Hydrophilic anthropogenic markers for quantification of wastewater contamination in ground-and surface WATERS

Authors

  • Maren Kahle,

    1. Plant Protection Chemistry, Agroscope Changins-Wädenswil Research Station ACW, Schloss, P.O. Box 185, CH-8820 Wädenswil, Switzerland
    Search for more papers by this author
  • Ignaz J. Buerge,

    1. Plant Protection Chemistry, Agroscope Changins-Wädenswil Research Station ACW, Schloss, P.O. Box 185, CH-8820 Wädenswil, Switzerland
    Search for more papers by this author
  • Markus D. Müller,

    1. Plant Protection Chemistry, Agroscope Changins-Wädenswil Research Station ACW, Schloss, P.O. Box 185, CH-8820 Wädenswil, Switzerland
    Search for more papers by this author
  • Thomas Poiger

    Corresponding author
    1. Plant Protection Chemistry, Agroscope Changins-Wädenswil Research Station ACW, Schloss, P.O. Box 185, CH-8820 Wädenswil, Switzerland
    • Plant Protection Chemistry, Agroscope Changins-Wädenswil Research Station ACW, Schloss, P.O. Box 185, CH-8820 Wädenswil, Switzerland
    Search for more papers by this author

  • Published on the Web 8/14/2009.

Abstract

Hydrophilic, persistent markers are useful to detect, locate, and quantify contamination of natural waters with domestic wastewater. The present study focused on occurrence and fate of seven marker candidates including carbamazepine (CBZ), 10,11-dihydro-10,11-dihydroxycarbamazepine (DiOH-CBZ), primidone (PMD), crotamiton (CTMT), N-acetyl-4-aminoantipyrine (AAA), N-formyl-4-aminoantipyrine (FAA), and benzotriazole (BTri) in wastewater treatment plants (WWTPs), lakes, and groundwater. In WWTPs, concentrations from 0.14 μg/L to several micrograms per liter were observed for all substances, except CTMT, which was detected at lower concentrations. Loads determined in untreated and treated wastewater indicated that removal of the potential markers in WWTPs is negligible; only BTri was partly eliminated (average 33%). In lakes, five compounds, CBZ, DiOH-CBZ, FAA, AAA, and BTri, were consistently detected in concentrations of 2 to 70 ng/L, 3 to 150 ng/L, less than the limit of quantification to 30 ng/L, 2 to 80 ng/L, and 11 to 920 ng/L, respectively. Mean per capita loads in the outflows of the lakes suggested possible dissipation in surface waters, especially of AAA and FAA. Nevertheless, concentrations of CBZ, DiOH-CBZ, and BTri correlated with the actual anthropogenic burden of the lakes by domestic wastewater, indicating that these compounds are suitable for quantification of wastewater contamination in lakes. Marker candidates were also detected in a number of groundwater samples. Carbamazepine concentrations up to 42 ng/L were observed in aquifers with significant infiltration of river water, receiving considerable wastewater discharges from WWTPs. Concentration ratios between compounds indicated some elimination of BTri and DiOH-CBZ during subsurface passage or in groundwater, while CBZ and PMD appeared to be more stable and thus are promising wastewater markers for groundwater. The wastewater burden in groundwater, estimated with the markers CBZ and PMD, reached up to 6%.

INTRODUCTION

Natural waters require rigorous protection from contamination by xenobiotic chemicals and pathogenic microbes, which may originate from domestic, industrial, traffic, or agricultural activities. For the detection of contamination by domestic wastewater, numerous chemical markers have been suggested in the recent literature [1–13]. An ideal marker should allow the unambiguous recognition of the source and the quantification of the magnitude of a pollution [14]. For domestic wastewater, constant per capita loads are a further prerequisite for a good marker as well as sufficient quantities to permit analytical quantification after dilution in the environment. To trace pollution caused by domestic wastewater, constituents and human metabolites of pharmaceuticals, personal care products, household chemicals, and food may be suitable candidates.

