in the Urban Water Cycle. Environmental Science & Technology 44(13), 4877-4883. 10.1021/es903823a Nature (2015) Why interdisciplinary research matters. Nature 525, 305. 10.1038/525305a Neumann, M., Schulz, R., Schafer, K., Muller, W., Mannheller, W. and Liess, M. (2002) The significance of entry routes as point and non-point sources of pesticides in small streams. Water Research 36(4), 835-842. 10.1016/s0043-1354(01)00310-4 NORMAN (2019), Network of reference laboratories, research centres and related organisations for monitoring of emerging environmental substances (accessed on 31.7.2019). https://www.norman-network.net/ Ort, C., Lawrence, M.G., Rieckermann, J. and Joss, A. (2010) Sampling for Pharmaceuticals and Personal Care Products (PPCPs) and Illicit Drugs in Wastewater Systems: Are Your Conclusions Valid? A Critical Review. Environmental Science & Technology 44(16), 6024-6035. 10.1021/es100779n Peter, K.T., Tian, Z., Wu, C., Lin, P., White, S., Du, B., McIntyre, J.K., Scholz, N.L. and Kolodziej, E.P. (2018) Using High-Resolution Mass Spectrometry to Identify Organic Contaminants Linked to Urban Stormwater Mortality Syndrome in Coho Salmon. Environmental Science & Technology 52(18), 10317-10327. 10.1021/acs.est.8b03287 Reimann, C., Filzmoser, P., Garret, R.G. and Dutter, R. (2008) Statistical Data Analysis Explained: Applied Environmental Statistics with R, pp. 233-247, John Wiley&Sons Ltd., Chichester, UK. Ruff, M., Mueller, M.S., Loos, M. and Singer, H.P. (2015) Quantitative target and systematic non-target analysis of polar organic micro-pollutants along the river Rhine using high- resolution mass-spectrometry – Identification of unknown sources and compounds. Water Research 87(Supplement C), 145-154. https://doi.org/10.1016/j.watres.2015.09.017 Samanipour, S., Kaserzon, S., Vijayasarathy, S., Jiang, H., Choi, P., Reid, M.J., Mueller, J.F. and Thomas, K.V. (2019a) Machine learning combined with non-targeted LC-HRMS analysis for a risk warning system of chemical hazards in drinking water: A proof of concept. Talanta 195, 426-432. https://doi.org/10.1016/j.talanta.2018.11.039 Samanipour, S., Martin, J.W., Lamoree, M.H., Reid, M.J. and Thomas, K.V. (2019b) Letter to the Editor: Optimism for Nontarget Analysis in Environmental Chemistry. Environmental Science & Technology 53(10), 5529-5530. 10.1021/acs.est.9b01476 Schäfer, R.B., von der Ohe, P.C., Kühne, R., Schüürmann, G. and Liess, M. (2011) Occurrence and Toxicity of 331 Organic Pollutants in Large Rivers of North Germany over a Decade (1994 to 2004). Environmental Science & Technology 45(14), 6167-6174. 10.1021/es2013006

132

Chapter 5 – Synthesis, conclusions and future perspective

Schollée, J.E., Schymanski, E.L., Avak, S.E., Loos, M. and Hollender, J. (2015) Prioritizing Unknown Transformation Products from Biologically-Treated Wastewater Using High- Resolution Mass Spectrometry, Multivariate Statistics, and Metabolic Logic. Analytical Chemistry 87(24), 12121-12129. 10.1021/acs.analchem.5b02905 Schymanski, E.L., Jeon, J., Gulde, R., Fenner, K., Ruff, M., Singer, H.P. and Hollender, J. (2014) Identifying Small Molecules via High Resolution Mass Spectrometry: Communicating Confidence. Environmental Science & Technology 48(4), 2097-2098. 10.1021/es5002105 Schymanski, E.L. and Williams, A.J. (2017) Open Science for Identifying “Known Unknown” Chemicals. Environmental Science & Technology 51(10), 5357-5359. 10.1021/acs.est.7b01908 UFZ (2018a), Helmholtz Centre for Environmental Research, Nationwide monitoring of small streams (KgM) (accessed on 31.7.2019). https://www.ufz.de/kgm/index.php?en=44480 UFZ (2018b), Helmholtz Centre for Environmental Research, SOLUTIONS for present and future emerging pollutants in land and water resources management (accessed on 31.7.2019). https://www.solutions-project.eu/ University of Duisburg-Essen (2018), MARS - Managing Aquatic ecosystems and water resources under multiple stress (accessed on 31.7.2019). http://www.mars-project.eu/ Vieno, N.M., Tuhkanen, T. and Kronberg, L. (2005) Seasonal Variation in the Occurrence of Pharmaceuticals in Effluents from a Sewage Treatment Plant and in the Recipient Water. Environmental Science & Technology 39(21), 8220-8226. 10.1021/es051124k Wittmer, I.K., Bader, H.P., Scheidegger, R., Singer, H., Lück, A., Hanke, I., Carlsson, C. and Stamm, C. (2010) Significance of urban and agricultural land use for biocide and pesticide dynamics in surface waters. Water Research 44(9), 2850-2862. http://dx.doi.org/10.1016/j.watres.2010.01.030 Wollschläger, U., Attinger, S., Borchardt, D., Brauns, M., Cuntz, M., Dietrich, P., Fleckenstein, J.H., Friese, K., Friesen, J., Harpke, A., Hildebrandt, A., Jäckel, G., Kamjunke, N., Knöller, K., Kögler, S., Kolditz, O., Krieg, R., Kumar, R., Lausch, A., Liess, M., Marx, A., Merz, R., Mueller, C., Musolff, A., Norf, H., Oswald, S.E., Rebmann, C., Reinstorf, F., Rode, M., Rink, K., Rinke, K., Samaniego, L., Vieweg, M., Vogel, H.-J., Weitere, M., Werban, U., Zink, M. and Zacharias, S. (2016) The Bode hydrological observatory: a platform for integrated, interdisciplinary hydro-ecological research within the TERENO Harz/Central German Lowland Observatory. Environmental Earth Sciences 76(1), 29. 10.1007/s12665-016-6327-5 WWAP (2017), UNESCO World Water Assessment Programme, The United Nations world water development report, 2017: Wastewater: the untapped resource. 978-92-3-100201-4. Paris, France.

133

134 Appendix A – Temporal chemical and risk patterns

Appendix A - Characterization and risk assessment of seasonal and weather dynamics in organic pollutant mixtures from discharge of a separate sewer system

A1 Discharge data of the WWTP and rain sewer

Table S1: Discharge and temperature data from the WWTP on the respective sampling days

Date Discharge volume Temperature Temperature Temperature [m³] [°C] - [°C] - [°C] - Influent Aeration tank Effluent 19.05.2015 10751 14.4 15.1 14.7 20.05.2015 10789 14.6 15.4 14.9 21.05.2015 10753 14.7 15.6 14.7 22.05.2015 10458 14.5 15.8 15.2 23.05.2015 9888 14.1 16.4 15.2 24.05.2015 8936 13.5 16.5 14.9 25.05.2015 9048 13.3 16.4 15 27.07.2015 10374 14.1 16 15.1 28.07.2015 9945 17.4 19.4 18.5 29.07.2015 10004 17.7 18.9 18.1 30.07.2015 10202 17.8 18.3 17.5 31.07.2015 10267 18.1 18.5 16.3 01.08.2015 10060 18.2 18.5 17.4 02.08.2015 8731 17.3 19.1 17.6 03.08.2015 8185 16.7 19.4 18 05.10.2015 9362 17.4 19.4 18.7 06.10.2015 9688 16.8 17.2 16.2 07.10.2015 10042 17.2 17.7 16.7 08.10.2015 12298 17.3 17.9 17.9 09.10.2015 10341 17.6 18.1 17.1 10.10.2015 11788 17.4 18.3 17.9 11.10.2015 9580 16.6 18.1 16.8 12.10.2015 8735 15.7 17.8 15.5 01.02.2016 11913 16 16.4 14.9 02.02.2016 14065 11 11.1 11 03.02.2016 12866 11.8 11.3 10.6 04.02.2016 12423 11.8 11.2 10.2 05.02.2016 12031 11.7 11.2 10.9 06.02.2016 11133 11.6 11.3 10.2 07.02.2016 10843 11 11.5 10.5 08.02.2016 14422 10.9 11.6 10.2

Appendix A – Temporal chemical and risk patterns

Table S2: Discharge data from the rain sewer in dry and rain weather during the sampling period

Month Discharge Discharge Rain sample Monthly volume [m³] - volume [m³] – available (Y precipitation Dry weather Rain weather = yes; N = Wernigerode [mm] no) April 2015 4473 1869 Y 33.1 May 2015* 3161 1456 N 15.3 June 2015* 2968 2891 N 27.9 July 2015 3714 10577 Y 78.8 August 2015 2795 15146 Y 79.1 September 2015 2838 1659 Y 34.6 October 2015 4131 4423 N 48.0 November 2015 3904 3563 N 55.9 December 2015 1629 1873 N 12.7 January 2016 5769 4378 Y 49.1 February 2016 5214 6115 N 53.3 March 2016 3268 785 N 19.5 April 2016 3015 2884 N No information May 2016 3157 11128 Y No information *No chemical analysis performed for May and June 2015 due to transport damage of samples. Precipitation data available at: https://www.wetterkontor.de/de/wetter/deutschland/monatswerte- station.asp (accessed 24.01.2017)

Appendix A – Temporal chemical and risk patterns

A2 Methods

A2.1 List of target compounds analyzed on QTrap 6500

Table S3: List of target compounds analyzed in WWTP and rain sewer samples. (DP = declustering potential; CE = collision energy; CXP = collision exit potential; MDL = method detection limit).

Compound Internal standard Q1 mass Q3 mass DP CE CXP LC-MS MDL [V] [V] [V] method [ng L-1] 10,11-Dihydro-10,11-dihydroxycarbamazepine IS_Imidacloprid-D4 270.944 179.900 86 39 24 2 5 270.944 235.800 86 17 12 10,11-Dihydro-10-hydroxycarbamazepine IS_DEET-D7 255.043 237.100 26 13 14 2 19 255.043 194.000 26 29 12 1H-Benzotriazole IS_Benzotriazole-D4 119.998 65.000 106 29 8 2 6.9 119.998 91.900 106 25 26 2(4-morpholinyl)benzothiazole IS_Atrazine-13C3 220.977 177.100 96 31 10 1 0.5 220.977 109.000 96 47 12 2-(Methylthio)benzothiazole IS_Metolachlor-D6 181.946 166.800 51 33 20 1 14.2 181.946 109.000 51 49 12 2,4-Dichlorophenoxyacetic acid IS_Mecoprop-D3 218.900 160.700 -20 -18 -13 3 14.2 220.900 162.900 -20 -18 -13 2,4-Dinitrophenol IS_Nitrophenol-D4 182.910 109.000 -40 -38 -11 3 16.1 182.910 108.800 -40 -36 -13 2,6-Dichlorobenzamide IS_Cotinine-d3 190.022 173.000 96 25 16 2 1.3 190.022 144.900 96 39 14 2-Aminobenzimidazole IS_Cotinine-d3 133.984 91.900 86 31 10 2 5.7 133.984 92.900 86 33 10 2-Hydroxycarbamazepine IS_DEET-D7 252.854 210.100 101 27 6 2 2 252.854 208.000 101 31 10

Appendix A – Temporal chemical and risk patterns

Compound Internal standard Q1 mass Q3 mass DP CE CXP LC-MS MDL [V] [V] [V] method [ng L-1] 2-Mercaptobenzothiazole IS_IsoproturonD3 167.854 135.000 71 33 14 1 29.5 167.854 123.900 130 44 10 2-Methylbenzothiazole IS_Sulfamethoxazole-D4 149.947 109.000 76 31 12 1 16 149.947 64.900 76 45 10 2-Naphthalenesulfonic acid IS_Mono-isobutyl- 207.925 144.000 -80 -32 -9 3 22.4 phthalate-d4 207.925 79.900 -80 -54 -9 2-Octyl-4-isothiazolin-3-one IS_Metolachlor-D6 214.047 101.900 76 21 12 1 1.2 214.047 43.100 76 39 10 3-Iodopropynyl butylcarbamate IS_DEET-D7 281.907 164.900 1 33 20 2 12 281.907 57.000 1 13 6 4+5-Methyl-1H-benzotriazole IS_Imidacloprid-D4 134.016 76.900 91 33 20 2 9 134.016 79.000 91 27 10 4-Fluorobenzoylproponic acid IS_Nitrophenol-D4 195.100 151.000 -25 -14 -9 3 15 195.100 95.000 -25 -28 -11 Acesulfame IS_Acesulfame-D4 161.873 81.800 -20 -18 -9 3 1.9 161.873 77.700 -20 -46 -9 Acetamiprid IS_Imidacloprid-D4 223.100 126.100 86 27 8 1 1.3 223.100 90.000 86 45 10 Acetylsulfamethoxazole IS_Sulfamethoxazole-D4 295.933 134.000 131 31 16 1 4.6 295.933 64.900 131 57 16 Ambroxol IS_Metolachlor-D6 378.837 263.900 1 27 16 2 2.7 378.837 104.000 1 71 12 Amitriptyline IS_Metolachlor-D6 278.085 91.200 86 29 10 2 2.1 278.085 104.900 86 29 18 Atorvastatin IS_Mecoprop-D3 557.169 278.100 -100 -58 -17 3 19 557.169 397.100 -100 -40 -19 Atrazine IS_Atrazine-13C3 216.005 174.000 36 23 20 2 3.3

Appendix A – Temporal chemical and risk patterns

Compound Internal standard Q1 mass Q3 mass DP CE CXP LC-MS MDL [V] [V] [V] method [ng L-1] 216.005 96.000 36 31 10 Azithromycin IS_Metolachlor-D6 749.474 591.400 1 39 20 2 50 749.474 158.100 1 47 18 Azoxystrobin IS_IsoproturonD3 404.113 372.000 6 19 18 1 0.4 404.113 329.100 6 41 16 Bentazone IS_Nitrophenol-D4 238.891 131.900 -45 -36 -9 3 1.2 238.891 196.800 -45 -28 -17 Benzocaine IS_Atrazine-13C3 166.015 138.000 11 15 12 1 1.2 166.015 120.000 11 25 14 Bicalutamide IS_Diclofenac-D4 428.912 255.000 -70 -22 -9 3 0.5 428.912 178.900 -70 -18 -11 Bifonazol IS_DEET-D7 311.014 242.900 1 15 12 1 0.8 311.014 165.000 1 61 8 Bisoprolol IS_DEET-D7 326.389 116.100 1 25 14 2 0.7 326.389 99.000 1 10 29 Bisphenol A IS_Mecoprop-D3 226.934 212.000 -85 -24 -11 3 16.1 226.934 133.000 -85 -34 -9 Bisphenol S IS_Nitrophenol-D4 249.000 107.900 -130 -36 -7 3 1.3 249.000 155.900 -130 -30 -13 Boscalid IS_Metolachlor-D6 343.054 306.900 111 27 16 1 1.9 343.054 270.900 111 43 14 Bromoxynil IS_Nitrophenol-D4 273.800 78.900 -45 -58 -9 3 13.2 275.800 80.900 -45 -58 -9 Butylparaben IS_Mecoprop-D3 192.967 91.900 -60 -34 -11 3 13.5 192.967 136.000 -60 -22 -9 Caffeine IS_Caffeine-D3 195.087 138.000 36 27 18 1 14 195.087 110.000 36 31 14

Appendix A – Temporal chemical and risk patterns

Compound Internal standard Q1 mass Q3 mass DP CE CXP LC-MS MDL [V] [V] [V] method [ng L-1] Carbamazepine IS_DEET-D7 237.165 194.000 61 29 10 2 4.7 237.165 179.100 61 47 20 Carbendazim IS_Imidacloprid-D4 192.015 159.900 61 27 8 2 1 192.015 131.900 61 41 14 Cetirizine IS_Atrazine-13C3 389.039 201.000 51 25 12 1 2.2 389.039 165.000 51 87 18 Chloridazone IS_Imidacloprid-D4 222.095 103.900 61 29 12 1 3.9 222.095 91.900 61 33 16 Chlorotoluron IS_DEET-D7 213.103 71.900 1 21 16 2 24 213.103 46.000 1 35 12 Chloroxuron IS_Metolachlor-D6 291.074 72.000 91 23 8 2 3.6 291.074 46.100 91 51 12 Citalopram IS_Isoproturon-D3 325.091 108.900 111 33 14 2 2.2 325.091 262.100 111 27 16 Clomazone IS_Isoproturon-D3 240.170 124.900 16 27 20 1 0.6 240.170 89.000 16 63 10 Cotinine IS_Cotinine-D3 177.035 79.900 116 29 14 1 2.5 177.035 98.000 116 29 12 Cybutryn (Irgarol) IS_Metolachlor-D6 253.869 198.000 1 25 22 2 1.2 253.869 90.900 1 35 12 Cyclamate IS_Acesulfame-D4 178.000 80.000 -110 -28 -9 3 24.2 178.000 96.000 -110 -30 -11 DCOIT IS_Tri-n- 281.935 170.000 86 21 8 1 12.5 butylphosphate-D27 281.935 43.000 86 41 12 DEET IS_DEET-D7 192.088 118.900 101 23 14 1 2 192.088 90.900 101 39 10 Desethylatrazine IS_Imidacloprid-D4 188.011 146.000 86 23 16 2 2.7

Appendix A – Temporal chemical and risk patterns

Compound Internal standard Q1 mass Q3 mass DP CE CXP LC-MS MDL [V] [V] [V] method [ng L-1] 188.011 103.900 86 35 12 Desethylterbutylazine IS_DEET-D7 202.061 145.900 91 21 18 2 1.9 202.061 103.900 91 39 18 Desisopropylatrazine IS_Caffeine-D3 174.017 103.900 96 31 12 2 2.4 174.017 131.900 96 23 14 Diazinon IS_Diazinon-D10 304.993 169.000 136 27 20 1 0.3 304.993 153.000 136 27 8 Diclofenac IS_Mecoprop-D3 293.850 249.800 -20 -16 -31 3 6.9 293.850 213.900 -20 -28 -15 Difenoconazole IS_Metolachlor-D6 406.012 251.100 111 31 14 1 0.6 406.012 337.000 111 23 18 Diflufenican IS_Tri-n- 395.069 265.900 86 29 14 2 0.6 butylphosphate-D27 395.069 238.100 86 51 12 Dimethachlor ESA IS_Mecoprop-D3 299.979 120.800 -85 -28 -13 3 16.1 299.979 80.000 -85 -60 -9 Dimethachlor OA IS_Mecoprop-D3 249.965 178.200 -25 -14 -7 3 8.2 249.965 130.100 -25 -32 -9 Dimethoate IS_Imidacloprid-D4 230.057 198.800 16 13 10 2 1 230.057 124.900 16 27 14 Diphenhydramine IS_Metolachlor-D6 255.890 167.000 141 23 24 2 22.5 255.890 152.100 141 45 12 Diuron IS_Isoproturon-D3 232.935 71.900 96 47 8 2 3.3 232.935 46.100 96 39 12 Ebastin IS_Tri-n- 470.225 167.100 86 37 20 2 11.7 butylphosphate-D27 470.225 203.100 86 41 10 Enalapril IS_DEET-D7 376.994 234.100 51 25 14 1 0.7 376.994 117.000 51 47 14

Appendix A – Temporal chemical and risk patterns

Compound Internal standard Q1 mass Q3 mass DP CE CXP LC-MS MDL [V] [V] [V] method [ng L-1] Epoxiconazole IS_Tebuconazole-D9 330.048 121.000 1 25 14 1 1.2 330.048 75.000 1 97 12 Ethofumesat IS_Metolachlor-D6 287.112 120.900 86 21 16 1 2.3 287.112 121.200 86 21 8 Ethylparaben IS_Mecoprop-D3 165.000 136.000 -55 -20 -11 3 6.6 165.000 91.800 -55 -30 -11 Fenpropimorph IS_Tri-n- 304.351 147.100 11 39 18 2 0.7 butylphosphate-D27 304.351 117.000 11 75 14 Fenuron IS_Imidacloprid-D4 165.059 72.000 46 39 12 2 0.5 165.059 46.000 46 22 16 Finasteride IS_Metolachlor-D6 373.150 317.200 131 31 18 1 4.6 373.150 305.000 131 41 16 Fipronil IS_Diclofenac-D4 434.879 330.000 -75 -24 -13 3 0.4 434.879 249.800 -75 -38 -13 Fipronil desulfinyl IS_Mecoprop-D3 386.887 351.000 -50 -20 -19 3 0.3 386.887 281.900 -50 -46 -17 Fipronil sulfide IS_Mecoprop-D3 418.852 382.900 -50 -18 -19 3 0.4 418.852 261.800 -50 -38 -13 Fipronil sulfone IS_Mecoprop-D3 450.817 414.900 -55 -24 -19 3 0.4 450.817 281.900 -55 -38 -15 Flufenacet IS_Metolachlor-D6 363.983 194.300 31 15 14 2 0.5 363.983 152.000 31 27 20 Flurtamone IS_IsoproturonD3 334.141 247.000 71 37 12 1 0.3 334.141 178.100 71 57 22 Flusilazole IS_Tebuconazole-D9 316.141 246.900 16 25 12 1 0.9 316.141 164.900 16 33 22 Gabapentin IS_Benzotriazole-D4 172.073 154.000 51 19 10 1 3

Appendix A – Temporal chemical and risk patterns

Compound Internal standard Q1 mass Q3 mass DP CE CXP LC-MS MDL [V] [V] [V] method [ng L-1] 172.073 137.000 51 21 16 Imidacloprid IS_Imidacloprid-D4 256.059 175.100 6 29 20 2 4.6 256.059 209.000 6 19 34 Isoproturon IS_Isoproturon-D3 207.130 72.000 76 23 10 2 4.9 207.130 46.000 76 33 4 Lenacil IS_Isoproturon-D3 235.193 153.000 21 23 10 2 3.8 235.193 136.000 21 41 14 Lidocaine IS_Metolachlor-D6 235.091 86.000 81 21 12 2 0.5 235.091 58.000 81 51 10 Linuron IS_Metolachlor-D6 249.027 160.000 46 21 8 2 4.3 249.027 182.000 46 23 30 Loperamide IS_Metolachlor-D6 477.091 266.000 101 35 14 2 1.4 477.091 210.200 101 67 20 Losartan IS_Diclofenac-D4 421.068 178.900 -75 -30 -9 3 4.6 421.068 127.000 -75 -34 -9 MCPA IS_Mecoprop-D3 198.897 140.800 -25 -20 -13 3 0.8 200.900 142.900 -25 -20 -13 Mecoprop IS_Mecoprop-D3 212.900 140.800 -15 -22 -15 3 1.5 214.900 142.800 -15 -22 -15 Melperon IS_Isoproturon-D3 264.074 165.100 71 27 18 2 1.4 264.074 122.900 71 41 16 Metamitron IS_Benzotriazole-D4 203.042 175.100 96 23 18 1 3.3 203.042 42.000 96 63 12 Metazachlor IS_DEET-D7 277.989 134.000 21 29 16 1 0.7 277.989 210.000 21 15 6 Metazachlor ESA IS_Mecoprop-D3 321.964 120.900 -90 -28 -5 3 4.5 321.964 148.100 -90 -32 -15

Appendix A – Temporal chemical and risk patterns

Compound Internal standard Q1 mass Q3 mass DP CE CXP LC-MS MDL [V] [V] [V] method [ng L-1] Metformin IS_Cotinine-D3 130.069 60.000 1 17 16 2 1.4 130.069 71.000 1 27 8 Methiocarb IS_Metolachlor-D6 226.182 169.200 36 13 10 1 0.9 226.182 121.000 36 23 16 Metolachlor IS_Metolachlor-D6 284.022 251.900 36 21 12 1 1 284.022 176.100 36 35 10 Metolachlor ESA IS_Mecoprop-D3 328.002 120.900 -90 -30 -7 3 8.2 328.002 79.900 -90 -72 -9 Metolachlor OA IS_Mecoprop-D3 277.988 206.200 -35 -16 -17 3 3.8 277.988 174.100 -35 -22 -5 Metoprolol IS_DEET-D7 268.073 116.000 136 25 12 2 2.7 268.073 77.000 136 77 8 Metoprolol acid IS_Caffeine-D3 268.075 144.800 61 33 16 1 1.7 268.075 190.900 61 25 10 Miconazole IS_Metolachlor-D6 414.860 158.900 136 37 8 1 2.4 414.860 123.000 136 87 12 Myclobutanil IS_Tebuconazole-D9 289.038 69.900 71 21 20 1 0.8 289.038 125.000 71 43 14 N-Acetyl-4-aminoantipyrine IS_Benzotriazole-D4 246.014 228.100 81 19 24 1 2.7 246.014 204.000 81 17 4 N-cyclohexl-2-benzothiazole-sulfenamide IS_Benzotriazole-D4 264.943 166.000 71 31 10 1 2.5 264.943 183.000 71 23 20 N-cyclohexyl-2-benzothiazole-amine IS_Atrazine-13C3 233.022 150.800 76 27 18 1 0.7 233.022 108.900 76 57 12 N-Formyl-4-aminoantipyrine IS_Benzotriazole-D4 232.051 214.200 41 19 12 1 3.5 232.051 103.900 41 27 12 Nitrendipine IS_Metolachlor-D6 361.035 329.100 56 17 16 1 24

S10

Appendix A – Temporal chemical and risk patterns

Compound Internal standard Q1 mass Q3 mass DP CE CXP LC-MS MDL [V] [V] [V] method [ng L-1] 361.035 315.200 56 15 18 Oxybutynin IS_Tri-n- 358.085 142.200 1 31 6 2 4.5 butylphosphate-D27 358.085 340.000 1 25 18 Paroxetine IS_DEET-D7 330.017 191.900 66 29 10 2 4.7 330.017 69.900 66 49 8 Pendimethaline IS_Tri-n- 282.187 211.900 26 15 24 2 1.2 butylphosphate-D27 282.187 118.000 26 27 12 Perfluorobutanoic acid IS_Nitrophenol-D4 212.825 168.900 -10 -12 -11 3 8.15 212.825 168.300 -10 -16 -55 Pethoxamid IS_Metolachlor-D6 296.234 131.100 16 25 18 1 0.7 296.234 250.100 16 15 12 Phenazone IS_Imidacloprid-D4 189.034 77.100 1 53 10 2 2.5 189.034 56.000 1 51 8 Phenylbenzimidazole sulfonic acid IS_4-Nonylphenol-D4 272.904 192.900 -85 -40 -11 3 1.9 272.904 79.900 -85 -78 -9 Picolinafen IS_Tri-n- 377.112 238.000 76 35 12 2 1.2 butylphosphate-D27 377.112 144.900 76 69 16 Picoxystrobin IS_Diazinon-D10 368.026 205.100 26 13 22 1 2.9 368.026 145.000 26 31 34 Pindolol IS_Imidacloprid-D4 249.101 116.200 61 25 12 2 3.2 249.101 172.100 61 25 10 Pipamperone IS_DEET-D7 376.103 165.000 106 37 10 2 2.3 376.103 291.000 106 27 14 Pirimicarb IS_Imidacloprid-D4 239.149 182.100 1 21 10 1 0.7 239.149 72.000 1 25 14 p-Nitrophenol IS_Nitrophenol-D4 137.885 107.800 -55 -24 -5 3 21.1 137.885 91.900 -55 -32 -9

S11

Appendix A – Temporal chemical and risk patterns

Compound Internal standard Q1 mass Q3 mass DP CE CXP LC-MS MDL [V] [V] [V] method [ng L-1] Pravastatin IS_Mecoprop-D3 423.112 321.000 -110 -22 -15 3 16.9 423.112 303.200 -110 -24 -5 Prochloraz IS_Metolachlor-D6 375.964 307.900 1 15 16 1 1 375.964 265.800 1 23 14 Promethazin IS_DEET-D7 285.008 86.000 36 21 10 2 8.3 285.008 198.000 36 35 10 Propamocarb IS_Benzotriazole-D4 189.179 101.900 61 23 12 2 0.9 189.179 144.100 61 17 6 Propiconazole IS_Tebuconazole-D9 342.091 158.900 81 39 18 1 1.6 342.091 41.100 81 59 18 Propoxycarbazone IS_Mono-isobutyl- 397.100 156.100 -25 -16 -9 3 10 phthalate-D4 397.100 113.100 -25 -38 -5 Propranolol IS_Isoproturon-D3 260.126 116.000 141 25 16 2 2.7 260.126 183.000 141 21 20 Propylparaben IS_Mecoprop-D3 179.000 92.000 -55 -32 -10 3 15.5 178.891 136.000 -55 -22 -10 Prosulfocarb IS_Tri-n- 252.222 91.000 1 27 10 2 2.1 butylphosphate-D27 252.222 127.900 1 17 14 Prothioconazole-desthio IS_Metolachlor-D6 312.121 70.000 26 49 18 2 3.9 312.121 124.900 26 41 16 p-Toluenesulfonamide IS_Caffeine-D3 172.000 90.900 71 23 10 2 12 172.000 154.900 71 11 18 Pyraclostrobin IS_Metolachlor-D6 388.055 194.000 11 15 24 1 0.8 388.055 163.000 11 33 26 Quinmerac IS_Caffeine-D3 222.073 203.800 16 23 10 1 2.1 222.073 141.000 16 43 16 Saccharin IS_Acesulfame-D4 182.000 106.000 -25 -24 -19 3 15

