Prospective environmental risk assessment of mixtures in wastewater treatment plant effluents – Theoretical considerations and experimental verification

Accepted Manuscript Prospective environmental risk assessment of mixtures in wastewater treatment plant effluents – Theoretical considerations and experimental verification Anja Coors, Pia Vollmar, Frank Sacher, Christian Polleichtner, Enken Hassold, Daniela Gildemeister, Ute Kühnen PII:

S0043-1354(18)30321-X

DOI:

10.1016/j.watres.2018.04.031

Reference:

WR 13728

To appear in:

Water Research

Received Date: 16 January 2018 Revised Date:

12 April 2018

Accepted Date: 13 April 2018

Please cite this article as: Coors, A., Vollmar, P., Sacher, F., Polleichtner, C., Hassold, E., Gildemeister, D., Kühnen, U., Prospective environmental risk assessment of mixtures in wastewater treatment plant effluents – Theoretical considerations and experimental verification, Water Research (2018), doi: 10.1016/j.watres.2018.04.031. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT 1

Prospective environmental risk assessment of mixtures in wastewater treatment plant

2

effluents – theoretical considerations and experimental verification

3

RI PT

4 5

Anja Coors 1*, Pia Vollmar 1, Frank Sacher 2 Christian Polleichtner 3, Enken Hassold 4,

6

Daniela Gildemeister 4, Ute Kühnen 4

SC

7 1

ECT Oekotoxikologie GmbH, Boettgerstrasse 2-14, 65439 Flörsheim/Main

9

2

DVGW-Technologiezentrum Wasser (TZW), Karlsruher Straße 84, 76139 Karlsruhe

10

3

UBA – German Environment Agency, Schichauweg 58, 12307 Berlin

11

4

UBA – German Environment Agency, Wörlitzer Platz 1, 06844 Dessau-Roßlau

M AN U

8

13

TE D

12

Abstract

15

The aquatic environment is continually exposed to a complex mixture of chemicals, whereby

16

effluents of wastewater treatment plants (WWTPs) are one key source. The aim of the

17

present study was to investigate whether environmental risk assessments (ERAs)

18

addressing individual substances are sufficiently protective for such coincidental mixtures.

19

Based on a literature review of chemicals reported to occur in municipal WWTP effluents and

20

mode-of-action considerations, four different types of mixtures were composed containing

21

human pharmaceuticals, pesticides, and chemicals regulated under REACH. The

22

experimentally determined chronic aquatic toxicity of these mixtures towards primary

23

producers and the invertebrate Daphnia magna could be adequately predicted by the

24

concept of concentration addition, with up to 5-fold overestimation and less than 3-fold

25

underestimation of mixture toxicity. Effluents of a municipal WWTP had no impact on the

AC C

EP

14

1

ACCEPTED MANUSCRIPT predictability of mixture toxicity and showed no adverse effects on the test organisms.

27

Predictive ERAs for the individual mixture components based on here derived predicted no

28

effect concentrations (PNECs) and median measured concentrations in WWTP effluents

29

(MCeff) indicated no unacceptable risk for any of the individual chemicals, while MCeff/PNEC

30

summation indicated a possible risk for multi-component mixtures. However, a refined

31

mixture assessment based on the sum of toxic units at species level indicated no

32

unacceptable risks, and allowed for a safety margin of more than factor 10, not taking into

33

account any dilution of WWTP effluents by surface waters. Individual substances, namely

34

climbazole, fenofibric acid and fluoxetine, were dominating the risks of the investigated

35

mixtures, while added risk due to the mixture was found to be low with the risk quotient being

36

increased by less than factor 2. Yet, uncertainty remains regarding chronic mixture toxicity in

37

fish, which was not included in the present study. The number and identity of substances

38

composing environmental mixtures such as WWTP effluents is typically unknown. Therefore,

39

a mixture assessment factor is discussed as an option for a prospective ERA of mixtures of

40

unknown composition.

SC

M AN U

TE D EP

42

AC C

41

RI PT

26

2

ACCEPTED MANUSCRIPT Introduction

44

Wastewater treatment plants (WWTPs) represent systems where chemicals from various

45

sources merge and, after their degradation and transformation, are released as mixture into

46

the environment. For the aquatic environment, effluents of WWTPs thereby represent a key

47

point source for unintentional and coincidental mixtures of chemicals. The environmental risk

48

assessment (ERA) for chemicals has traditionally been conducted on a single-substance

49

basis within different regulatory frameworks (e.g. human pharmaceuticals, pesticides,

50

biocides, and chemicals regulated under REACH). Such a single-substance ERA is

51

extending towards an ERA of intentional mixtures in some of these frameworks (JRC 2014).

52

Yet, there is currently no regulation addressing a prospective ERA of coincidental mixtures of

53

micro-pollutants as represented by WWTP effluents (Kienzler et al. 2016).

54

The summation of risk quotients such as the ratios of predicted environmental concentration

55

and predicted no effect concentration (PEC/PNEC ratios) provides a simple and pragmatic

56

approach for a prospective mixture risk assessment and has been proposed for different

57

regulatory frameworks (e.g. ECHA 2014, Frische et al. 2014, JRC 2014, Backhaus 2016).

58

This approach is generally more conservative than applying the concept of concentration

59

addition (CA) separately for identical endpoints (or species) as explained in detail by

60

Backhaus and Faust (2012). Hence, applying CA based on the sum of toxic units (STU)

61

calculated separately for identical endpoints or at least trophic levels could represent a

62

refinement step if unacceptable risk could not be excluded by PEC/PNEC summation (JRC

63

2014). A few studies applied this tiered approach already to WWTP effluents or surface

64

water samples using predicted (Escher et al. 2011) or measured substance concentrations

65

(Backhaus & Karlsson 2014, Thomaidi et al. 2015) as exposure estimates. These and other

66

studies addressing mixture toxicity in the aquatic environment (e.g. Santos et al. 2013,

67

Ginebreda et al. 2014) relied on acute toxicity endpoints due to unavailability of chronic

68

endpoints for many of the individual mixture components. Yet, the regulatory ERA of human

69

pharmaceuticals, as one key group of biologically active micro-pollutants in wastewater, is

70

required to be based on chronic toxicity data (EMA 2006). The most sensitive species and

AC C

EP

TE D

M AN U

SC

RI PT

43

3

ACCEPTED MANUSCRIPT the resulting PNEC can differ considerably when acute instead of chronic endpoints are

72

used. This is illustrated by the dataset for human pharmaceuticals from Vestel et al. (2016),

73

where in 32 out of 60 cases the chronic and the acute PNEC for the same substance were

74

derived from species of different trophic levels. So far, there is some, though still limited

75

evidence that CA provides reasonably good estimates not only for acute but also for chronic

76

effects in aquatic organisms (Hermens et al. 1984, Coors et al. 2014, Hassold & Backhaus

77

2014, Watanabe et al. 2016). The present study therefore focussed on chronic aquatic

78

toxicity, also in view of the almost constant discharge from WWTPs.

79

The aim of the present study was to experimentally investigate the applicability of CA for a

80

predictive ERA of mixtures of micro-pollutants in WWTP effluents and to compare single-

81

substance and mixture ERA based on currently proposed approaches. For this purpose,

82

mixtures of up to ten substances known or expected to occur in WWTP effluents were tested

83

for aquatic chronic toxicity in two primary producers and one invertebrate. Additionally, the

84

influence of WWTP effluent on the predictability of mixture toxicity was investigated.

85

Measured concentrations in WWTP effluents (MCeff) served as exposure estimate in the

86

single-substance and mixture assessments.

87

Material & Methods

88

Selection of substances and mixtures

89

The rationale for the selection of the individual components in the investigated mixtures was

90

dominated by exposure considerations. An extensive review of literature studies (published

91

between 2002 and 2012) produced a list of more than 500 chemicals reported to frequently

92

occur in the effluents of municipal WWTPs and surface waters in Europe. Eight substances

93

were selected among these for the present study: the fungicidal human pharmaceutical

94

fluconazole, the anti-depressant fluoxetine, the beta-blocker metoprolol, the fungicide

95

climbazole (regulated under REACH and contained e.g. in anti-dandruff shampoo), the

96

flame-retardant tris(2-chloropropyl) phosphate (TCPP), the corrosion inhibitor 5-

97

methylbenzotriazole (5-MBT), the fungicidal preservative methylparaben (used in personal

AC C

EP

TE D

M AN U

SC

RI PT

71

4

ACCEPTED MANUSCRIPT care products), and the fungicide propiconazole (authorized as biocide and plant protection

99

product in the European Union, EU). Three more human pharmaceuticals were selected for

100

which no or few published data on occurrence in European municipal WWTP effluents were

101

available, but consumption data indicated potential presence in the environment. These were

102

the calcium channel blocker amlodipine, the lipid-modifying agent fenofibric acid (an active

103

pharmaceutical ingredient (API) on its own as well as the active metabolite of fenofibrate),

104

and the antibiotic linezolid. In addition to the occurrence in WWTP effluents, the

105

(dis)similarity of the intended mode of action (MoA) of the compounds was considered during

106

the selection process. In order to challenge CA-based mixture predictions, compounds were

107

preferably selected which are known inhibitors of the detoxifying cytochrome P450 system,

108

because this may increase the likelihood of synergistic interactions in the mixture

109

(Cedergreen 2014). The assumed or known sensitivity of the pre-selected test species

110

(primary producers and one aquatic invertebrate) were additionally taken into account for

111

selecting the compounds and composing the individual test mixtures (aiming to include the

112

most relevant chemicals in a mixture test with a given species). Since testing mixtures in fish

113

was beyond the scope of the present study, no substances were selected for which fish can

114

be assumed to be particularly sensitive (e.g. estrogens or suspected endocrine disruptors).