Depending on the information needed, different kinds of marker substances are useful. For example, coprostanol, linear alkylbenzenes, and additional lipophilic markers have been proposed to assess the magnitude of sediment contamination by discharges of domestic wastewater [1]. Caffeine and cotinine were shown to be suitable hydrophilic markers for untreated domestic wastewater since they are extensively degraded in wastewater treatment plants (WWTPs) with efficiencies of typically >98% [3,12]. Caffeine has been used for the quantification of discharges of raw wastewater, primarily from overflows of combined sewer systems in Switzerland [15].

To investigate contamination of surface waters and especially groundwater by treated and untreated domestic wastewater, hydrophilic and persistent markers are necessary. Ideally, these compounds are not significantly retained or degraded in WWTPs, in surface waters, during infiltration, and in groundwater.

The aim of the present study was to evaluate the suitability of a set of compounds as hydrophilic, persistent markers of domestic wastewater in surface and groundwaters. Seven candidate compounds were selected from a much larger number of pharmaceuticals, personal care products, and household chemicals. Important criteria were their physico-chemical properties, data on usage and environmental concentrations, and information on their elimination behavior in WWTPs and natural waters, so far reported in the literature. Some of the compounds have already been proposed as markers by other authors.

The anticonvulsant carbamazepine (CBZ) has been detected in wastewater [5,7,10,11,16–19], surface waters [7,11,13,16,20,21], and groundwater [5,13,22–24]. The compound showed little degradation or retention in WWTPs [5,18,19], during bank filtration [22,23], or in sediment transport studies [25]. Fenz et al. [24] used CBZ as a marker for groundwater contamination through sewer exfiltration. The main human metabolite of CBZ, 10,11-dihydro-10,11-dihydroxycarbamazepine (DiOH-CBZ) was also shown to be present in wastewater [17]. It was not removed during wastewater treatment [17].

Similar to CBZ, the anticonvulsant primidone (PMD) was readily transported during bank filtration [22,23]. N-Acetyl-4-aminoantipyrine (AAA) and N-formyl-4-aminoantipyrine (FAA), both metabolites of the analgesic metamizole, have been observed in wastewater, with no or only partial removal in WWTPs [26]. Along the River Elbe (Germany) and its tributaries, FAA and AAA were frequently found [21]. Both substances were also detected in groundwater after bank filtration of lake water ([27]; www.blac.de/servlet/is/2146/P-2c.pdf). The antipruritic drug crotamiton (CTMT) was found in five WWTPs of Tokyo [10]. Crotamiton was persistent during secondary treatment in WWTPs [10]. The compound was detected in rivers, groundwater, and estuaries and was recommended as a conservative marker [13].

Finally, benzotriazole (BTri) is used for silver protection in dishwasher detergents, but has also industrial applications as anticorrosive, e.g., in aircraft de-icer and anti-icer fluids [28]. This compound was found in WWTPs, rivers, and lakes [28,29]. The elimination of BTri in WWTPs seemed to be relatively low [29,30].

With the current study, additional data is provided on occurrence and fate of the selected compounds in WWTPs, surface waters, and groundwater. The elimination behavior in WWTPs was studied with analyses in influent and effluent wastewater (flow-proportional 24-h composite samples). To investigate their suitability as markers for quantitative application in surface waters, concentrations in a series of Swiss lakes were correlated with their actual anthropogenic burden. Per capita loads determined in the outflows of lakes were compared with those in WWTP effluents to evaluate their persistence in these lakes. The marker candidates were finally analyzed in groundwater samples from a regular monitoring program. In particular, we focused on concentration ratios between marker compounds in wastewater, surface waters, and groundwater, in order to gain additional information on their relative elimination behavior in surface waters, through subsurface passage or in groundwater.

MATERIALS AND METHODS

Chemicals

4-Acetylaminoantipyrine (purity, 97%), 1H-benzotriazole (99%), carbamazepine, N-ethyl-O-crotonotoluidide (CTMT, 97%), 4-formylaminoantipyrine, and primidone were purchased from Sigma-Aldrich Chemie; 10,11-dihydro-10,11-dihydroxycarbamazepine from μ-Mol was kindly provided by T.A. Ternes, Federal Institute of Hydrology, Koblenz, Germany. Structures and some physicochemical properties of the compounds investigated are shown in the Supporting Information (Table S1; http://dx.doi.org/10.1897/08–606.S1). Carbamazepine-D10 (ring-D10, 98%) from CDN Isotopes (Pointe-Claire, QC, Canada) was used as an internal standard.