S12

Appendix A – Temporal chemical and risk patterns

Compound Internal standard Q1 mass Q3 mass DP CE CXP LC-MS MDL [V] [V] [V] method [ng L-1] 182.000 41.900 -25 -54 -19 Scopolamine-N-butyl IS_DEET-D7 360.083 194.100 81 31 10 2 1.4 360.083 103.000 81 67 12 Sertraline IS_Metolachlor-D6 306.065 275.000 1 17 14 2 1.8 306.065 158.900 1 37 20 Simazine IS_DEET-D7 202.079 124.000 96 25 12 2 8.5 202.079 132.000 96 25 54 Spiroxamine IS_Tri-n- 298.316 144.000 46 27 22 2 0.8 butylphosphate-D27 298.316 100.000 46 39 10 Sulfamethoxazole IS_Sulfamethoxazole-D4 253.940 155.900 86 23 8 1 1.1 253.940 91.900 86 35 10 Tebuconazole IS_Tebuconazole-D9 308.209 70.000 76 55 8 2 2.8 308.209 124.900 76 49 18 Terbuthylazine IS_Metolachlor-D6 230.002 173.900 121 21 8 2 1.9 230.002 68.100 121 47 32 Terbuthylazine-2-hydroxy IS_Benzotriazole-D4 212.192 156.000 6 21 20 1 0.5 212.192 86.000 6 31 10 Terbutryn IS_Metolachlor-D6 242.064 186.000 96 27 8 2 0.7 242.064 90.900 96 35 10 Tetracaine IS_Metolachlor-D6 265.091 175.900 71 21 22 2 0.7 265.091 71.900 71 33 8 Thiabendazole IS_Caffeine-D3 201.990 175.000 121 35 20 1 0.6 201.990 131.000 121 45 14 Thiacloprid IS_Imidacloprid-D4 252.992 126.000 66 29 16 1 0.4 252.992 90.000 66 47 12 Thiamethoxam IS_Benzotriazole-D4 291.955 211.100 31 17 22 1 1.3 291.955 180.900 31 29 46

S13

Appendix A – Temporal chemical and risk patterns

Compound Internal standard Q1 mass Q3 mass DP CE CXP LC-MS MDL [V] [V] [V] method [ng L-1] Tramadol IS_Imidacloprid-D4 264.062 58.000 61 47 16 2 4.4 264.062 42.000 61 109 10 Triclosan IS_Triclosan-D3 287.000 35.000 -35 -40 -15 3 23.1 287.000 142.000 -35 -44 -9 Trifloxystrobin IS_Tri-n- 409.115 186.100 6 23 10 1 0.8 butylphosphate-D27 409.115 206.000 6 19 20 Trimethoprim IS_Imidacloprid-D4 291.005 230.200 101 31 14 2 2.2 101 31 14 291.005 123.100 IS_4-Nonylphenol-D4 223.078 109.900 -115 -28 -13 -115 -50 -11 223.078 123.000 IS_Acesulfame-D4 166.000 86.000 -10 -20 -11 -10 -46 -9 166.000 78.000 IS_Atrazine-13C3 219.100 177.000 106 23 8 106 47 8 219.100 69.900 IS_Benzotriazole-D4 124.086 69.000 106 31 8 106 49 10 124.086 41.100 IS_BisphenolA-13C12 239.000 224.000 -85 -24 -11 -85 -40 -11 239.000 223.000 IS_Caffeine-D3 198.135 138.000 36 27 18 36 31 14 198.135 110.000 IS_Cotinine-D3 180.095 80.000 56 29 10 56 31 18 180.095 101.000 IS_DEET-D7 199.192 126.000 121 23 6 121 41 10 199.192 98.000 IS_Diazinon-D10 314.926 170.000 131 29 26 131 29 20 314.926 154.000 IS_Diclofenac-D4 297.775 253.900 -15 -16 -23

S14

Appendix A – Temporal chemical and risk patterns

Compound Internal standard Q1 mass Q3 mass DP CE CXP LC-MS MDL [V] [V] [V] method [ng L-1] -15 -30 -13 297.775 216.900 IS_Imidacloprid-D4 259.989 213.000 36 21 12 36 27 10 259.989 179.100 IS_IsoproturonD3 210.131 74.900 91 21 8 91 31 12 210.131 49.000 IS_Mecoprop-D3 215.900 143.900 -25 -18 -7 -25 -18 -7 217.900 145.900 IS_Metolachlor-D6 290.166 258.100 56 19 12 56 35 10 290.166 182.100 IS_Mono-isobutyl-phthalate-D4 224.993 81.000 -30 -22 -9 -30 -14 -11 224.993 181.100 IS_Nitrophenol-D4 141.927 112.000 -50 -24 -7 -50 -32 -11 141.927 95.900 IS_Sulfamethoxazole-D4 258.106 95.900 111 35 10 111 31 14 258.106 111.900 IS_Tebuconazole-D9 317.046 70.000 51 61 18 51 57 14 317.046 124.900 IS_Triclosan-D3 290.000 35.000 -35 -40 -15 -35 -40 -15 292.000 35.000 IS_Tri-n-butylphosphate-D27 294.303 101.900 66 25 26 294.303 82.900 130 44 10

S15

Appendix A – Temporal chemical and risk patterns

A2.2 Sample preparation and LC-MS/MS settings for target screening of WWTP and rain sewer samples The solvent gradient is shown in Table S4. Table S5 provides parameter settings for the ion source in positive and negative ionization mode.

Table S4: LC solvent gradient for chromatographic separation of target compounds

Time [min] Solvent A [%] Solvent B [%] 0 95 5 1 95 5 6.2 5 95 11.4 5 95 11.7 95 5 16 95 5

Table S5: Ion source parameters setting in positive and negative ionization mode

Parameter Positive mode Negative mode Curtain gas 50 50 Collision gas Medium Medium IonSpray Voltage 2900 -2800 Temperature 400 400 Ion source gas 1 60 60 Ion source gas 2 60 60 Entrance potential 10 -10

S16

Appendix A – Temporal chemical and risk patterns

A2.3 Chemical screening LVSPE samples Cartridges were freeze-dried and eluted with 100 ml of ethyl acetate and 100 ml of methanol, followed by 100 ml of methanol containing 1% of formic acid and 100 ml of methanol with 2% 7N ammonia solution. The extracts were evaporated to dryness under nitrogen and reconstituted to 1.5 ml in methanol resulting in a concentration factor of 1000. For chemical analysis, aliquots of 100 µL were taken from each sample and brought to a final volume of 200 µL methanol/water (70:30) including 10 µL of a mixture of 31 isotope-labelled internal standards (1 µg ml-1). Chemical screening was performed on an UltiMate 3000 LC system (Thermo Scientific) coupled to a hybrid quadrupole - Orbitrap MS (Q ExactiveTM Plus, Thermo Scientific) with a HESI source. For chromatographic separation, a Kinetex 2.6 μm EVO C18 (50x2.1 mm) column equipped with a pre-column (C18 EVO 5.x2.1 mm) and an inline filter was used. The injection volume was 5 µL. The column temperature was 40°C and the LC gradient is shown in Table S6.

Table S6: LC solvent gradient for analysis of LVSPE samples with LC-HRMS

Time [min] Flow rate [µL] Solvent A [%] Solvent B [%] Solvent C [%] Water + 0.1% MeOH + 0.1% Acetone/ formic acid formic acid Isopropanol (50:50) 0 0.3 95 5 0 1 0.3 95 5 0 13 0.3 0 100 0 24 0.3 0 100 0 24.1 0.35 5 10 85 26.2 0.35 5 10 85 26.3 0.35 95 5 0 31.9 0.35 95 5 0 32.0 0.3 95 5 0 For quantification, method-matched calibrations standard were prepared with pristine river water from Wormsgraben. The spiked standards were extracted by SPE to match the sample preparation procedure and concentration factor of the LVSPE samples. The eleven-point calibration ranged from 0.1 to 1000 ng L-1. LVSPE samples were evaluated for the main risk driving compounds identified in the WWTP and rain sewer samples. The software TraceFinder 3.2 (Thermo Scientific) was used for evaluation of the LVSPE data. Details on MDLs and ionization modes are shown in Table S7. For MS/MS analysis, a full scan experiment (100-1000 m/z) at a nominal resolving power of 70,000 (referenced to m/z 200) was combined with data-independent (DIA) MS/MS experiments at a nominal resolving power of 35,000. For DIA experiments, broad isolation windows of about 50 (i.e., m/z ranges 97-147, 144-194, 191-241, 238-288, 285-335, 332- 382, 379-429, 426-476) and 280 (i.e., m/z ranges 460-740, 730-1010), respectively, were used for data acquisition.

S17

Appendix A – Temporal chemical and risk patterns

Table S7: Risk driving compounds analyzed in LVSPE river samples including information on ionization mode and MDL [ng L-1]

Compound Ionization mode MDL [ng L-1] Internal standard Decyltrimethylammonium- Amitriptyline Positive 1 D30 Carbendazim Positive 1 Carbendazim-D4 Citalopram Positive 0.7 Verapamil-D6 Diazinon Positive 0.7 Diazinon-D10 Diclofenac Negative 1 Diclofenac-D4 Dimethoate Positive 1 Imidacloprid-D4 Diuron Positive 1 Isoproturon-D3 Fipronil Negative 0.5 Diclofenac-D4 Mono-isobutylphthalate- MCPA Negative 1 D4 Metolachlor Positive 0.7 Metolachlor-D6 N-Formyl-4- Positive 3 Deisopropylatrazine-D5 aminoantipyrine Spiroxamine Positive 2 Atrazine-13C3 Terbuthylazine Positive 1 Bezafibrate-D4 Terbutryn Positive 1 Atrazine-13C3

S18

Appendix A – Temporal chemical and risk patterns

A2.4 Acute effect concentrations of target compounds

In case of missing measured effect concentrations, predicted effect concentrations (ECP) were used for the calculation of toxic units. ECPs were provided for all BQE were based on read-across or QSAR models (according to Busch et al. (2016)). Data from read-across models was preferred but was not always available for every compound, especially for fish.

Table S8: Acute effect concentrations for each compound and BQE. (“PR” = predicted effect concentration via read-across; “PE” = predicted effect concentration via ECOSAR; no letter = measured effect concentration).

Compound Use class LC50 fish LC50 daphnia EC50 algae [mg L-1] [mg L-1] [mg L-1] 10,11-Dihydro-10,11-dihydroxycarbamazepine TP of carbamazepine 2.78E+02 (PE) 1.18E+03 (PE) 2.29E+00 (PR) 10,11-Dihydro-10-hydroxycarbamazepine TP of carbamazepine 6.50E+01 (PE) 2.15E+01 (PE) 1.61E+00 (PR) 1H-Benzotriazole wastewater marker 2.41E+01 (PE) 5.67E+01 (PE) 1.10E+01 (PR) 2(4-morpholinyl)benzothiazole industrial chemical 4.17E+01 (PE) 2.53E+01 (PE) 2.47E+01 (PR) 2-(methylthio)benzothiazole industrial chemical 2.25E+00 (PE) 4.92E-01 (PE) 2.38E+00 (PR) 2,4-Dinitrophenol industrial chemical 5.55E-01 3.09E+00 1.24E+01 TP of carbendazim 2-Aminobenzimidazole (fungicide) 6.49E+01 (PE) 2.01E+00 (PE) 5.23E+00 (PR) 2-Hydroxycarbamazepine TP of carbamazepine 8.26E+00 (PE) 5.90E+01 (PR) 1.12E+01 (PR) 2-Naphthalenesulfonic acid industrial chemical 4.86E+01 2.94E+02 (PE) 1.24E+00 (PR) 2-Octyl-4-isothiazolinone biocide 3.07E-02 (PE) 3.99E-02 (PR) 1.30E+00 (PR) 4+5-Methyl-1H-benzotriazole industrial chemical 1.56E+01 (PE) 1.51E+02 (PR) 2.16E+00 (PR) Acesulfame food constituent 2.57E+03 (PE) 5.82E+02 (PE) 9.93E-01 (PR) Acetyl-Sulfamethoxazole TP of sulfamethoxazole 1.36E+02 (PE) 6.03E+01 (PE) 1.93E+01 (PR) Ambroxol pharmaceutical 1.32E+01(PE) 1.79E+00 (PE) 1.14E+00 (PE) Amitriptyline pharmaceutical 7.80E-01 9.42E-01 1.86E+00 (PR) Atrazine legacy herbicide 9.13E-01 1.24E-01 7.86E-03 Azoxystrobin fungicide 4.87E-01 5.94E-02 2.09E-02 Bentazon herbicide 1.48E+02 (PE) 1.49E+02 5.00E-02 Bicalutamide pharmaceutical 1.66E+01 (PE) 1.18E+01 (PE) 5.26E-01 (PR) Bisoprolol pharmaceutical 3.75E+01 (PE) 4.01E+00 (PE) 2.39E+00 (PR)

S19

Appendix A – Temporal chemical and risk patterns

Compound Use class LC50 fish LC50 daphnia EC50 algae [mg L-1] [mg L-1] [mg L-1] Boscalid fungicide 2.79E+00 3.81E+00 5.96E-01 Carbamazepine pharmaceutical 3.63E+01 1.11E+02 1.65E+01 Carbendazim biocide 1.21E-02 2.51E-02 4.20E-01 Cetirizine pharmaceutical 1.24E+01 (PE) 2.27E+00 (PR) 2.72E+00 (PR) Citalopram pharmaceutical 3.70E+00 (PE) 3.92E+00 3.75E+00 (PR) Clomazone herbicide 7.54E+00 2.20E-01 1.79E-01 Cotinine Nicotine TP 2.71E+02 (PE) 1.29E+01 (PR) 8.09E+00 (PR) Cyclamate food constituent 8.14E+02 (PE) 2.33E+02 (PE) 1.01E+01 (PR) Desethylatrazine TP of atrazine 5.60E+01 (PE) 1.26E+02 (PR) 3.08E-02 (PR) Desethylterbuthylazine TP of terbuthylazine 3.94E+01 (PE) 7.86E+01 (PR) 1.54E-02 (PR) Desisopropylatrazine TP of atrazine 8.19E+01 (PE) 2.07E+02 (PR) 6.77E-04 Diazinon biocide 3.00E-02 1.00E-05 1.00E+00 Diclofenac pharmaceutical 1.01E+01 1.86E+01 6.46E+00 DEET insect repellent 7.20E+01 6.12E+01 (PR) 4.05E+00 (PR) Difenoconazole fungicide 2.83E-01 1.50E-01 8.29E-03 Diflufenican herbicide 5.36E-01 (PE) 4.07E-01 (PR) 3.31E-01 (PR) Dimethachlor-OA TP of dimetholachlor 4.60E+02 (PE) 2.44E+02 (PE) 5.26E-01 (PR) Dimethoate insecticide 1.30E-01 2.24E-03 1.01E+01 Diuron biocide 2.06E-01 3.82E-01 6.94E-04 Enalapril pharmaceutical 7.95E+01 (PE) 4.47E+01 (PE) 1.51E+00 (PR) Epoxiconazole fungicide 3.72E+00 (PE) 7.77E+00 (PR) 1.23E+00 (PR) Ethofumesate herbicide 5.65E+00 2.50E+00 2.90E+00 Fenpropimorph fungicide 2.05E-01 4.20E+00 (PR) 5.99E-03 Fenuron legacy herbicide 1.84E+02 (PR) 1.29E+00 (PR) 1.13E+00 (PR) Fipronil biocide 3.80E-02 2.53E-04 1.24E-01 Fipronil desulfinyl TP of fipronil - Fipronil sulfide TP of fipronil 2.04E-02 (PE) 4.12E-01 (PR) 7.94E-01 (PR) Fipronil sulfone TP of fipronil 6.31E+01 2.72E+00 (PR) 8.18E-01 (PR) Flufenacet herbicide 5.65E-01 1.30E+00 3.01E-03

S20

Appendix A – Temporal chemical and risk patterns

Compound Use class LC50 fish LC50 daphnia EC50 algae [mg L-1] [mg L-1] [mg L-1] Flurtamone herbicide 2.80E-02 (PE) 2.18E+01 (PR) 4.44E-01 (PR) Flusilazole legacy fungicide 3.74E-01 (PE) 4.61E-01 (PE) 6.03E-01 Gabapentin pharmaceutical 5.07E+02 (PE) 1.10E+03 3.12E+00 (PR) Imidacloprid insecticide 5.30E+01 5.43E-02 1.22E+01 (PR) Isoproturon biocide 1.20E+01 (PE) 9.81E-01 (PR) 6.47E-03 Lenacil herbicide 1.66E+01 (PE) 2.27E+01 (PR) 1.69E-03 Lidocaine pharmaceutical 8.03E+00 (PE) 5.81E+00 (PE) 3.95E+01 Loperamide pharmaceutical 5.87E-01 (PE) 3.99E-01 (PE) 5.00E-01 (PE) Losartan pharmaceutical 1.38E-01 (PE) 2.06E-01 (PE) 2.46E+00 (PR) MCPA herbicide 1.50E+00 2.51E+00 4.48E+00 Mecoprop herbicide 1.02E+01 1.44E+02 (PR) 1.00E+01 (PE) Melperon pharmaceutical 3.12E+00 (PE) 2.50E-01 (PE) 4.60E-01 Metamitron herbicide 3.26E+00 (PE) 2.32E+02 (PR) 2.42E-02 Metazachlor herbicide 8.14E-01 (PE) 2.29E+01 (PR) 5.00E-01 Metazachlor-ESA TP of metazachlor 7.44E+00 (PE) 9.51E+01 (PE) 4.30E+00 (PR) Metformin pharmaceutical 2.14E+02 (PE) 1.35E+03 3.20E+02 Metolachlor herbicide 4.10E-02 5.05E+00 2.60E-02 Metoprolol pharmaceutical 1.00E+02 9.71E+00 7.30E+00 Metoprolol acid pharmaceutical TP 1.46E+02 (PE) 6.97E+01 (PR) 2.93E+01 (PR) Myclobutanil fungicide 1.29E+00 2.40E-01 8.30E-01 N-Acetyl-4-aminoantipyrine pharmaceutical TP 8.70E-01 (PE) 1.03E+01 (PE) 2.37E+00 (PR) N-cyclohexl-2-benzothiazole-sulfenamide industrial chemical - - N-Formyl-4-aminoantipyrine pharmaceutical TP 9.07E-01 (PE) 9.82E+00 (PE) 2.63E+00 (PR) Nitrendipine pharmaceutical 1.27E+01 (PE) 2.07E+00 (PE) 1.36E+00 (PE) Pethoxamid herbicide 1.11E+00 (PE) 3.55E+01 (PR) 9.13E-02 (PR) Phenazone pharmaceutical 1.78E-01 (PE) 6.52E+00 (PE) 2.97E+00 (PR) Phenylbenzimidazolesulfonic acid UV stabilizer 7.88E+01 (PE) 3.82E+02 (PE) 4.49E+00 (PR) Pipamperone pharmaceutical 5.57E+01 (PE) 3.05E+01 (PE) 9.03E-01 (PR) Pirimicarb insecticide 1.24E+01 2.68E-02 1.20E+02

S21

Appendix A – Temporal chemical and risk patterns

Compound Use class LC50 fish LC50 daphnia EC50 algae [mg L-1] [mg L-1] [mg L-1] Prochloraz fungicide 1.74E+00 (PE) 4.07E+00 (PR) 3.92E-01 (PR) Propamocarb fungicide 9.22E+01 1.05E+02 2.61E+00 (PR) Propiconazole fungicide 1.00E+00 8.13E-01 5.84E-02 Propranolol pharmaceutical 1.18E+01 9.19E-01 5.80E+00 TP of prothioconazole Prothioconazole-desthio (fungicide) 1.54E+01 (PE) 1.71E+01 (PR) 3.54E+00 (PR) p-Toluenesulfonamide industrial chemical 1.76E+02 (PE) 1.92E+02 (PR) 6.25E+01 (PR) Pyraclostrobin fungicide 6.09E-03 2.19E-03 1.70E-03 Quinmerac herbicide 2.99E+02 (PE) 1.91E+02 (PE) 1.75E+01 (PE) Saccharin food constituent 1.07E+00 (PE) 2.56E+01 (PR) 1.29E+00 (PR) Scopolamine-N-butyl pharmaceutical - Sertraline pharmaceutical 7.81E-01 (PE) 1.77E+00 (PR) 1.07E+00 (PR) Simazine legacy herbicide 2.50E+00 5.19E+01 (PR) 3.08E-02 Spiroxamine fungicide 1.40E+00 4.59E+01 (PR) 3.79E-04 Sulfamethoxazole pharmaceutical 5.79E+02 3.54E+01 5.79E-01 Tebuconazole fungicide 4.63E+00 3.82E-01 3.90E-01 Terbuthylazine herbicide 2.71E+00 1.09E-01 1.30E-03 Terbuthylazine-2-hydroxy TP of terbuthylazine 1.14E+02 (PE) 5.03E+01 (PE) 4.16E+00 (PR) Terbutryn biocide 9.11E+00 (PE) 2.30E+01 (PR) 5.92E-03 (PR) Thiabendazole fungicide 5.60E-01 4.21E-01 9.00E+00 (PR) Thiacloprid insecticide 2.00E-02 2.23E-02 4.57E+01 Tramadol pharmaceutical 2.25E+01 (PE) 1.38E+02 (PR) 1.76E+01 (PR) Trimethoprim pharmaceutical 2.90E+02 (PE) 6.83E+01 (PR) 7.50E-01 (PR)

A2.5 Sublethal effect concentrations of target compounds Lowest-observable-effect-concentrations (LOEC) were derived from the ECOTOX database (https://cfpub.epa.gov/ecotox/). The query focused on effects on growth/population, reproduction and behavior. The lowest LOEC was selected for each species and compound

(Table S9). Compounds without reported effect data were not considered for the calculation of TUsub.

S22

Appendix A – Temporal chemical and risk patterns

Table S9: Sublethal effect concentrations of target compounds

Compound name Species Effect LOEC Author Title Source Year [µg L-1]

2,4-Dinitrophenol Subchronic Toxicities of Industrial Center for Lake Superior and Agricultural Chemicals to Environmental Studies, Fathead Call,D.J., and D.L. Fathead Minnows (Pimephales University of Wisconsin, Minnow Growth 183 Geiger promelas). Volume I Superior, WI:318 p. 1992 Limiting Values for the Noxious Effects of Water Pollutant Material to Blue Algae (Microcystis aeruginosa) and Green Algae (Scenedesmus quadricauda) in Cell Propagation Inhibition Tests TR-80-0201, Literature Bringmann,G., and R. (Grenzwerte der Schadwirkung Research Company, Annandale, Green Algae Population 16000 Kuhn Wasser) VA:39 p. 1978 Amitriptyline Growth Inhibition and Coordinated Physiological Regulation of Yang,M., W. Qiu, J. Zebrafish (Danio rerio) Embryos Chen, J. Zhan, C. Pan, Upon Sublethal Exposure to Danio rerio Growth 0.1 X. Lei, and M. Wu Antidepressant Amitriptyline Aquat. Toxicol.151:68-76 2014 Atrazine Evaluation of Potential Mechanisms of Atrazine-Induced Reproductive Richter,C.A., D.M. Impairment in Fathead Minnow Fathead Papoulias, J.J. Whyte, (Pimephales promelas) and Environ. Toxicol. Chem.35(9): Minnow Reproduction 0.5 and D.E. Tillitt Japanese Medaka (Oryzias latipes) 2230-2238 2016 Palma,P., V.L. Palma, Assessment of the Pesticides C. Matos, R.M. Atrazine, Endosulfan Sulphate and Fernandes, A. Bohn, Chlorpyrifos for Juvenoid-Related Daphnia A.M.V.M. Soares, and Endocrine Activity Using Daphnia magna Reproduction 500 I.R. Barbosa magna Chemosphere76(3): 335-340 2009 Effects of Atrazine on Selenastrum Report to CIBA-GEIGY University of capricornutum, Lemna minor, and Corporation, Greensboro, Green Algae Growth 1 Mississippi Elodea canadensis NC:122 p. 1991 Azoxystrobin Pesticide Ecotoxicity Database Environmental Fate and Effects Fathead U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Minnow Growth 193 Protection Agency Database (EEDB)) D.C.: 1992 Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Reproduction 84 Protection Agency Database (EEDB)) D.C.: 1992 Van Wijngaarden,R.P.A., D.J.M. Belgers, M.I. Chronic Aquatic Effect Assessment Environ. Toxicol. Chem.33(12): Green Algae Population 0.98 Zafar, A.M. Matser, for the Fungicide Azoxystrobin 2775-2785 2014

S23

Appendix A – Temporal chemical and risk patterns

Compound name Species Effect LOEC Author Title Source Year [µg L-1] M.C. Boerwinkel, and G.H.P. Arts Bicalutamide Panter,G.H., Y.C. Effects of the Anti-Androgen, Glennon, J. Robinson, Bicalutamide, in a Reduced Life- Fathead A. Hargreaves, and R. Cycle Study with the Fathead Minnow Reproduction 92.111 Murray-Smith Minnow (Pimephales promelas) Aquat. Toxicol.114/115:31-38 2012 Boscalid Pesticide Ecotoxicity Database Environmental Fate and Effects Rainbow U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Trout Growth 241 Protection Agency Database (EEDB)) D.C.: 1992 Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Reproduction 2630 Protection Agency Database (EEDB)) D.C.: 1992 Carbamazepine Gene-Class Analysis of Expression Patterns Induced by Psychoactive Pharmaceutical Exposure in Fathead Minnow (Pimephales Comp. Biochem. Physiol. C Fathead Thomas,M.A., P.P. promelas) Indicates Induction of Toxicol. Pharmacol.155(1): 109- Minnow Behavior 100 Joshi, and R.D. Klaper Neuronal Systems 120 2012 Single and Combined Toxicity of Dietrich,S., F. Ploessl, Pharmaceuticals at Environmentally Daphnia F. Bracher, and C. Relevant Concentrations in Daphnia magna Reproduction 0.5 Laforsch magna - A Multigenerational Study Chemosphere79(1): 60-66 2010 Haase,S.M., P. Panas, Effects of Carbamazepine on Two T. Rath, and B. Microalgae Species Differing in Water Air Soil Pollut.226(338): Green Algae Population 100 Huchzermeyer Stress Resistance 12 p. 2015 Carbendazim Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Reproduction 6.6 Protection Agency Database (EEDB)) D.C.: 1992 Impact of the Fungicide Van den Brink,P.J., J. Carbendazim in Freshwater Hattink, F. Bransen, E. Microcosms. II. Zooplankton, Van Donk, and T.C.M. Primary Producers and Final Green Algae Population 330 Brock Conclusions Aquat. Toxicol.48(2-3): 251-264 2000 Citalopram Olsen,K.H., K. Ask, H. Effects of the SSRI Citalopram on Olsen, I. Porsch- Behaviours Connected to Stress Endler's Hallstrom, and S. and Reproduction in Endler Guppy, Guppy Behavior 0.2 Hallgren Poecilia wingei Aquat. Toxicol.148:113-121 2014 Henry,T.B., J.W. Acute and Chronic Toxicity of Five Kwon, K.L. Armbrust, Selective Serotonin Reuptake Environ. Toxicol. Chem.23(9): C.dubia Reproduction 4000 and M.C. Black Inhibitors in Ceriodaphnia dubia 2229-2233 2004

S24

Appendix A – Temporal chemical and risk patterns

Compound name Species Effect LOEC Author Title Source Year [µg L-1] Clomazone Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Growth 4380 Protection Agency Database (EEDB)) D.C.: 1992 DEET Effects of Triclocarban, N,N-Diethyl- Zenobio,J.E., B.C. meta-Toluamide, and a Mixture of Sanchez, L.C. Pharmaceuticals and Personal Care Fathead Archuleta, and M.S. Products on Fathead Minnows Environ. Toxicol. Chem.33(4): Minnow MPH 0.6 Sepulveda (Pimephales promelas) 910-919 2014 Minderhout,T., T.Z. DEET: A Semi-Static Life-Cycle Daphnia Kendall, and H.O. Toxicity Test with the Cladoceran Project 319A-128. Wildlife magna Growth 7500 Krueger (Daphnia magna) International:82 p. 2008 Desisopropylatrazine Liu,Z.Z., Y.Y. Wang, Z.H. Zhu, E.L. Yang, Atrazine and Its Main Metabolites X.Y. Feng, Z.W. Fu, Alter the Locomotor Activity of Danio rerio Behavior 30 and Y.X. Jin Larval Zebrafish (Danio rerio) Chemosphere148:163-170 2016 Diazinon Pesticide Ecotoxicity Database Environmental Fate and Effects Fathead U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Minnow Reproduction 1.82 Protection Agency Database (EEDB)) D.C.: 1992 Assessment of the Toxicity of a Sanchez,M., M.D. Pesticide with a Two-Generation Comp. Biochem. Physiol. C Daphnia Ferrando, E. Sancho, Reproduction Test Using Daphnia Comp. Pharmacol. magna Growth 0.00005 and E. Andreu magna Toxicol.124(3): 247-252 1999 Chemical Cocktails in Aquatic Systems: Pesticide Effects on the Response and Recovery of > 20 Algae Population 6.7 Hua,J., and R. Relyea Animal Taxa Environ. Pollut.189:18-26 2014 Diclofenac Prostaglandins in the Zebrafish Ovary: Role, Regulation, and Modulation by Environmental Ph.D.Thesis, University of Danio rerio Reproduction 100 Lister,A.L. Pharmaceuticals Guelph, Ontario, Canada:245 p. 2009 Single and Combined Toxicity of Dietrich,S., F. Ploessl, Pharmaceuticals at Environmentally Daphnia F. Bracher, and C. Relevant Concentrations in Daphnia magna Growth 0.36 Laforsch magna - A Multigenerational Study Chemosphere79(1): 60-66 2010 Lawrence,J.R., B. Zhu, Molecular and Microscopic G.D.W. Swerhone, J. Assessment of the Effects of Roy, V. Tumber, M.J. Caffeine, Acetaminophen, Waiser, E. Topp, and Diclofenac, and Their Mixtures on Environ. Toxicol. Chem.31(3): Algae Population 5 D.R. Korber River Biofilm Communities 508-517 2012 Difenoconazole