115

In addition to the individual selected substances, the following conceptionally different

116

mixtures were experimentally tested in the present study:

118 119 120

SC

M AN U

TE D

EP

A. Three components with similar MoA: mixture of the three azoles fluconazole,

AC C

117

RI PT

98

propiconazole and climbazole to represent mixtures of components with a similar

intended MoA (inhibition of ergosterol biosynthesis by binding to C-14 demethylase) that are regulated in different regulatory frameworks. This mixture was tested at an

121

equipotent concentration ratio of the components in green algae and the aquatic

122

invertebrate D. magna, and, as only mixture, also in the water lentil L. minor because

123

of its known high sensitivity for azoles (Richter et al. 2013)

5

ACCEPTED MANUSCRIPT 124

B. Three components with dissimilar MoA: mixture of climbazole, amlodipine, and

125

metoprolol to represent mixtures of components with a dissimilar intended MoA. This

126

mixture was tested only in green algae at an equipotent concentration ratio.

127

C. Multiple components (9-10) at an equipotent concentration ratio: mixture of the three azoles (climbazole, fluconazole, and propiconazole), three chemicals regulated under

129

REACH (TCPP, 5-MBT, and methylparaben), and two more pharmaceuticals

130

(fluoxetine and metoprolol). When tested in green algae, it contained additionally

131

amlodipine and the antibiotic linezolid (10 components in total), but fenofibric acid

132

when tested in D. magna (9 components in total). This difference in mixture

133

composition reflects the differences in sensitivity of the two species towards these

134

three substances.

M AN U

SC

RI PT

128

D. Multiple components (9-10) at an exposure-related concentration ratio: mixture of the

136

same components and tested in the same species as in (c) with the difference that

137

the relative proportions of the components were based on their median

138

concentrations in WWTP effluents as evidenced by the literature review.

139

In order to evaluate whether the WWTP effluent matrix and the plethora of substances

140

therein would interfere with the mixture toxicity predictions (as proposed e.g. by Frische et al.

141

2009), several of the above listed mixture compositions (a, b, and c) were tested in parallel

142

with and without municipal WWTP effluent mixed into the medium.

143

Exposure estimates – Determination of MCeff values

144

Since production volumes of chemicals regulated under REACH are confidential, predicted

145

environmental concentrations (PECs) were not available for several substances of the

146

present study. Therefore, measured concentrations in municipal WWTP effluents (MCeff)

147

were compiled and used in the ERA. No factor accounting for dilution by surface waters was

148

applied. The data compiled from literature studies were complemented by repeated

149

measurements of the effluent of a municipal WWTP, which was also used in some mixture

150

experiments. Details on the investigated WWTP and the analytical results are provided in the

AC C

EP

TE D

135

6

ACCEPTED MANUSCRIPT supplements. From all compiled data, median and 90% percentiles were derived as

152

descriptors of the statistical distribution of the concentrations of each chemical. Non-detects

153

were treated as zero values in the statistical evaluation.

154

Test substances

155

Test substances were purchased from Dr. Ehrenstorfer, Sigma-Aldrich, Tokyo Chemical

156

Industry, or Toronto Research Chemicals at a purity of >95%. For some pharmaceuticals,

157

salts were used as test item in the tests (see details in supplements). Yet, all concentrations

158

and toxicity estimates relate to the active moiety of the API, i.e., the free base or acid and not

159

to the salt. Single-substance and mixture tests were conducted within a period of 2 years

160

during which regular tests with reference substances demonstrated unaltered sensitivity of

161

the used test organisms. In all tests, the substances were directly dissolved in the test

162

medium, without using dispersants or solvents.

163

Growth inhibition tests with the water plant Lemna minor

164

Growth inhibition tests with L. minor were performed according to OECD guideline 221. In a

165

geometric dilution series with a spacing factor not exceeding 3.2, six concentrations (three

166

replicates each) were tested along with a control (six replicates). The ErC10 (concentration

167

with 10% effect on growth rate) based on frond numbers and determined after 7 days of

168

static exposure was used as key endpoint. Temperature was between 22°C and 25.4°C, and

169

pH ranged from 5.7 to 7.7 in the absence and from 8.5 to 9.4 in the presence of WWTP

170

effluent. The pH value changed only in one individual test (with mixture A) with 1.8 units

171

slightly more than the limit of 1.5 units allowed by the guideline. As these deviations are only

172

minor, they are not deemed to invalidate the test results. All tests in the present study fulfilled

173

the validity criteria of the OECD 221 guideline with regard to the doubling time of frond

174

number in the control (less than 2.5 days, corresponding to an about 7-fold increase of frond

175

number in seven days and a growth rate of 0.275 per day).

176

Growth inhibition tests with the green algae Raphidocelis subcapitata

AC C

EP

TE D

M AN U

SC

RI PT

151

7

ACCEPTED MANUSCRIPT In all tests, R. subcapitata (formerly known as Pseudokirchneriella subcapitata or

178

Selenastrum capricornutum) was statically exposed for a test period of 72 hours according to

179

OECD guideline 201. ErC10 was selected as key endpoint. Concentration-response tests

180

were conducted with seven concentration levels (three replicates each), separated by a

181

spacing factor of 3.2 or less, and a control (six replicates). Each replicate vessel (100 ml test

182

solution) was inoculated with 0.5*104 cells/ml taken from an exponentially growing pre-

183

culture. Vessels were continuously shaken during the exposure under permanent light at an

184

intensity of 60-120 µE m-2 s-1 as confirmed by measurements. Across all tests, temperature

185

ranged between 20.2 and 24°C and pH between 7.3 and 10.2. Due to strong growth, the

186

change of the pH during the exposure period occasionally exceeded in the controls and at

187

low test concentrations the maximum of 1.5 pH units allowed by the guideline (maximum

188

change of 2.4 units). Since no impact on growth was observed, these slight deviations from

189

the prescribed pH range are not deemed to have any impact on the reliability of the results.

190

All tests of the present study fulfilled the validity criteria of the OECD 201 guideline with

191

regard to the mean biomass increase (at least 16-fold induction) and the coefficient of

192

variation of growth rate in the controls (equal or less than 7%).

193

Reproduction tests with the freshwater crustacean Daphnia magna

194

D. magna reproduction tests were conducted according to OECD guideline 211 as semi-

195

static exposure with eleven replicate vessels for the control and ten replicate vessels for

196

each of the five to seven concentration levels, separated by a spacing factor between 1.4

197

and 3.2. EC10 of reproduction, measured as cumulative number of living offspring produced

198

per surviving female within 21 days, was selected as key endpoint. Tests started with one

199

individual D. magna (<24 h old) in each replicate. Feeding, exposure conditions and

200

measurements were in accordance with OECD 211. The temperature range prescribed by

201

the guideline (18 to 22°C controlled at ±1°C) was occasionally slightly exceeded (minimum of

202

16.0°C, maximum of 24.2°C), while oxygen content, pH (7.7 to 8.8), and light intensity were

203

in all tests in the recommended range as confirmed by repeated measurements. Survival and

204

number of living offspring per female were recorded daily. All tests in the present study

AC C

EP

TE D

M AN U

SC

RI PT

177

8

ACCEPTED MANUSCRIPT fulfilled the validity criteria of the OECD guideline 211, i.e. 20% or less mortality of control

206

females until the end of the test and at least 60 offspring per surviving female in the control.

207

This last criterion was achieved in the test with metoprolol only after prolongation of the test

208

period until day 22.