Water samples

Wastewater samples were obtained from nine municipal WWTPs in the region of Zürich, Switzerland (Supporting Information, Table S2; http://dx.doi.org/10.1897/08–606.S1). The plants operate with a mechanical, biological (activated sludge, mostly with nitrification and partially with denitrification), and chemical treatment (phosphate coprecipitation by iron salts, without chlorination) and subsequent sand filtration (with flocculation) ([31]; www.abwasser.zh.ch/internet/bd/awel/gs/aw/de/ara.html). In WWTP Wädenswil, 50% of wastewater passes a membrane filtration. The WWTPs serve populations of between 10,000 and 370,000 (Supporting Information, Table S2). Influent and effluent wastewaters were collected flow-proportionally during 24 h. Generally, influent samples were taken after the primary sedimentation basin; in Zürich and Männedorf, raw wastewater was collected. Treated effluent samples were taken after sand filtration. All WWTPs were sampled once whereby influent and effluent were sampled during the same 24-h period irrespective of the hydraulic retention time.

Surface water was sampled from eight lakes in the Swiss Midland region. The lakes differed with respect to population in the catchment area, morphology, hydraulics, and chemical and biological characteristics (Supporting Information, Table S3; http://dx.doi.org/10.1897/08–606.S1) ([32]; www.bafu.admin.ch/publikationen/publikation/00361/index.html?lang=de, [33]; www.gewaesserqualitaet.zh.ch/internet/bd/awel/gs/gq/de/doku/dok.html). While the lakes are stratified during the warmer season (April-November), they are usually mixed down to the lake bottom in winter. Water samples from these lakes were taken in winter at their outflow at 0 to 1 m depth so that the corresponding samples may be considered as representative for the whole lake. Groundwater samples were obtained from 15 pumping stations located in Canton of Zurich, Switzerland (Supporting Information, Table S4; http://dx.doi.org/10.1897/08–606.S1). For the present study, each lake and each groundwater pumping station was sampled once.

All water samples were filled into glass bottles and stored at 4°C in the dark. Before analysis, lake and groundwater samples were fortified with internal standard (100 μg/L in methanol) to spike levels of 100 ng/L. Wastewater was filtered through glass microfiber filters (GF/F, Whatman International). Analyses were done within 1 to 5 d.

Online solid-phase extraction-liquid chromatography-tandem mass spectrometry

All substances were analyzed with high-performance liquid chromatography (HPLC) coupled to tandem mass spectrometry (HPLC pump, Agilent 1100 Series, Agilent Technologies; HTS PAL autosampler, CTC Analytics; API 4000 triple quadrupole mass spectrometer, Applied Biosystems). Direct coupling of an online solid-phase extraction cartridge (two stacked Gemini C18 cartridge pre-columns, 4 × 3 mm, 5 μm, Phenomenex®) to the liquid chromatography (LC)-tandem mass spectrometry (MS) was accomplished using a column switching technique [34].

Sample volumes of 1 ml were injected into a loop. Samples were then transferred from the loop to the extraction cartridge using an auxiliary HPLC pump (Jasco PU980, Jasco®) and purified water as mobile phase (BTri: purified water with 0.1% formic acid) at a flow rate of 1 ml/min during 90 s. After valve switching, the enriched compounds were eluted directly to the analytical column with the mobile phase used for analysis. The compounds were separated on a Gemini 5 μC18 110A(150 × 2 mm, 5 μm, Phenomenex) HPLC column, fitted with a cartridge precolumn (4 × 2 mm) filled with the same stationary phase. For BTri, which was analyzed separately, the mobile phases were LCMS-grade water (Carl Roth; solvent A) and methanol (solvent B), both with 0.1% formic acid. For all other substances, the mobile phases were LCMS-grade water with 1 mM ammonium acetate (solvent A) and methanol (solvent B). Initial solvent composition was 90% A to 10% B. At the beginning of the elution phase, the content of B was quickly increased to 50% within 1 min, then with a linear gradient to 90% B within 9 min, followed by a 2-min isocratic hold. Afterwards, the initial composition was reestablished within 1 min, and the column was allowed to equilibrate for 7 min prior to the consecutive injection. The flow rate was set to 0.2 ml/min.