S25

Appendix A – Temporal chemical and risk patterns

Compound name Species Effect LOEC Author Title Source Year [µg L-1] Pesticide Ecotoxicity Database Environmental Fate and Effects Fathead U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Minnow Growth 3.7 Protection Agency Database (EEDB)) D.C.: 1992 Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Reproduction 13 Protection Agency Database (EEDB)) D.C.: 1992 Dimethoate Embryo and Fingerling Toxicity of Dimethoate and Effect on Fecundity, Viability, Hatchability and Ansari,S., and B.A. Survival of Zebrafish, Danio rerio World J. Fish Mar. Sci.3(2): 167- Danio rerio Reproduction 24.68 Ansari (Cyprinidae) 173 2011 Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Reproduction 100 Protection Agency Database (EEDB)) D.C.: 1992 Dimethoate and Quinalphos Toxicity: Pattern of Photosynthetic Blue-Green Mohapatra,P.K., and Pigment Degradation and Recovery Algae Population 11463 U. Schiewer in Synechocystis sp. PCC 6803 Algol. Stud.99:79-94 2000 Diuron Pesticide Ecotoxicity Database Environmental Fate and Effects Fathead U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Minnow Reproduction 61.8 Protection Agency Database (EEDB)) D.C.: 1992 Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Growth 130 Protection Agency Database (EEDB)) D.C.: 1992 Selective Real-Time Herbicide Monitoring by an Array Chip Podola,B., and M. Biosensor Employing Diverse Green Algae Population 0.5 Melkonian Microalgae J. Appl. Phycol.17(3): 261-271 2005 Ethofumesate Pesticide Ecotoxicity Database Environmental Fate and Effects Fathead U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Minnow Growth 4170 Protection Agency Database (EEDB)) D.C.: 1992 Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Growth 2500 Protection Agency Database (EEDB)) D.C.: 1992 Fenpropimorph Pesticide Ecotoxicity Database Environmental Fate and Effects Rainbow U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Trout Growth 0.16 Protection Agency Database (EEDB)) D.C.: 1992 Fipronil Pesticide Ecotoxicity Database Environmental Fate and Effects Sheephead U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, minnow Reproduction 0.24 Protection Agency Database (EEDB)) D.C.: 1992

S26

Appendix A – Temporal chemical and risk patterns

Compound name Species Effect LOEC Author Title Source Year [µg L-1] Chevalier,J., E. Exploration of Daphnia Behavioral Harscoet, M. Keller, P. Effect Profiles Induced by a Broad Daphnia Pandard, J. Cachot, Range of Toxicants with Different Environ. Toxicol. Chem.34(8): magna Behavior 1.4 and M. Grote Modes of Action 1760-1769 2015 Overmyer,J.P., D.R. Rouse, J.K. Avants, A.W. Garrison, M.E. DeLorenzo, K.W. Chung, P.B. Key, W.A. Toxicity of Fipronil and Its J. Environ. Sci. Health Part B Wilson, and M.C. Enantiomers to Marine and Pestic. Food Contam. Agric. Green Algae Population 500 Black Freshwater Non-Targets Wastes42(5): 471-480 2007 Flufenacet Pesticide Ecotoxicity Database Environmental Fate and Effects Rainbow U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Trout Growth 334 Protection Agency Database (EEDB)) D.C.: 1992 Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Reproduction 12800 Protection Agency Database (EEDB)) D.C.: 1992 Flusilazole Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Growth 570 Protection Agency Database (EEDB)) D.C.: 1992 Imidacloprid Pesticide Ecotoxicity Database Environmental Fate and Effects Rainbow U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Trout Growth 1200 Protection Agency Database (EEDB)) D.C.: 1992 Evidence for Links Between Feeding Inhibition, Population Daphnia Agatz,A., and C.D. Characteristics, and Sensitivity to Environ. Sci. Technol.47(16): magna Population 178 Brown Acute Toxicity for Daphnia magna 9461-9469 2013 Comparative Toxicity of Imidacloprid and Its Transformation Product 6- Chloronicotinic Acid to Non-Target Malev,O., R.S. Aquatic Organisms: Microalgae Klobucar, E. Fabbretti, Desmodesmus subspicatus and Pestic. Biochem. Physiol.104(3): Green Algae Population 127800 and P. Trebse Amphipod Gammarus fossarum 178-186 2012 Isoproturon Sensitivity, Variability, and Recovery of Functional and Structural Knauer,K., and U. Endpoints of an Aquatic Community Ecotoxicol. Environ. Saf.78:178- Green Algae Population 14 Hommen Exposed to Herbicides 183 2012 MCPA Daphnia Martins,J., M.L. Phototactic Behavior in Daphnia Ecotoxicol. Environ. Saf.67(3): magna Behavior 5625 Soares, M.L. Saker, L. magna Straus as an Indicator of 417-422 2007

S27

Appendix A – Temporal chemical and risk patterns

Compound name Species Effect LOEC Author Title Source Year [µg L-1] OlivaTeles, and V.M. Toxicants in the Aquatic Vasconcelos Environment Comparative Study of the Effects of MCPA, Butylate, Atrazine, and Caux,P.Y., L. Menard, Cyanazine on Selenastrum Green Algae Population 6 and R.A. Kent capricornutum Environ. Pollut.92(2): 219-225 1996 Mecoprop Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Reproduction 50800 Protection Agency Database (EEDB)) D.C.: 1992 Metolachlor Pesticide Ecotoxicity Database Environmental Fate and Effects Fathead U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Minnow Growth 56 Protection Agency Database (EEDB)) D.C.: 1992 Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Growth 707 Protection Agency Database (EEDB)) D.C.: 1992 Bioavailability and Toxicity of Agricultural Chemicals in Runoff Final report for IAG from MSEA Sites: Potential Impacts DW14935600-01-2. on Non-Target Aquatic Organisms: Environmental Research Fairchild,J., S. An Aquatic Hazard Assessment of Laboratory, U.S. Environmental Ruessler, M. Nelson, Four Herbicides Using Six Species Protection Agency, Duluth, Green Algae Population 75 and P. Haverland of Algae and Five Sp MN:123 p. 1994 Metoprolol Single and Combined Toxicity of Dietrich,S., F. Ploessl, Pharmaceuticals at Environmentally Daphnia F. Bracher, and C. Relevant Concentrations in Daphnia magna Growth 1.2 Laforsch magna - A Multigenerational Study Chemosphere79(1): 60-66 2010 Phenazone Phytotoxicity and Antioxidative Enzymes of Green Microalga (Desmodesmus subspicatus) and Bisewska,J., E.I. Duckweed (Lemna minor) Exposed J. Environ. Sci. Health Part B Sarnowska, and Z.H. to Herbicides MCPA, Chloridazon Pestic. Food Contam. Agric. Green Algae Population 1700 Tukaj and Their Mixtures Wastes47(8): 814-822 2012 Pirimicarb Pesticide Ecotoxicity Database Environmental Fate and Effects Fathead U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Minnow Growth 10000 Protection Agency Database (EEDB)) D.C.: 1992 Prochloraz Zhang,X., M. Hecker, Responses of the Medaka HPG Japanese P.D. Jones, J. Axis PCR Array and Reproduction Environ. Sci. Technol.42(17): Medaka Reproduction 30 Newsted, D. Au, R. to Prochloraz and Ketoconazole 6762-6769 2008

S28

Appendix A – Temporal chemical and risk patterns

Compound name Species Effect LOEC Author Title Source Year [µg L-1] Kong, R.S.S. Wu, and J.P. Giesy Propamocarb Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Growth 12300 Protection Agency Database (EEDB)) D.C.: 1992 Propiconazole Pesticide Ecotoxicity Database Environmental Fate and Effects Fathead U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Minnow Growth 210 Protection Agency Database (EEDB)) D.C.: 1992 Soetaert,A., L.N. Moens, K. Van der Ven, K. Van Leemput, Molecular Impact of Propiconazole Comp. Biochem. Physiol. C Daphnia B. Naudts, R. Blust, on Daphnia magna Using a Comp. Pharmacol. magna Growth 1000 and W.M. De Coen Reproduction-Related cDNA Array Toxicol.142(1-2): 66-76 2006 Comparative Sensitivity of Eight Freshwater Phytoplankton Species Ma,J., J. Chen, P. to Isoprocarb, Propargite, Pol. J. Environ. Stud.17(4): 525- Green Algae Population 100 Wang, and S. Tong Flumetralin and Propiconazol 529 2008 Propranolol The Toxicity Potential of Pharmaceuticals Found in the Madureira,T.V., M.J. Douro River Estuary (Portugal): Rocha, C. Cruzeiro, Assessing Impacts on Gonadal M.H. Galante, R.A.F. Maturation with a Histopathological Monteiro, and E. and Stereological Study of Zebrafish Danio rerio Reproduction 31.8 Rocha Ovary and Testis After Sub-Acu Aquat. Toxicol.105(3/4): 292-299 2011 Dzialowski,E.M., P.K. Physiological and Reproductive Daphnia Turner, and B.W. Effects of beta Adrenergic Receptor Arch. Environ. Contam. magna Reproduction 110 Brooks Antagonists in Daphnia magna Toxicol.50(4): 503-510 2006 Liu,Q.T., T.D. Williams, R.I. Comparative Aquatic Toxicity of Cumming, G. Holm, Propranolol and Its M.J. Hetheridge, and Photodegradaded Mixtures: Algae Environ. Toxicol. Chem.28(12): Green Algae Population 156 R. Murray-Smith and Rotifer Screening 2622-2631 2009 Pyraclostrobin Pesticide Ecotoxicity Database Environmental Fate and Effects Fathead U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Minnow Growth 8.37 Protection Agency Database (EEDB)) D.C.: 1992 Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Reproduction 8 Protection Agency Database (EEDB)) D.C.: 1992 Sertraline Fathead Valenti,T.W.,Jr., G.G. Human Therapeutic Plasma Levels Environ. Sci. Technol.46(4): Minnow Behavior 3 Gould, J.P. Berninger, of the Selective Serotonin Reuptake 2427-2435 2012

S29

Appendix A – Temporal chemical and risk patterns

Compound name Species Effect LOEC Author Title Source Year [µg L-1] K.A. Connors, N.B. Inhibitor (SSRI) Sertraline Decrease Keele, K.N. Prosser, Serotonin Reuptake Transporter and B.W. Brooks Binding and Shelter-Seeking Behavior in Adult Male Fathead Minnows Exposures to a Selective Serotonin Reuptake Inhibitor (SSRI), Lamichhane,K., S.N. Sertraline Hydrochloride, over Garcia, D.B. Huggett, Multiple Generations: Changes in D.L. DeAngelis, and Life History Traits in Ceriodaphnia Ecotoxicol. Environ. C.dubia Growth 4.8 T.W. La Point dubia Saf.101:124-130 2014 Aquatic Ecotoxicity of the Selective Minagh,E., R. Hernan, Serotonin Reuptake Inhibitor K. O'Rourke, F.M. Sertraline Hydrochloride in a Battery Ecotoxicol. Environ. Saf.72(2): Green Algae Population 75 Lyng, and M. Davoren of Freshwater Test Species 434-440 2009 Simazine Liu,Z.Z., Y.Y. Wang, Z.H. Zhu, E.L. Yang, Atrazine and Its Main Metabolites X.Y. Feng, Z.W. Fu, Alter the Locomotor Activity of Danio rerio Behavior 100 and Y.X. Jin Larval Zebrafish (Danio rerio) Chemosphere148:163-170 2016 Effects of the Herbicide Simazine Daphnia upon Production in a Two Member Ph.D.Thesis, Utah State pulex Reproduction 100 Carter,J.G. Aquatic Food Chain University, Logan, UT:213 p. 1981 Selective Real-Time Herbicide Monitoring by an Array Chip Podola,B., and M. Biosensor Employing Diverse Green Algae Population 2 Melkonian Microalgae J. Appl. Phycol.17(3): 261-271 2005 Spiroxamine Pesticide Ecotoxicity Database Environmental Fate and Effects U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Danio rerio Growth 6.1 Protection Agency Database (EEDB)) D.C.: 1992 Pesticide Ecotoxicity Database Environmental Fate and Effects Green 2.57 U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Algae Population (EC50) Protection Agency Database (EEDB)) D.C.: Sulfamethoxazole The Toxicity Potential of Pharmaceuticals Found in the Douro River Estuary (Portugal): Madureira,T.V., M.J. Evaluation of Impacts on Fish Liver, Rocha, C. Cruzeiro, I. by Histopathology, Stereology, Rodrigues, R.A.F. Vitellogenin and CYP1A Monteiro, and E. Immunohistochemistry, After Sub- Environ. Toxicol. Danio rerio Morphology 533 Rocha Acute Expo Pharmacol.34(1): 34-45 2012 Effects of Selected Pharmaceuticals Daphnia on Growth, Reproduction and Fresenius Environ. Bull.22(9): magna Reproduction 120 Lu,G., Z. Li, and J. Liu Feeding of Daphnia magna 2583-2589 2013

S30

Appendix A – Temporal chemical and risk patterns

Compound name Species Effect LOEC Author Title Source Year [µg L-1] Growth-Inhibiting Effects of 12 Yang,L.H., G.G. Ying, Antibacterial Agents and Their H.C. Su, J.L. Stauber, Mixtures on the Freshwater M.S. Adams, and M.T. Microalga Pseudokirchneriella Environ. Toxicol. Chem.27(5): Green Algae Population 800 Binet subcapitata 1201-1208 2008 Tebuconazole Pesticide Ecotoxicity Database Environmental Fate and Effects Rainbow U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Trout Growth 25 Protection Agency Database (EEDB)) D.C.: 1992 Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Reproduction 230 Protection Agency Database (EEDB)) D.C.: 1992 Terbuthylazine Perez,J., I. Domingues, M. Synergistic Effects Caused by Monteiro, A.M.V.M. Atrazine and Terbuthylazine on Soares, and S. Chlorpyrifos Toxicity to Early-Life Environ. Sci. Pollut. Res. Danio rerio Behavior 13000 Loureiro Stages of the Zebrafish Danio rerio Int.20(7): 4671-4680 2013 Growth Rate of Pseudokirchneriella subcapitata Exposed to Herbicides Perez,J., I. Found in Surface Waters in the Domingues, A.M.V.M. Alqueva Reservoir (Portugal): A Soares, and S. Bottom-up Approach Using Binary Green Algae Population 77 Loureiro Mixtures Ecotoxicology20(6): 1167-1175 2011 Terbutryn Gonzalez-Barreiro,O., Removal of Triazine Herbicides 6.03 C. Rioboo, C. Herrero, from Freshwater Systems Using Green Algae Population (LOEL) and A. Cid Photosynthetic Microorganisms Environ. Pollut.144(1): 266-271 2006 Thiabendazole Pesticide Ecotoxicity Database Environmental Fate and Effects Fathead U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Minnow Growth 230 Protection Agency Database (EEDB)) D.C.: 1992 Thiacloprid Pesticide Ecotoxicity Database Environmental Fate and Effects Rainbow U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, Trout Growth 1900 Protection Agency Database (EEDB)) D.C.: 1992 Pesticide Ecotoxicity Database Environmental Fate and Effects Daphnia U.S. Environmental (Formerly: Environmental Effects Division, U.S.EPA, Washington, magna Reproduction 1050 Protection Agency Database (EEDB)) D.C.: 1992 Tramadol Le,T.H., E.S. Lim, S.K. Toxicity Evaluation of Verapamil Daphnia Lee, J.S. Park, Y.H. and Tramadol Based on Toxicity magna Reproduction 8500 Kim, and J. Min Assay and Expression Patterns of Environ. Toxicol.26(5): 515-523 2011

S31

Appendix A – Temporal chemical and risk patterns

Compound name Species Effect LOEC Author Title Source Year [µg L-1] Dhb, Vtg, Arnt, CYP4, and CYP314 in Daphnia magna Trimethoprim The Toxicity Potential of Pharmaceuticals Found in the Madureira,T.V., M.J. Douro River Estuary (Portugal): Rocha, C. Cruzeiro, Assessing Impacts on Gonadal M.H. Galante, R.A.F. Maturation with a Histopathological Monteiro, and E. and Stereological Study of Zebrafish Danio rerio Reproduction 157 Rocha Ovary and Testis After Sub-Acu Aquat. Toxicol.105(3/4): 292-299 2011 De Liguoro,M., V. Di Leva, M. Dalla Bona, R. Merlanti, G. Daphnia Caporale, and G. Sublethal Effects of Trimethoprim Ecotoxicol. Environ. Saf.82:114- magna Reproduction 6250 Radaelli on Four Freshwater Organisms 121 2012 Growth-Inhibiting Effects of 12 Yang,L.H., G.G. Ying, Antibacterial Agents and Their H.C. Su, J.L. Stauber, Mixtures on the Freshwater M.S. Adams, and M.T. Microalga Pseudokirchneriella Environ. Toxicol. Chem.27(5): Green Algae Population 40000 Binet subcapitata 1201-1208 2008

S32

Appendix A – Temporal chemical and risk patterns

A3 Results

A3.1 Results of factor analysis of mixed data and k-means clustering with reduced data set from rain sewer samples In order to identify the influence of weather conditions and seasons on the pollutant concentrations in the rain sewer effluent, a factor analysis of mixed data (FAMD) was performed using the R package FactoMineR (function ‘FAMD’) (Le et al., 2008). The analysis was based on 21 compounds, which were detected in all samples. FAMD and k-means clustering confirmed the precipitation-related pattern dominating the seasonal influence. Rain samples were separated from dry samples (Figure S1). Again, the October sample was distinct from all other samples. The organic micropollutants clustered into two groups resembling rain discharge (group 1) and dry discharge (group 2) (Figure S2).

Figure S 1: Output of factor analysis for mixed data. Rain-weather samples (“R”) colored in blue were separated from dry-weather samples (“D”) colored in red

S33

Appendix A – Temporal chemical and risk patterns

Figure S 2: Emission groups of reduced chemical data set in rain sewer effluent based on loads [mg d-1] detected in each sample. Data was standardized prior to PCA and clustering; zeros treated by glog transformation. Ellipses represent 95% confidence interval. Full compound names and details are given in Table S11.

S34

Appendix A – Temporal chemical and risk patterns

A3.2 Loads in WWTP and rain sewer effluents

Table S 10: Loads [mg d-1] of micropollutants emitted from WWTP at different sampling days. WV = average within-week variation, BV = between-week variation. Emission groups according to Figure 5 (group 1= seasonal/random, group 2= constant, group 3= season-specific, group 4 = constant/random).

Compound Abbreviation 19.05. 20.05. 21.05. 22.05. 23.05. 24.05. 25.05. 26.05. 10,11-Dihydro-10,11-dihydroxycarbamazepine CMZ_DiOH 28570 27269 30907 49870 31101 27390 27830 29537 10,11-Dihydro-10-hydroxycarbamazepine CMZ_10OH 1870 1715 2103 2014 2751 1470 1095 1291 1H-Benzotriazole BZ 24190 48398 78776 63595 45366 26030 21427 19428 2-Aminobenzimidazole 2ABZ 169 149 208 0 174 181 176 181 2-Hydroxycarbamazepine CMZ_2OH 727 727 742 721 1154 668 642 786 2-Naphthalene sulfonic acid NSA 317 1117 1168 337 0 898 522 478 4+5-Methyl-1H-benzotriazole MBZ 16034 17102 19647 27662 20363 13498 12530 15337 Acesulfame ACE 188312 183508 207135 146043 127962 101602 108097 113607 Acetylsulfamethoxazole ASMZ 551 330 448 376 337 411 400 443 Ambroxol AMX 147 128 128 221 174 82 72 89 Amitriptyline ATP 645 593 607 295 582 228 265 582 Azoxystrobin AZ 47 32 30 33 20 25 49 27 Bentazon BTN 65 52 46 30 63 31 36 39 Bicalutamide BCA 1550 1380 1525 1298 1733 1130 1205 1624 Bisoprolol BSP 2457 2245 2758 2855 2890 2175 2097 2272 Boscalid BC 4353 6908 10344 6391 4132 4317 4123 4228 Carbamazepine CMZ 11994 10526 11994 12236 15194 10972 10120 12100 Carbendazim CBZ 234 219 317 415 281 261 260 235 Cetirizine CZ 3639 3611 3052 3073 1563 2381 2700 2855 Citalopram CP 1765 2283 3949 4091 5123 3138 2877 3401 Clomazone CL 6 8 10 0 0 0 0 0 DEET DEET 1852 1929 1558 1102 676 1139 1107 1572 Desethylterbutylazine TBA_DE 0 0 0 28 40 68 70 83 Desisopropylatrazine ATZ_DIP 0 0 0 0 0 0 0 0 Diazinon DZ 0 0 0 0 0 0 0 0 Diclofenac DCF 43972 52182 64945 42005 32590 29840 35965 33671 Difenoconazole DFC 0 0 0 0 0 0 0 0

S35

Appendix A – Temporal chemical and risk patterns

Compound Abbreviation 19.05. 20.05. 21.05. 22.05. 23.05. 24.05. 25.05. 26.05. Diflufenican DFF 33 31 95 54 57 20 17 38 Dimethachlor OA DMC_OA 0 0 0 0 0 0 0 0 Dimethoate DMT 13 17 59 96 48 46 31 20 Enalapril EP 0 0 0 0 0 0 0 0 Epoxiconazole EXC 3476 3641 3502 3258 2987 1949 2181 2020 Ethofumesat ETF 914 1312 1348 1028 967 737 745 694 Fenpropimorph FPP 29 30 32 23 22 10 8 11 Fenuron FE 172 160 177 227 203 160 167 186 Fipronil FIP 150 141 146 141 152 114 112 142 Fipronil desulfinyl FIP_Desulfinyl 4 0 3 4 0 3 3 4 Fipronil sulfide FIP_Sulfide 0 0 0 0 0 0 0 0 Fipronil sulfone FIP_Sulfone 36 31 46 47 38 33 27 35 Flufenacet FFC 56 50 39 27 27 12 9 7 Flurtamone FT 16 31 39 24 15 14 17 15 Flusilazole FZ 18 24 25 33 29 15 18 22 Gabapentin GP 20408 20787 27102 20245 24477 11351 14536 14503 Imidacloprid IDC 154 140 138 199 175 172 123 175 Isoproturon IP 0 65 100 76 74 57 50 0 Lenacil LC 137 189 288 205 193 146 141 135 Lidocaine LDC 884 865 1194 1252 1479 1109 1051 1221 Loperamide LPA 0 0 0 0 0 0 0 0 Losartan LOS 279 183 260 306 433 180 174 147 MCPA MCPA 26504 96709 191793 93144 60032 63161 62448 42383 Mecoprop MCP 82 81 107 71 93 40 62 51 Melperon MEL 10309 9295 10468 8586 14161 13254 9625 10546 Metamitron MET 2402 3786 4590 2617 3201 1899 1700 1818 Metazachlor ESA MZA_ESA 223 170 261 276 259 347 159 246 Metformin MF 8190 5575 7269 6585 3852 3403 3111 2639 Metolachlor MC 0 525 4856 3934 6106 7588 7661 6594 Metoprolol MTP 4422 4713 5080 17228 5059 4374 6194 1743 Metoprolol Acid MTP_A 6406 5568 4808 3504 3421 3045 2979 3988

S36

Appendix A – Temporal chemical and risk patterns

Compound Abbreviation 19.05. 20.05. 21.05. 22.05. 23.05. 24.05. 25.05. 26.05. Myclobutanil MY 28 60 68 11 0 15 21 70 N-Acetyl-4-aminoantipyrine NAA 3406 2766 3415 2228 3180 1934 1746 2244 N-cyclohexl-2-benzothiazole-sulfenamide NCBS 12281 11680 14199 8030 14839 12186 11845 11489 N-Formyl-4-aminoantipyrine NFA 37011 39592 59136 41052 46800 31916 38807 39025 Nitrendipin ND 0 0 0 0 0 0 0 0 Pethoxamid PX 0 0 0 44 103 94 112 54 Phenazone PHZ 1514 1266 1487 3659 1371 1211 1086 1221 Phenylbenzimidazole sulfonic acid PBI_SA 21192 20055 26470 16964 15578 14103 15763 17801 Pipamperone PPP 10616 16300 31163 41719 47672 23958 20004 17192 Pirimicarb PC 0 9 0 12 0 0 0 11 Prochloraz PZ 0 0 0 0 0 0 0 0 Propamocarb PPM 0 0 0 0 0 0 0 0 Propiconazole PPC 696 890 1349 1222 1298 3103 4097 3614 Propranolol PPN 623 671 781 736 783 518 511 711 Prothioconazole-desthio PTC_desthio 74 100 168 83 138 76 66 47 p-Toluenesulfonamide TSA 372 553 828 1510 3571 1014 360 272 Pyraclostrobin PYC 0 0 0 0 0 0 0 0 Quinmerac QM 230 198 193 83 92 147 239 173 Saccharin SAC 2695 4039 3808 1981 837 3190 2833 3285 Scopolamine-N-butyl SCA 0 0 0 0 0 0 0 0 Sertraline SER 0 0 0 0 0 0 0 0 Spiroxamine SPX 350 335 350 116 274 89 78 172 Sulfamethoxazole SMZ 1580 1238 1264 824 630 902 1178 1361 Tebuconazole TBC 197 196 346 303 370 582 704 735 Terbuthylazine TBA 0 192 1920 2024 2826 2813 3278 3068 Terbuthylazine-2-hydroxy TBA_OH 66 65 112 83 126 113 135 153 Terbutryn TBY 89 81 122 136 184 85 76 115 Thiabendazole TIB 93 94 98 77 104 59 82 77 Thiacloprid TC 38 39 71 51 25 31 40 36 Tramadol TRA 8263 7921 9634 13659 9177 8126 8520 9417 Trimethoprim TMP 914 865 915 1022 626 706 776 815

S37

Appendix A – Temporal chemical and risk patterns

Table S 10 continued

Abbreviation 27.7. 28.7. 29.7. 30.7. 31.7. 01.8. 02.8. 03.8. 05.10. 06.10. 07.10. 08.10. 09.10. 10.10. 11.10. 12.10.