209

Testing mixtures in absence and presence of WWTP effluents

210

All mixtures were tested as geometric dilution series of a stock solution, i.e. at a fixed ratio of

211

the mixture component concentrations. In order to test mixtures in parallel in absence and

212

presence of WWTP effluent, 24-h composite samples from a municipal WWTP were

213

collected and either used in the tests within 12 h (algae) or stored refrigerated for up to 5

214

days (D. magna, L. minor). Each mixture concentration level in the dilution series with WWTP

215

effluent consequently contained 50% effluent in the test solution (i.e., the effluent was not

216

diluted along the mixture dilution series). From a twofold concentrated stock solution

217

prepared in twofold concentrated culture medium of the respective test organism, one

218

dilution series was prepared using deionized water and culture medium (absence of effluent),

219

and a second series with identical dilution steps using the WWTP effluent and standard

220

culture medium (presence of effluent). Thereby, it was ensured that both dilution series

221

(presence and absence of effluent) originated from the same stock solution and contained

222

essential elements and nutrients at least at the concentrations prescribed by the guideline.

223

There were always two test substance-free controls: standard culture medium (medium

224

control) and twofold concentrated standard culture medium diluted 1:1 with effluent (effluent

225

control).

226

Analytical measurements of chemicals

227

In all single-substance and mixture tests, the concentrations of all test substances in the

228

exposure solutions at test start were verified by analytical measurement of the lowest, a

229

medium and the highest concentration level. In some cases, substances were additionally

230

analyzed after 2-3 three days of exposure to confirm constant concentrations during the test.

231

Analysis of the test substances was done by direct injection into a HPLC-MS-MS system

AC C

EP

TE D

M AN U

SC

RI PT

205

9

ACCEPTED MANUSCRIPT (liquid chromatograph 1260 Infinity from Agilent Technologies (Waldbronn, Germany)

233

coupled via an electrospray interface to an API 5500 tandem mass spectrometer (AB Sciex,

234

Langen, Germany). If needed (due to low test concentrations), solid-phase extraction on a

235

polymeric material (Strata-X from Phenomenex, Aschaffenburg, Germany) was used for pre-

236

concentration. Quantification was done against an external calibration in test medium.

237

Data analysis

238

Data of concentration-dependent growth rate and reproduction were fitted by non-linear

239

modelling with the free software R version 3.2.2 (R Development Core Team 2013) using the

240

most recent version of the package “drc” (Ritz et al. 2015). In most cases, a 3-parameter log-

241

logistic function achieved good fits, while in some cases a 5-parameter or 3-parameter

242

Weibull model produced better fits. Effect concentrations with 10% and 50% effect were

243

derived from the fitted models, and their 95% confidence intervals were obtained with the

244

implemented function “ED” of the “drc” package using the delta method and the t-distribution.

245

Concentration-response curves determined in absence and presence of WWTP effluent were

246

compared with each other using the ratio test (Wheeler et al. 2006) for each pair of fitted

247

model parameters. If the parameters of two concentration-response curves did not

248

significantly differ in the ratio test (two-sided, alpha=0.05), the curves were deemed to not

249

differ significantly. Lowest-observed-effect-concentrations (LOEC) and no-observed-effect-

250

concentrations (NOEC) are additionally reported in the supplements.

251

Toxicity estimates were derived based on nominal concentrations and corrected for

252

measured concentrations if they deviated by more than 20% from nominal test

253

concentrations.

254

Mixture calculations

255

The toxicity of each mixture was predicted for the EC10 according to the CA concept as

AC C

EP

TE D

M AN U

SC

RI PT

232

,

1

= ∑

∑ ,

10

ACCEPTED MANUSCRIPT 256

With Ci as concentration of each compound i in the mixture. In addition, the toxic units (TUi)

257

and the related sum of toxic units (STU) as a measure for the relative contribution of each

258

component to the overall toxicity were calculated as

,

RI PT

=

As measure for the compliance between predicted and observed toxicity, the Model

260

Deviation Ratio (MDR) introduced by Belden et al. (2007) was calculated for each toxicity

261

estimate as

M AN U

=

SC

259

An MDR above 1 indicates that the toxicity of the mixture is underestimated by the CA

263

prediction, while an MDR below 1 indicates that it is overestimated. An MDR between 0.5

264

and 2 (i.e. indicating a deviation between prediction and observation of up to factor 2) has

265

been interpreted as indicating compliance with CA given the inherent variability of biological

266

test results (Belden et al. 2007, Cedergreen at al. 2008, Coors & Frische 2011, Cedergreen

267

2014). MDR values for acute toxicity (based on EC50 for algal growth inhibition) were

268

calculated additionally and reported in the supplements.

269

Derivation of PNECs and risk quotients

270

Based on the results of the present study and literature data, PNECs were derived for the 11

271

substances using an assessment factor (AF) adapted to the data available for species from

272

different trophic levels (ECB 2003). Risk quotients were calculated as ratio between MCeff

273

and PNEC for the single substances. For the mixtures, MCeff/PNEC ratios were summed for

274

a first-tier assessment, while summed MCeff/EC10 ratios for individual test organisms served

275

as risk quotients in a second-tier assessment.

276

Results & Discussion

277

In the following, the derived estimates for exposure and effects to be used in a risk

278

assessment are described and discussed first. Subsequently, predictability of chronic mixture

AC C

EP

TE D

262

11

ACCEPTED MANUSCRIPT toxicity and the differences between a single-substance and a mixture risk assessment are

280

addressed.

281

Concentrations of the selected chemicals in WWTP effluents

282

Table 1 summarizes the key results from the statistical evaluation of the compiled occurrence

283

data. For amlodipine, fenofibric acid, and linezolid, the present study is among the first to

284

report concentrations in a municipal WWTP effluent. The antibiotic linezolid was only

285

detected once, while amlodipine and fenofibric acid were detected in all four analyzed

286

samples in a concentration above the limit of detection (0.01 µg/L). Concentrations measured

287

in the present study (details provided in the supplements) were in the range of those reported

288

in literature, e.g. for fenofibric acid and metoprolol (Ternes et al. 2007, Rosal et al. 2010) as

289

well as for TCPP and fluconazole (Loos et al. 2013). This confirms that the effluent used in

290

the toxicity tests can be regarded as sufficiently representative for municipal WWTP

291

effluents. Fluoxetine represented an outlier since the concentrations reported by Loos et al.

292

(2013) for European WWTP effluents were about 10-fold lower than those compiled for

293

mostly North American WWTP effluents due to a lack of published data on fluoxetine

294

concentrations in European wastewaters. The concentrations measured in effluent samples

295

of one municipal WWTP in the present study (median of 0.024 µg/L), however, did not

296

support a generally 10-fold lower fluoxetine concentration in European compared to North

297

American WWTP effluents. Methylparaben was not detected in the effluents measured in the

298

present study, which agrees with the frequent non-detects reported in the literature.

299

Confirmation of actual test concentrations and mixture compositions in the biotests

300

Chemical analysis (detailed results in the supplement) of test solutions at test start confirmed

301

less than 20% deviation from nominal concentrations in all single-substance tests of the

302

present study except for propiconazole and metoprolol tested in L. minor. In the mixture

303

tests, concentrations of some components deviated occasionally slightly more than 20% from

304

the nominal concentrations. In all these cases, observed (and predicted) toxicity estimates

305

were accordingly corrected and relate to measured initial concentrations. Constant

AC C

EP

TE D

M AN U

SC

RI PT

279

12

ACCEPTED MANUSCRIPT concentrations under test conditions were checked in one multi-component test with algae

307

and all three multi-component tests with D. magna. The results indicated that fluconazole,

308

climbazole, propiconazole, TCPP, 5-MBT, metoprolol, fenofibric acid (except in one test),

309

and linezolid did not dissipate from the test solutions since at least 80% of initial measured

310

concentration was detected after 2 to 3 days of exposure. This is in agreement with previous

311

findings for fluconazole, climbazole (Richter et al. 2013, 2016), and metoprolol (Dzialowski et

312

al. 2006). Hence, it can be assumed that these substances are generally stable under the

313

exposure conditions, and that toxicity estimates can be based on nominal or initial measured

314

concentrations. Amlodipine and fluoxetine showed some dissipation (35 to 68 % recovery of

315

initial measured concentrations), while methylparaben concentrations in aged exposure

316

solutions were consistently below the limit of detection. However, the dissipation of these

317

substances during the exposure did not impact the mixture toxicity predictions based on

318

initial measured concentrations, because there is no reason to expect that dissipation

319

differed between single-substance and mixture tests. Only the single-substance estimate for

320

fluoxetine takes into account concentrations measured at the end of the exposure (Oakes et

321

al. 2010); therefore, fluoxetine concentrations in the mixture tests were as well corrected for

322

time-weighted average concentrations.

323

Aquatic chronic toxicity of the individual substances and derived PNECs

324

The toxicity estimates of the individual substances that were used in mixture predictions are

325

summarized in Table 2. For most substances, complete concentration-response curves were

326

obtained (see supplements), and EC10 values with mostly very tight confidence intervals

327

could be derived. No effects were detected up to the highest tested concentration of

328

methylparaben and 5-MBT on D. magna reproduction, and of metoprolol on growth of L.