The mass spectrometer was equipped with a turbo ion spray source and operated in positive mode (ion spray voltage, 5 kV, 450°C) using multiple reaction monitoring. Ion transitions optimized for the marker candidates, limits of quantification, and reproducibility information are given in Supporting Information, Table S5 (http://dx.doi.org/10.1897/08–606.S1). Peak areas were integrated and calibration curves were performed with area ratio of analyte to the labeled internal standard versus concentration ratio. To account for matrix effects, in wastewater samples all marker compounds were quantified by standard addition. For BTri, standard addition was applied in all matrices.

RESULTS AND DISCUSSION

Occurrence and elimination in WWTPs

The substances, selected as potential chemical markers for the present study, were detected in all WWTPs investigated in the region of Zürich (Table 1). Their concentrations ranged from 0.3 μg/L to several micrograms per liter, except PMD with concentrations always below 0.5 μg/L and CTMT below 0.15 μg/L. Concentrations up to 24 μg/L were observed for BTri, and thus much higher than for the other substances.

Generally, the concentrations determined in treated wastewater were similar to those in untreated wastewater (Table 1). Deviations (positive and negative) as in WWTPs Männedorf, Meilen, and Wetzikon may be due to the fact that influent and effluent samples were taken during the same 24-h periods and thus did not correspond exactly to the same water package. An exception was BTri with consistently lower concentrations in effluent compared to influent wastewater. The results indicate that BTri was removed in WWTPs to some extent (33% average removal) whereas elimination of all other compounds was low to negligible. Consequently, all compounds are expected to regularly enter surface waters with treated wastewater.

The mean substance loads in untreated and treated wastewater, normalized for the population serviced by the plants, are also listed in Table 1. For CBZ, a mean influent per capita load of 0.44 mg/person/d was observed. This value is in excellent agreement with the expected load of 0.45 mg/person/d, estimated from the amount of CBZ prescribed in the year 2000 (4,065 kg, [35]) normalized to the population of Switzerland (7.5 million persons, 2006, see www.bfs.admin.ch), considering typical excretion rates of CBZ of 2% (renal) and 28% (with feces, www.kompendium.ch). In the human body, CBZ is extensively metabolized, mainly to CBZ-10,11-epoxide and further to DiOH-CBZ. Conjugation of DiOH-CBZ to the corresponding glucuronides was also observed [17]. For DiOH-CBZ, the average measured load (0.73 mg/person/d) was similar to the load calculated using consumption and pharmacological data (0.45 mg/person/d, renal excretion rate 30% relative to parent, including conjugates; www.kompendium.ch). The per capita loads observed for CBZ and DiOH-CBZ are within the range of previously reported data [16–18]. Removal of the compounds in WWTPs was low or insignificant in most studies published in literature [5,17,18].

For BTri, a normalized load of approximately 5 mg/person/d was previously reported in a main sewer of an urban area in the vicinity of Zurich [36], which is similar to the mean per capita load in untreated wastewater determined in the present study (4.3 mg/person/d). Voutsa et al. [29] observed BTri concentrations of 13 to 75 μg/L and 11 to 100 μg/L in the primary and secondary effluent of Swiss WWTPs, respectively. The elimination rates were up to 62% [29]. In a German WWTP, untreated and treated wastewater had BTri concentrations of 12 and 9.6 μg/L, respectively [30].

Comparable to our results, PMD concentrations of 0.20 and 0.11 μg/L were reported for two WWTP effluents in the United States [22]. In six WWTPs in Berlin, influent concentrations of AAA from 6.6 to 15 μg/L and effluent concentrations from 1.8 to 6 μg/L were observed [26]. The respective concentrations for FAA were 1.4 to 3.2 μg/L (influent) and 1.5 to 2.6 μg/L (effluent). While AAA was partly eliminated, no removal of FAA was found in that study [26].

Crotamiton concentrations from 0.25 to 0.97 μg/L observed in WWTPs in Tokyo, Japan [10] were an order of magnitude higher than our values. Differences in use may be the reason. The reported removal efficiency of CTMT in Japanese WWTPs was less than 45% [10].