CMZ_DiOH 24350 14911 14288 20735 22575 13541 16336 26559 14094 25505 25233 15821 17743 18261 19327 17846 CMZ_10OH 994 942 791 1227 994 682 689 1127 1523 1541 1766 1762 1589 1144 721 2145 BZ 25761 24854 20347 29965 28889 18697 12541 22510 37215 41943 31629 39423 35069 25877 32867 33307 2ABZ 166 152 116 188 188 132 118 187 147 173 149 182 185 127 109 174 CMZ_2OH 750 608 458 722 625 458 481 776 843 766 890 880 857 601 474 1038 NSA 0 0 0 0 328 0 313 306 0 0 0 0 0 0 0 0 MBZ 13571 8770 7100 10291 11094 7172 8084 11744 8725 13348 12300 11534 11032 14303 17059 11617 ACE 7780 5685 4793 8885 10083 6658 6546 11392 11161 11913 10958 16615 15002 11696 9600 17060 ASMZ 193 215 114 184 176 129 139 82 156 205 91 234 130 152 113 132 AMX 131 105 76 146 128 103 115 90 272 338 348 391 421 285 210 263 ATP 540 507 417 488 415 399 391 475 679 770 755 792 733 445 465 601 AZ 8 0 0 7 5 7 3 9 29 17 16 35 44 20 11 0 BTN 20 14 13 0 22 14 20 23 0 16 0 0 0 0 0 29 BCA 2213 1768 1409 2539 1829 1685 1724 2521 514 529 639 605 817 606 566 539 BSP 1389 1291 1254 1648 1221 1202 996 1053 1699 1452 1684 1914 1641 881 922 1829 BC 140 104 93 159 147 121 106 158 229 163 158 218 290 189 167 225 CMZ 12163 7278 9594 10526 11510 7170 6884 11678 11898 9621 8457 11644 7365 7705 12502 10110 CBZ 273 228 138 174 138 98 91 145 105 157 105 202 232 142 101 191 CZ 1252 1001 795 1216 967 770 846 1123 719 777 709 833 891 620 507 799 CP 1564 1415 1333 1645 1321 1447 1116 1435 2240 2274 2011 2679 2944 1787 1982 2328 CL 0 0 0 0 0 0 0 0 18 18 21 38 58 33 21 41 DEET 196 147 55 687 306 48 91 141 185 158 106 130 117 146 155 319 TBA_DE 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ATZ_DIP 0 0 0 0 0 0 0 0 0 34 57 0 29 26 22 36 DZ 8 7 5 6 5 4 5 5 47 0 78 85 80 50 4 5 DCF 25677 21350 17410 28100 28196 16833 16535 34019 29081 28496 33116 33262 38577 24866 24392 38409 DFC 14 8 0 11 7 6 6 7 27 25 34 35 30 26 0 0

S38

Appendix A – Temporal chemical and risk patterns

Abbreviation 27.7. 28.7. 29.7. 30.7. 31.7. 01.8. 02.8. 03.8. 05.10. 06.10. 07.10. 08.10. 09.10. 10.10. 11.10. 12.10. DFF 41 40 26 45 35 30 24 32 24 24 31 26 35 25 0 10 DMC_OA 0 0 0 0 0 0 0 0 261 311 326 398 443 289 243 400 DMT 0 0 0 0 0 0 0 0 33 0 0 0 31 0 0 0 EP 23 20 9 10 0 6 0 8 0 0 0 0 0 0 0 0 EXC 51 43 27 38 33 19 26 33 20 0 16 0 14 14 0 0 ETF 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 FPP 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 FE 88 83 55 77 67 46 36 66 51 67 55 93 102 69 51 106 FIP 162 126 114 211 155 167 131 181 40 43 49 51 57 40 37 32 FIP_Desulfinyl 4 3 0 3 4 3 0 5 0 0 0 3 0 0 0 3 FIP_Sulfide 8 6 4 9 6 6 6 7 7 7 6 11 9 8 9 7 FIP_Sulfone 69 56 38 85 61 80 64 78 47 55 70 85 84 51 59 51 FFC 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 FT 0 5 0 0 0 3 0 3 23 22 24 27 20 16 0 0 FZ 0 0 0 0 0 0 0 0 14 16 17 14 15 15 0 0 GP 14097 20907 41075 39649 16258 4961 3865 7423 22133 29218 26060 38286 30490 10371 12372 21230 IDC 64 61 0 87 65 0 0 93 129 140 0 119 124 0 88 234 IP 0 0 0 0 0 0 45 48 0 0 0 0 0 0 0 0 LC 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 LDC 1107 964 818 1000 984 797 752 937 1054 1126 1086 1493 1583 966 839 1321 LPA 31 22 20 24 19 21 17 18 189 196 145 212 192 100 94 115 LOS 111 73 364 1025 967 636 217 115 320 330 312 241 218 203 288 197 MCPA 338 123 73 119 70 163 875 1539 61 75 80 66 93 77 70 138 MCP 117 87 45 44 30 21 18 16 0 0 0 0 50 26 45 34 MEL 1028 1091 1943 4720 6076 5075 2726 3284 19042 13269 6213 8794 5855 2351 3027 3034 MET 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 MZA_ESA 218 170 190 226 190 123 196 206 84 55 78 173 0 143 93 220 MF 1853 1729 1595 1767 1357 1072 1207 1342 2753 3055 2793 4600 3320 2474 1952 3099 MC 58 49 26 49 43 33 42 51 0 0 0 0 0 0 0 0 MTP 8836 7541 8169 9250 7478 6456 4746 6059 6575 5951 7360 6617 5822 3695 3173 6950

S39

Appendix A – Temporal chemical and risk patterns

Abbreviation 27.7. 28.7. 29.7. 30.7. 31.7. 01.8. 02.8. 03.8. 05.10. 06.10. 07.10. 08.10. 09.10. 10.10. 11.10. 12.10. MTP_A 2927 2353 1548 2196 1937 1456 1287 2215 3251 3451 2885 4626 3695 2998 2204 3722 MY 20 21 10 25 40 14 16 21 36 38 46 38 50 41 16 13 NAA 7737 6167 5641 8555 5768 4544 4572 3786 11574 14044 10404 13575 9776 3267 4323 6804 NCBS 1082 1018 2043 5400 5139 4363 2949 2970 38148 29570 11411 16891 11369 1886 3433 3687 NFA 16505 11507 11991 17486 15291 11779 10517 13579 34053 39289 35370 34224 34089 11474 15559 24649 ND 0 0 0 0 0 0 0 0 340 278 693 693 653 344 40 474 PX 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 PHZ 684 556 373 573 470 392 432 547 523 743 562 1030 840 458 450 772 PBI_SA 12436 7518 6761 10296 9796 6005 7035 14038 9228 10320 9677 9181 10170 7369 5593 9530 PPP 25669 57994 43218 40588 26364 17296 9116 9548 2161 1759 1418 5045 6629 3478 4417 7490 PC 22 24 18 22 20 19 15 24 13 14 0 10 0 0 10 21 PZ 0 0 0 0 0 0 0 0 25 31 24 22 28 21 0 0 PPM 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 PPC 8565 6562 4222 4930 4562 3422 2537 3303 5256 10181 18225 15810 14563 10441 9012 9921 PPN 457 378 347 467 401 338 316 470 452 550 602 573 605 458 411 618 PTC_desthio 0 0 0 0 0 0 0 0 0 0 0 69 86 69 63 105 TSA 1193 629 795 560 581 240 462 397 0 0 0 1705 5509 3559 3320 1121 PYC 0 0 0 0 0 0 0 0 21 20 26 22 22 17 0 0 QM 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 SAC 1350 1064 778 980 1496 524 677 778 364 460 766 942 1469 807 206 640 SCA 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 SER 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 SPX 11 9 0 0 0 7 0 0 0 0 0 0 0 0 0 0 SMZ 466 343 279 363 289 273 339 449 1054 1163 825 842 975 707 671 895 TBC 2589 2225 1521 1867 1868 1239 1198 1365 1698 3361 6083 9803 8755 7426 8683 6401 TBA 92 80 40 62 76 51 46 68 0 0 0 0 0 0 0 0 TBA_OH 117 87 61 78 74 58 79 83 111 112 97 116 141 50 60 89 TBY 108 89 66 137 142 113 91 103 227 291 267 338 425 223 188 250 TIB 94 61 42 74 76 59 53 72 260 258 231 309 318 230 191 291 TC 25 18 43 58 32 16 10 15 6 5 0 0 0 0 0 0

S40

Appendix A – Temporal chemical and risk patterns

Abbreviation 27.7. 28.7. 29.7. 30.7. 31.7. 01.8. 02.8. 03.8. 05.10. 06.10. 07.10. 08.10. 09.10. 10.10. 11.10. 12.10. TRA 4706 3652 3474 4217 4708 3977 3847 5800 4669 6215 4518 6910 6990 3852 3496 6384 TMP 395 445 330 359 332 328 309 422 412 573 434 514 540 357 269 501

Table S 10 continued

Compound Abbreviation 01.2. 02.2. 03.2. 04.2. 05.2. 06.2. 07.2. 08.2. WW [%] BW [%] Emission group 10,11-Dihydro-10,11-dihydroxycarbamazepine CMZ_DiOH 16567 16052 15061 18121 11255 15476 12257 26314 25 37 2 10,11-Dihydro-10-hydroxycarbamazepine CMZ_10OH 3012 2668 2424 2608 2709 1904 1853 3708 25 44 2 1H-Benzotriazole BZ 20228 20273 22539 27383 18878 20796 15714 39984 31 46 2 2-Aminobenzimidazole 2ABZ 123 112 104 112 127 83 74 162 26 30 2 2-Hydroxycarbamazepine CMZ_2OH 680 622 546 608 634 518 557 833 20 23 2 2-Naphthalene sulfonic acid NSA 0 0 0 0 0 0 0 0 103 183 1 4+5-Methyl-1H-benzotriazole MBZ 9530 9878 10062 15667 11417 14578 11003 21983 26 34 2 Acesulfame ACE 19747 11563 11856 13380 13924 9533 9925 22917 28 138 3 Acetylsulfamethoxazole ASMZ 262 262 180 134 207 199 158 539 33 55 4 Ambroxol AMX 724 700 692 602 807 533 564 1045 26 82 3 Amitriptyline ATP 869 845 818 757 761 594 620 931 21 31 2 Azoxystrobin AZ 14 17 15 19 19 16 11 19 46 70 4 Bentazon BTN 0 0 0 0 0 0 0 0 91 119 1 Bicalutamide BCA 720 636 615 691 561 571 603 802 16 53 2 Bisoprolol BSP 2396 2243 1938 1958 1981 1768 1897 3013 19 32 2 Boscalid BC 105 103 131 144 141 107 95 166 25 173 3 Carbamazepine CMZ 7726 7682 8754 8312 6171 7524 6672 14116 21 24 2 Carbendazim CBZ 101 102 87 106 128 106 72 174 31 46 2 Cetirizine CZ 655 537 468 621 790 564 512 805 20 76 3 Citalopram CP 2779 2633 2965 3564 3239 2807 2451 8256 28 52 2 Clomazone CL 0 0 0 0 0 0 0 0 93 176 1 DEET DEET 0 0 0 0 1311 1536 648 451 75 111 1 Desethylterbutylazine TBA_DE 0 0 0 0 0 0 0 0 96 253 1

S41

Appendix A – Temporal chemical and risk patterns

Compound Abbreviation 01.2. 02.2. 03.2. 04.2. 05.2. 06.2. 07.2. 08.2. WW [%] BW [%] Emission group Desisopropylatrazine ATZ_DIP 41 51 34 43 34 30 0 57 61 131 1 Diazinon DZ 6 0 0 5 5 0 4 7 64 190 1 Diclofenac DCF 31046 27380 27477 28916 29955 22488 19395 44975 25 33 2 Difenoconazole DFC 0 0 0 0 0 0 0 0 59 156 1 Diflufenican DFF 65 48 54 55 53 44 44 64 37 48 4 Dimethachlor OA DMC_OA 0 0 0 0 0 0 0 0 22 181 3 Dimethoate DMT 0 0 0 0 0 12 13 27 135 162 1 Enalapril EP 48 23 24 16 17 12 1 0 87 164 1 Epoxiconazole EXC 0 0 0 0 0 0 0 0 55 179 1 Ethofumesat ETF 0 0 0 0 0 0 0 0 26 183 3 Fenpropimorph FPP 0 0 0 0 0 0 0 0 47 197 3 Fenuron FE 69 69 58 57 68 58 38 98 25 56 2 Fipronil FIP 45 49 43 47 35 35 50 52 16 58 2 Fipronil desulfinyl FIP_Desulfinyl 0 0 0 0 0 0 0 0 100 116 1 Fipronil sulfide FIP_Sulfide 0 8 10 6 0 5 5 6 37 74 4 Fipronil sulfone FIP_Sulfone 43 49 56 48 29 34 51 31 22 33 2 Flufenacet FFC 0 0 0 0 0 0 0 0 67 217 1 Flurtamone FT 0 0 0 0 0 0 0 0 84 119 1 Flusilazole FZ 0 0 0 0 0 0 0 0 44 124 3 Gabapentin GP 9185 8439 8556 9914 13606 8817 7146 17220 45 56 4 Imidacloprid IDC 124 102 73 0 94 126 78 96 55 63 4 Isoproturon IP 0 0 0 0 0 0 0 0 127 184 1 Lenacil LC 0 0 0 0 0 0 0 0 29 185 3 Lidocaine LDC 1359 1359 1285 1321 1569 1064 886 1442 17 21 2 Loperamide LPA 33 32 27 32 26 27 24 32 21 128 3 Losartan LOS 143 124 119 106 223 99 119 273 48 81 4 MCPA MCPA 71 0 39 0 0 0 0 0 104 212 1 Mecoprop MCP 110 73 65 36 35 0 0 0 80 79 4 Melperon MEL 3874 2939 2459 2067 1829 1685 1585 2147 46 77 4 Metamitron MET 0 0 0 0 0 0 0 0 38 190 3

S42

Appendix A – Temporal chemical and risk patterns

Compound Abbreviation 01.2. 02.2. 03.2. 04.2. 05.2. 06.2. 07.2. 08.2. WW [%] BW [%] Emission group Metazachlor ESA MZA_ESA 202 0 0 189 196 168 134 226 43 49 4 Metformin MF 3030 3107 2257 2538 2723 1994 2120 9549 40 64 4 Metolachlor MC 0 0 0 0 0 0 0 0 44 212 3 Metoprolol MTP 8019 7427 6566 6742 7131 5395 5318 10823 37 40 2 Metoprolol Acid MTP_A 3126 2411 2430 2422 2887 1899 2030 5464 30 40 2 Myclobutanil MY 0 0 0 0 0 0 0 0 54 94 4 N-Acetyl-4-aminoantipyrine NAA 3158 2781 2582 2978 3233 2124 2042 6129 35 66 4 N-cyclohexl-2-benzothiazole-sulfenamide NCBS 3429 2724 2298 2039 1717 1304 1326 1648 49 106 4 N-Formyl-4-aminoantipyrine NFA 15558 13545 14539 17355 19224 16109 13619 27633 25 52 2 Nitrendipin ND 2932 3364 3381 2385 2292 1794 1371 1144 45 154 3 Pethoxamid PX 0 0 0 0 0 0 0 0 94 251 1 Phenazone PHZ 749 922 579 720 1051 789 527 2232 41 70 4 Phenylbenzimidazole sulfonic acid PBI_SA 4525 5616 5314 6359 14821 4846 4442 7583 31 51 4 Pipamperone PPP 2946 2870 2316 2607 5787 8961 8504 9932 56 97 4 Pirimicarb PC 0 0 0 0 0 0 0 0 82 111 1 Prochloraz PZ 0 0 0 0 0 0 0 0 64 214 1 Propamocarb PPM 17 16 13 15 14 12 10 19 21 180 3 Propiconazole PPC 1299 5253 4930 4356 3566 2220 1765 2590 47 83 4 Propranolol PPN 750 557 515 679 713 0 533 1009 24 34 2 Prothioconazole-desthio PTC_desthio 0 0 0 0 0 0 0 0 65 136 1 p-Toluenesulfonamide TSA 716 2683 1193 277 410 254 0 4516 96 118 4 Pyraclostrobin PYC 0 0 0 0 0 0 0 0 64 214 1 Quinmerac QM 0 0 0 0 0 0 0 0 34 188 3 Saccharin SAC 794 758 300 519 397 0 273 1558 52 87 4 Scopolamine-N-butyl SCA 24 25 0 0 0 16 0 29 56 275 1 Sertraline SER 33 28 27 23 24 25 24 28 12 178 3 Spiroxamine SPX 0 0 0 0 0 0 0 0 98 200 1 Sulfamethoxazole SMZ 592 565 412 401 440 333 362 705 24 51 2 Tebuconazole TBC 345 1286 784 565 1153 885 622 1035 40 119 3 Terbuthylazine TBA 0 0 0 0 0 0 0 0 46 205 3

S43

Appendix A – Temporal chemical and risk patterns

Compound Abbreviation 01.2. 02.2. 03.2. 04.2. 05.2. 06.2. 07.2. 08.2. WW [%] BW [%] Emission group Terbuthylazine-2-hydroxy TBA_OH 27 20 23 20 20 14 12 37 30 53 4 Terbutryn TBY 34 43 48 54 109 73 57 117 33 66 4 Thiabendazole TIB 98 66 58 73 85 50 43 85 21 71 3 Thiacloprid TC 0 0 0 0 0 0 0 0 93 117 1 Tramadol TRA 4308 4129 3442 3841 4973 4054 2904 6513 22 42 2 Trimethoprim TMP 589 655 479 524 642 523 373 841 19 37 2

Table S 11: Loads [mg month-1] of micropollutants emitted from rain sewer at different sampling months and weather conditions. Emission groups according to Figure 7(group 1 = illicit connections, group 2 = surface runoff).

Compound Abbreviation 04/15 rain 07/15 rain 08/15 rain 09/15 rain 01/16 rain 05/16 rain 04/15 dry 07/15 dry 08/15 dry 10,11-Dihydro-10,11-dihydroxycarbamazepine CMZ_DiOH 0 0 0 0 44 0 616 654 1876 10,11-Dihydro-10-hydroxycarbamazepine CMZ_10OH 0 0 0 0 0 0 538 0 99 1H-Benzotriazole BZ 850 2473 2422 724 1753 1565 2795 1692 1640 2(4-morpholinyl)benzothiazole BTZ_morph 19 110 142 35 264 126 22 17 10 2-(Methylthio)benzothiazole BTZ_methylthio 250 4035 1362 231 2301 1098 175 373 155 2,4-Dinitrophenol DNP 0 0 20407 722 5321 23748 0 1174 541 2-Hydroxycarbamazepine CMZ_2OH 0 0 0 0 1.7 0 15 17 41 2-Octyl-4-isothiazolin-3-one OTZ 0 0 0 0 92.94 0 0 0 0 4+5-Methyl-1H-benzotriazole MBZ 407 1648 1636 379 658 816 183 218 392 Acesulfame ACE 332 1573 2846 603 593 701 8822 10368 19952 Acetylsulfamethoxazole ASMZ 0 0 0 0 734 0 0 358 59 Ambroxol AMX 7 0 0 0 7 0 0 17 29 Amitriptyline ATP 0 0 0 0 0 0 10 25 32 Atrazine ATZ 0 0 0 0 7.7 0 31 0 31 Bicalutamide BCA 0 0 0 0 0 0 4.7 3.2 3.4 Bisoprolol BSP 0 0 18 11 11 2.7 257 280 391 Carbamazepine CMZ 0 0 0 0 6.3 0 95 178 79 Carbendazim CBZ 133 987 1295 101 101 285 130 359 180 Cetirizine CZ 0 0 0 0 0 0 0 14 0

S44

Appendix A – Temporal chemical and risk patterns

Compound Abbreviation 04/15 rain 07/15 rain 08/15 rain 09/15 rain 01/16 rain 05/16 rain 04/15 dry 07/15 dry 08/15 dry Citalopram CP 0 0 0 0 0 0 119 65 183 Clomazone CL 0 0 173 8 19 9 0 0 0 Cotinine CTN 602 1183 1519 276 3714 1962 634 635 460 Cyclamate CYC 1152 3550 4016 449 2402 1347 14478 32249 10228 Desethylatrazine ATZ_DE 0 0 0 0 0 0 47 40 33 Desethylterbutylazine TBA_DE 10 698 96 2.7 0 2831 0 60 6.6 Desisopropylatrazine ATZ_DIP 0 0 0 0 0 37 0 0 0 Diclofenac DCF 0 0 0 54 0 0 130 577 1576 Dimethoate DMT 1.3 1053 154 2.8 94 219 0 20 0.2 Diuron DIU 36 340 1566 177 62 874 44 68 55 Enalapril EP 0 0 0 0 40 0 39 15 23 Epoxiconazole EXC 8.5 68 65 5.7 26 138 6.2 3.0 3.6 Fenpropimorph FPP 5.4 12 26 0 0 7.0 0 2.7 1.9 Fenuron FE 4.9 47 38 11 14 30 15 20 20 Fipronil FIP 8.0 34 44 6.6 3.1 5.8 11 10 7.2 Fipronil sulfone FIP_Sulfone 1.9 13 15 1.7 1.6 4.5 0 2.1 1.0 Flufenacet FFC 15 34 15 3.5 12 165 6.0 3.6 0 Gabapentin GP 0 0 0 0 0 0 6460 9743 6777 Isoproturon IP 143 34 2316 26 15 18 11 0 47 Lidocaine LDC 10 30 41 8 12 25 21 14 12 Losartan LOS 19 171 234 59 112 179 46 173 266 MCPA MCPA 254 475 1442 45 14 226 0 117 0 Mecoprop MCP 144 1094 1184 139 129 299 254 252 135 Melperon MEL 4.3 8.9 2.6 0.9 4.1 3.8 1.5 17 0.9 Metazachlor MZC 0 0 228 26 18 0 0 0 14 Metazachlor ESA MZC_ESA 0 0 0 0 0 0 0 0 0 Metformin MF 1492 3003 4229 5130 1550 1618 29989 53481 64169 Metoprolol MTP 49 196 296 6366 108 72 1476 1835 1531 Metoprolol_Acid MTP_A 147 330 528 314 2032 301 5371 8046 6854 N-Acetyl-4-aminoantipyrine NAA 0 69 110 112 0 0 5233 0 0

S45

Appendix A – Temporal chemical and risk patterns

Compound Abbreviation 04/15 rain 07/15 rain 08/15 rain 09/15 rain 01/16 rain 05/16 rain 04/15 dry 07/15 dry 08/15 dry N-cyclohexyl-2-benzothiazole-amine NCBS 30 157 234 31 617 179 13 15 6.5 N-Formyl-4-aminoantipyrine NFA 0 0 0 33 72 0 1313 1102 913 Phenazone PHZ 0 0 0 0 0 0 32 45 56 Phenylbenzimidazole sulfonic acid PBI_SA 22 205 469 134 229 1642 567 1378 1501 Propamocarb PPM 3.8 148 94 5.5 10 79 0 13 6.4 Propiconazole PPC 43 542 0 0 396 353 32 44 0 Propranolol PPN 0 0 0 0 0 0 0 0 0 Prothioconazole-desthio PTC_desthio 88 252 0 0 7 695 0 40 0 p-Toluenesulfonamide TSA 383 2760 3527 830 530 5259 0 481 212 Simazine SIZ 0 0 0 0 0 0 94 49 59 Sulfamethoxazole SMZ 0 0 0 0 136 0 0 0 0 Tebuconazole TBC 33 238 152 14 12 111 26 28 14 Terbuthylazine-2-hydroxy TBA_OH 46 640 735 95 306 510 28 113 125 Terbutryn TBY 123 556 4028 526 385 668 47 81 146 Terbutylazine TBA 0 228 64 11 3.7 1900 0.00 17 0 Thiabendazole TIB 8.8 48 117 15 89 54 5.6 12 6.0 Tramadol TRA 3.0 9.4 51 14 10 14 115 415 238 Trimethoprim TMP 0 0 0 0 31 0 0 94 81

Table S 11 continued

Compound Abbreviation 09/15 10/15 11/15 12/15 01/16 02/16 03/16 04/16 05/16 Emission dry dry dry dry dry dry dry dry dry group 10,11-Dihydro-10,11- CMZ_DiOH 556 8759 1120 315 707 1416 590 381 369 1 dihydroxycarbamazepine 10,11-Dihydro-10- CMZ_10OH 273 0 420 60 0 334 124 146 0 1 hydroxycarbamazepine 1H-Benzotriazole BZ 1794 10945 2472 458 1507 972 386 526 1277 1 2(4-morpholinyl)benzothiazole BTZ_morph 14 21 27 14 77 37 21 22 15 2 2-(Methylthio)benzothiazole BTZ_ 194 2264 687 338 992 599 626 602 666 2 methylthio 2,4-Dinitrophenol DNP 1466 5255 2228 0 880 948 585 2343 1164 2

S46

Appendix A – Temporal chemical and risk patterns

Compound Abbreviation 09/15 10/15 11/15 12/15 01/16 02/16 03/16 04/16 05/16 Emission dry dry dry dry dry dry dry dry dry group 2-Hydroxycarbamazepine CMZ_2OH 25 157 41 12 24 34 22 9 9 1 2-Octyl-4-isothiazolin-3-one OTZ 0 0 22 6 76 22 17 0 0 1 4+5-Methyl-1H-benzotriazole MBZ 266 631 490 71 542 394 90 221 319 2 Acesulfame ACE 15980 82226 19904 5046 11137 9812 9337 8369 12812 1 Acetylsulfamethoxazole ASMZ 243 3570 77 1106 591 0 73 74 68 1 Ambroxol AMX 80 950 249 12 44 32 26 32 87 1 Amitriptyline ATP 33 154 39 9.6 9.4 5.3 6.5 13 14 1 Atrazine ATZ 24 27 36 8.1 24.2 19 12 14 18 1 Bicalutamide BCA 11 43 0.7 0.9 0.3 39 0 5.5 11 1 Bisoprolol BSP 341 2356 409 190 155 138 167 175 487 1 Carbamazepine CMZ 181 203 122 29 163 195 36 25 23 1 Carbendazim CBZ 78 228 260 44 82 48 15 20 44 2 Cetirizine CZ 0 230 0 40 33 121 55 71 326 1 Citalopram CP 169 206 97 42 45 28 56 39 37 1 Clomazone CL 9.7 9.1 0.0 2.9 0.0 0 0 0 0 2 Cotinine CTN 648 335 960 826 1606 1004 1386 150 1911 2 Cyclamate CYC 15221 133403 11877 8626 26297 17950 21588 8312 4104 1 Desethylatrazine ATZ_DE 28 0 37 14 47 28 23 21 20 1 Desethylterbutylazine TBA_DE 5.2 9.3 0 0 0 0 0 0 62 2 Desisopropylatrazine ATZ_DIP 0 0 0 9.1 14 21 12 7.9 10 1 Diclofenac DCF 644 3312 906 212 925 862 337.7 194 665 1 Dimethoate DMT 1.3 696 20149 39 85 25 9.7 17 14 2 Diuron DIU 100 158 118 6.5 112 68 9.1 41 34 2 Enalapril EP 40 749 45 66 115 81 114 60 87 1 Epoxiconazole EXC 3.3 5.8 2.7 6.6 9.3 0.0 0.0 9.3 21.4 2 Fenpropimorph FPP 0 3.6 3.6 0.0 0.0 0.0 0.0 0.0 0.0 2 Fenuron FE 17 55 20 7.7 23.0 12.4 7.7 8.4 16.1 1 Fipronil FIP 7.7 14 12 1.4 2.3 0.4 0.0 0.6 1.2 2 Fipronil sulfone FIP_Sulfone 1.2 1.8 2.4 0.9 1.3 0.3 0.0 0.0 0.2 2

S47

Appendix A – Temporal chemical and risk patterns

Compound Abbreviation 09/15 10/15 11/15 12/15 01/16 02/16 03/16 04/16 05/16 Emission dry dry dry dry dry dry dry dry dry group Flufenacet FFC 0 69 68 2.1 5.6 2.9 0.8 1.2 8.2 2 Gabapentin GP 11012 121115 13969 6895 16686 6232 13615 13754 30114 1 Isoproturon IP 21 50 75 3.3 14.3 9.9 2.0 5.6 4.3 2 Lidocaine LDC 24 110 24 28 27 53 20 36 15 1 Losartan LOS 515 2936 333 65 97 94 80 72 341 1 MCPA MCPA 0 69 30 20 0 46 0 13 68 2 Mecoprop MCP 83 141 178 30 366 190 48 47 101 2 Melperon MEL 2.9 9.8 6.9 4.1 15 4.5 2.0 2.1 2.5 2 Metazachlor MZC 31 19 0.0 2.9 0 0 0 0 0 2 Metazachlor ESA MZC_ESA 0 127 77 9.2 0 107 0 28 32 1 Metformin MF 72230 369946 83845 31110 19814 17892 28687 32296 52609 1 Metoprolol MTP 1971 14201 2123 1074 874 863 686 820 1518 1 Metoprolol_Acid MTP_A 7113 75965 10298 12936 10042 6894 11825 8972 15602 1 N-Acetyl-4-aminoantipyrine NAA 0 0 0 4328 9568 4707 8253 3358 10804 1 N-cyclohexyl-2-benzothiazole-amine NCBS 17 31 33 16 134 87 39 31 21 2 N-Formyl-4-aminoantipyrine NFA 2491 15581 2287 1362 1623 2953 2554 993 3589 1 Phenazone PHZ 109 738 108 44 77 65 52 29 64 1 Phenylbenzimidazole sulfonic acid PBI_SA 727 9696 1578 508 214 567 342 441 1722 1 Propamocarb PPM 9.7 108 14 12 17 9.6 5.8 5.6 7.7 2 Propiconazole PPC 0 0 69 32 124 95 41 51 63 2 Propranolol PPN 27 14 24 13 17 15 9 17 30 1 Prothioconazole-desthio PTC_desthio 0 0 0 1.8 0 0 0 7.1 37 2 p-Toluenesulfonamide TSA 401 841 301 53 154 165 81 89 102 2 Simazine SIZ 78 0 43 21 60 63 48 23 33 1 Sulfamethoxazole SMZ 0 1184 48 129 144 22 0 0 0 1 Tebuconazole TBC 15 59 26 3.1 9.0 6.4 2.0 9.4 12 2 Terbuthylazine-2-hydroxy TBA_OH 125 168 105 38 140 108 83 110 196 2 Terbutryn TBY 389 857 598 36 477 304 66 113 88 2 Terbutylazine TBA 0 0 0 0.2 0 0 0 0 35 2

S48

Appendix A – Temporal chemical and risk patterns

Compound Abbreviation 09/15 10/15 11/15 12/15 01/16 02/16 03/16 04/16 05/16 Emission dry dry dry dry dry dry dry dry dry group Thiabendazole TIB 15 52 27 66 72 154 31 15 21 2 Tramadol TRA 371 427 453 76 58 185 81 72 116 1 Trimethoprim TMP 40 1327 198 164 63 19 71 34 21 1

A3.3 Concentration [ng L-1] of risk driving compounds in LVSPE samples

Table S 12: Concentrations [ng L-1] of 14 main risk driving compounds in LVSPE samples taken during the sampling period April-November 2015

April May June July August October/November Amitriptyline 6.3 7.3 6.1 17.0 31.7 26.5 Carbendazim 4.1 2.2 4.5 10.2 18.8 10.9 Citalopram 17.3 20.5 84.8 94.7 91.8 89.9 Diazinon 0 0 0 0 0 0 Diclofenac 248.5 252.6 427.9 623.3 1240.4 931.1 Dimethoate 0 0 1.0 1.2 1.1 18.5 Diuron 0 0 0 2.7 1.6 0 Fipronil 0.6 0.7 1.7 3.0 6.0 2.5 MCPA 8.4 8.5 223.2 51.7 19.6 5.4 Metolachlor 0 2.0 49.0 22.2 1.9 0.7 N-Formyl-4-aminoantipyrine 315.2 259.6 513.8 745.8 1110.0 711.5 Spiroxamine 1.0 4.3 1.6 2.2 0 0 Terbuthylazine 2.4 1.0 6.6 24.4 5.3 0 Terbutryn 1.1 1.3 3.9 5.5 6.8 8.0

S49

Appendix A – Temporal chemical and risk patterns

References – Appendix A Busch, W., Schmidt, S., Kühne, R., Schulze, T., Krauss, M. and Altenburger, R. (2016) Micropollutants in European rivers: A mode of action survey to support the development of effect- based tools for water monitoring. Environmental Toxicology and Chemistry 35(8), 1887-1899. 10.1002/etc.3460 Le, S., Josse, J. and Husson, F. (2008) FactoMineR: An R Package for Multivariate Analysis. Journal of Statistical Software 25(1), 1-18.