329

minor. The EC10 of TCPP for D. magna reproduction was slightly extrapolated beyond the

330

lowest tested concentration based on a clear concentration-response curve. For some

331

substances (as indicated in Table 2), estimates for the key endpoints were taken from the

332

literature.

AC C

EP

TE D

M AN U

SC

RI PT

306

13

ACCEPTED MANUSCRIPT There are no data available regarding long-term fish toxicity for TCPP, 5-MBT,

334

methylparaben, amlodipine, fenofibric acid, and linezolid, while the present study provides

335

chronic endpoints for two trophic levels. Hence, an AF of 100 was applied to the here

336

determined lowest EC10 of these substances to derive the PNEC. Similarly, no long-term fish

337

toxicity data are available in the literature for the azoles climbazole and fluconazole, while

338

acute toxicity is considerably less for fish than for primary producers (Chen & Ying 2015).

339

The water plant L. minor clearly represented the most sensitive endpoint for both fluconazole

340

and climbazole. Since chronic toxicity data are available for two trophic levels, among them

341

the presumably most sensitive one, an AF of 50 was used for deriving these PNECs. For

342

propiconazole, none of the EC10 values derived in the present study was lower than the

343

NOEC listed for fish in the most recent regulatory assessment of propiconazole as biocide

344

(SC 2015). The PNEC was accordingly derived from this NOEC using an AF of 10, because

345

chronic data are available for three trophic levels. For metoprolol, Moermond & Smit (2016)

346

recently proposed an environmental quality standard (EQS) of 62 µg/L relying on a 9-day

347

toxicity test in Daphnia as most sensitive endpoint with an AF of 50 as this Daphnia test was

348

classified as semi-chronic only. The EC10 for Daphnia reproduction determined in the present

349

study is lower than chronic toxicity data for other trophic levels (Moermond and Smit 2016

350

and present study), which supports using this guideline-conform Daphnia EC10 with an AF of

351

10 (Table 1). Similar to metoprolol, there are numerous studies available on the (chronic)

352

aquatic toxicity of fluoxetine, justifying an AF of 10 applied to the estimate for algal toxicity as

353

most sensitive endpoint (Oakes et al. 2010). Overall, the PNECs for the individual

354

substances ranged from 0.1 µg/L to 610 µg/L.

355

Predictability of mixture toxicity for chronic endpoints in absence and presence of WWTP

356

effluent

357

The equipotent three-component mixtures caused clear concentration-dependent responses

358

of the key endpoints (Fig. 1, A-D). There was no significant difference between the

359

parameters of the models fitted to responses observed in absence and presence of WWTP

360

effluent (all p>0.05, only tested with primary producers). There was also no adverse effect of 14

AC C

EP

TE D

M AN U

SC

RI PT

333

ACCEPTED MANUSCRIPT the diluted WWTP effluent on primary producers as demonstrated by the controls. Note that

362

the background concentrations in the WWTP effluent (Table S2) were not included in the

363

mixture predictions, since they were well below the test concentrations. Most concentration-

364

response curves for the three-component mixtures did not differ much from the respective

365

curves predicted by CA (dotted lines in Fig. 1). The MDR values determined as quantitative

366

measure of deviation (Table 3) indicated up to 5-fold overestimation of mixture toxicity with

367

the greatest deviations being observed in algae and the smallest ones in D. magna. Mixture

368

toxicity was underestimated, though by less than a factor of 3, for one mixture tested in

369

algae. It is remarkable that the only case of mixture toxicity underestimation by CA occurred

370

for the mixture of components with dissimilar MoA (Fig. 1D), for which according to theory

371

(Altenburger et al. 2004) the IA model would be more appropriate while the CA model should

372

rather overestimate mixture toxicity. This indicates that the slight deviation between

373

prediction and observation was not related to the choice of the mixture model. Possible

374

reasons are rather inherent variability of biological responses and the relatively poor model fit

375

at the low effect level, which could not be further improved beyond the finally applied 4-

376

parameter Weibull model. Correction for initial measured concentrations of the mixture

377

components resulted mostly in MDR values closer to 1, i.e. better compliance between

378

predicted and observed toxicity of the three-component mixtures.

379

The tests with the multi-component mixtures in green algae and Daphnia achieved also clear

380

concentration-response curves (Fig. 2A-D). The toxicity of these mixtures was in no case

381

underestimated and less than 3-fold overestimated by the CA prediction (Fig. 2, Table 3). As

382

with the 3-component mixtures, the compliance between predicted and observed toxicity

383

slightly increased when measured test concentrations were used (data not shown). In

384

accordance with the 3-component mixture tests, no influence of the WWTP effluent on the

385

concentration-response curve was observed for R. subcapitata. For D. magna, however, only

386

two of the three parameters did not differ significantly (curve steepness and EC50), while the

387

upper limit as third parameter differed significantly (p<0.001). Hence, at control or low

388

exposure conditions the reproduction of D. magna was significantly enhanced by the

AC C

EP

TE D

M AN U

SC

RI PT

361

15

ACCEPTED MANUSCRIPT presence of 50% WWTP effluent in the test solution. This finding is in agreement with

390

previous studies and most likely relates to improved food conditions for the filter-feeder D.

391

magna due to the bacterial loading of the effluent as discussed elsewhere (Schlüter-Vorberg

392

et al. 2017).

393

MDR values calculated for acute toxicity of all tested mixtures (i.e., based on EC50 values,

394

Table S8 in Supplements) indicated good compliance with CA predictions (less than 3-fold

395

deviation). The present study hence demonstrates that the aquatic toxicity of mixtures can

396

not only be predicted for acute endpoints (as it has been repeatedly shown, see e.g. reviews

397

by Cedergreen 2014 and Backhaus 2016), but also for (sub)chronic endpoints in various

398

species. The deviation between the predicted and observed chronic mixture toxicity appears

399

to be in the range of that reported for acute endpoints, where the majority of cases deviated

400

less than two-fold (Belden et al. 2007, Coors & Frische 2011). A recent study of Watanabe et

401

al. (2016) investigated the toxicity of 10 wastewater-born substances in green algae,

402

Ceriodaphnia dubia and fish embryos, and also concluded that low-effect level (sub)chronic

403

endpoints can be sufficiently well predicted by CA. Toxicity was equally predictable by CA for

404

mixtures with similar as well as dissimilar intended MoA in the present study. While this may

405

be simply due to the fact that the predictions of these two models often do not greatly differ

406

(Junghans et al. 2006), it may also indicate that some or all of the substances actually

407

exhibited a similar MoA in the tested non-target organisms, i.e. non-specific baseline toxicity

408

as it has been discussed for micro-pollutants in wastewater (Escher et al. 2011). The matrix

409

of wastewater effluent and the multitude of other substances present at low concentrations

410

did not affect the performance of the test organisms nor did it impact the predictability of

411

chronic mixture toxicity in the here conducted tests. Overall, it can therefore be concluded

412

that the CA model can be applied for mixture risk predictions of WWTP effluents with

413

reasonably good reliability. This holds at least for key chronic endpoints of primary producers

414

and crustaceans, which are required within the ERA of human pharmaceuticals. However,

415

this conclusion does not necessarily extend to other species. Particularly for fish, there is still

416

a lack of chronic mixture toxicity studies as supporting evidence for the applicability of CA.

AC C

EP

TE D

M AN U

SC

RI PT

389

16

ACCEPTED MANUSCRIPT Risk quotients for single-substances and their mixture

418

The median MCeff/PNEC ratios were below 1 for all single substances, i.e. no unacceptable

419

risk was indicated on a single-substance level (Table 4). When using the more conservative

420

90% percentile of the MCeff, the risk quotient was above 1 only in the case of climbazole. The

421

sum of the MCeff/PNEC ratios for the mixture of all eleven substances was above 1, when

422

using the median as well as the 90% percentile MCeff. Hence, a possible risk was identified

423

for the mixture by the first-tier MCeff/PNEC summation approach, while a single-substance

424

assessment indicated mostly no unacceptable risk for WWTP effluents.