Marker candidates in lakes

Carbamazepine, DiOH-CBZ, AAA, FAA, and BTri were detected in all lakes investigated, whereas PMD and CTMT, which were present in wastewater in lower concentrations, were found less frequently (Table 2). Consistent with WWTP effluent data, the highest concentrations of all investigated compounds were found for BTri (up to 920 ng/L).

The calculated mean per capita loads of the marker candidates in the outflows of the lakes (Table 2) were somewhat lower (two to three times) than the mean per capita loads in WWTP effluents (Table 1), especially for AAA and FAA, indicating possible dissipation in surface waters (e.g., hydrolysis, photolysis). However, the dissipation processes must be slow since the mean water residence time in the investigated lakes is more than one year (Supporting Information, Table S3).

The apparent slow dissipation of the marker candidates in surface waters can be compared to available data from laboratory experiments on photolysis and degradation in water-sediment systems. For CBZ, a solar photodegradation half-life of 5 d was observed in bi-distilled water at 40°N latitude in spring [37]. The half-life increased in the presence of dissolved organic matter and decreased with nitrate. Photolysis of CBZ may thus play a role in clear, shallow, nitrate-rich waters and in the epilimion of stratified waterbodies during summer. However, on an annual average, photolysis is supposed to be insignificant in the investigated lakes because the photic zone in these lakes is thin (< 1 m) compared to the depth of the water bodies (18–105 m, Supporting Information, Table S3). In a water-sediment study, slow dissipation of CBZ (time for 50% dissipation, DT50, 328 d) and DiOH-CBZ (DT90, >365 d) was observed [38]. For all other substances, no comparable information on dissipation processes was found.

Quantitative suitability of marker candidates in lakes

The concentrations of CBZ, DiOH-CBZ, BTri, and AAA correlated reasonably well (r2 = 0.86–0.90) with the actual anthropogenic burden of the investigated lakes, expressed as ratio of population (P) per water throughflow (Q) (Fig. 1, AAA and BTri not shown). This suggests that the substances enter surface waters primarily associated with domestic wastewater and that they reflect contamination in a quantitative sense. Correlations with P/Q have previously been shown for other compounds, like caffeine, nicotine derivatives, and polycyclic musk compounds, which are suitable anthropogenic markers for domestic wastewater contamination in lakes as well [3,4,12]. Data points for some lakes (Sempachersee, Baldeggersee, and Hallwilersee) consistently were below the correlation lines. These three lakes have fairly high water residence times (15, 5.6, and 3.8 years, respectively, Supporting Information, Table S3) and the lower concentrations probably are due to slow degradation.

Table Table 1.. Concentrations and per capita loads of marker candidates in wastewater treatment plants (WWTPs), Canton of Zurich, Switzerlanda
 CBZDiOH-CBZPMDFAAAAACTMTBTri
WWTPinf.eff.inf.eff.inf.eff.inf.eff.inf.eff.inf.eff.inf.eff.
  1. a inf. = influent; eff. = effluent of WWTP; CBZ = carbamazepine; DiOH-CBZ = 10,11-dihydro-10,11-dihydroxycarbamazepine; PMD = primidone; FAA = N-formyl-4-aminoantipyrine; AAA = N-acetyl-4-aminoantipyrine; CTMT = crotamiton; BTri = benzotriazole.

  2. b Semiquantitative value, because of too low standard additions.

  3. c Raw wastewater was collected in these while primary effluent was sampled in all other WWTPs.

  4. d Qualifier below limit of detection (determined by signal to noise ratio ≥ 3:1), values not used for calculation of mean.

  5. e NQ = substance detected, but not quantifiable with standard addition.

  6. f Analyzed without online-solid-phase extraction.

  7. g ND = not detected.

  8. h Per capita loads calculated with wastewater throughput and population serviced (Supporting Information, Table S2; http://dx.doi.org/10.1897/08–606.S1).

  9. i Mean (±standard deviation [STDV] of samples), PMD influent was calculated without Männedorf.

  10. j For calculations only WWTP Wädenswil, Wetzikon, Gossau, and Uster were used.