Appendix B – Spatial chemical patterns

Appendix B - Unraveling longitudinal pollution patterns of organic micropollutants in a small river by nontarget screening and cluster analysis

B1 Physicochemical parameters of sampling spots Physicochemical parameters of the sampling spots were recorded at the time of sampling. The discharges of official gauges were provided for the sampling day by the State Office for Flood Protection and Water Management Saxony-Anhalt (LHW, 2019). The discharge increased from 0.041 m³ s-1 in the pristine headwaters to 0.34 m³ s-1 at mouth of the river.

Table S 13: Physicochemical parameters of sampling spots. MOBICOS = mobile aquatic mesocosms

Sampling spot pH Electric conductivity O2 Temperature [µS cm-1] [°C] 1 3 MOBICOS Wernigerode 7.8 106 10.1 10.0 2 9 UPS RW drainage 7.9 540 9.8 12.6 3 11 UPS Barrenbach 7.8 596 9.4 12.3 4 13 UPS Silstedter Bach 7.9 682 9.3 12.0 5 15 UPS WWTP Silstedt 8.0 722 9.7 11.9 6 17 DS WWTP Silstedt 7.8 933 8.8 13.6 7 18 UPS Derenburg 7.9 890 9.3 13.1 8 21 DS Derenburg 7.7 898 8.7 12.9 9 22 Pegel Mahndorf 8.0 927 9.8 12.9 10 26 UPS RW Drainage 8.0 1006 10.1 13.0 11 31 DS WWTP Halberstadt 7.8 1114 9.8 14.1 12 34 UPS Asse 8.3 1082 11.1 14.8 13 36 a UPS Weir Gr. 8.3 1102 10.7 15.0 Quenstedt 14 38 MOBICOS Nienhagen 8.9 1079 11.5 15.4 15 40 UPS Salzgraben 8.5 1109 11.0 15.5 16 42 Mouth 8.6 1102 10.5 14.0

S51

Appendix B – Spatial chemical patterns

B2 Material and Methods

B2.1 Chemicals and Reagents Chemicals and reagents are listed in Table S14. Standards for candidate confirmation not listed here were provided by partner institutes.

Table S 14: Chemicals and reagents used for LC-HRMS analysis and standards for identified compounds

Chemicals and Reagents Supplier Methanol (LCMS grade) Honeywell Water (LCMS grade) Fisher Chemical Acetone (LCMS grade) Honeywell Isopropanol (LCMS grade) Merck Formic Acid (99 %) Honeywell Ammonium formate Sigma-Aldrich Ammonium acetate Sigma-Aldrich Ammonium bicarbonate Sigma-Aldrich Ammonia solution Sigma-Aldrich Deuterium oxide (99.9 atom %) Sigma-Aldrich Methan-d1-ol (99.8 atom %) Sigma-Aldrich Azelaic acid Carbolution /AKSci Embelin Carbolution /AKSci Methocarbamol Geyer/J&K Triacetin Geyer/J&K Triethylene glycol monomethyl ether Carbolution/AKSci N-lauroylethanolamine (N-(2-HYDROXYETHYL)DODECANAMIDE) Geyer/Chempur

B2.2 LC-HRMS analysis Chromatographic separation was performed on a Kinetex 2.6 μm EVO C18 (50x2.1 mm) column equipped with a pre-column (C18 EVO 5x2.1 mm) and an inline filter. The column temperature was 40°C. The same LC solvent gradient was applied for all fullscan and data dependent-MS² experiments (Table S15). Ionization was performed with a HESI source.

S52

Appendix B – Spatial chemical patterns

Table S 15: Solvent gradient for LC analysis

Time [min] Flow rate [mL] Solvent A [%] Solvent B [%] Solvent C [%] Water + 0.1% MeOH + 0.1% Acetone/ formic acid formic acid Isopropanol (50:50) 0 0.3 95 5 0 1 0.3 95 5 0 13 0.3 0 100 0 24 0.3 0 100 0 24.1 0.35 5 10 85 26.2 0.35 5 10 85 26.3 0.35 95 5 0 31.9 0.35 95 5 0 32.0 0.3 95 5 0 *MeOH = methanol

Parameters and setting for the fullscan experiments and the dd-MS² experiments for positive and negative ionization mode are listed in Table S16. For dd-MS² experiments an inclusion list of the prioritized masses was provided for ionization modes. The nominal resolving power in dd-MS2 experiments was 70,000 (referenced to m/z 200) in fullscan mode and 35,000 (referenced to m/z 200) in data-dependent-MS2 scans.

Table S 16: Parameters and setting of Q ExactiveTM Plus (Thermo Fisher) for fullscan experiments

Parameter Positive mode Negative mode Sheath gas flow rate 45 25 Aux gas flow rate 1 1 Spray voltage [kV] 3.8 3.5 Capillary temperature [°C] 300 300 S-lens RF level 70 70 Aux gas heater temperature 300 280 [°C] Scan range 100-1500 m/z 100-1500 m/z Resolution (referenced to 200 140,000 140,000 m/z)

S53

Appendix B – Spatial chemical patterns

B2.3 MZmine parameters The settings of MZmine parameters are listed in Table S17. The peak list was annotated with 372 target compounds in positive and 106 compounds in negative ionization mode. The target compounds supported the interpretation of the identified pollution patterns by cluster analysis.

Table S 17: Settings of MZmine parameters

Step Parameter Setting Mass detection Mass detector Centroid Noise level 5e3 ADAP chromatogram building Min group size of # of scans 8 Group intensity threshold 1e4 Min highest intensity 5e3 m/z tolerance 0.001 m/z or 7 ppm Smoothing Filter width 7 Chromatogram deconvolution Algorithm Local minimum search Chromatographic threshold 60 % Search minimum in RT range 0.10 min Minimum relative height 30 % Minimum absolute height 5e4 Min ration of peak top/edge 2.3 Peak duration range 0.1 – 5 min Join aligner m/z tolerance 0.001 m/z or 7 ppm Weight for m/z 70 Retention time tolerance 0.3 (absolute) min Weight for RT 30 Custom database search m/z tolerance 0.001 m/z or 7 ppm Retention time tolerance 0.4 (absolute) min Gap filling Intensity tolerance 30 % m/z tolerance 0.001 m/z or 7 ppm Retention time tolerance 0.15 (absolute) min RT correction yes

S54

Appendix B – Spatial chemical patterns

B2.4 Settings of R ‘nontarget’ package

Table S 18: Settings of parameters in R ‘nontarget’ package

Step Parameter ESIpos ESIneg Pattern search Used "13C","37Cl","81Br","15N","34 "13C","37Cl","81Br","15N","34S","1 isotopes S","13C","37Cl","81Br","15N"," 3C","37Cl","81Br","15N","34S" 34S" Cutint 50000 50000 rttol [min] ±0.05 ±0.05 mztol[ppm] 2 2 Mzfrac 0.5 0.3 Inttol 0.1 0.15 Adduct search Used "M+H", "M+Na", "M+K", "M-H","M+Cl","M+FA-H", "M-2H" adducts "M+NH4", "M+CH3OH+H", "M+2H", "M+H+NH4", "M+2Na", "2M+Na", "2M+NH4" rttol [min] 0.08 0.05 mztol [ppm] 2 2 Homolgue series search Used 14.0157, 30.0106, 44.0262, 14.0157, 30.0106, 44.0262, homologue 58.0419, 74.0188, 49.9968, 58.0419, 74.0188, 49.9968, series 7.0079, 15.0053, 22.0131, 7.0079, 15.0053, 22.0131, 29.0210, 37.0094, 24.9984 29.0210, 37.0094, 24.9984 elements C","H","O","","F" C","H","O","","F" minmz [u] 7 7 maxmz [u] 80 80 minrt [min] 0 0 maxrt [min] 3 3 mztol [ppm] 2.5 2.5 rttol [min] 0.2 0.2 minlength 4 4 R2 0.96 0.96 Spar 0.5 0.5

S55

Appendix B – Spatial chemical patterns

B2.5 Script of cluster analysis by R ‘kml’ package The script for cluster analysis using the R ‘kml’ package is provided below. Prior to cluster analysis, the data matrix containing all peaks and river samples (in correct spatial order) is transformed into a longitudinal object. The cluster analysis is performed for two to six clusters. Details and explanations on further parameters are described elsewhere (Genolini et al., 2015). The customized distance function ‘diss.CORT’ function from the R ‘TSclust’ package was used (Montero and Vilar, 2014). The appropriate number of cluster is selected by plotting the results for all quality criteria for the different numbers of clusters using the “plotAllCriterion” command. Afterwards, the cluster assignment for each peak can be extracted and as well as the probabilities for each peak of belonging to the different clusters.

#LongData object with normalized data event_long <- clusterLongData(event_sc1, varNames = "Peak Height")

#perform kml analysis kml(event_long, nbClusters = 2:6, nbRedrawing = 10,toPlot= "none", parAlgo = parALGO(saveFreq= 100,startingCond= "kmeans-", distanceName = "ignore", distance = diss.CORT))

#plot all criteria to choose cluster number plotAllCriterion(event_long)

#plot overview results for selected number of clusters plot(event_long,3)

#extract cluster members cluster_name <-getClusters(event_long,3) cluster_name2 <- as.data.frame(cluster_name)

#get probabilities for cluster fitting/ cluster evaluation cluster_probability <-getBestPostProba(event_long, nbCluster =3, clusterRank = 1) cluster_probability2 <- as.data.frame(cluster_probability)

## get cluster probabilities for all cluster cluster_allprobs2 <- event_long@c3[[1]]@postProba

S56

Appendix B – Spatial chemical patterns

B2.6 Structure elucidation

B2.6.1 Determination of molecular formulas in XCalibur QualBrowser The elements C, H, O, N, P, I, F were included as well as S, Cl and Br if suggested by isotopic pattern in the mass spectrum. The number of C, N, S atoms was adjusted based on intensity giving a tolerance of ± 4 atoms. The number of P and I were restricted to 1 atom and for F to 3 atoms. Overall, a mass tolerance between 5 ppm and 7 ppm (in case of few candidates) was used. The RDB equivalents were set down to -2.5 in presence of [M+NH4]+ adducts.

B2.6.2 MetFrag settings The MetFrag (Ruttkies et al., 2016) v.2.3 web tool was used for retrieval of candidate structures based on MS² fragment lists. ChemSpider (Royal Society of Chemistry, 2015) was used as a database. The fragment peak match relative deviation was set to 5 ppm. The absolute deviation was set to 0.001 u. The maximum tree depth was 2. For further ranking of candidates the spectral similarity score term was used. Furthermore, the ChemSpider (Royal Society of Chemistry, 2015) database score terms were selected, which ranked the candidates according to hits in different data reference counts.

B2.6.3 Hydrogen-deuterium exchange Hydrogen-deuterium exchange (HDX) experiments were performed in order to identify the number of exchangeable hydrogen atoms, i.e., hydrogen atoms attached to a heteroatom. The LC gradient program for the HDX experiments was according to Table S15. The eluents were exchanged for deuterium oxide and methan-d1-ol both containing 0.1% formic acid. The resolution was 70,000 referenced to m/z 200. The resulting spectra were evaluated with Xcalibur QualBrowser for m/z shifts resulting from an exchange of a hydrogen atom by a deuterium atom (i.e., 1.0063 m/z).

B2.6.4 pH-dependent LC retention The retention time of ionizable compounds depends on their charge, which varies with the pH of the eluent. The unknown compounds were analyzed at two additional pH values of the LC eluent (6.4 and 10). Shifts in retention time with different pH thus indicated certain functional groups. The gradient program was according to Table S15. Water and methanol with 2.5 mM ammonium acetate were used as eluents in the pH 6.4 method. The pH 10 method involved the eluents water and methanol with 2.5 mM ammonium bicarbonate. The desired pH in both methods was adjusted using ammonia solution in water and methanol. The settings of the Q Exactive Plus instrument

S57

Appendix B – Spatial chemical patterns

(Thermo Scientific) were according to Table S16. The resulting spectra were evaluated with Xcalibur QualBrowser for respective retention time shifts.

B3 Results

B3.1 Performance of replicates analysis The performance (i.e., repeatability) of the detection of peaks by chemical analysis and peak picking was evaluated by performing replicate analysis of selected samples from different spots in the river (i.e., Holt13, Holt22 and Holt38) as well as Holt17 (Figure S3A-D). Overall, a detection frequency greater than 80% was observed for 91-96% of the peaks. For the target compounds, 97% of peaks were detected in all replicates. Peaks with single detections were mostly observed at intensities close to the cutoff values and at retention times less than 1 minute. These peaks were also not removed by blank correction and likely resulted from difficult peak shapes that were not always detected by the peak picking algorithm. Still, these peaks only represented about 4% of the total number of peaks. The cluster analysis is robust enough to not be disturbed by this amount of false positives (section 3.3.1.2).

S58

Appendix B – Spatial chemical patterns

FigureFigure SS 326: Detection: Detection frequency frequency of of peaks peaks in in replicates replicates of of sample sample A) A) Holt13, Holt13, B) D B) Holt22, C) Holt38 andHolt22, D) Holt17 C) Holt38 and D) Holt17. S59

Appendix B – Spatial chemical patterns

B3.2 Effect of normalization of peak heights by internal standards on main clusters The peak heights were normalized with peak heights of internal standards. This normalization did not change the main patterns but led to shift individual peaks into different patterns (Figure S4). Thus, for most peaks the variations among the samples were higher than the variation of internal standards (i.e., matrix effects).

Figure S 4: Results of cluster analysis on internal standard normalized data (left), and not-normalized data (right).

B3.3 Quality criteria cluster analysis One challenge of partitioning cluster analysis is that the number of clusters has to be pre-selected by the user. The ‘kml’ package offers a number of quality criteria which support the identification of the optimal number of clusters. Thus, the number of clusters at which the most quality criteria are maximized, is the preferred number of clusters (Genolini et al., 2015). Maximized quality criteria indicate that the within-cluster variation is minimized and the between-cluster variation is maximized. If several numbers of clusters were possible candidates based on the quality criteria, the cluster analysis was performed with different number of clusters. The results were checked for plausibility using annotated target compounds. Overall, the lowest reasonable number of clusters was targeted to identify general pollution patterns and main pollution sources.

S60

Appendix B – Spatial chemical patterns

B3.3.1 Quality criteria of cluster analysis on whole data set The quality criteria for the cluster analysis on the whole data set suggested three clusters. However, the performance of two and four clusters was also tested for plausibility.

Figure S 5: Quality criteria of cluster analysis on whole data set. Line numbers refer to different quality criteria. 1) Calinski-Harabatz, 2) Calinski-Harabatz2, 3) Calinski -Harabatz3, 4) Ray-Turi, 5) Davies-Bouldin

S61

Appendix B – Spatial chemical patterns

B3.3.2 Quality criteria of cluster analysis on subclusters a. WW subpattern

The quality criteria for the cluster analysis on the main wastewater (“WW”) pattern suggested four clusters.

Figure S 6: Quality criteria of cluster analysis on WW subpatterns. Line numbers refer to different quality criteria. 1) Calinski-Harabatz, 2) Calinski-Harabatz2, 3) Calinski -Harabatz3, 4) Ray-Turi, 5) Davies-Bouldin.

S62

Appendix B – Spatial chemical patterns

b.) BR subpattern

The quality criteria for the cluster analysis on the main Bode River (“BR”) pattern suggested two to four clusters. These numbers of clusters were tested for plausibility.

Figure S 7: Quality criteria of cluster analysis on BR subpatterns. Line numbers refer to different quality criteria. 1) Calinski-Harabatz, 2) Calinski-Harabatz2, 3) Calinski -Harabatz3, 4) Ray-Turi, 5) Davies-Bouldin.

S63

Appendix B – Spatial chemical patterns

c.) DRI subpattern

The quality criteria for the cluster analysis on the diffuse and random input (“DRI”) main pattern suggested more than four clusters.

Figure S 8: Quality criteria of cluster analysis on DRI subpatterns. Line numbers refer to different quality criteria. 1) Calinski-Harabatz, 2) Calinski-Harabatz2, 3) Calinski-Harabatz3, 4) Ray-Turi, 5) Davies-Bouldin.

S64

Appendix B – Spatial chemical patterns

B3.4 Identification of ions of interests

B3.4.1 Intensities of target and prioritized unknown compounds in river samples

Table S 19: Intensities of target and prioritized unknown compounds in river samples belonging to main WW pattern

m/z rt Compound Name Use Pattern Sub- Holt3 Holt9 Holt11 Holt13 Holt15 probability pattern 271.1075 8.3 10,11-Dihydro-10,11- Target pharmaceutical TP 0.9997 WW1 0 0 0 0 0 dihydroxycarbamazepine of carbamazepine 255.1127 8.4 10,11-Dihydro-10- Target pharmaceutical TP 0.8310 WW1 0 0 0 0 0 hydroxycarbamazepine of carbamazepine 120.0557 4.0 1H-Benzotriazole Target industrial compound 0.9998 WW1 0 0 6.0E+04 0 0 221.0743 9.7 2(4-morpholinyl)benzothiazole Target industrial compound 1.0000 WW2 0 0 0 0 0 218.9628 10.6 2-4-Dichlorophenoxyacetic acid Target herbicide 0.9996 WW2 0 0 0 0 0 134.0712 1.2 2-Aminobenzimidazole Target fungicide TP 1.0000 WW2 0 0 0 0 0 of carbendazim 253.097 8.7 2-Hydroxycarbamazepine Target pharmaceutical TP 0.9946 WW1 0 0 0 0 0 of carbamazepine 195.9131 10.3 3-5-6-Trichloro-2-pyridinol Target insecticide TP of 1.0000 WW3 0 0 0 0 0 chlorpyrifos and chlorpyrifos-methyl 123.0916 0.9 4-(Dimethylamino)pyridine Target industrial compound 0.9824 WW1 0 0 0 0 0 151.0866 0.9 4'-Aminoacetanilide Target industrial compound 0.9994 WW1 0 0 0 0 0 204.1131 1.6 4-Aminoantipyrine Target pharmaceutical TP 1.0000 WW3 0 0 0 0 0 of metamizol 137.0709 0.9 4-Aminobenzamide Target industrial compound 0.9958 WW1 0 0 0 0 0 136.0506 2.0 4-Hydroxybenzotriazole Target industrial compound 0.9921 WW1 0 0 0 0 0 134.0712 7.4 5-Methyl-1H-benzotriazole Target industrial compound 0.9998 WW1 0 6.5E+04 1.3E+05 1.2E+05 1.4E+05 176.0706 7.3 7-Amino-4-methylcoumarin Target TP of 7-Diethylamino- 1.0000 WW3 0 0 0 0 0 4-methylcoumarin 232.1331 11.6 7-Diethylamino- Target industrial compound 0.9997 WW3 0 0 0 0 0 4-methylcoumarin 161.9867 1.0 Acesulfame Target sweetener 0.9116 WW3 1.3E+03 3.4E+03 2.3E+03 2.6E+03 3.7E+03 152.0706 1.5 Acetaminophen Target pharmaceutical 0.9027 WW2 0 0 0 0 0 152.1433 2.6 Amantadine Target pharmaceutical 0.9999 WW1 0 0 0 0 0 278.1902 9.5 Amitriptyline Target pharmaceutical 1.0000 WW3 0 0 0 0 0 749.5153 7.7 Azithromycin Target pharmaceutical 1.0000 WW3 0 0 0 0 0 239.05 9.7 Bentazon Target herbicide 0.9986 WW1 0 0 0 0 0

S65

Appendix B – Spatial chemical patterns

m/z rt Compound Name Use Pattern Sub- Holt3 Holt9 Holt11 Holt13 Holt15 probability pattern 307.0287 10.3 Benzophenone-4 Target UV stabilizer 0.6719 WW2 0 6.1E+03 2.1E+04 1.6E+04 1.8E+04 360.1014 11.7 Bezafibrate Target pharmaceutical 0.9980 WW1 0 0 0 0 0 429.0543 11.2 Bicalutamide Target pharmaceutical 0.9997 WW1 0 0 0 0 0 326.2323 7.7 Bisoprolol Target pharmaceutical 0.7883 WW2 0 0 0 0 0 237.1021 9.9 Carbamazepine Target pharmaceutical 0.9998 WW1 0 1.9E+05 9.1E+04 8.4E+04 7.3E+04 192.0766 2.2 Carbendazim Target biocide 0.6677 WW3 0 0 0 0 0 380.0691 12.3 Celecoxib Target pharmaceutical 0.9959 WW1 0 0 0 0 0 389.1624 10.4 Cetirizine Target pharmaceutical 0.9999 WW1 0 0 0 0 0 244.9083 13.2 Chlorothalonil-4-hydroxy Target fungicide TP of 0.8991 WW1 0 0 0 0 0 chlorothalonil 332.1399 6.1 Ciprofloxacin Target pharmaceutical 1.0000 WW3 4.7E+03 5.5E+03 2.5E+03 0 0 325.1709 8.2 Citalopram Target pharmaceutical 0.9593 WW1 0 0 0 0 0 748.4835 10.4 Clarithromycin Target pharmaceutical 0.9998 WW1 6.7E+04 6.1E+04 6.5E+04 6.3E+04 6.8E+04 322.0662 11.6 Clopidogrel Target pharmaceutical 0.9488 WW1 0 0 0 0 0 361.2006 10.3 Cortisone Target steroid 1.0000 WW2 0 0 0 0 0 204.1383 11.3 Crotamiton Target pharmaceutical 0.9979 WW1 0 0 0 0 0 292.1208 11.7 Cyproconazole Target fungicide 1.0000 WW3 0 0 0 0 0 192.1382 10.4 DEET Target insect repellent 0.9974 WW2 0 9.3E+05 6.5E+05 4.8E+05 4.5E+05 325.2273 8.2 Denatonium Target industrial compound 0.8785 WW1 0 0 0 0 0 631.8031 1.5 Diatrizoate Target X-ray contrast 1.0000 WW1 0 0 0 0 0 medium 294.0099 12.5 Diclofenac Target pharmaceutical 0.9973 WW1 0 0 0 0 0 302.1056 10.0 Dimethachlor ESA Target herbicide TP of 0.9999 WW1 0 0 0 0 0 dimethachlor 252.123 9.6 Dimethachlor OA Target herbicide TP of 0.9771 WW3 0 0 0 0 0 dimethachlor 256.1694 9.1 Diphenhydramine Target pharmaceutical 0.9527 WW1 0 0 0 0 0 304.1213 11.1 Ethofumesate Target herbicide 0.6207 WW2 0 5.8E+04 0 0 0 194.1175 11.7 Ethyl-4-dimethylaminobenzoate Target industrial compound 1.0000 WW4 1.7E+05 0 0 0 0 165.0559 9.3 Ethylparaben Target plasticizer 0.5561 WW3 0 0 0 0 0 165.1022 6.1 Fenuron Target herbicide 0.5131 WW3 0 0 0 0 0 434.932 12.6 Fipronil Target insecticide 0.9999 WW1 0 0 0 0 0 450.927 13.0 Fipronil sulfone Target insecticide TP 1.0000 WW1 0 0 0 0 0

S66

Appendix B – Spatial chemical patterns

m/z rt Compound Name Use Pattern Sub- Holt3 Holt9 Holt11 Holt13 Holt15 probability pattern of fipronil 307.1111 7.5 Fluconazole Target pharmaceutical 0.9963 WW1 0 0 0 0 0 364.0734 12.2 Flufenacet Target herbicide 1.0000 WW2 0 0 0 0 0 262.0872 9.6 Flumequine Target pharmaceutical 1.0000 WW3 0 0 0 0 6.6E+04 329.001 9.3 Furosemide Target pharmaceutical 0.7830 WW1 7.5E+03 0 0 0 5.3E+03 172.1331 1.4 Gabapentin Target pharmaceutical 0.5375 WW2 0 0 0 0 0 295.9574 1.7 Hydrochlorothiazide Target pharmaceutical 0.9928 WW1 0 0 0 0 0 256.0594 6.6 Imidacloprid Target insecticide 0.9944 WW2 0 0 0 0 0 356.07 12.6 Indometacin Target pharmaceutical 0.9993 WW1 0 0 0 0 0 207.1492 10.4 Isoproturon Target herbicide 0.9484 WW2 0 0 5.6E+04 6.4E+04 8.7E+04 255.1015 11.1 Ketoprofen Target pharmaceutical 0.9986 WW2 6.4E+04 0 0 0 5.0E+04 235.1803 2.4 Lidocaine Target pharmaceutical 0.9998 WW1 0 0 0 0 0 423.1695 11.1 Losartan Target pharmaceutical 0.9979 WW1 0 0 0 0 0 199.0169 10.7 MCPA Target herbicide 0.6239 WW2 5.5E+03 0 0 5.2E+03 0 213.0327 11.5 Mecoprop Target herbicide 0.9536 WW4 1.1E+04 0 0 0 9.4E+03 240.1033 13.2 Mefenamic acid Target pharmaceutical 1.0000 WW2 0 0 0 0 0 264.1757 7.0 Melperon Target pharmaceutical 1.0000 WW3 0 0 0 0 0 180.1746 7.7 Memantine Target pharmaceutical 0.5029 WW2 0 0 0 0 0 280.1542 10.4 Metalaxyl Target fungicide 0.8166 WW3 0 0 0 5.3E+04 0 322.0874 9.3 Metazachlor ESA Target herbicide TP of 1.0000 WW1 0 0 6.0E+03 6.2E+03 2.0E+04 metazachlor 130.1088 0.8 Metformin Target pharmaceutical 0.9108 WW2 0 2.0E+05 1.2E+05 1.2E+05 5.4E+05 455.1786 6.0 Methotrexate Target pharmaceutical 1.0000 WW3 0 0 0 0 5.9E+04 330.1367 12.0 Metolachlor ESA Target herbicide TP of 0.9998 WW1 0 0 0 0 0 metolachlor 268.1905 6.4 Metoprolol Target pharmaceutical 0.9763 WW2 0 0 0 0 0 268.1543 6.4 Metoprolol Acid Target pharmaceutical TP 0.9766 WW3 1.9E+04 1.3E+04 8.4E+03 6.8E+03 9.3E+03 of metoprolol 194.046 4.3 N-Acetyl mesalazine Target pharmaceutical TP 0.9998 WW1 0 0 0 0 0 of mesalazine 246.1235 5.1 N-Acetyl-4-aminoantipyrine Target pharmaceutical TP 0.9999 WW1 0 0 0 0 0 of metamizol 231.1015 11.2 Naproxen Target pharmaceutical 0.7425 WW1 0 0 0 0 0

S67

Appendix B – Spatial chemical patterns

m/z rt Compound Name Use Pattern Sub- Holt3 Holt9 Holt11 Holt13 Holt15 probability pattern 214.0895 10.1 N-Butylbenzenesulfonamide Target plasticizer 0.9998 WW2 0 0 0 0 0 232.1079 4.6 N-Formyl-4-aminoantipyrine Target pharmaceutical TP 0.9274 WW1 0 0 0 0 0 of metamizol 362.1507 5.8 Ofloxacin Target pharmaceutical 1.0000 WW3 1.6E+04 1.2E+04 6.1E+03 3.3E+03 3.6E+03 287.058 10.6 Oxazepam Target pharmaceutical 0.9939 WW1 0 0 0 0 0 151.0263 0.9 Oxypurinol Target pharmaceutical TP 0.9898 WW1 0 0 0 0 0 of allopurinol 189.1022 6.5 Phenazone Target pharmaceutical 0.9999 WW2 0 0 0 0 0 376.2395 8.9 Pipamperone Target pharmaceutical 0.9166 WW3 0 0 0 0 0 356.2428 14.0 Piperonyl butoxide Target pesticide component 0.9999 WW2 0 0 0 0 0 423.2396 11.0 Pravastatin Target pharmaceutical 0.9980 WW1 6.6E+03 0 0 5.3E+03 6.0E+03 342.0769 12.7 Propiconazole Target fungicide 1.0000 WW3 0 0 0 0 0 260.1644 7.8 Propranolol Target pharmaceutical 0.9995 WW1 0 0 0 0 0 179.0717 10.3 Propylparaben Target plasticizer 1.0000 WW2 0 0 0 0 0 252.1413 13.5 Prosulfocarb Target herbicide 0.9967 WW2 0 0 0 0 0 315.1487 1.2 Ranitidine Target pharmaceutical 0.9842 WW2 0 0 0 0 0 837.5313 10.6 Roxithromycin Target pharmaceutical 1.0000 WW3 0 0 0 0 0 273.1265 1.0 Sotalol Target pharmaceutical 0.5878 WW2 0 0 0 0 0 327.0098 9.4 Sulcotrione Target herbicide 0.9997 WW1 0 0 0 0 0 254.0592 6.4 Sulfamethoxazole Target pharmaceutical 0.9738 WW1 0 0 0 0 0 250.0643 2.6 Sulfapyridine Target pharmaceutical 0.9945 WW1 0 0 0 0 0 308.1522 12.6 Tebuconazole Target fungicide 1.0000 WW3 0 0 0 0 0 301.0744 10.8 Temazepam Target pharmaceutical 0.9621 WW1 0 0 0 0 0 242.1433 10.4 Terbutryn Target biocide 0.9930 WW1 0 0 0 0 0 264.1956 5.9 Tramadol Target pharmaceutical 0.9988 WW1 0 0 0 0 6.8E+05 286.9442 13.4 Triclosan Target biocide 0.9993 WW1 0 0 0 0 0 291.1449 3.5 Trimethoprim Target pharmaceutical 0.8375 WW1 0 0 0 0 0 436.234 12.1 Valsartan Target pharmaceutical 0.9995 WW2 0 1.3E+05 5.3E+04 5.7E+04 5.4E+04 455.2903 9.1 Verapamil Target pharmaceutical 0.9999 WW3 0 0 0 0 0 154.1226 8.5 Gabapentin-lactam Unknown pharmaceutical TP 1.0000 WW1 0 0 0 0 0 of gabapentin 204.1019 10.0 4-Methyl-7-ethylaminocoumarin Unknown TP of 7-Diethylamino- 1.0000 WW3 0 0 0 0 0 4-methylcoumarin