425

The species-specific risk quotients (Table 4), which represent the toxic units of the individual

426

substances at the 90% percentile MCeff, were all below 1. Hence, all of the substances would

427

exhibit less than 10% effect on any of the species at a concentration equal to a conservative

428

estimate of their municipal WWTP effluent concentration. The sum of the single-substance

429

risk quotients was still below 1 for all three species, which indicates likewise that the mixtures

430

of 7 to 10 compounds (depending on species) would induce less than 10% effect. The

431

theoretical expectation for no measurable effects was confirmed by the experimental testing

432

of mixtures composed of some of the eleven compounds, as no adverse effects occurred at

433

the MCeff concentration range. Diluted WWTP effluent samples (used as controls) also

434

exhibited no adverse effects on any of the tested species. In terms of a risk assessment,

435

extrapolation to WWTP effluents in general can be assumed covered by using the

436

conservative 90% percentile MCeff value as exposure estimate. Regarding extrapolation to

437

other species, however, an assessment factor should be applied. The summed species-

438

specific risk quotients (Table 4) allowed for a safety margin (i.e., potential size of an AF) of

439

1000 (D. magna) or about 12 (green algae) before a 10% inhibition would be expected,

440

which is greater than the appropriate AF in case of a complete dataset.

441

The here calculated risk quotients for single substances as well as mixtures do not include a

442

factor accounting for dilution of WWTP effluent in receiving surface waters. Hence.

443

regardless of the question whether the current standard dilution factor of 10 is appropriate

AC C

EP

TE D

M AN U

SC

RI PT

417

17

ACCEPTED MANUSCRIPT (Link et al. 2017), risks for the aquatic environment resulting from combined effects of the

445

considered compounds were already low without assuming dilution by surface water.

446

The mixture risk was driven by only few substances: fluoxetine in the case of green algae,

447

fenofibric acid in the case of D. magna, and climbazole in the case of L. minor. This finding is

448

in line with other publications that reported mixture risks to be dominated by few of the

449

mixture’s components (e.g. Price et al. 2012, Backhaus & Karlsson 2014, Gustavsson et al.

450

2017). The maximum cumulative ratio (MCR) introduced by Price & Han (2011) relates the

451

risk quotient of the most “risky” single substance to that of the mixture in order to quantify the

452

added risks due to combined effects. In the present study, the MCR values ranged from 1.01

453

to 1.48 (i.e. less than twofold increase due to combined effects), which indicates low concern

454

for mixture risks. However, the selection of compounds (their identity, number and

455

concentration) is crucially determining the outcome of any mixture assessment. Based on

456

current knowledge, deriving a representative or a priority mixture of micro-pollutants in

457

WWTP effluents appears an impossible task. It would require reliable effluent concentration

458

data for all chemicals possibly occurring in wastewater, of which many may not have been

459

subject of WWTP effluent monitoring programs yet. This is illustrated by fenofibric acid,

460

which has so far not been prioritized for environmental risks in any study, but has been

461

ranked third based on individual risk quotients in the present study. More so far unknown

462

micro-pollutants may occur in WWTP effluents at relevant concentrations.

463

One possibility to cover the additional risk of unknown components in a non-defined mixture

464

would be a mixture assessment factor (MAF) as discussed earlier (e.g. Backhaus 2016). The

465

MAF would be applied in every single-substance assessment as an additional assessment

466

factor to account for ‘mixture uncertainty’. Apart from lacking justification to ‘blame’ each

467

substance in the same way for potential additional mixture risk independently of its actual or

468

potential contribution, the problem arises to establish an appropriate size of such a MAF. The

469

so far available scarce evidence from mixture risk assessment studies indicates an at

470

maximum 5-fold greater risk of environmental mixtures compared to their most ‘risky’ single

AC C

EP

TE D

M AN U

SC

RI PT

444

18

ACCEPTED MANUSCRIPT component (Price et al. 2012, Backhaus & Karlsson 2014, Carvalho et al. 2014). This would

472

support an MAF of up to 5 or lower (i.e., up to 2) when based on the chronic aquatic mixture

473

toxicity evaluated in the present study.

474

Conclusions

475

The present study provides ample evidence that the chronic aquatic toxicity of mixtures of

476

diverse substances can be predicted by CA with less than 3-fold underestimation, at least

477

with regard to primary producers and invertebrates. Acknowledging that uncertainty remains

478

regarding the predictability of chronic mixture toxicity in fish, reliable prospective ERAs for

479

mixtures of chemicals present in WWTP effluents could be conducted based on CA. Yet,

480

mixture toxicity concepts such as CA can only address mixtures that are clearly defined in

481

terms of components and their concentrations. Hence, the identification of all possibly

482

relevant chemicals and their concentrations is warranted for a prospective quantitative ERA

483

of WWTP effluents as mixtures. This is hardly possible in view of the currently available data,

484

and may not be feasible in the future either given the great number of chemicals possibly

485

occurring in WWTP effluents. As substitute, applying a mixture assessment factor to all

486

single-substance ERAs could account for combined effects from simultaneous exposures

487

without the need to define all components the actual environmental mixtures. However, it

488

must be acknowledged that the present study (considering up to 11 components in a mixture

489

and applying worst-case assumptions) did not indicate substantial (i.e. less than two-fold)

490

additional risk due to the mixture beyond that already posed by some of the individual

491

mixture components.

SC

M AN U

TE D

EP

AC C

492

RI PT

471

493

Acknowledgements

494

This study was financially supported by the German Environment Agency through the project

495

FKZ 3712 64 419. The views expressed herein are those of the authors and do not

496

necessarily represent the policy of the Agency. 19

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

497

20

ACCEPTED MANUSCRIPT References

499

Altenburger, R., Walter, H., Grote, M. 2004. What contributes to the combined effect of a mixture? Environ. Sci. Technol.

500

38, 6353-6362.

501

Backhaus, T., 2016. Environmental risk assessment of pharmaceutical mixtures: demands, gaps, and possible bridges. The

502

AAPS Journal 18(4), 804-813.

503

Backhaus, T., Faust, M., 2012. Predictive environmental risk assessment of chemical mixtures: a conceptual framework.

504

Environ. Sci. Technol. 46, 2564-2573.

505

Backhaus, T., Karlsson, M., 2014. Screening level mixture risk assessment of pharmaceuticals in STP effluents. Water Res.

506

49, 157-165.

507

Belden, J.B., Gilliom, R.J., Lydy, M.J., 2007. How well can we predict the toxicity of pesticides mixtures to aquatic life? Integr.

508

Environ. Assess. Manag. 3, 364-372.

509

Bendz, D., Paxeus, N.A., Ginn, T.R., Loge, F.J., 2005. Occurrence and fate of pharmaceutically active compounds in the

510

environment, a case study: Hoje River in Sweden. J. Hazard. Mater. 122, 195-204.

511

Benijts, T., Lambert, W., De Leenheer, A., 2004. Analysis of multiple endocrine disruptors in environmental waters via wide-

512

spectrum solid-phase extraction and dual-polarity ionization LC-ion trap-MS/MS. Anal. Chem. 76, 704-711.

513

Blanco, E., Casais, M.C., Mejuto, M.C., Cela, R., 2009. Combination of off-line solid-phase extraction and on-column sample

514

stacking for sensitive determination of parabens and p-hydroxybenzoic acid in waters by non-aqueous capillary

515

electrophoresis. Anal. Chim. Acta 647, 104-111.

516

Canosa, P., Rodriguez, I., Rubi, E., Bollain, M.H., Cela, R., 2006. Optimisation of a solid-phase microextraction method for the

517

determination of parabens in water samples at the low ng per litre level. J. Chrom. A 1124, 3-10.

518

Carvalho, R.N., Arukwe, A., Ait-Aissa, S. et al. 2014. Mixtures of chemical pollutants at European legislation safety

519

concentrations: how safe are they? Toxicol. Sci. 141, 218-233.

520

Cedergreen, N., 2014. Quantifying synergy: A systematic review of mixture toxicity studies within environmental toxicology.

521

PLoS ONE 9(5), e96580.

522

Cedergreen, N., Munch-Christensen, A., Kamper, A., Kudsk, P., Mathiassen, S.K., Streibig, J., Sorensen, H., 2008. A review of

523

independent action compared to concentration addition as reference models for mixtures of compounds with different

524

molecular target sites. Environ. Toxicol. Chem. 27, 1621-1632.

525

Chen, Z.-F., Ying, G.-G., 2015. Occurrence, fate and ecological risk of five typical azole fungicides as therapeutic and personal

526

care products in the environment: a review. Environ. Int. 84, 142-153.

AC C

EP

TE D

M AN U

SC

RI PT

498

21

ACCEPTED MANUSCRIPT Chen, Z.-F., Ying, G.-G., Lai, H.-J., Chen, F., Su, H.-C., Liu, Y.-S., Peng, F.-Q., Zhao, J.-L., 2012. Determination of biocides in

528

different environmental matrices by use of ultra-high-performance liquid chromatography-tandem mass spectrometry.

529

Anal. Bioanal. Chem. 404, 3175–3188.

530

Coors, A., Frische, T., 2011. Predicting the aquatic toxicity of commercial pesticide mixtures. Environmental Sciences Europe

531

23:22.