  11. k Calculated using the mean per capita wastewater discharge in Canton Zurich of 0.53 m3/person/d [31].

 __________(μg/L)__________
Horgen0.960.981.551.420.270.180.560.571.081.200.030.0510b8b
Zürichc0.891.101.411.540.220.191.360.993.772.680.050.0513b8b
Männedorfc1.620.741.592.780.33d0.421.610.554.551.380.07NQe24b7b
Küsnachf1.030.791.361.23NDgNQe0.710.771.851.85NDgNDg13b9b
Meilenf0.770.361.430.73NDgNDg0.610.322.090.95NDgNDe18b8b
Wädenswil0.660.551.481.360.250.140.380.361.371.64NQeNQe144
Wetzikon0.671.031.872.150.330.340.850.932.773.540.050.101511
Gossau2.372.014.624.900.230.250.370.431.291.96NQeNQe1711
Uster1.131.232.342.400.230.190.740.642.082.310.140.041312
Per capita load (mg/person/d)h
Mean (±STDV)i0.440.370.730.760.090.100.350.251.000.750.030.024.322.89
 ± 0.19± 0.14± 0.18± 0.34± 0.03± 0.06± 0.23± 0.12± 0.65± 0.30± 0.02± 0.005± 0.80± 1.69j
Mean concentration (μg/L)k
Mean (±STDV)i0.840.701.381.430.170.180.660.481.881.420.060.0408.25.5
 ± 0.36± 0.26± 0.34± 0.64± 0.06± 0.11± 0.43± 0.23± 1.23± 0.57± 0.04± 0.009± 1.5± 3.2
Table Table 2.. Concentrations, per capita loads, and concentration ratios of marker candidates in lakes in the Swiss midland region. For abbreviations refer to Table 1
Lake CBZDiOH-CBZPMDFAAAAACTMTBTri
  1. a ND = not detected.

  2. b Substance detected, but below limit of quantification.

  3. c Quantifier above limit of quantification (determined by signal to noise ratio ≥ 5:1), qualifier below limit of detection (signaknoise ≥ 3:1), values not used for mean calculation.

  4. d Per capita loads calculated with annual mean water throughflow and population in catchment area (Supporting Information, Table S3; http://dx.doi.org/10.1897/08–606.S1).

  5. e Mean (±standard deviation [STDV] of samples), for PMD the value of Greifensee is given, BTri was calculated without Walensee.

  6. f Ratio was calculated for each WWTP or lake; the mean values were calculated using these single values.

  7. g Only Greifensee, PMD concentrations in other lakes too low.

  __________(ng/L)__________
WalenseeMarch 10, 200823NDa<LOQb2NDa11c
VierwaldstätterseeMarch 17, 200846NDa<LOQb4<LOQb39
ZürichseeFebruary 2, 200811173c614NDa117
SempacherseeMarch 17, 2008511<LOQb<LOQb5NDa70
HallwilerseeMarch 17, 20089184c<LOQb8<LOQb56
BaldeggerseeMarch 17, 200815275c826NDa139
PfäffikerseeFebruary 19, 200821594c8236336
GreifenseeFebruary 19, 2008671531431766917
Per capita load (mg/person/d)d
Mean (±STDV)e 0.19 ± 0.110.35 ± 0.150.040.09 ± 0.030.22 ±0.110.03 ± 0.011.89 ± 1.01
Mean concentration ratiosf  DiOH-CBZ/CBZPMD/CBZFAA/CBZAAA/CBZ BTri/CBZ
Lakes  2.0 ± 0.40.21g0.5 ± 0.071.1 ± 0.26 11.5 ± 3.2
Treated wastewater  2.1 ± 0.70.26 ±0.150.7 ± 0.242.2 ± 0.80 8.1 ± 2.4

In the investigated lakes, the concentration ratios DiOH-CBZ/CBZ, PMD/CBZ, BTri/CBZ, and FAA/CBZ were in the same range as in treated wastewater, indicating similar behavior of the substances in the aquatic environment (Table 2). Therefore, plotted together, CBZ and DiOH-CBZ concentrations in lakes were in line with their mean concentrations in treated wastewater (Table 1, Fig. 2). In the corresponding plot of BTri versus CBZ data points for BTri consistently were above the line indicating the ratio expected from WWTP effluents (Fig. 2). Industrial inputs of BTri (see Introduction section and Giger et al. [28]) may explain the somewhat higher concentrations in lakes. On the other hand, the deviations may also indicate a higher persistence of BTri compared to CBZ in surface waters. The lower concentration ratios of AAA/CBZ in lakes compared to WWTP effluents again point to preferential dissipation of AAA in surface waters.