S68

Appendix B – Spatial chemical patterns

m/z rt Compound Name Use Pattern Sub- Holt3 Holt9 Holt11 Holt13 Holt15 probability pattern 212.2008 11.9 C13H25NO Unknown 0.9995 WW1 0 0 0 0 0 214.1437 8.1 C11H19NO3 Unknown 1.0000 WW2 0 0 0 0 0 214.1437 8.1 C11H19NO3 Unknown 0.9996 WW2 0 0 0 0 0 223.144 9.2 C12H18N2O2 Unknown 0.9135 WW3 0 0 0 0 0 242.1021 7.4 Methocarbamol Unknown pharmaceutical 0.9799 WW1 0 0 0 0 0 249.1847 11.2 C16H24O2 Unknown 1.0000 WW1 0 0 0 0 0 256.0149 5.2 Lamotrigine Unknown pharmaceutical 0.6724 WW3 0 0 0 0 0 265.0729 9.3 Valsartan acid [M+H]+ Unknown pharmaceutical TP 0.9983 WW1 0 0 0 0 0 of valsartan 267.0871 9.3 Valsartan acid [M+H]- Unknown pharmaceutical TP 0.9977 WW1 0 0 0 0 0 of valsartan 269.1283 11.6 C16H16N2O2 Unknown 0.9373 WW3 0 0 0 0 0 271.0649 9.4 C6-SPC Unknown surfactant 0.9817 WW2 5.2E+04 1.0E+05 5.4E+04 4.5E+04 1.8E+05 274.201 7.7 C14H27NO4 Unknown 0.9994 WW2 0 0 0 0 0 285.0806 10.6 C7-SPC [M-H]- Unknown surfactant 0.8395 WW1 1.0E+05 3.8E+05 1.7E+05 1.3E+05 3.4E+05 304.1213 10.6 C7-SPC [M+NH4]+ Unknown surfactant 0.9584 WW1 0 0 0 0 0 327.1337 11.9 C18H18N2O4 Unknown 0.9999 WW3 0 0 0 0 0 429.2397 11.2 Irbesartan Unknown pharmaceutical 0.7602 WW1 0 0 0 0 0 447.2136 9.0 Olmesartan Unknown pharmaceutical 0.9013 WW2 0 0 0 0 0

S69

Appendix B – Spatial chemical patterns

Table 19 continued

Compound Name Holt17 Holt18 Holt21 Holt22 Holt26 Holt31 Holt34 Holt36a Holt38 Holt40 Holt42

10,11-Dihydro-10,11-dihydroxycarbamazepine 6.2E+06 4.7E+06 3.5E+06 4.0E+06 3.8E+06 6.7E+06 6.1E+06 5.6E+06 4.3E+06 4.3E+06 2.7E+06 10,11-Dihydro-10-hydroxycarbamazepine 9.7E+05 7.3E+05 5.4E+05 5.8E+05 5.5E+05 1.6E+06 1.6E+06 1.3E+06 8.0E+05 7.6E+05 5.1E+05 1H-Benzotriazole 4.9E+06 3.5E+06 2.5E+06 2.8E+06 2.6E+06 4.6E+06 4.2E+06 3.8E+06 2.7E+06 2.7E+06 2.1E+06 2(4-morpholinyl)benzothiazole 0 0 0 0 0 5.0E+04 0 0 0 0 0 2-4-Dichlorophenoxyacetic acid 0 0 0 9.7E+03 0 1.9E+04 1.4E+04 1.3E+04 9.9E+03 0 0 2-Aminobenzimidazole 0 0 0 0 0 1.1E+05 1.1E+05 1.0E+05 0 0 0 2-Hydroxycarbamazepine 2.5E+06 1.9E+06 1.4E+06 1.6E+06 1.5E+06 2.6E+06 2.3E+06 2.1E+06 1.4E+06 1.4E+06 9.8E+05 3-5-6-Trichloro-2-pyridinol 1.3E+04 8.2E+03 8.6E+03 6.2E+03 0 7.9E+03 0 0 0 0 0 4-(Dimethylamino)pyridine 1.9E+05 1.6E+05 0 1.5E+05 1.2E+05 2.9E+05 2.5E+05 2.2E+05 1.5E+05 1.4E+05 1.7E+05 4'-Aminoacetanilide 2.5E+05 2.0E+05 1.8E+05 2.0E+05 1.9E+05 3.2E+05 3.3E+05 3.1E+05 2.4E+05 2.2E+05 2.3E+05 4-Aminoantipyrine 2.5E+06 4.7E+05 2.4E+05 6.1E+05 6.1E+05 2.8E+05 1.5E+05 1.2E+05 1.3E+05 1.5E+05 8.6E+04 4-Aminobenzamide 1.8E+05 0 0 0 0 0 1.7E+05 1.8E+05 0 0 1.7E+05 4-Hydroxybenzotriazole 1.7E+05 1.1E+05 6.2E+04 5.5E+04 5.0E+04 1.5E+05 1.2E+05 1.1E+05 5.6E+04 5.3E+04 0 5-Methyl-1H-benzotriazole 1.3E+07 9.6E+06 7.3E+06 8.2E+06 7.6E+06 1.4E+07 1.2E+07 1.1E+07 8.3E+06 8.1E+06 5.7E+06 7-Amino-4-methylcoumarin 2.5E+07 1.7E+07 1.3E+07 1.5E+07 1.3E+07 6.7E+06 5.0E+06 4.9E+06 5.7E+06 5.3E+06 4.3E+06 7-Diethylamino-4-methylcoumarin 1.8E+07 1.3E+07 9.9E+06 1.1E+07 1.0E+07 7.3E+06 7.3E+06 6.8E+06 7.0E+06 6.7E+06 5.2E+06 Acesulfame 3.3E+04 2.7E+04 1.9E+04 2.2E+04 2.5E+04 1.9E+04 2.6E+04 2.7E+04 3.2E+04 3.0E+04 3.3E+04 Acetaminophen 2.5E+05 0 0 0 0 3.2E+05 3.3E+05 0 0 0 2.4E+05 Amantadine 1.4E+05 1.2E+05 7.5E+04 8.9E+04 7.9E+04 1.5E+05 1.4E+05 1.3E+05 1.0E+05 9.6E+04 7.5E+04 Amitriptyline 1.2E+05 8.6E+04 6.8E+04 5.6E+04 0 7.5E+04 0 0 0 0 0 Azithromycin 5.3E+04 0 0 0 0 0 0 0 0 0 0 Bentazon 1.2E+04 9.1E+03 1.2E+04 3.1E+04 2.1E+04 2.9E+04 0 4.4E+04 5.0E+04 5.7E+04 4.9E+04 Benzophenone-4 3.0E+05 2.1E+05 1.7E+05 1.9E+05 2.0E+05 6.2E+05 5.4E+05 4.5E+05 2.8E+05 3.0E+05 2.8E+05 Bezafibrate 5.8E+04 4.2E+04 3.3E+04 3.3E+04 2.9E+04 6.2E+04 5.7E+04 4.2E+04 3.3E+04 3.0E+04 2.0E+04 Bicalutamide 3.1E+05 1.8E+05 1.5E+05 1.7E+05 1.2E+05 3.5E+05 3.2E+05 2.2E+05 1.6E+05 1.8E+05 2.0E+05 Bisoprolol 9.8E+05 6.7E+05 4.5E+05 4.4E+05 3.7E+05 2.0E+06 1.6E+06 1.5E+06 6.0E+05 6.6E+05 6.3E+05 Carbamazepine 2.1E+07 1.6E+07 1.2E+07 1.4E+07 1.3E+07 2.6E+07 2.4E+07 2.2E+07 1.6E+07 1.6E+07 1.1E+07 Carbendazim 8.5E+04 6.5E+04 6.0E+04 0 5.6E+04 0 6.8E+04 6.0E+04 0 0 9.6E+04

S70

Appendix B – Spatial chemical patterns

Compound Name Holt17 Holt18 Holt21 Holt22 Holt26 Holt31 Holt34 Holt36a Holt38 Holt40 Holt42 Celecoxib 3.0E+04 0 1.5E+04 1.5E+04 1.4E+04 3.7E+04 3.0E+04 2.5E+04 1.9E+04 1.5E+04 1.6E+04 Cetirizine 5.6E+05 3.9E+05 341521 3.3E+05 2.9E+05 5.5E+05 5.5E+05 5.2E+05 3.4E+05 3.6E+05 2.6E+05 Chlorothalonil-4-hydroxy 1.5E+04 0 1.1E+04 9.9E+03 7.8E+03 1.4E+04 7.6E+03 7.9E+03 0 0 9.8E+03 Ciprofloxacin 4.8E+04 2.9E+04 1.5E+04 1.1E+04 5.9E+03 1.1E+04 9.0E+03 1.5E+04 6.9E+03 6.6E+03 0 Citalopram 8.2E+05 6.0E+05 4.7E+05 4.6E+05 3.3E+05 7.0E+05 4.7E+05 4.3E+05 2.6E+05 2.5E+05 2.8E+05 Clarithromycin 1.6E+05 1.3E+05 1.2E+05 1.1E+05 1.1E+05 1.8E+05 1.5E+05 1.5E+05 1.1E+05 1.2E+05 9.5E+04 Clopidogrel 9.2E+04 7.4E+04 0 5.9E+04 0 1.3E+05 1.1E+05 8.8E+04 5.5E+04 0 0 Cortisone 0 0 0 0 0 5.2E+04 0 0 0 0 0 Crotamiton 7.4E+04 5.8E+04 5.0E+04 5.5E+04 0 8.9E+04 8.2E+04 7.6E+04 6.0E+04 5.8E+04 0 Cyproconazole 9.2E+05 6.5E+05 4.9E+05 4.3E+05 2.0E+05 1.3E+05 1.2E+05 1.1E+05 1.2E+05 1.0E+05 0 DEET 8.2E+05 6.3E+05 5.3E+05 5.6E+05 5.7E+05 2.1E+06 1.8E+06 1.8E+06 9.6E+05 9.2E+05 9.7E+05 Denatonium 6.2E+05 4.2E+05 3.3E+05 3.8E+05 3.3E+05 1.1E+06 8.8E+05 7.2E+05 4.6E+05 4.5E+05 5.9E+05 Diatrizoate 2.3E+05 1.9E+05 1.5E+05 1.7E+05 1.9E+05 2.1E+05 2.6E+05 2.3E+05 2.1E+05 2.1E+05 1.4E+05 Diclofenac 1.2E+06 7.8E+05 6.3E+05 7.0E+05 5.7E+05 1.7E+06 1.4E+06 1.1E+06 6.0E+05 5.8E+05 4.2E+05 Dimethachlor ESA 1.6E+06 1.2E+06 8.9E+05 9.4E+05 9.4E+05 1.7E+06 1.6E+06 1.6E+06 1.1E+06 1.1E+06 7.7E+05 Dimethachlor OA 3.5E+05 2.6E+05 2.0E+05 2.4E+05 2.2E+05 1.8E+05 1.8E+05 1.8E+05 2.0E+05 2.1E+05 0 Diphenhydramine 1.0E+04 1.4E+04 7.7E+03 5.9E+03 4.7E+03 1.3E+04 7.9E+03 8.7E+03 6.3E+03 4.7E+03 6.9E+03 Ethofumesate 1.3E+06 9.6E+05 7.0E+05 7.9E+05 7.1E+05 3.4E+06 2.8E+06 2.8E+06 1.4E+06 1.3E+06 9.2E+05 Ethyl-4-dimethylaminobenzoate 0 0 0 0 0 0 0 1.7E+05 0 0 0 Ethylparaben 8.3E+03 0 6.9E+03 0 0 0 7.9E+03 0 0 0 7.7E+03 Fenuron 2.2E+05 1.9E+05 1.3E+05 1.5E+05 8.7E+04 1.4E+05 1.3E+05 1.1E+05 1.2E+05 1.2E+05 1.1E+05 Fipronil 3.0E+04 2.0E+04 1.8E+04 1.9E+04 1.6E+04 3.6E+04 3.4E+04 2.3E+04 2.0E+04 1.8E+04 1.8E+04 Fipronil sulfone 2.0E+04 0 1.1E+04 1.8E+04 1.4E+04 1.5E+04 2.1E+04 1.9E+04 1.9E+04 1.6E+04 1.7E+04 Fluconazole 1.9E+05 1.5E+05 1.3E+05 1.3E+05 1.3E+05 2.5E+05 2.3E+05 2.2E+05 1.6E+05 1.6E+05 1.1E+05 Flufenacet 0 0 0 0 0 5.7E+04 0 0 0 0 0 Flumequine 8.6E+04 0 0 0 0 0 0 0 0 0 5.1E+04 Furosemide 4.2E+05 2.5E+05 2.0E+05 2.1E+05 1.6E+05 4.1E+05 2.3E+05 1.7E+05 5.7E+04 5.7E+04 4.8E+04 Gabapentin 1.7E+06 1.4E+06 1.0E+06 1.0E+06 8.7E+05 3.5E+06 3.2E+06 2.8E+06 1.6E+06 1.6E+06 1.2E+06 Hydrochlorothiazide 1.3E+05 1.0E+05 7.5E+04 8.4E+04 7.5E+04 1.6E+05 1.3E+05 9.3E+04 5.1E+04 5.3E+04 3.5E+04 Imidacloprid 5.2E+04 0 0 0 0 8.8E+04 7.1E+04 7.4E+04 0 0 0

S71

Appendix B – Spatial chemical patterns

Compound Name Holt17 Holt18 Holt21 Holt22 Holt26 Holt31 Holt34 Holt36a Holt38 Holt40 Holt42 Indometacin 1.4E+04 1.1E+04 9.4E+03 6.9E+03 1.2E+04 1.8E+04 1.7E+04 1.2E+04 8.0E+03 1.3E+04 0 Isoproturon 7.9E+04 6.0E+04 7.3E+04 8.4E+04 5.8E+04 1.1E+05 1.2E+05 1.0E+05 6.9E+04 7.5E+04 7.7E+04 Ketoprofen 5.8E+04 0 0 0 0 1.3E+05 5.4E+04 5.0E+04 0 0 5.5E+04 Lidocaine 2.0E+05 1.4E+05 9.7E+04 1.1E+05 9.6E+04 2.0E+05 2.0E+05 1.7E+05 9.9E+04 1.0E+05 8.6E+04 Losartan 1.7E+05 1.3E+05 9.6E+04 1.0E+05 8.4E+04 2.5E+05 2.2E+05 2.1E+05 1.0E+05 9.3E+04 9.5E+04 MCPA 8.3E+03 9.3E+03 9.3E+03 6.0E+03 7.2E+03 2.5E+04 2.0E+04 1.7E+04 8.0E+03 1.2E+04 1.4E+04 Mecoprop 7.6E+03 5.7E+03 7.0E+03 0 6.8E+03 1.1E+04 6.7E+03 0 0 0 6.2E+03 Mefenamic acid 0 0 0 0 0 6.9E+03 0 0 0 0 0 Melperon 1.8E+06 1.4E+06 9.2E+05 1.0E+06 7.5E+05 6.3E+05 4.4E+05 4.3E+05 2.8E+05 2.3E+05 2.5E+05 Memantine 1.4E+05 1.0E+05 7.7E+04 8.3E+04 7.9E+04 2.8E+05 2.3E+05 2.1E+05 1.2E+05 1.2E+05 1.0E+05 Metalaxyl 2.1E+06 1.6E+06 1.4E+06 1.5E+06 1.5E+06 1.1E+06 1.1E+06 1.0E+06 1.3E+06 1.3E+06 8.0E+05 Metazachlor ESA 3.2E+04 2.8E+04 4.4E+04 4.8E+04 5.8E+04 4.7E+04 1.1E+04 5.5E+04 6.6E+04 7.2E+04 6.1E+04 Metformin 2.5E+05 2.8E+05 1.6E+05 2.1E+05 2.7E+05 3.7E+05 3.7E+05 3.1E+05 2.1E+05 1.9E+05 1.2E+05 Methotrexate 9.3E+04 0 0 0 0 0 0 0 0 0 0 Metolachlor ESA 6.4E+05 4.5E+05 3.3E+05 3.6E+05 3.6E+05 7.5E+05 6.8E+05 5.8E+05 4.1E+05 4.1E+05 3.0E+05 Metoprolol 5.2E+06 3.7E+06 2.7E+06 2.8E+06 2.4E+06 1.5E+07 1.2E+07 1.1E+07 4.7E+06 4.6E+06 4.5E+06 Metoprolol Acid 9.5E+04 8.0E+04 5.4E+04 8.3E+04 8.6E+04 2.8E+04 7.4E+04 8.1E+04 6.3E+04 6.7E+04 6.4E+04 N-Acetyl mesalazine 1.0E+04 8.3E+03 7.6E+03 7.0E+03 7.4E+03 1.1E+04 8.2E+03 9.4E+03 7.6E+03 8.3E+03 6.6E+03 N-Acetyl-4-aminoantipyrine 2.6E+06 1.9E+06 1.7E+06 1.7E+06 1.7E+06 3.2E+06 2.8E+06 2.3E+06 1.7E+06 1.7E+06 1.3E+06 Naproxen 5.0E+04 0 0 0 0 5.8E+04 51321 0 0 0 0 N-Butylbenzenesulfonamide 0 0 0 0 0 7.5E+05 744812 6.5E+05 0 0 5.9E+05 N-Formyl-4-aminoantipyrine 6.0E+06 4.0E+06 2.9E+06 3.1E+06 3.1E+06 1.2E+07 8.8E+06 7.6E+06 4.3E+06 4.6E+06 3.5E+06 Ofloxacin 2.0E+04 1.8E+04 7.3E+03 9.0E+03 6.1E+03 5.1E+03 7.4E+03 1.2E+04 7.1E+03 5.6E+03 6.3E+03 Oxazepam 8.0E+04 6.7E+04 5.0E+04 5.3E+04 5.7E+04 1.4E+05 1.3E+05 1.2E+05 7.5E+04 7.3E+04 5.4E+04 Oxypurinol 3.6E+05 2.3E+05 1.6E+05 1.8E+05 1.5E+05 3.6E+05 3.3E+05 2.3E+05 1.5E+05 1.6E+05 2.3E+05 Phenazone 1.2E+06 1.3E+06 1.3E+06 9.9E+05 1.1E+06 1.6E+07 6.8E+06 6.8E+06 3.4E+06 3.2E+06 2.0E+06 Pipamperone 7.0E+04 9.8E+04 4.4E+04 6.2E+04 6.3E+04 4.5E+04 4.9E+04 4.5E+04 4.9E+04 5.1E+04 4.9E+04 Piperonyl butoxide 0 0 0 0 0 8.2E+04 0 6.9E+04 0 0 0 Pravastatin 9.4E+03 6.9E+03 9.3E+03 8.2E+03 7.2E+03 8.0E+03 9.6E+03 1.1E+04 8.7E+03 9.0E+03 1.1E+04 Propiconazole 1.2E+07 8.0E+06 6.1E+06 5.3E+06 2.3E+06 1.8E+06 1.7E+06 1.7E+06 1.4E+06 1.2E+06 9.0E+05

S72

Appendix B – Spatial chemical patterns

Compound Name Holt17 Holt18 Holt21 Holt22 Holt26 Holt31 Holt34 Holt36a Holt38 Holt40 Holt42 Propranolol 2.5E+05 1.9E+05 1.4E+05 1.5E+05 1.2E+05 2.7E+05 1.9E+05 1.7E+05 9.8E+04 1.0E+05 1.1E+05 Propylparaben 0 0 0 0 0 1.1E+04 0 0 0 0 0 Prosulfocarb 0 0 0 0 1.8E+05 3.1E+05 1.6E+05 1.8E+05 8.7E+04 1.2E+05 1.1E+05 Ranitidine 9.5E+04 6.3E+04 0 0 0 1.9E+05 1.4E+05 1.3E+05 0 0 5.5E+04 Roxithromycin 2.5E+06 1.6E+06 1.2E+06 1.1E+06 8.6E+05 5.0E+05 4.4E+05 4.4E+05 4.3E+05 3.8E+05 4.3E+05 Sotalol 2.8E+05 2.3E+05 1.7E+05 1.8E+05 1.7E+05 6.3E+05 5.7E+05 5.4E+05 3.3E+05 3.1E+05 2.7E+05 Sulcotrione 1.4E+04 8.4E+03 8.4E+03 6.9E+03 1.5E+04 2.0E+04 1.0E+04 1.1E+04 1.4E+04 1.3E+04 6.4E+03 Sulfamethoxazole 3.6E+05 2.9E+05 2.2E+05 2.7E+05 2.6E+05 5.9E+05 5.0E+05 4.8E+05 3.3E+05 3.4E+05 2.4E+05 Sulfapyridine 2.9E+05 2.2E+05 1.6E+05 2.0E+05 1.7E+05 2.4E+05 2.1E+05 1.8E+05 1.5E+05 1.5E+05 1.2E+05 Tebuconazole 6.2E+06 4.0E+06 2.9E+06 2.7E+06 1.4E+06 9.5E+05 9.6E+05 8.8E+05 8.6E+05 7.8E+05 5.6E+05 Temazepam 1.1E+05 8.2E+04 0 0 6.2E+04 1.4E+05 1.3E+05 1.1E+05 8.2E+04 7.9E+04 0 Terbutryn 2.2E+05 2.0E+05 1.5E+05 1.4E+05 1.1E+05 3.8E+05 3.1E+05 2.7E+05 2.1E+05 2.2E+05 1.7E+05 Tramadol 2.5E+06 1.7E+06 1.3E+06 1.5E+06 1.4E+06 3.6E+06 2.8E+06 2.6E+06 1.5E+06 1.5E+06 1.4E+06 Triclosan 1.5E+04 0 1.2E+04 1.2E+04 9.6E+03 1.4E+04 1.4E+04 1.6E+04 8.4E+03 8.2E+03 0 Trimethoprim 1.3E+05 8.8E+04 5.7E+04 5.6E+04 0 1.1E+05 9.6E+04 9.1E+04 0 0 0 Valsartan 5.2E+05 4.5E+05 2.8E+05 3.0E+05 2.9E+05 3.4E+06 3.4E+06 2.6E+06 1.0E+06 1.0E+06 8.7E+05 Verapamil 1.7E+05 1.2E+05 8.2E+04 7.0E+04 0 1.0E+05 6.4E+04 6.4E+04 0 0 0 Gabapentin-lactam 1.6E+07 1.1E+07 8.4E+06 9.1E+06 8.5E+06 1.5E+07 1.4E+07 1.3E+07 9.9E+06 9.9E+06 7.0E+06 4-Methyl-7-ethylaminocoumarin 2.6E+07 1.8E+07 1.4E+07 1.6E+07 1.4E+07 6.1E+06 4.1E+06 3.8E+06 4.4E+06 3.9E+06 3.2E+06 C13H25NO 7.1E+06 4.8E+06 3.8E+06 4.3E+06 3.9E+06 8.1E+06 7.3E+06 6.6E+06 4.5E+06 4.7E+06 3.6E+06 C11H19NO3 0 8.9E+05 6.8E+05 0 0 6.9E+06 6.0E+06 5.8E+06 2.3E+06 2.4E+06 7.6E+05 C11H19NO3 1.3E+06 8.9E+05 6.6E+05 7.2E+05 6.5E+05 6.8E+06 6.1E+06 5.5E+06 2.2E+06 2.2E+06 1.6E+06 C12H18N2O2 7.2E+06 4.9E+06 3.5E+06 3.7E+06 3.5E+06 4.0E+06 3.6E+06 3.5E+06 3.1E+06 3.2E+06 2.3E+06 Methocarbamol 3.6E+06 2.8E+06 2.1E+06 2.3E+06 2.2E+06 5.7E+06 4.9E+06 4.5E+06 3.0E+06 2.9E+06 2.0E+06 C16H24O2 3.5E+07 2.4E+07 1.8E+07 2.0E+07 1.8E+07 3.2E+07 3.0E+07 2.7E+07 2.0E+07 2.0E+07 1.5E+07 Lamotrigine 2.0E+07 1.3E+07 9.1E+06 1.1E+07 1.1E+07 9.6E+06 7.8E+06 7.6E+06 8.0E+06 8.2E+06 6.0E+06 Valsartan acid [M+H]+ 9.7E+06 6.7E+06 5.6E+06 6.4E+06 6.0E+06 1.3E+07 1.1E+07 9.6E+06 6.9E+06 7.8E+06 5.5E+06 Valsartan acid [M+H]- 2.2E+07 1.6E+07 1.2E+07 1.4E+07 1.2E+07 2.8E+07 2.7E+07 2.3E+07 1.7E+07 1.6E+07 1.1E+07 C16H16N2O2 3.5E+06 2.8E+06 2.3E+06 2.5E+06 2.1E+06 1.4E+06 1.5E+06 1.3E+06 1.4E+06 1.2E+06 9.2E+05 C6-SPC 2.2E+06 1.4E+06 1.2E+06 1.3E+06 1.3E+06 6.9E+06 5.5E+06 4.8E+06 2.4E+06 2.3E+06 5.1E+06

S73

Appendix B – Spatial chemical patterns

Compound Name Holt17 Holt18 Holt21 Holt22 Holt26 Holt31 Holt34 Holt36a Holt38 Holt40 Holt42 C14H27NO4 1.1E+06 7.6E+05 6.5E+05 5.4E+05 4.0E+05 5.1E+06 4.9E+06 4.6E+06 1.6E+06 1.8E+06 1.5E+06 C7-SPC [M-H]- 7.1E+06 4.1E+06 3.5E+06 3.6E+06 3.6E+06 1.4E+07 1.3E+07 1.0E+07 5.3E+06 5.6E+06 4.4E+06 C7-SPC [M+NH4]+ 2.9E+06 5.5E+05 4.1E+05 1.4E+06 1.3E+06 5.5E+06 4.6E+06 4.5E+06 2.1E+06 2.1E+06 1.7E+06 C18H18N2O4 3.8E+06 2.7E+06 1.9E+06 2.1E+06 1.6E+06 9.2E+05 9.2E+05 7.9E+05 8.0E+05 7.6E+05 5.9E+05 Irbesartan 5.2E+06 3.9E+06 2.6E+06 2.9E+06 2.5E+06 8.9E+06 8.0E+06 7.1E+06 3.9E+06 3.9E+06 3.3E+06 Olmesartan 1.7E+06 1.3E+06 9.8E+05 9.9E+05 1.0E+06 4.4E+06 3.5E+06 3.6E+06 1.8E+06 1.9E+06 1.6E+06