532

Coors, A., Weisbrod, B., Schoknecht, U., Sacher, F., Kehrer, A., 2014. Predicting acute and chronic effects of wood

533

preservative products in Daphnia magna and Pseudokirchneriella subcapitata based on the concept of concentration

534

addition. Environ. Toxicol. Chem. 33, 382-393.

535

Dzialowski, E.M., Turner, P.K., Brooks, B.W., 2006. Physiological and reproductive effects of beta adrenergic receptor

536

antagonists in Daphnia magna. Arch. Environ. Contam. Toxicol. 50, 503–510.

537

ECB, 2003. Technical Guidance Document on Risk Assessment, Part II. European Chemicals Bureau.

538

ECHA, 2014. Transitional Guidance on the Biocidal Products Regulation – Transitional Guidance on mixture toxicity

539

assessment for biocidal products for the in environment. European Chemicals Agency, Helsinki, May 2014.

540

EMA, 2006. Guideline on the environmental risk assessment of medicinal products for human use. Committee for Medicinal

541

Products for Human Use, European Medicines Agency, London, 01 June 2006. Doc. Ref. EMEA/CHMP/SWP/4447/00.

542

Escher, B.I., Baumgartner, R., Koller, M., Treyer, K., Lienert, J., McArdell, C.S., 2011. Environmental toxicology and risk

543

assessment of pharmaceuticals from hospital wastewaters. Water Res. 45, 75-92.

544

Frische, T., Faust, M., Meyer, W., Backhaus, T., 2009. Toxic masking and synergistic modulation of the estrogenic activity of

545

chemical mixtures in a yeast estrogen screen (YES). Environ. Sci. Pollut. Res. 16, 593-603.

546

Frische, T., Matezki, S., Wogram, J., 2014. Environmental risk assessment of pesticide mixtures under regulation

547

1107/2009/EC: a regulatory review by the German Federal Environment Agency (UBA). J. Verbrauch. Lebensm. 9, 377-389.

548

García-López, M., Rodríguez, I., Cela, R., 2010. Mixed-mode solid-phase extraction followed by liquid chromatography–

549

tandem mass spectrometry for the determination of tri- and di-substituted organophosphorus species in water samples. J.

550

Chrom. A. 1217, 1476–1484.

551

Ginebreda, A., Kuzmanovic, M., Guasch, H., López de Alda, M., López-Doval, J., Munoz, I., Ricart, M., Romaní, A.M., Sabater,

552

S., Barceló, D., 2014. Assessment of multi-chemical pollution in aquatic ecosystems using toxic units: Compound

553

prioritization, mixture characterization and relationships with biological descriptors. Sci. Total Environ. 468-469, 715-723.

554

Gonzalez-Marino, I., Quintana, J.B., Rodriguez, I., Cela, R., 2009. Simultaneous determination of parabens, triclosan and

555

triclocarban in water by liquid chromatography/electrospray ionisation tandem mass spectrometry. Rapid Commun. Mass

556

Spectrom. 23, 1756–1766.

AC C

EP

TE D

M AN U

SC

RI PT

527

22

ACCEPTED MANUSCRIPT Gustavsson, M., Kreuger, J., Bundschuh, M., Backhaus, T., 2017. Pesticide mixtures in the Swedish streams: Environmental

558

risks, contributions of individual compounds and consequences of single-substance oriented risk mitigation. Sci. Total

559

Environ. 598, 973-983.

560

Hassold, E., Backhaus, T., 2014. The predictability of mixture toxicity of demethylase inhibiting fungicides to Daphnia magna

561

depends on life-cycle parameters. Aquat. Toxicol. 152, 205-214.

562

Hedgespeth, M.L., Sapozhnikova, Y., Pennington, P., Clum, A., Fairey, A., Wirth, E., 2012. Pharmaceuticals and personal care

563

products (PPCPs) in treated wastewater discharges into Charleston Harbor, South Carolina. Sci. Total Environ. 437, 1-9.

564

Hermens, J., Canton, H., Janssen, P., De Jong, R., 1984. Quantitative structure-activity relationships and toxicity studies of

565

mixtures of chemicals with anaesthetic potency: acute lethal and sublethal toxicity to Daphnia magna. Aquat. Toxicol. 5,

566

143-154.

567

JRC, 2014. Assessment of mixtures – review of regulatory requirements and guidance. Report EUR 26675 EN. European

568

Commission, Joint Research Centre, Ispra, Italy.

569

Junghans, M., Backhaus, T., Faust, M., Scholze, M., Grimme, L.H., 2006. Application and validation of approaches for the

570

predictive hazard assessment of realistic pesticide mixtures. Aquat. Toxicol. 76, 93-110.

571

Kahle, M., Buerge, I.J., Hauser, A., Muller, M.D., Poiger, T., 2008. Azole fungicides: occurrence and fate in wastewater and

572

surface waters. Environ. Sci. Technol. 42, 7193-7200.

573

Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2008. Multiresidue methods for the analysis of pharmaceuticals, personal

574

care products and illicit drugs in surface water and wastewater by solid-phase extraction and ultra performance liquid

575

chromatography-electrospray tandem mass spectrometry. Anal. Bioanal. Chem. 391, 1293-1308.

576

Kienzler, A., Bopp, S.K., van der Linden, S., Berggren, E., Worth, A., 2016. Regulatory assessment of chemical mixtures:

577

Requirements, current approaches and future perspectives. Regul. Toxicol. Pharmacol. 80, 321-334.

578

Lee, H.B., Sarafin, K., Peart, T.E., 2007. Determination of beta-blockers and beta2-agonists in sewage by solid-phase

579

extraction and liquid chromatography-tandem mass spectrometry. J. Chrom. A 1148, 158-167.

580

Lindberg, R.H., Fick, J., Tysklind, M., 2010. Screening of antimycotics in Swedish sewage treatment plants-waters and sludge.

581

Water Res. 44, 649-657.

582

Link, M., von der Ohe, P., Voß C., Schäfer, R.B. (2017): Comparison of dilution factors for German wastewater treatment

583

plant effluents in receiving streams to the fixed dilution factor from chemical risk assessment. Sci. Total Environ. 598, 805-

584

813

585

Loos, R., Carvalho, R., Antonio, D.C., Comero, S., Locoro, G., Tavazzi, S., Paracchini, B., Ghiani, M., Lettieri, T., Blaha, L.,

586

Jarosova, B., Voorspoels, S., Servaes, K., Haglund, P., Fick, J., Lindberg, R.H., Schwesig, D., Gawlik, B.M., 2013. EU-wide

AC C

EP

TE D

M AN U

SC

RI PT

557

23

ACCEPTED MANUSCRIPT monitoring survey on emerging polar organic contaminants in wastewater treatment plant effluents. Water Res. 47, 6475-

588

6487.

589

Moermond, C.T.A., Smit, C.E., 2016. Derivation of water quality standards for carbamazepine, metoprolol, and metformin

590

and comparison with monitoring data. Environ. Toxicol. Chem. 35, 882-888.

591

Oakes, K.D., Coors, A., Escher, B.I., Fenner, K., Garric, J., Gust, M., Knacker, T., Küster, A., Kussatz, C., Metcalfe, C.D.,

592

Monteiro, S., Moon, T.W., Mennigen, J.A., Parrott, J., Péry, A.R.R., Ramil, M., Roennefahrt, I., Tarazona, J.V., Sánchez-

593

Argüello, P., Ternes, T.A., Trudeau, V.L., Boucard, T., Van Der Kraak, G.J., Servos, M.R., 2010. An environmental risk

594

assessment for the serotonin re-uptake inhibitor fluoxetine - a case study utilizing the European risk assessment

595

framework. Integrat. Environ. Assess. Manag. 6, Suppl. 1, 524-539.

596

Pedrouzo, M., Borrull, F., Marce, R.M., Pocurull, E., 2009. Ultra-high-performance liquid chromatography-tandem mass

597

spectrometry for determining the presence of eleven personal care products in surface and wastewaters. J. Chrom. A 1216,

598

6994-7000.

599

Price, P., Han, X., Junghans, M., Kunz, P., Watts, C., Leverett, D. 2012. An application of a decision tree for assessing effects

600

from exposures to multiple substances to the assessment of human and ecological effects from combined exposures to

601

chemicals observed in surface waters and waste water effluents. Environmental Sciences Europe 24:34.

602

Price P, Han X (2011) Maximum Cumulative Ratio (MCR) as a tool for assessing the value of performing a cumulative risk

603

assessment. Int. J. Environ. Res. Public Health 8, 2212-2225.

604

Price, P., Han, X., Junghans, M., Kunz, P., Watts, C., Leverett, D. 2012. An application of a decision tree for assessing effects

605

from exposures to multiple substances to the assessment of human and ecological effects from combined exposures to

606

chemicals observed in surface waters and waste water effluents. Environmental Sciences Europe 24:34.