Results of the present study indicate that, in summary, CBZ, DiOH-CBZ, and BTri are regularly discharged to surface waters associated with treated domestic wastewater in sufficient amounts to allow for their detection even in weakly polluted water bodies. Their concentrations in lakes correlated reasonably well with the actual anthropogenic burden (expressed as P/Q), and these compounds thus are suitable quantitative markers of domestic wastewater in surface waters. Taking into account the mean concentrations of CBZ, DiOH-CBZ, and BTri in treated wastewater and their respective limits of quantification in surface water samples, wastewater fractions of ≥0.15, 0.35, and 0.28% can be traced, respectively.

Marker candidates in groundwater

Groundwater samples were obtained from a regular monitoring program in the area of Zürich, Switzerland. Nine pumping stations were located in valleys of rivers with moderate to high wastewater burden (such as the rivers Glatt and Limmat [39]), whereas the remaining six pumping stations were located in rural areas (Supporting Information, Table S4). Marker candidates were detected in seven of these 15 groundwater samples (Table 3): Carbamazepine was found in seven samples, BTri in six, and PMD and DiOH-CBZ in four samples, while AAA, FAA, and CTMT were not detected in any of the samples.

For CTMT, concentrations in wastewater were lower than for the other compounds, so that the sensitivity of the analytical method may not have been sufficient for detection in groundwater. Compared to the other substances, AAA and FAA seemed to be effectively removed on the way to groundwater. This is in accordance with previously reported data. Massmann et al. [40] observed an average removal of 96% for AAA and 89% for FAA in the first few meters of infiltration flow below an artificial recharge pond. The authors postulated the importance of oxygen in the biodegradation of FAA and AAA [40]. Batch experiments with filter material (microbially active clay) revealed complete degradation of FAA and AAA within 8 d, while sorption of both substances was negligible [26]. However, in a further study, AAA and FAA were still observed in groundwater after bank filtration of lake water and a flow time of several months [27]. These data on elimination of AAA and FAA somewhat contradict the minimal elimination of the same compounds in WWTPs observed here. However, residence times in WWTPs are much shorter (several hours) than those in the published studies on bank filtration (days). Further, microbial processes in the subsurface soil may well be quite different than in activated sludge treatment.

Figure Fig. 1..

Concentration of carbamazepine (CBZ; •) and 10,11-dihydro-10,11-dihydroxycarbamazepine (DiOH-CBZ; ○) in Swiss lakes plotted versus ratio population in the catchment area per throughflow of water (P/Q). The second abscissa indicates the approximate wastewater burden, derived from P/Q and the mean per capita wastewater discharge in the Canton of Zürich of 0.5 m3/persons/d [31].

The highest concentrations of CBZ, BTri, and PMD were observed in aquifers with significant infiltration of river water, receiving considerable wastewater discharges from WWTPs [39]. Pumping stations PS 1 to 3 all are located in the lower Glatt valley; PS 4, in the Limmat valley (Table 3). The presence of these markers in groundwater may thus indicate infiltration of treated domestic wastewater via surface water (indirect inputs), but may also indicate untreated wastewater, e.g., from a leaky sewer (direct inputs).

Concentration ratios of markers in groundwater

Ideally, a chemical marker is not significantly retained or degraded in groundwater and thus its concentration is only affected by dilution. Likewise, concentration ratios of a set of (ideal) marker compounds should not change significantly from those in wastewater. In fact, in the groundwater samples investigated, the concentration ratios PMD/CBZ were in the same range as in treated or untreated wastewater (Table 3). This is also illustrated in a plot of PMD versus CBZ concentrations in groundwater samples together with the mean concentrations in treated wastewater (Fig. 3). This indicates that the two compounds primarily enter groundwater associated with wastewater and that they behave similarly on the pathway to groundwater and in groundwater.