S74

Appendix B – Spatial chemical patterns

Table S 20: Intensities of target and prioritized unknown compounds in river samples belonging to main BR pattern

m/z rt Compound Name Use Pattern Sub- Holt3 Holt9 Holt11 Holt13 Holt15 probability pattern 152.0837 0.6 (3-Chloro-2-hydroxypropyl)- Target industrial compound 1.0000 BR1 0 0 0 0 0 trimethylammonium 152.0164 6.0 1,2-Benzisothiazolinone Target biocide 1.0000 BR1 0 0 0 0 0 213.9641 7.1 2-Benzothiazolesulfonic acid Target industrial compound 1.0000 BR1 0 0 2.7E+04 4.1E+04 5.0E+04 207.0124 8.6 2-Naphthalene sulfonic acid Target industrial compound 1.0000 BR1 9.8E+04 0 0 0 0 136.0216 8.3 Benzothiazole Target industrial compound 1.0000 BR1 0 3.8E+05 0 0 0 249.023 7.8 Bisphenol S Target plasticizer 1.0000 BR1 1.2E+05 0 0 1.6E+04 5.1E+03 213.0327 11.0 Clofibric acid Target pharmaceutical 1.0000 BR1 6.7E+03 0 0 0 0 177.1022 0.9 Cotinine Target TP of nicotine 0.9998 BR2 6.6E+04 1.3E+05 1.0E+05 1.3E+05 2.0E+05 114.0662 0.8 Creatinine Target human metabolite 0.9998 BR1 1.1E+06 0 0 0 0 251.0466 11.2 Diphenylphosphate Target industrial compound 0.8326 BR2 0 0 0 0 0 330.08 12.2 Epoxiconazole Target fungicide 1.0000 BR1 0 0 0 0 0 363.2163 10.2 Hydrocortisone Target steroid 1.0000 BR1 0 0 0 0 0 230.175 11.0 Icaridin Target insect repellent 0.9995 BR1 0 0 0 0 0 114.1277 0.9 Mepiquat Target plant growth 0.9724 BR2 0 0 0 0 0 regulator 149.9774 2.9 Methylchloroisothiazolinone Target biocide 1.0000 BR1 0 0 0 0 0 321.1331 11.1 Mycophenolic acid Target pharmaceutical 1.0000 BR2 5.3E+04 0 0 0 5.4E+04 279.1443 8.2 Pentoxifylline Target pharmaceutical 1.0000 BR4 0 0 0 0 0 212.9796 8.1 Perfluorobutanoic acid Target industrial compound 1.0000 BR1 0 0 0 0 0 275.0482 5.0 Phenylbenzimidazolesulfonic acid Target UV stabilizer 0.9999 BR1 0 0 0 5.2E+04 0 240.1804 6.7 Tetraglyme Target industrial compound 0.4645 BR2 0 0 0 0 0 399.2504 13.7 Tri(butoxyethyl)phosphate Target industrial compound 0.9825 BR3 0 5.4E+05 0 4.9E+05 8.1E+05 277.1281 9.5 Triethylcitrate Target industrial compound 1.0000 BR1 0 0 0 0 0 284.961 9.4 Tris(2-chloroethyl)phosphate Target industrial compound 1.0000 BR4 0 0 0 0 0 187.0979 8.8 Azelaic acid Unknown PCP, industrial 1.0000 BR1 0 0 0 0 0 compound 476.3065 7.9 Decaethylene Glycol Unknown surfactant 1.0000 BR1 4.9E+05 1.4E+06 1.2E+06 1.6E+06 5.9E+05 163.1328 7.9 Diethylene glycol monobutyl ether Unknown surfactant 1.0000 BR1 0 0 0 0 0 344.2276 6.6 Heptaethylene Glycol Unknown surfactant 1.0000 BR1 7.9E+05 1.6E+06 1.6E+06 2.1E+06 7.9E+05

S75

Appendix B – Spatial chemical patterns

m/z rt Compound Name Use Pattern Sub- Holt3 Holt9 Holt11 Holt13 Holt15 probability pattern 300.2014 5.7 Hexaethyllene Glycol Unknown surfactant 1.0000 BR1 1.3E+06 1.6E+06 8.4E+05 1.1E+06 6.4E+05 432.2802 7.5 Nonaethylene Glycol Unknown surfactant 1.0000 BR1 6.4E+05 1.5E+06 1.4E+06 1.9E+06 6.8E+05 155.1542 1.4 C19H32N6 Unknown 1.0000 BR1 0 0 0 0 0 388.2539 7.1 Octaethylene Glycol Unknown surfactant 1.0000 BR1 6.1E+05 1.6E+06 1.7E+06 2.1E+06 7.6E+05 236.1127 7.2 Triacetin Unknown industrial compound 1.0000 BR1 0 0 0 0 0

Table S20 continued

Compound Name Holt17 Holt18 Holt21 Holt22 Holt26 Holt31 Holt34 Holt36a Holt38 Holt40 Holt42 (3-Chloro-2-hydroxypropyl)trimethylammonium 0 0 0 0 0 0 0 0 0 0 1.7E+06 1,2-Benzisothiazolinone 0 0 0 0 0 0 0 0 0 0 3.0E+06 2-Benzothiazolesulfonic acid 2.4E+05 1.8E+05 2.0E+05 1.4E+05 2.0E+05 2.6E+05 2.1E+05 1.9E+05 1.6E+05 1.6E+05 1.5E+06 2-Naphthalene sulfonic acid 0 0 0 0 0 7.2E+04 5.7E+04 5.2E+04 0 0 1.2E+06 Benzothiazole 0 0 0 0 0 0 0 0 0 0 5.3E+05 Bisphenol S 0 0 5.9E+03 0 1.8E+04 5.8E+03 0 9.3E+03 6.1E+03 6.5E+03 1.5E+05 Clofibric acid 0 0 0 0 0 0 0 0 0 0 5.3E+03 Cotinine 4.4E+05 3.3E+05 3.5E+05 2.9E+05 3.4E+05 4.5E+05 5.0E+05 4.8E+05 3.0E+05 3.3E+05 7.1E+05 Creatinine 0 0 0 0 1.1E+05 0 0 0 0 0 2.5E+05 Diphenylphosphate 6.8E+04 0 0 0 0 0 0 0 0 0 1.6E+05 Epoxiconazole 0 0 0 0 0 0 0 0 0 0 5.7E+04 Hydrocortisone 0 0 0 0 0 0 0 0 0 0 3.6E+05 Icaridin 0 0 0 0 4.0E+05 2.7E+05 2.6E+05 4.3E+05 0 0 1.4E+06 Mepiquat 6.1E+04 0 0 5.4E+04 0 5.9E+04 5.5E+04 5.2E+04 0 0 1.9E+05 Methylchloroisothiazolinone 0 0 0 0 0 0 0 0 0 0 5.4E+04 Mycophenolic acid 8.5E+04 0 0 0 0 0 0 0 0 0 5.2E+04 Pentoxifylline 0 0 0 0 0 7.1E+04 5.8E+04 0 0 0 6.9E+04 Perfluorobutanoic acid 0 0 0 0 0 0 0 0 0 0 2.3E+05 Phenylbenzimidazolesulfonic acid 4.2E+05 3.1E+05 2.2E+05 2.6E+05 2.4E+05 5.6E+05 4.7E+05 4.3E+05 2.9E+05 2.7E+05 2.3E+06 Tetraglyme 2.5E+05 1.9E+05 1.2E+05 1.3E+05 2.6E+05 4.2E+05 3.7E+05 3.5E+05 2.2E+05 2.3E+05 1.4E+06

S76

Appendix B – Spatial chemical patterns

Compound Name Holt17 Holt18 Holt21 Holt22 Holt26 Holt31 Holt34 Holt36a Holt38 Holt40 Holt42 Tri(butoxyethyl)phosphate 0 0 6.2E+05 0 6.1E+05 9.6E+05 9.8E+05 8.8E+05 5.0E+05 0 1.6E+06 Triethylcitrate 0 0 0 0 0 0 0 0 0 0 1.1E+06 Tris(2-chloroethyl)phosphate 0 0 0 0 0 4.2E+05 3.9E+05 3.5E+05 0 0 8.9E+05 Azelaic acid 0 0 0 0 0 0 0 0 0 0 6.6E+07 Decaethylene Glycol 4.2E+05 3.8E+05 0 0 7.8E+05 3.5E+05 3.4E+05 6.5E+05 3.5E+05 4.3E+05 4.3E+07 Diethylene glycol monobutyl ether 0 0 0 0 0 0 0 0 0 0 3.4E+07 Heptaethylene Glycol 7.5E+05 9.6E+05 4.8E+05 0 1.6E+06 5.2E+05 7.7E+05 1.2E+06 5.7E+05 8.3E+05 9.3E+07 Hexaethyllene Glycol 5.4E+05 8.3E+05 3.9E+05 2.5E+05 1.4E+06 5.9E+05 7.3E+05 8.9E+05 4.9E+05 4.0E+05 5.6E+07 Nonaethylene Glycol 4.8E+05 4.8E+05 3.8E+05 0 1.0E+06 4.0E+05 5.3E+05 7.8E+05 4.2E+05 5.8E+05 5.6E+07 C19H32N6 7.7E+04 6.9E+04 0 0 5.7E+04 6.9E+04 7.4E+04 6.4E+04 5.0E+04 0 4.1E+07 Octaethylene Glycol 5.6E+05 5.5E+05 3.6E+05 0 1.1E+06 4.7E+05 6.1E+05 9.9E+05 4.8E+05 6.0E+05 8.7E+07 Triacetin 0 1.2E+06 1.1E+06 1.2E+06 0 1.6E+06 1.1E+06 1.1E+06 1.3E+06 1.2E+06 6.3E+07

S77

Appendix B – Spatial chemical patterns

Table S 21: Intensities of target and prioritized unknown compounds in river samples belonging to main DRI pattern

m/z rt Compound Name Use Pattern probability Holt3 Holt9 Holt11 Holt13 Holt15

216.1012 10.3 Atrazine Target legacy pesticide 0.8198 0 7.3E+04 6.3E+04 7.0E+04 9.4E+04 178.0545 3.9 Cyclamate Target sweetener 0.9766 2.3E+02 2.5E+02 3.7E+02 2.2E+02 3.1E+02 253.051 9.5 Daidzein Target phytoestrogen 0.9854 0 0 0 0 5.4E+03 188.0702 7.2 Desethylatrazine Target herbice TP of atrazine 0.8631 0 7.5E+04 6.1E+04 6.3E+04 5.2E+04 202.0858 9.1 Desethylterbutylazine Target herbice TP of terbutylazine 0.7420 0 7.1E+04 5.7E+04 6.3E+04 5.2E+04 386.9651 12.4 Fipronil desulfinyl Target insecticide TP of fipronil 0.7278 0 0 0 0 0 269.0459 10.1 Genistein Target phytoestrogen 0.9637 0 0 0 0 8.2E+03 103.0614 0.5 Guanylurea Target pharmaceutical TP of metformin 0.9606 0 0 0 0 0 288.2531 13.2 Lauryl diethanolamide Target surfactant 0.9984 0 1.4E+06 0 1.1E+06 1.2E+06 278.1401 11.7 Metolachlor OA Target herbice TP of metolachlor 0.9999 0 0 0 0 0 359.1262 11.9 Nitrendipin Target pharmaceutical 0.9622 0 0 1.7E+04 0 0 181.992 1.2 Saccharin Target sweetener 0.9010 7.5E+02 2.0E+02 7.1E+02 1.7E+02 2.8E+03 230.1169 11.2 Terbuthylazine Target herbicide 0.5142 1.5E+05 0 6.3E+04 0 5.1E+04 272.2581 12.5 Lauryl betaine Unknown surfactant 0.9999 1.6E+06 2.6E+07 4.2E+06 1.8E+07 1.4E+07 293.1762 12.0 C17H26O4 Unknown 0.9999 0 0 0 0 0 244.227 13.3 N-laurylethnaolamine Unknown surfactant 1.0000 0 1.2E+07 0 6.9E+06 3.2E+06 371.3265 13.7 Myristamidopropyl betaine Unknown surfactant 0.9999 0 1.1E+07 0 7.3E+06 7.5E+06 345.2745 12.6 C19H32N6 Unknown 1.0000 0 2.3E+07 5.6E+05 1.2E+07 7.3E+06 344.2986 12.7 Lauramidopropyl betaine Unknown surfactant 1.0000 0 8.8E+06 4.4E+05 4.9E+06 3.0E+06 165.1121 1.8 Triethylene glycol monomethyl ether Unknown surfactant 0.9999 0 0 0 0 0 265.1482 22.0 Lauryl sulfate Unknown surfactant 0.9999 0 0 0 0 0

S78

Appendix B – Spatial chemical patterns

Table S 21 continued

Compound Name Holt17 Holt18 Holt21 Holt22 Holt26 Holt31 Holt34 Holt36a Holt38 Holt40 Holt42

Atrazine 6.2E+04 6.1E+04 7.0E+04 6.4E+04 8.0E+04 7.3E+04 8.0E+04 7.4E+04 6.3E+04 6.8E+04 0 Cyclamate 0 0 4.0E+02 0 4.7E+02 4.3E+02 4.3E+02 5.0E+02 2.9E+02 2.5E+02 2.7E+02 Daidzein 0 0 0 0 0 0 0 0 0 0 0 Desethylatrazine 5.7E+04 0 0 0 0 5.3E+04 0 0 0 0 0 Desethylterbutylazine 5.5E+04 5.4E+04 5.0E+04 0 5.6E+04 7.9E+04 7.9E+04 7.3E+04 6.3E+04 6.4E+04 0 Fipronil desulfinyl 0 0 0 0 0 0 0 6.5E+03 0 0 0 Genistein 5.4E+03 0 7.4E+03 0 5.6E+03 6.6E+03 0 0 0 0 5.7E+03 Guanylurea 0 0 0 0 0 0 0 0 5.8E+04 0 0 Lauryl diethanolamide 0 0 0 0 0 0 5.8E+05 7.2E+05 0 0 7.5E+05 Metolachlor OA 0 0 0 0 5.7E+03 0 0 0 0 0 0 Nitrendipin 0 0 0 0 0 0 0 0 0 0 0 Saccharin 1.4E+02 2.1E+02 1.5E+02 0 2.3E+02 1.8E+02 1.8E+02 1.6E+02 1.5E+02 1.6E+02 0 Terbuthylazine 0 0 0 0 0 0 5.6E+04 0 0 5.1E+04 0 Lauryl betaine 7.4E+06 5.8E+06 2.8E+06 2.1E+06 3.7E+06 1.9E+06 2.2E+06 2.8E+06 1.8E+06 2.5E+06 6.4E+06 C17H26O4 0 0 0 0 1.4E+07 0 0 0 0 0 0 N-laurylethnaolamine 6.1E+05 6.2E+05 0 0 7.5E+05 0 0 0 0 0 6.5E+05 Myristamidopropyl betaine 0 0 0 0 0 0 0 0 0 0 0 C19H32N6 1.5E+06 1.2E+06 2.2E+05 1.8E+05 8.2E+05 9.2E+04 1.8E+05 3.7E+05 6.4E+04 1.9E+05 3.8E+05 Lauramidopropyl betaine 8.2E+05 5.6E+05 0 0 7.3E+05 0 5.5E+05 4.4E+05 0 0 3.7E+06 Triethylene glycol monomethyl ether 0 0 0 0 8.6E+06 0 2.6E+05 2.6E+05 0 0 0 Lauryl sulfate 0 0 0 0 3.1E+06 0 0 2.7E+06 0 0 0

S79

Appendix B – Spatial chemical patterns

B3.4.2 MS2 spectra of identified compounds (level 1) and standards

Gabapentin-lactam [M+H]+: m/z 154.1226 Smiles: O=C1CC2(CN1)CCCCC2 InChIKey: JAWPQJDOQPSNIQ-UHFFFAOYSA-N

Sample Holt17 Reference standard RT: 8.24 min RT: 8.15 min

Figure S 9: MS2 spectra of gabapentin-lactam in original sample and reference standard.

4-methyl-7-ethylaminocoumarin [M+H]+: m/z 204.1019 Smiles: CCNC1=CC=C2C(C)=CC(=O)OC2=C1 InChIKey: OTNIKUTWXUODJZ-UHFFFAOYSA-N

Sample Holt17 Reference standard RT: 9.78 min RT: 9.73 min

Figure S 10: MS2 spectra of 4-methyl-7-ethylaminocoumarin in original sample and reference standard.

Appendix B – Spatial chemical patterns

Lamotrigine [M+H]+: m/z 256.0149 Smiles: NC1=NC(N)=C(N=N1)C1=CC=CC(Cl)=C1Cl InChIKey: PYZRQGJRPPTADH-UHFFFAOYSA-N

Sample Holt17 Reference standard RT: 4.12 min RT: 3.81 min

Figure S 11: MS2 spectra of lamotrigine in original sample and reference standard.

Methocarbamol [M+H]+: m/z 242.1021 Smiles: COC1=C(OCC(O)COC(N)=O)C=CC=C1 InChIKey: GNXFOGHNGIVQEH-UHFFFAOYSA-N

Sample Holt29 Reference standard RT: 7.25 min RT: 7.16 min

Figure S 12: MS2 spectra of methocarbamol in original sample and reference standard.

S81

Appendix B – Spatial chemical patterns

Valsartan acid [M+H]+: m/z 267.0871 Smiles: OC(=O)C1=CC=C(C=C1)C1=CC=CC=C1C1=NN=NN1 InChIKey: USAWIVMZUYOXCF-UHFFFAOYSA-N

Sample Holt29 Reference standard RT: 9.16 min RT: 9.14 min

Figure S 13: MS2 spectra of valsartan acid in original sample and reference standard in positive ionization.

Valsartan acid [M-H]-: m/z 265.0729

Sample Holt29 Reference standard RT: 9.16 min RT: 9.10 min

Figure S 14: MS2 spectra of valsartan acid in original sample and reference standard in negative ionization.

S82

Appendix B – Spatial chemical patterns

Irbesartan [M+H]+: m/z 429.2397 Smiles: YOSHYTLCDANDAN-UHFFFAOYSA-N InChIKey: CCCCC1=NC2(CCCC2)C(=O)N1CC1=CC=C(C=C1)C1=CC=CC=C1C1=NN=NN1

Sample Holt29 Reference standard RT: 11.0 min RT: 10.92 min

Figure S 15: MS2 spectra of irbesartan in original sample and reference standard.

Olmesartan [M+H]+: m/z 447.2136 Smiles: CCCC1=NC(=C(N1CC1=CC=C(C=C1)C1=C(C=CC=C1)C1=NN=NN1)C(O)=O)C(C)(C)O InChIKey: VTRAEEWXHOVJFV-UHFFFAOYSA-N

Sample Holt29 Reference standard RT: 8.95 min RT: 8.93 min

Figure S 16: MS2 spectra of olmesartan in original sample and reference standard.

S83

Appendix B – Spatial chemical patterns

Triethylene glycol monoethyl ether [M+H]+: m/z 165.1121 Smiles: COCCOCCOCCO InChIKey: JLGLQAWTXXGVEM-UHFFFAOYSA-N

Sample Holt26 Reference standard RT: 1.65 min RT: 1.55 min

Figure S 17: MS2 spectra of triethylene glycol monoethyl ether in original sample and reference standard.

N-lauroylethanolamine [M+H]+: m/z 244.227 Smiles: CCCCCCCCCCCC(=O)NCCO InChIKey: QZXSMBBFBXPQHI-UHFFFAOYSA-N

Sample Holt9 Reference standard RT: 13.03 min RT: 12.97 min

Figure S 18: MS2 spectra of n-lauroylethanolamine in original sample and reference standard.

S84

Appendix B – Spatial chemical patterns

Lauryl betaine [M+H]+: m/z 272.2581 Smiles: CCCCCCCCCCCC[N+](C)(C)CC(O)=O InChIKey: DVEKCXOJTLDBFE-UHFFFAOYSA-N

Sample Holt9 Reference standard RT: 12.24 min RT: 12.21 min

Figure S 19: MS2 spectra of lauryl betaine in original sample and reference standard.

Lauramidopropyl betaine [M+H]+: m/z 343.2953 Smiles: CCCCCCCCCCCC(=O)NCCC[N+](C)(C)CC(O)=O InChIKey: MRUAUOIMASANKQ-UHFFFAOYSA-N

Sample Holt9 Reference standard RT: 12.41 min RT: 12.36 min

Figure S 20: MS2 spectra of lauramidopropyl betaine in original sample and reference standard.

S85

Appendix B – Spatial chemical patterns

Myristamidopropyl betaine [M+H]+: m/z 371.3265 Smiles: CCCCCCCCCCCCCC(=O)NCCC[N+](C)(C)CC(O)=O InChIKey: QGCUAFIULMNFPJ-UHFFFAOYSA-N

Sample Holt9 Reference standard RT: 13.48 min RT: 13.43 min

Figure S 21: MS2 spectra of myristamidopropyl betaine in original sample and reference standard.

Diethylene glycol monobutyl ether [M+H]+: m/z 163.1328 Smiles: CCCCOCCOCCO InChIKey: OAYXUHPQHDHDDZ-UHFFFAOYSA-N

Sample Holt42 Reference standard RT: 7.89 min RT: 7.46 min

Figure S 22: MS2 spectra of diethylene glycol monobutyl ether in original sample and reference standard.

S86

Appendix B – Spatial chemical patterns

Triacetin [M+NH4]+: m/z 236.1127 Smiles: CC(=O)OCC(COC(C)=O)OC(C)=O InChIKey: URAYPUMNDPQOKB-UHFFFAOYSA-N

Sample Holt42 Reference standard RT: 6.91 min RT: 6.77 min

Figure S 23: MS2 spectra of triacetin in original sample and reference standard.

Hexaethylene glycol (PEG-6) [M+NH4]+: m/z 300.2014 Smiles: C(COCCOCCOCCOCCOCCO)O InChIKey: IIRDTKBZINWQAW-UHFFFAOYSA-N

Sample Holt42 Reference standard RT: 5.24 min RT: 4.96 min

Figure S 24: MS2 spectra of hexaethylene glycol in original sample and reference standard.

S87

Appendix B – Spatial chemical patterns

Heptaethylene glycol (PEG-7) [M+NH4]+: m/z 344.2276 Smiles: C(COCCOCCOCCOCCOCCOCCO)O InChIKey: XPJRQAIZZQMSCM-UHFFFAOYSA-N

Sample Holt42 Reference standard RT: 6.41 min RT: 6.34 min

Figure S 25: MS2 spectra of heptaethylene glycol in original sample and reference standard.

Octaethylene glycol (PEG-8) [M+NH4]+: m/z 388.2539 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCO InChIKey: GLZWNFNQMJAZGY-UHFFFAOYSA-N

Sample Holt42 Reference standard RT: 7.00 min RT: 6.94 min

Figure S 26: MS2 spectra of octaethylene glycol in original sample and reference standard.

S88

Appendix B – Spatial chemical patterns

Nonaethylene glycol (PEG-9) [M+NH4]+: m/z 432.2802 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCOCCO InChIKey: YZUUTMGDONTGTN-UHFFFAOYSA-N

Sample Holt42 Reference standard RT: 7.40 min RT: 7.37 min

Figure S 27: MS2 spectra of nonaethylene glycol in original sample and reference standard.

Decaethylene glycol (PEG-10) [M+NH4]+: m/z 476.3065 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCOCCOCCO InChIKey: DTPCFIHYWYONMD-UHFFFAOYSA-N

Sample Holt42 Reference standard RT: 7.73 min RT: 7.69 min

Figure S 28: MS2 spectra of decaethylene glycol in original sample and reference standard.

S89

Appendix B – Spatial chemical patterns

Azelaic acid [M-H]-: m/z 187.0979 Smiles: OC(=O)CCCCCCCC(O)=O InChIKey: BDJRBEYXGGNYIS-UHFFFAOYSA-N

Sample Holt42 Reference standard RT: 8.57 min RT: 8.48 min

Figure S 29: MS2 spectra of azelaic acid in original sample and reference standard.

S90

Appendix B – Spatial chemical patterns

B3.4.3 MS2 spectra of level 2 compounds

- - [C9H9O3S] [C8H7O3S] [O S]- - - 3 [O S] 183.0123 [C9H9O3S] 3 1 [C H O S]- 8 8 7 3 8 9 1 Figure S 30: Chromatograms7 and MS2 spectra of C6-SPC and C7-SPC. Diagnostic fragments matching previous studies8 (Gonsior et al., 2011, Lara-Martín et al., 2011) are marked. 3

S91

Appendix B – Spatial chemical patterns

The MS2 spectrum for lauryl sulfate was extracted from MassBank (Horai et al., 2010) (MassBank Record AU406656). The spectrum is registered under splash10-0002-9010000000- 3a02b86200a1303cde64. Alygizakis et al. (2019) performed the chemical analysis an LC-ESI- QTOF.

Figure S 31: MS2 spectra for lauryl sulfate from MassBank (top) and from original sample (bottom).

S92

Appendix B – Spatial chemical patterns

B3.4.4 pH-dependent LC retention time shifts of m/z 274.2010 At pH 2.6, only one high intensity peak was detected for m/z 274.2010 at 7.7 min. At pH 6.4, the peak separated into two peaks at 6.93 min and 8.87 min. Further increase of the pH led to further reduced retention time of one of the peaks (4.67 min). A reduction of retention time with increasing pH suggests the presence of an acidic group in the structure. Due to the absence of sulfur this indicated a carboxylic acid. An increase of retention time with increasing pH suggested the presence of a basic group for the second peak.

pH = 2.6

pH = 6.4

pH = 10

Figure S 32: pH-dependent LC retention time shift of m/z 274.2010.

S93

Appendix B – Spatial chemical patterns

References – Appendix B Alygizakis, N. A.; Gago-Ferrero, P.; Hollender, J.; Thomaidis, N. S. Untargeted time- pattern analysis of LC-HRMS data to detect spills and compounds with high fluctuation in influent wastewater. Journal of Hazardous Materials 2019, 361, 19-29. https://doi.org/10.1016/j.jhazmat.2018.08.073 Genolini, C.; Alacoque, X.; Sentenac, M.; Arnaud, C. kml and kml3d: R Packages to Cluster Longitudinal Data. Journal of Statistical Software 2015, 65, (4), 34. 10.18637/jss.v065.i04 Gonsior, M.; Zwartjes, M.; Cooper, W. J.; Song, W.; Ishida, K. P.; Tseng, L. Y.; Jeung, M. K.; Rosso, D.; Hertkorn, N.; Schmitt-Kopplin, P. Molecular characterization of effluent organic matter identified by ultrahigh resolution mass spectrometry. Water Research 2011, 45, (9), 2943- 2953. https://doi.org/10.1016/j.watres.2011.03.016 Horai, H.; Arita, M.; Kanaya, S.; Nihei, Y.; Ikeda, T.; Suwa, K.; Ojima, Y.; Tanaka, K.; Tanaka, S.; Aoshima, K.; Oda, Y.; Kakazu, Y.; Kusano, M.; Tohge, T.; Matsuda, F.; Sawada, Y.; Hirai, M. Y.; Nakanishi, H.; Ikeda, K.; Akimoto, N.; Maoka, T.; Takahashi, H.; Ara, T.; Sakurai, N.; Suzuki, H.; Shibata, D.; Neumann, S.; Iida, T.; Tanaka, K.; Funatsu, K.; Matsuura, F.; Soga, T.; Taguchi, R.; Saito, K.; Nishioka, T. MassBank: a public repository for sharing mass spectral data for life sciences. Journal of Mass Spectrometry 2010, 45, (7), 703-714. 10.1002/jms.1777 Montero, P.; Vilar, J. A. TSclust: An R Package for Time Series Clustering. Journal of Statistical Software 2014, 62, (1), 43. 10.18637/jss.v062.i01 Lara-Martín, P. A.; González-Mazo, E.; Brownawell, B. J. Multi-residue method for the analysis of synthetic surfactants and their degradation metabolites in aquatic systems by liquid chromatography–time-of-flight-mass spectrometry. Journal of Chromatography A 2011, 1218, (30), 4799-4807. https://doi.org/10.1016/j.chroma.2011.02.031 LHW (2019), Datenportal Gewässerkundlicher Landesdienst Sachsen-Anhalt (Database State Waterways Service) (accessed on 28.01.2019). http://gldweb.dhi-wasy.com/gld-portal/ Royal Society of Chemistry, ChemSpider. http://www.chemspider.com/ Ruttkies, C.; Schymanski, E. L.; Wolf, S.; Hollender, J.; Neumann, S. MetFrag relaunched: incorporating strategies beyond in silico fragmentation. Journal of cheminformatics 2016, 8, 3-3. 10.1186/s13321-016-0115-9

S94

Appendix C – Precipitation-related chemical patterns

Appendix C - Identification of complex pollutant patterns and representative organic micropollutants during heavy rain events in a small stream

C1 Material and Methods

C1.1 Chemicals and reagents Table S 22: Chemicals and reagents used for chemical analysis by LC-HRMS

C1.2 Settings LC-HRMS Parameters and settings for chromatographic separation are provided in Table S23. A Kinetex 2.6 μm EVO C18 (50x2.1 mm) column equipped with a pre-column (C18 EVO 5x2.1 mm) and an inline filter was used. The column temperature was 40°C. Parameters and settings for fullscan experiments in positive and negative ionization mode on the Q ExactiveTM Plus (Thermo Fisher) instrument are presented in Table S24. Data-dependent MS² experiments were run for prioritized ions of interests in positive ionization mode. The nominal resolving power was 70,000 (referenced to m/z 200) in fullscan mode and 35,000 (referenced to m/z 200) in dd-MS2 scans.

Table S 23: Solvent Gradient elution program for chromatographic separation on LC

Appendix C – Precipitation-related chemical patterns

Table S 24: Settings of Q ExactiveTM Plus (Thermo Fisher) for fullscan experiments

C1.3 Settings MZmine MZmine v2.32 was used for automated peak picking in the event samples. The parameters and settings of the MZmine software are presented in Table S25. For meta-alignment of clustered peaklists, the retention time tolerance in the join aligner was increased to 0.6 min to account for normal retention time shifts between analysis batches.

Table S 25: Parameters and settings of the MZmine software

Appendix C – Precipitation-related chemical patterns

Step Parameter Setting m/z tolerance 0.001 m/z or 7 ppm Retention time tolerance 0.4 (absolute) min Gap filling Intensity tolerance 30 % m/z tolerance 0.001 m/z or 7 ppm Retention time tolerance 0.15 (absolute) min RT correction yes

C1.4 Settings R package ‘nontarget’ R ‘nontarget’ was used for identification of isotopes in the peaklists exported from MZmine for each event. Moreover, it was used for the identification of adducts and homologue series in clustered peaklists. The settings are presented in Table S26.

Table S 26: Parameters and setting for R ‘nontarget’

S97

Appendix C – Precipitation-related chemical patterns

C1.5 Settings R package ‘kml’ The script of the R package ‘kml’ is provided below (Genolini et al., 2015). The appropriate number of cluster is chosen based on a consensus score of the incorporated quality criteria (section C2.3). After clustering, original peaklists with cluster assignment are extracted.