607

R Development Core Team, 2013. R: A language and environment for statistical computing. Version 3.0.1. R Foundation for

608

Statistical Computing, Vienna, Austria. www.R-project.org.

609

Reemtsma, T., Miehe, U., Duennbier, U., Jekel, M., 2010. Polar pollutants in municipal wastewater and the water cycle:

610

Occurrence and removal of benzotriazoles. Water Res. 44, 596-604.

611

Richter, E., Roller, E., Kunkel, U., Ternes, T.A., Coors, A., 2016. Phytotoxicity of wastewater-born micropollutants –

612

characterisation of three antimycotics and a cationic surfactant. Environ. Pollut. 208, 512-522.

613

Richter, E., Wick, A., Ternes, T.A., Coors, A., 2013. Ecotoxicity of climbazole, a fungicide contained in anti-dandruff shampoo.

614

Environ. Toxicol. Chem. 32, 2816-2825.

615

Ritz, C., Baty, F., Streibig, J.C., Gerhard, D., 2015. Dose-response analysis using R. PLoS ONE 10(12): e0146021.

616

http://dx.doi.org/10.1371/journal.pone.0146021.

AC C

EP

TE D

M AN U

SC

RI PT

587

24

ACCEPTED MANUSCRIPT Rodil, R., Quintana, J.B., Lopez-Mahia, P., Muniategui-Lorenzo, S., Prada-Rodriguez, D., 2009. Multi-residue analytical

618

method for the determination of emerging pollutants in water by solid-phase extraction and liquid chromatography-

619

tandem mass spectrometry. J. Chrom. A 1216, 2958-2969.

620

Rosal, R., Rodriguez, A., Perdigon-Melon, J.A., Petre, A., Garcia-Calvo, E., Gomez, M.J., Aguera, A., Fernandez-Alba, A.R.,

621

2010. Occurrence of emerging pollutants in urban wastewater and their removal through biological treatment followed by

622

ozonation. Water Res. 44, 578-588.

623

Santos, L.H.M.L.M., Gros, M., Rodriguez-Mozaz, S., Delerue-Matos, C., Pena, A., Barceló, D., Montenegro, M.C.B.S.M., 2013.

624

Contribution of hospital effluents to the load of pharmaceuticals in urban wastewaters: Identification of ecologically

625

relevant pharmaceuticals. Sci. Total Environ. 461-462, 302-316.

626

SC, 2015. Assessment report Propiconazole Product type 7 (film preservatives). Evaluation of active substances according to

627

Regulation (EU) n°528/2012 concerning the making available on the market and use of biocidal products. Not yet finalised

628

in the Standing Committee (SC) on Biocidal Products. January 2015.

629

Schlüter-Vorberg, L., Knopp, G., Cornel, P., Ternes, T., Coors, A. 2017, Survival, reproduction, growth, and parasite

630

resistance of aquatic organisms exposed on-site to wastewater treated by advanced treatment processes. Aquat. Toxicol.

631

186, 171-179.

632

Ternes, T.A., Bonerz, M., Herrmann, N., Teiser, B., Andersen, H.R. 2007. Irrigation of treated wastewater in Braunschweig,

633

Germany: an option to remove pharmaceuticals and musk fragrances. Chemosphere 66, 894-904.

634

Thomaidi, V.S., Stasinakis, A.S., Borova, V.L., Thomaidis, N.S., 2015. Is there a risk for the aquatic environment due to the

635

existence of emerging organic contaminants in treated domestic wastewater? Greece as a case-study. J. Hazard. Mater.

636

283, 740-747.

637

Unceta, N., Sampedro, M.C., Abu Bakar, N.K., Gomez-Caballero, A., Goicolea, M.A., Barrio, R.J. 2010. Multi-residue analysis

638

of pharmaceutical compounds in wastewaters by dual solid-phase microextraction coupled to liquid chromatography

639

electrospray ionization ion trap mass spectrometry. J. Chrom. A 1217, 3392-3399.

640

Universität Dortmund, Fachbereich Chemietechnik, Lehrstuhl Umwelttechnik, 2003. Studies on the entry and elimination of

641

hazardous substances in municipal sewage treatment plants Part 1 (Untersuchungen zum Eintrag und zur Elimination von

642

gefährlichen Stoffen in kommunalen Kläranlagen Teil 1). Final report on a research project. Dortmund.

643

Van de Steene, J.C., Stove, C.P., Lambert, W.E., 2010. A field study on 8 pharmaceuticals and 1 pesticide in Belgium: removal

644

rates in waste water treatment plants and occurrence in surface water. Sci. Total Environ. 408, 3448-3453.

645

Vestel, J., Caldwell, D.J., Constantine, L., d’Aco, V.J., Davidson, T., Dolan, D.G., Millard, S.P., Murray-Smith, R., Parke, N.,

646

Ryan, J.J., Straub, J.O., Wilson, P., 2016. Use of acute and chronic ecotoxicity data in environmental risk assessment of

647

pharmaceuticals. Environ. Toxicol. Chem. 35, 1201-1212.

AC C

EP

TE D

M AN U

SC

RI PT

617

25

ACCEPTED MANUSCRIPT Watanabe, H., Tamura, I., Abe, R., Takanobu, H., Nakamura, A., Suzuki, T., Hirose, A., Nishimura, T., Tatarazako, N., 2016.

649

Chronic toxicity of an environmentally relevant mixture of pharmaceuticals to three aquatic organisms (alga, daphnid, and

650

fish). Environ. Toxicol. Chem. 35, 996-1006.

651

Weiss, S., Jakobs, J., Reemtsma, T., 2006. Discharge of three benzotriazole corrosion inhibitors with municipal wastewater

652

and improvements by membrane bioreactor treatment and ozonation. Environ. Sci. Technol. 40, 7193-7199.

653

Weiss, S., Reemtsma, T., 2005. Determination of benzotriazole corrosion inhibitors from aqueous environmental samples by

654

liquid chromatography-electrospray ionization-tandem mass spectrometry. Anal. Chem. 77, 7415-7420.

655

Wheeler, M.W., Park, R.M., Bailer, A.J., 2006. Comparing median lethal concentration values using confidence interval

656

overlap or ratio tests. Environ. Toxicol. Chem. 25(5), 1441–1444.

657

Wick, A., Fink, G., Ternes, T.A., 2010. Comparison of electrospray ionization and atmospheric pressure chemical ionization

658

for multi-residue analysis of biocides, UV-filters and benzothiazoles in aqueous matrices and activated sludge by liquid

659

chromatography–tandem mass spectrometry. J. Chrom. A 1217, 2088–2103.

660

Writer, J.H., Ferrer, I., Barber, L.B., Thurman, E.M., 2013. Widespread occurrence of neuro-active pharmaceuticals and

661

metabolites in 24 Minnesota rivers and wastewaters. Sci. Total Environ. 461-462, 519-527.

662

Zorita, S., Mårtensson, L., Mathiasson, L., 2009. Occurrence and removal of pharmaceuticals in a municipal sewage

663

treatment system in the south of Sweden. Sci. Total Environ. 407, 2760-2770.

SC

M AN U

TE D EP AC C

664

RI PT

648

26

ACCEPTED MANUSCRIPT Table 1: Exposure estimates for the selected substances as concentrations in WWTP effluents reported in the literature and measured in the present study (median and 90% percentile). Detections were defined as values above the limit of quantification in the respective study. Median measured concentration in WWTP effluents (MCeff) in µg/L (90% percentile)

a,b,c,d,e,f

90 (91)

0.74 (2.3)

a,f,g,h,i

81 (83)

0.92 (1.8)

4 (13)

0 (0.0062)

9 (9)

0.15 (0.37)

24 (27)

0.012 (0.042)

5-MBT

Methylparaben

a,n,p

Propiconazole

a,n,q,s

16 (19)

a,c,e,f,t,u

89 (89)

Fluconazole Metoprolol

Fenofibric acid Fluoxetine

a

5 (5) 50 (59)

0.044 (0.086) 1.1 (2.2)

0.13 (0.24)

0.035 (0.078)

4 (4)

0.047 (0.074)

1 (6)

0 (0.018)

EP

present study; b García-López et al. (2010); c Bendz et al. (2005); d Rodil et al. (2009); e Universität Dortmund (2003); f unpublished own data; g Weiss et al. (2006); h Weiss & Reemtsma (2005); i Reemtsma et al. (2010); j Blanco et al. (2009); k Pedrouzo et al. (2009); l Benijts et al. (2004); m Canosa et al. (2006); n Chen et al. (2012); o Gonzalez-Marino et al. (2009); p Wick et al. (2010); q Kahle et al. (2008); r van de Steene et al. (2010); s Lindberg et al. (2010); t Kasprzyk-Hordern et al. (2008); u Lee et al. (2007); v Ternes et al. (2007), w Unceta et al. (2010); x Writer et al. (2013); y Hedgespeth et al. (2012); z Zorita et al. (2009)

AC C

a

a

a,v

a,w,x,y,z

Amlodipine Linezolid

a,p,q,r

TE D

Climbazole

M AN U

a,j,k,l,m,n,o

RI PT

TCPP

Number of detections (number of measurements)

SC

Substance

ACCEPTED MANUSCRIPT Table 2: Toxicity estimates and their 95% confidence intervals (CI) for the 11 individual substances determined for the key endpoints 10% inhibition of growth rate (ErC10) and 10% inhibition of reproduction (EC10) in the three test organisms. In addition, the mean (n=3-9) of the measured initial test concentration in relation to the nominal test concentration (% recovery) is provided. All toxicity estimates are based on nominal concentrations unless otherwise indicated. Predicted-No-Effect-Concentrations (PNECs) were derived based on data produced in the present study (indicated in bold) or literature data with the given assessment factor (AF) Substance

L. minor

R. subcapitata

D. magna

ErC10 (CI) in mg/L and [recovery]

ErC10 (CI) in mg/L and [recovery]

EC10 (CI) in mg/L and [recovery]

TCPP

44.2 (28.6-60.8) a [90.7%]

40.3 (36.8-43.7) a [102.3%]

2.87 (1.99-3.74) a,b [99.1%]

5-MBT

13.8 (10.3-17.4) a [98.0%]

39.9 (33.7-46.0) a [96.6%]

>10.0 [92.0%]

a

Methylparaben

16.3 (10.6-21.9) a [91.7%]

34.2 (20.3-48.1) a [99.2%]

>10.0 [91.6%]

a

Climbazole

0.008 (0.006-0.010) c,d [70.5%]

0.315 (0.223-0.407) e [85.0%]

Propiconazole

0.356 (0.276-0.428) a,f [132.2%]

Fluconazole

0.473 (0.369-0.577) c [121%]

Metoprolol

>156 [62.5%]

Fenofibric acid

n.a.

Fluoxetine

n.a.

0.001 (n.a.) [n.a.]

Amlodipine

n.a.

0.342 (0.116-0.568) a [118.4%] 0.183 (0.064-0.301) a [100.7%]

a

n.a.

RI PT

PNEC (µg/L)

28.7

100

100

100

58.4 i

0.454 (0.060-0.848) a [97.7%]

50

0.16

1.01 (0.49-1.53) g [82.6%]

0.869 (0.699-1.04) a [89.5%]

10

6.8 j

26.8 (9.7-43.7) a [91.8%]

20.0 (0-42.3) a [99.8%]

50

9.46

6.10 (0-17.1) a [91.1%]

10

610

100

0.4

10

0.1

n.a.

100

3.4

n.a.

100

18.3

M AN U

SC

100

TE D

EP

AC C

Linezolid

a,f

AF

22.5 (17.8-27.2) a [102.2%] >12.9 [129.0%]

a,f

0.04 (0.006-0.068) a [114.2%] d,h

0.113 (n.a.) [n.a.]

d,h

present study; b extrapolated beyond the lowest tested concentration of 5.62 mg/L; c Richter et al. 2016; d EC10 corrected for mean (initial and aged) measured concentrations; e Richter et al. 2013; f EC10 corrected for initial measured concentrations; g Coors et al. 2014 (recalculated); h Oakes et al. 2010 (recalculated); i based on survival of D. magna, 21 d, with LC50 of 5.84 in present study; j NOEC fish of 0.068 mg/L (SC 2015); n.a. not available

ACCEPTED MANUSCRIPT Table 3: Model Deviation Ratios (MDRs) determined for the tested mixtures as quantitative measure for the compliance between observed and CA-predicted mixture toxicity. MDR values are based on nominal concentrations, and MDR values based on measured concentrations are additionally given in brackets if the measured concentration of any mixture component deviated by more than 20% from the nominal concentration Endpoint *

MDR in absence of WWTP effluent

MDR in presence of WWTP effluent

(A) Equipotent mixture of three components with similar intended MoA

L. minor

0.93 (1.16)

0.61 (0.75)

R. subcapitata

0.27

0.19

D. magna

0.77 (1.08)

not tested

2.45 (2.98)

0.46 (0.61)

SC

RI PT

Mixture

R. subcapitata

(C) Multi-component mixture at an equipotent ratio

R. subcapitata

0.71 (0.67)

0.97 (0.79)

D. magna **

0.67 (0.83)

0.72 (0.90)

R. subcapitata

0.38 (0.43)

not tested

D. magna **

0.39 (0.44)

not tested

TE D

(D) Multi-component mixture toxicity-dominated by three of the components

M AN U

(B) Equipotent mixture of three components with dissimilar intended MoA

AC C

EP

* L. minor: EC10 of growth rate inhibition based on frond number (7 d); R. subcapitata: EC10 of growth rate inhibition (72 h); D. magna: EC10 of reproduction inhibition (21 d); ** MDR calculation excluding 5-MBT and methylparaben

ACCEPTED MANUSCRIPT Table 4: Risk quotients for single substances and mixtures based on predicted no effect concentrations (PNECs) or species-specific 10% effect concentrations (EC10) and median as well as 90% percentile of measured concentrations in WWTP effluents (MCeff). Dilution of WWTP effluents in receiving surface waters is not taken into account for the risk quotients. Substance

Median MCeff/PNEC

90% percentile MCeff/PNEC

90% percentile MCeff/EC10,algae -2

90% percentile MCeff/EC10,daphnia

[X*10-2]

90% percentile MCeff/EC10,lemna

[X*10-2]

RI PT

[X*10 ] 0.026

0.080

0.0057

5-MBT

0.009

0.018

0.0045

0

0.0001

<0.0001

Climbazole

0.938

2.312

0.1175

Propiconazole

0.002

0.006

0.0042

0.0005

0.0012

Fluconazole

0.005

0.009

0.0003

<0.0001

0.0018

Metoprolol

0.002

0.004

0.0098

0.0036

<0.0001

Fenofibric acid

0.325

0.600

<0.0001

0.0600

n.a.

Fluoxetine

0.350

0.780

7.80

0.0069

n.a.

Amlodipine

0.014

0.022

0.0216

n.a.

n.a.

0

0.001

0.0098

n.a.

n.a.

7.973

0.089

0.467

Sum for the mixture

1.67

AC C

EP

n.a.: not available

M AN U

Linezolid

TE D

Methylparaben

3.83

0.008

0.0005

0.0018

0.0013

<0.0001

<0.0001

0.0081

0.4625

SC

TCPP

ACCEPTED MANUSCRIPT Figure Labels

RI PT

Figure 1: Concentration-response curves for three-component mixtures in absence (full line, filled circles) and presence (dashed line, empty circles) of WWTP effluents. The dotted lines represent the respective response curves predicted by CA. Shown are mean responses related to summed nominal concentrations of the three mixture components, all present at an equipotent ratio. Number of replicates per treatment was three for primary producers and ten for D. magna. Climbazole, fluconazole, and propiconazole represent a mixture of components with a similar intended MoA tested in L. minor (A), D. magna (B), and R. subcapitata (C). Climbazole, metoprolol, and amlodipine represent a mixture of components with a dissimilar intended MoA tested in R. subcapitata (D).

AC C

EP

TE D

M AN U

SC

Figure 2: Concentration-response curves for multi-component mixtures in absence (full line, filled circles) and presence (dashed line, empty circles) of WWTP effluents. The dotted lines represent the respective response curves predicted by concentration addition. Shown are mean responses related to summed nominal concentrations of the mixture components. Number of replicates per treatment was three for algae and ten for D. magna. Climbazole, fluconazole, propiconazole, 5-MBT, TCPP, methylparaben, fluoxetine, metoprolol, amlodipine, and linezolid represent a 10-component mixture tested in R. subcapitata at an equipotent ratio (A) and at a ratio based on concentrations measured in WWTP effluents (C). Climbazole, fluconazole, propiconazole, 5-MBT, TCPP, methylparaben, fluoxetine, metoprolol, and fenofibric acid represent a 9-component mixture tested in D. magna at an about equipotent ratio (B) and at a ratio based on concentrations measured in WWTP effluents (D).

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Figure 1

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Figure 2

ACCEPTED MANUSCRIPT Chronic aquatic toxicity of so far untested pharmaceuticals and other chemicals

•

Chronic toxicity of mixtures can be predicted with less than 5-fold deviation

•

Initial assessment indicates environmental risk only for mixtures

•

Refinement indicates acceptable risk for mixtures, but uncertainties remain

•

Risk quotient for mixture less than twofold greater than that of single chemicals

AC C

EP

TE D

M AN U

SC

RI PT

•