#LongData object with normalized data event_long <- clusterLongData(event_sc1, varNames = "Peak Height")

#perform kml analysis kml(event_long, nbClusters = 2:6, nbRedrawing = 10,toPlot= "none", parAlgo = parALGO(saveFreq= 100,startingCond= "kmeans-", distanceName = "ignore", distance = diss.CORT))

#plot all criteria to choose cluster number plotAllCriterion(event_long)

#plot overview results for selected number of clusters plot(event_long,3)

#extract cluster members cluster_name <-getClusters(event_long,3) cluster_name2 <- as.data.frame(cluster_name)

#get probabilities for cluster fitting/ cluster evaluation cluster_probability <-getBestPostProba(event_long, nbCluster =3, clusterRank = 1) cluster_probability2 <- as.data.frame(cluster_probability)

## get cluster probabilities for all cluster cluster_allprobs2 <- event_long@c3[[1]]@postProba

S98

Appendix C – Precipitation-related chemical patterns

C2 Results

C2.1 Internal standards in samples The intensity of internal standards in samples of each event is presented in Figures S33-38. In general, no changes in the intensity of internal standards and thus changes in matrix effects were observed over the course of each event.

Figure S 33: Variation of internal standards in samples of E2905.

Figure S 34: Variation of internal standards in samples of E0106.

S99

Appendix C – Precipitation-related chemical patterns

Figure S 35: Variation of internal standards in samples of E1306.

Figure S 36: Variation of internal standards in samples of E2406.

S100

Appendix C – Precipitation-related chemical patterns

Figure S 37: Variation of internal standards in samples of E1307.

Figure S 38: Variation of internal standards in samples of E1709.

S101

Appendix C – Precipitation-related chemical patterns

C2.2 Discharge at catchment gauging stations during events Discharge is displayed for the two gauging station at the Holtemme River. Gauge 1 is upstream of the WWTP with CSO; gauge 2 is at the sampling station (Figure 20).

E2905

Figure S 39: Discharge at catchment gauges from 28.5.-30.5.2016.

E0106

Figure S 40: Discharge at catchment gauges from 31.5.-2.6.2016.

S102

Appendix C – Precipitation-related chemical patterns

E1306

Figure S 41: Discharge at catchment gauges from 12.-14.6.2016.

E2406

Figure S 42: Discharge at catchment gauges from 23. - 25.6.2016.

E1307

Figure S 43: Discharge at catchment gauges from 12.-14.7.2016.

S103

Appendix C – Precipitation-related chemical patterns

E1709

Figure S 44: Discharge at catchment gauges from 16.-18.9.2016.

S104

Appendix C – Precipitation-related chemical patterns

C2.3 Quality criteria cluster analysis The appropriate number of clusters was chosen based on a consensus score of all quality criteria. The quality criteria should be maximized indicating high between-cluster variability and low within- cluster variability (Genolini et al., 2015).

Figure S 45: Quality criteria of cluster analysis on data set E2905. Line numbers refer to different quality criteria. 1) Calinski-Harabatz, 2) Calinski-Harabatz2, 3) Calinski-Harabatz3, 4) Ray- Turi, 5) Davies-Bouldin.

Figure S 46: Quality criteria of cluster analysis on data set E0106. Line numbers refer to different quality criteria. 1) Calinski-Harabatz, 2) Calinski-Harabatz2, 3) Calinski-Harabatz3, 4) Ray- Turi, 5) Davies-Bouldin.

S105

Appendix C – Precipitation-related chemical patterns

Figure S 47: Quality criteria of cluster analysis on data set E1306. Line numbers refer to different quality criteria. 1) Calinski-Harabatz, 2) Calinski-Harabatz2, 3) Calinski-Harabatz3, 4) Ray- Turi, 5) Davies-Bouldin.

Figure S 48: Quality criteria of cluster analysis on data set E2406. Line numbers refer to different quality criteria. 1) Calinski-Harabatz, 2) Calinski-Harabatz2, 3) Calinski-Harabatz3, 4) Ray- Turi, 5) Davies-Bouldin.

S106

Appendix C – Precipitation-related chemical patterns

Figure S 49: Quality criteria of cluster analysis on data set E1307. Line numbers refer to different quality criteria. 1) Calinski-Harabatz, 2) Calinski-Harabatz2, 3) Calinski-Harabatz3, 4) Ray- Turi, 5) Davies-Bouldin.

Figure S 50: Quality criteria of cluster analysis on data set E1709. Line numbers refer to different quality criteria. 1) Calinski-Harabatz, 2) Calinski-Harabatz2, 3) Calinski-Harabatz3, 4) Ray- Turi, 5) Davies-Bouldin.

S107

Appendix C – Precipitation-related chemical patterns

C2.4 Physicochemical parameters measured at sampling station during events

Figure S 51: Physicochemical parameters during E2905.

Figure S 52: Physicochemical parameters during E0106.

S108

Appendix C – Precipitation-related chemical patterns

Figure S 53: Physicochemical parameters during E1306.

Figure S 54: Physicochemical parameters during E2406.

S109

Appendix C – Precipitation-related chemical patterns

Figure S 55: Physicochemical parameters during E1307.

Figure S 56: Physicochemical parameters during E1709.

S110

Appendix C – Precipitation-related chemical patterns

C2.5 Identification of representative compounds for rain event mixtures Peak heights of target and identified unknown compounds are presented in Table S27: Target and identified unknown organic micropollutants detected in surface water samples taken during heavy rain events (available under http://doi.org/10.5281/zenodo.3379087 or as csv. data file on CD).

C2.5.1 MS2 spectra of confirmed compounds (level 1)

Chenodeoxycholic acid [M+H]+: m/z 410.3262 Smiles: C[C@H](CCC(O)=O)[C@H]1CC[C@H]2[C@@H]3[C@H](O)C[C@@H]4C[C@H](O)CC[C@]4 (C)[C@H]3CC[C@]12C InChIKey: RUDATBOHQWOJDD-BSWAIDMHSA-N

Sample E2905_B14 Reference standard RT: 13.87 min RT: 13.79 min

Figure S 57: MS2 spectra of chenodeoxycholic acid in original sample and reference standard (hcd = 35).

Sample E2905_B14 Reference standard RT: 13.87 min RT: 13.79 min

Figure S 58: MS2 spectra of chenodeoxycholic acid in original sample and reference standard (hcd = 55).

S111

Appendix C – Precipitation-related chemical patterns

Lauyl betaine [M+H]+: m/z 271.251 Smiles: CC(=O)N(CCN(C(=O)C)C(=O)C)C(=O)C InChIKey: BGRWYDHXPHLNKA-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 12.37 min RT: 12.21 min

Figure S 59: MS2 spectra of lauryl betaine in original sample and reference standard.

Tetraacetylethylenediamine [M+Na]+: m/z 251.0099 Smiles: CC(=O)N(CCN(C(=O)C)C(=O)C)C(=O)C InChIKey: BGRWYDHXPHLNKA-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 6.48 min RT: 6.49 min

Figure S 60: MS2 spectra of tetraacetylethylenediamine in original sample and reference standard.

S112

Appendix C – Precipitation-related chemical patterns

Triethylene glycol (PEG-3) [M+NH4]+: m/z 168.1228 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCO InChIKey: GLZWNFNQMJAZGY-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 1.09 min RT:1.08 min

Figure S 61: MS2 spectra of triethylene glycol in original sample and reference standard.

Octaethylene glycol (PEG-8) [M+NH4]+: m/z 388.2538 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCO InChIKey: GLZWNFNQMJAZGY-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 7.10 min RT: 7.10 min

Figure S 62: MS2 spectra of octaethylene glycol in original sample and reference standard.

S113

Appendix C – Precipitation-related chemical patterns

Nonaethylene glycol (PEG-9) [M+NH4]+: m/z 432.2801 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCOCCO InChIKey: YZUUTMGDONTGTN-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 7.51 min RT: 7.50 min

Figure S 63: MS2 spectra of nonaethylene glycol in original sample and reference standard.

Decaethylene glycol (PEG-10) [M+NH4]+: m/z 476.3063 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCOCCOCCO InChIKey: DTPCFIHYWYONMD-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 7.83 min RT: 7.84 min

Figure S 64: MS2 spectra of decaethylene glycol in original sample and reference standard.

S114

Appendix C – Precipitation-related chemical patterns

Undecaethylene glycol (PEG-11) [M+NH4]+: m/z 520.3327 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCOCCOCCOCCO InChIKey: PSVXZQVXSXSQRO-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 8.10 min RT: 8.11 min

Figure S 65: MS2 spectra of undecaethylene glycol in original sample and reference standard.

Tridecaethylene glycol (PEG-13) [M+NH4]+: m/z 608.385 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCOCCOCCOCCOCCOCCO InChIKey: AKWFJQNBHYVIPY-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 8.10 min RT: 8.11 min

Figure S 66: MS2 spectra of tridecaethylene glycol in original sample and reference standard.

S115

Appendix C – Precipitation-related chemical patterns

Pentaglycol ether sulfate [M+NH4]+: m/z 424.1844 Smiles: OCCOCCOCCOCCOCCOCCOCCOS(O)(=O)=O InChIKey: JDFRNFXDFNZJQH-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 6.98 min RT: 6.93 min

Figure S 67: MS2 spectra of pentaglycol ether sulfate in original sample and reference standard.

Hexaglycol ether sulfate [M+NH4]+: m/z 468.2108 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCOS(O)(=O)=O InChIKey: QJIJGAHDHRUUCX-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 7.51 min RT: 7.47 min

Figure S 68: MS2 spectra of hexaglycol ether sulfate in original sample and reference standard.

S116

Appendix C – Precipitation-related chemical patterns

Heptaglycol ether sulfate [M+NH4]+: m/z 512.2371 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCOCCOS(O)(=O)=O InChIKey: MVTDSFRJTVUNEA-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 7.96 min RT: 7.93 min

Figure S 69: MS2 spectra of heptaglycol ether sulfate in original sample and reference standard.

Octaglycol ether sulfate [M+NH4]+: m/z 556.2633 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCOCCOCCOS(O)(=O)=O InChIKey: GAZVTRDRUQJIGC-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 8.31 min RT: 8.26 min

Figure S 70: MS2 spectra of octaglycol ether sulfate in original sample and reference standard.

S117

Appendix C – Precipitation-related chemical patterns

Nonaglycol ether sulfate [M+NH4]+: m/z 600.2893 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCOCCOCCOCCOS(O)(=O)=O InChIKey: TWHSEFKMFIDNSZ-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 8.62 min RT: 8.54 min

Figure S 71: MS2 spectra of nonaglycol ether sulfate in original sample and reference standard.

Decaglycol ether sulfate [M+NH4]+: m/z 644.3155 Smiles: OCCOCCOCCOCCOCCOCCOCCOCCOCCOCCOCCOCCOS(O)(=O)=O InChIKey: SQCRLUZZBXNMHC-UHFFFAOYSA-N

Sample E2905_B14 Reference standard RT: 8.90 min RT: 8.86 min

Figure S 72: MS2 spectra of decaglycol ether sulfate in original sample and reference standard.

S118

Appendix C – Precipitation-related chemical patterns

C2.5.2 Results of structure elucidation of prioritized peaks The results of structure elucidation of prioritized peaks are presented in Table S28. The identification of ions of interests focused on high- intensity compounds in the rain-related mixture as well as homologue series which were detected also in the small rain event E2406.

Table S 28: Identified ions of interests in rain-related mixture. Confidence levels are given accordingly to Schymanski et al. (2014).

m/z rt exact adduct molecular formula (Tentative) Compound name Main Level mass fragments 410.3262 13.8 392.293 [M+NH4]+ C24H40O4 Chenodeoxycholic acid 1 272.2582 12.3 271.251 [M+H]+ C16H35NO2 Lauryl betaine 1 251.1 6.2 228.1108 [M+Na]+ C10H16N2O4 Tetraacetylethylenediamine 1

Cn*2Hn*4+2Hn+1 PEG 168.123 1.0 150.0892 [M+NH4]+ C6H14O4 Triethylene gylcol PEG-3 1 388.2538 7.0 370.22 [M+NH4]+ C16H34O9 Octaethylene glycol PEG-8 1 432.2801 7.4 414.2463 [M+NH4]+ C18H38O10 Nonaethylene glycol PEG-9 1 476.3063 7.7 458.2725 [M+NH4]+ C20H42O11 Decathylene glycol PEG-10 1 520.3327 8.0 502.2989 [M+NH4]+ C22H46O12 Undecathylene glycol PEG-11 1 608.385 8.5 590.3512 [M+NH4]+ C26H54O14 Tridecathylene glycol PEG-13 1 AES, homologue series #7 and #4 (ESIneg) 424.1844 6.8 406.1514 [M+NH4]+ C14H30O11S Pentaglycol ether sulfate 1 468.2108 7.4 450.1778 [M+NH4]+ C16H34O12S Hexaglycol ether sulfate 1 512.2371 7.8 494.204 [M+NH4]+ C18H38O13S Heptagylcol ether sulfate 1 556.2633 8.2 538.2304 [M+NH4]+ C20H42O14S Octaglycol ether sulfate 1 600.2893 8.5 582.2555 [M+NH4]+ C22H46O15S Nonaglycol ether sulfate 1 644.3155 8.8 626.2817 [M+NH4]+ C24H50O16S Decaglycol ether sulfate 1

145.0971 0.9 144.0899 [M+H]+ C6H12N2O2 4 283.2629 14.3 282.2557 [M+H]+ C18H34O2 4 357.2782 13.8 356.271 [M+H]+ C24H36O2 4

CH3(CH2)9(CH2CH2O)nOH homologue series #7 440.358 13.6 422.3248 [M+NH4]+ C22H46O7 3,6,9,12,15,18-Hexaoxaoctacosan-1-ol 89,133,177 3 484.3843 13.7 466.3511 [M+NH4]+ C24H50O8 3,6,9,12,15,18,21-Heptaoxaoctacosan-1-ol 89,133,177 3 528.4106 13.7 510.3774 [M+NH4]+ C26H54O9 3,6,9,12,15,18,21,24-Octaoxatetratriacontan-1-ol 89,133,177 3 572.4366 13.7 554.4034 [M+NH4]+ C28H58O10 3,6,9,12,15,18,21,24,27-Nonaoxaheptatriacontan-1-ol 89,133,177 3

119

Appendix C – Precipitation-related chemical patterns

m/z rt exact adduct molecular formula (Tentative) Compound name Main Level mass fragments COOHCH2(CH2OCH2)nCH2- homologue series #7 COOH 444.244 7.9 426.2108 [M+NH4]+ C18H34O11 4,7,10,13,16,19,22-Heptaoxapentacosane-1,25-dioic acid 59,103,133,147 3 488.2699 8.2 470.2367 [M+NH4]+ C20H38O12 4,7,10,13,16,19,22,25-Octaoxaoctacosane-1,28-dioic acid 59,103,133,147 3 532.2964 8.5 514.2632 [M+NH4]+ C22H42O13 4,7,10,13,16,19,22,25,28-Nonaoxahentriacontane-1,31-dioic acid 59,103,133,147 3 576.3224 8.7 558.2892 [M+NH4]+ C24H46O14 4,7,10,13,16,19,22,25,28,31-Decaoxatetratriacontane-1,34-dioic acid 59,103,133,147 3 620.3485 9.0 602.3153 [M+NH4]+ C26H50O15 4,7,10,13,16,19,22,25,28,31,34-Undecaoxaheptatriacontane-1,37- 59,103,133,147 3 dioic acid 664.3745 9.1 646.3413 [M+NH4]+ C28H54O16 59,103,133,147 3

COOH(CH2CH2O)nCH3 homologue series #7 342.2119 7.4 324.1787 [M+NH4]+ C14H28O8 2,5,8,11,14,17-Hexaoxaicosan-20-oic acid 89,133,177 3 386.2381 7.5 368.2049 [M+NH4]+ C16H32O9 2,5,8,11,14,17,20-Heptaoxatricosan-23-oic acid 89,133,177 3 430.2644 7.9 412.2312 [M+NH4]+ C18H36O10 2,5,8,11,14,17,20,23-Octaoxahexacosan-26-oic acid 89,133,177 3 474.2907 8.2 456.2575 [M+NH4]+ C20H40O11 2,5,8,11,14,17,20,23,26-Nonaoxanonacosan-29-oic acid 89,133,177 3 518.3171 8.4 500.2839 [M+NH4]+ C22H44O12 2,5,8,11,14,17,20,23,26,29-Decaoxadotriacontan-32-oic acid 89,133,177 3 562.3431 8.7 544.3099 [M+NH4]+ C24H48O13 2,5,8,11,14,17,20,23,26,29,32-Undecaoxapentatriacontan-35-oic acid 89,133,177 3 606.3693 8.9 588.3361 [M+NH4]+ C26H52O14 2,5,8,11,14,17,20,23,26,29,32,35-Dodecaoxaoctatriacontan-38-oic 89,133,177 3 acid 650.3954 9.1 632.3622 [M+NH4]+ C28H56O15 2,5,8,11,14,17,20,23,26,29,32,35,38-Tridecaoxahentetracontan-41-oic 89,133,177 3 acid COOHCH2O(CH2CH2O)n-2- homologue series #7 CH2COOH 372.1861 6.4 354.1529 [M+NH4]+ C14H26O10 3,6,9,12,15,18-Hexaoxaicosane-1,20-dioic acid 59,103,133,147 3 416.2125 7.0 398.1793 [M+NH4]+ C16H30O11 3,6,9,12,15,18,21-Heptaoxatricosane-1,23-dioic acid 59,103,133,147 3 460.2385 7.5 442.2053 [M+NH4]+ C18H34O12 3,6,9,12,15,18,21,24-Octaoxahexacosane-1,26-dioic acid 59,103,133,147 3 504.2648 7.8 486.2316 [M+NH4]+ C20H38O13 3,6,9,12,15,18,21,24,27-Nonaoxanonacosane-1,29-dioic acid 59,103,133,147 3 548.2911 8.1 530.2579 [M+NH4]+ C22H42O14 59,103,133,147 3 592.3173 8.3 574.2841 [M+NH4]+ C24H44O15 59,103,133,147 3 636.3433 8.6 618.3101 [M+NH4]+ C26H48O16 3,6,9,12,15,18,21,24,27,30,33-Undecaoxapentatriacontane-1,35-dioic 59,103,133,147 3 acid 680.3696 8.8 662.3364 [M+NH4]+ C28H52O17 59,103,133,147 3 homologue series #7 + 363.1621 6.4 340.1729 [M+Na] C14H28O9 C3H6O loss 4 + 407.1886 7.0 384.1994 [M+Na] C16H32O10 C3H6O loss 4 + 451.2147 7.4 428.2255 [M+Na] C18H36O11 C3H6O loss 4 + 495.241 7.8 472.2518 [M+Na] C20H40O12 C3H6O loss 4 + 539.2673 8.0 516.2781 [M+Na] C22H44O13 C3H6O loss 4

S120

Appendix C – Precipitation-related chemical patterns

m/z rt exact adduct molecular formula (Tentative) Compound name Main Level mass fragments homologue series #7 324.2378 8.9 306.2046 [M+NH4]+ C15H30O6 59,88,115 4 382.2796 10.0 364.2464 [M+NH4]+ C18H36O7 59,88,115 4 440.3216 10.9 422.2884 [M+NH4]+ C21H42O8 59,88,115 4 498.3635 11.6 480.3303 [M+NH4]+ C24H48O9 59,88,115 4 556.4054 12.2 538.3722 [M+NH4]+ C27H54O10 59,88,115 4 614.447 12.7 596.4138 [M+NH4]+ C30H60O11 59,88,115 4 672.4889 13.1 654.4557 [M+NH4]+ C33H66O12 59,88,115 4 homologue series #7 466.1952 8.0 448.162 [M+NH4]+ 89,113,133 5 510.2217 8.4 492.1885 [M+NH4]+ 89,113,133 5 554.2477 8.7 536.2145 [M+NH4]+ 89,113,133 5 598.2738 9.0 580.2406 [M+NH4]+ 89,113,133 5

S121

Appendix C – Precipitation-related chemical patterns

C2.5.3 MS2 spectra of ions of interests in homologue series MS2 spectra of ions in detected homologue series in the rain-related mixture are presented in Figures S73 - 82. Spectra are presented for ions identified to level 3, 4 and 5 (Schymanski et al.,

2014). The fragments m/z 89, 133 and 177 correspond to a (CH2CH2O)n group with n = 2-4. The fragment m/z 58 represented a diagnostic loss of a C3H6O group in homologue series 5 (Figure S75).

Figure S 73: MS2 spectra of HS2. Main fragments = m/z 89,133 and 177.

122

Appendix C – Precipitation-related chemical patterns

Figure S 74: MS2 spectra of HS2. Main fragments = m/z 89,133 and 177.

Figure S 75: MS2 spectra of HS5. Diagnostic loss of m/z 58.

S123

Appendix C – Precipitation-related chemical patterns

Figure S 76: MS2 spectra of HS6. Main fragments = m/z 59,103, 133 and 147.

Figure S 77: MS2 spectra of HS8 (part 1). Main fragments = m/z 59, 88 and 115.

S124

Appendix C – Precipitation-related chemical patterns

Figure S 78: MS2 spectra of HS8 (part 2). Main fragments = m/z 59, 88 and 115.

Figure S 79: MS2 spectra of HS9 (part 1). Main fragments = m/z 89, 113, 133 and 177.

S125

Appendix C – Precipitation-related chemical patterns

Figure S 80: MS2 spectra of HS9 (part 2). Main fragments = m/z 89, 113, 133 and 177.

Figure S 81: MS2 spectra of HS10 (part 1). Main fragments = m/z 103, 133 and 147.

S126

Appendix C – Precipitation-related chemical patterns

Figure S 82: MS2 spectra of HS10 (part 2). Main fragments = m/z 103, 133 and 147.

S127

Appendix C – Precipitation-related chemical patterns

References - Appendix C Genolini, C., Alacoque, X., Sentenac, M. and Arnaud, C. (2015) kml and kml3d: R Packages to Cluster Longitudinal Data. Journal of Statistical Software 65(4), 34. 10.18637/jss.v065.i04 Schymanski, E.L., Jeon, J., Gulde, R., Fenner, K., Ruff, M., Singer, H.P. and Hollender, J. (2014) Identifying Small Molecules via High Resolution Mass Spectrometry: Communicating Confidence. Environmental Science & Technology 48(4), 2097-2098. 10.1021/es5002105

S128

Acknowledgement

Acknowledgements This dissertation has been supported by many people in many different ways. First of all, I would like thank the wonderful people in the department of Effect-Directed Analysis (WANA) for a great working environment in the last years. I am especially thankful for the support and guidance by my two supervisors at the UFZ; Dr. Werner Brack and Dr. Martin Krauss. Thank you giving me the opportunity to learn from your expert knowledge and great experience. It gave me confidence that it will all work out in the end. I would like to thank Prof. Dr. Henner Hollert for taking on my supervision at RWTH Aachen University.

A special thanks to the technical staff of WANA; Margit, Ola, Hubert and Jörg, who took a lot of work of my shoulders and kept the lab and instruments running. Jörg, thank you for the many wonderful sampling trips to the Holtemme River, interesting talks and teaching me how to drive a car. These trips were among the most enjoyable moments of this PhD.

The work at the Holtemme River was a wonderful case study for applying scientific methods in an interdisciplinary context with real-world challenges. I am very grateful for the open minds and voluntary support of the operators of the WWTPs, many people from the Environmental Agency of Saxony-Anhalt (LAU) and the State Office for flood protection and water management of Saxony-Anhalt (LHW) as well as local farmers. Working at the Holtemme River has also enabled me to work with many people from different UFZ departments. I especially want to thank Dr.-Ing. Ilona Bärlund and Dr. Markus Weitere for supporting and organizing the Holtemme Stakeholder Workshop with me.

Finally, I thank my friends many whom supported me from the Master’s thesis and throughout the PhD. Even though we were all heavily occupied with work, we managed to have some great times together – in and outside of Leipzig. Thanks to: Melis, Elisa, Matthias, Jörg, Pedro, Harry, Olli, Tanne and Riccardo (best Master’s thesis supervisor and office mate). I am also thankful to my friends all around Germany; Charlotte, Anne, Jasmin, Sarah, Vera and Christoph. Thanks for getting me out of the UFZ bubble from time to time. A very special thanks to Stefan for his continuous mental and scientific support. So far, our time together often revolved around the struggles as PhD students. Now, I’m very much looking forward to the time with you after the PhD, hopefully filled with more weekends off and travels around the world.

I deeply thank my family for their constant support, reassurance and encouragement. Thanks for giving me a place with food and comfort to escape to when necessary. I

Scientific contributions

Publications:

Kandie FJ, Krauss M, Beckers L-M, Massei R, Fillinger U, Becker J, et al. Occurrence and risk assessment of organic micropollutants in freshwater systems within the Lake Victoria South Basin, Kenya. Science of The Total Environment 2020; 714: 136748.

*Beckers, L.-M.; Busch, W.; Krauss, M.; Schulze, T.; Brack, W., Characterization and risk assessment of seasonal and weather dynamics in organic pollutant mixtures from discharge of a separate sewer system. Water Research 2018, 135, 122-133.

Massei, R.; Byers, H.; Beckers, L.-M.; Prothmann, J.; Brack, W.; Schulze, T.; Krauss, M., A sediment extraction and cleanup method for wide-scope multitarget screening by liquid chromatography–high-resolution mass spectrometry. Analytical and Bioanalytical Chemistry 2017

*Publication which is part of this dissertation.

Platform presentations:

Beckers, L.-M.; Busch, W.; Krauss, M.; Schulze, T.; Brack, W. Characterization and exposure assessment of seasonal and weather dynamics in pollutant mixtures from wastewater discharge. International Conference on Urban Drainage (ICUD) 2017, Prague

Beckers, L.M., Busch, W.; Inostroza, P, Krauss, M., Muschket, M., Schulze, T., Schmitt-Jansen, M., Brack, W. Unraveling spatial and temporal chemical stress in small rivers– Case Study: Holtemme River (Saxony-Anhalt). Deutsche Gesellschaft für Limnologie 2018, Cottbus

Beckers, L.-M., Brack, W, Krauss, M., Müller, E., Schulze, T. Unravelling longitudinal pollution patterns in freshwaters by non-target screening and cluster analysis. SETAC Europe 2018, Rome

Curriculum vitae

Liza-Marie Beckers Konradstraße 31, 04315 Leipzig [email protected] *23.03.1990, Berlin, Germany

Employment

Since 02/2018 Scientist Helmholtz Centre for Environmental Research, Leipzig, Germany Department of Effect-Directed Analysis

06/2018-08/2019 Scientist in LVSPE-EVENT Project Funded by AiF (Arbeitsgemeinschaft industrieller Forschungsvereinigungen) e.V.

„Verteilereinheit basierend auf stromlosen Magnetventilen als Erweiterung des großvolumigen Festphasenextraktionsgerätes (LVSPE) für Wasserproben“

02/2015 – 01/2018 PhD student Helmholtz Centre for Environmental Research, Leipzig, Germany Department of Effect-Directed Analysis

Education

02/2015 -01/2020 PhD student Helmholtz Centre for Environmental Research, Leipzig, Germany Department of Effect-Directed Analysis

RWTH Aachen University, Aachen, Germany Institute for Biology V (Environmental Research), Department of Ecosystem Analysis

Title: “Novel approaches to identify drivers of chemical stress in small rivers”

Defence: 15.01.2020

04/2014 – 02/2015 Master student Helmholtz Centre for Environmental Research, Leipzig, Germany Department of Effect-Directed Analysis Preparation of Master’s thesis

Title: “Sediment contamination with pharmaceuticals in major European estuaries”

III

Curriculum vitae

10/2013 - 02/2015 Master of Science University of Hohenheim, Stuttgart, Germany Environmental Science - Soil, Water an Biodiversity (EnvEuro) Host University of Double Degree Programme Final grade: 1.2 (“very good”)

09/2012 - 02/2015 Master of Science University of Copenhagen, Denmark Environmental Science - Soil, Water an Biodiversity (EnvEuro) Double Degree Programme Final grade: 1.2 (“very good”)

09/2011 - 07/2012 Bachelor of Science (Hons) Robert Gordon University, Aberdeen, United Kingdom Forensic and Analytical Science Host University of Double Degree Programme Degree: First Degree

10/2009 - 07/2012 Bachelor of Science Hochschule Bonn-Rhein-Sieg, Campus Rheinbach, Germany Naturwissenschaftliche Forensik Double Degree Programme Final grade: 1.4 (“very good”)

11/2011 - 04/2012 Bachelor student James Hutton Institute, Aberdeen, United Kingdom

Title: “Development and Application of a Method for the Determination of the Elemental Concentration of Forensic Soil Samples”

09/2003 - 08/2009 Tagore-Schule Secondary School, Berlin A-levels, June 2009 Grade: 1.4

Curriculum vitae

Awards and scholarships

10/05/2017 HIGRADE Conference Poster Prize Awarded by HIGRADE – Helmholtz Interdisciplinary Graduate school for Environmental Research 3rd Place

Title: “Wrong time, wrong place – Unravelling temporal pollution and risk pattern in wastewater discharge”

14/11/2015 ELLS Prize for Excellenrt Master Theses 2015 Awarded by Prof. Dr. Prof. h.c. Dr. h.c. Klaus Macharzina former president of the University of Hohenheim Grant: 1,000€

09/2011-07/2012 Scholarship holder of DAAD (German Academic Exchange Service) one year scholarship at Robert Gordon University Aberdeen, United Kongdom

Leipzig,

Place, Date Signatur