To quantify the wastewater burden of groundwater samples (indicated in Fig. 3), measured concentrations in groundwater were compared with the mean concentrations in treated (or untreated) wastewater (Table 1). Assuming that no degradation of the compounds occurred, a wastewater burden of 6% (using CBZ or PMD) was calculated for groundwater of pumping station 1, for example.

Figure Fig. 2..

10,11-Dihydro-10,11-dihydroxycarbamazepine (DiOH-CBZ; ○) and benzotriazole (BTri; ▵) versus carbamazepine (CBZ) concentrations in lakes. The lines indicate the mean concentration ratio in treated wastewater (Table 2).

Previously reported data suggest that CBZ and PMD are persistent in or on the way to groundwater. Studies of bank filtration systems in Germany and the United States reported, irrespective of the redox stage, no or only slight removal during groundwater recharge, even though groundwater travel time in the subsurface was up to several years [22,23,40]. No significant removal of CBZ was observed after infiltration of treated wastewater into unsaturated soil and subsequently into groundwater in Austria [5]. In column experiments with sediment under unsaturated conditions, CBZ was not eliminated from percolating water [25]. The sorption coefficients (Kd) of CBZ in two sediments (organic carbon content ≤0.2%) was below 6 L/kg [41]. Fenz et al. [24] already suggested CBZ as a marker for sewer exfiltration of wastewater to groundwater and calculated a loss of 1% of the dry weather flow on a city-wide scale in a field study.

While literature data and our results indicate that PMD and CBZ are quite persistent and, consequently, the concentration ratios PMD/CBZ are similar in wastewater and in groundwater, the concentration ratios DiOH-CBZ/CBZ and BTri/CBZ were lower in most groundwater samples compared to wastewater (Table 3). BTri and DiOH-CBZ thus seemed to be eliminated to a considerable extent through a soil or subsurface passage or in groundwater. Exceptions like for pumping station 5 (DiOH-CBZ/CBZ) and pumping station 4 (BTri/CBZ), with similar ratios in groundwater and untreated wastewater (Table 3), may indicate short pathways or industrial inputs of BTri.

Table Table 3.. Concentrations and concentration ratios of marker candidates in groundwater of pumping stations (PS) in the Canton of Zurich. For abbreviations refer to Table 1
Sample no.CBZ (ng/L)DiOH-CBZ (ng/L)PMD (ng/L)BTri (ng/L)Ratio DiOH-CBZ/CBZRatio PMD/CBZRatio BTri/CBZ
  1. a Quantifier above limit of quantification (determined by S/N ratio ≥ 5:1), qualifier below limit of detection (S/N ≥ 3:1) (S/N = signal to noise).

  2. b ND = not detected.

  3. c Substance detected, but below limit of quantification (Supporting Information, Table S5; http://dx.doi.org/10.1897/08–606.S1).

  4. d Ratio was calculated for each WWTP, the mean values were calculated using these single values.

PS 1422.811160.10.260.4
PS 2341.8a8.4690.10.252.1
PS 321NDb6.3260.301.3
PS 45.01.1a1.6a770.20.3215.3
PS 53.55.6NDNDb1.6
PS 60.97NDbND<LOQc
PS 70.19aNDbND<LOQc
Mean concentration ratiosd
Untreated wastewater   1.8 ± 0.40.28 ±0.1415.4 ± 7.4
Treated wastewater   2.1 ± 0.70.26 ±0.158.1 ± 2.4

CONCLUSIONS

Results of the present study demonstrated that all investigated marker candidates continuously reach the aquatic environment, with domestic wastewater as the major source. The compounds were quite persistent in WWTPs. Only BTri was eliminated from wastewater to some extent (on average 33%). However, BTri concentrations in treated wastewater were still high, so that this substance was also found in lakes and groundwater samples.

Figure Fig. 3..

Correlation between primidone (PMD) and carbamazepine (CBZ) concentrations in groundwater samples. The line indicates the mean concentration ratio in treated wastewater (Table 3). The second abscissa indicates the approximate wastewater burden, calculated from a mean CBZ concentration of =700 ng/L in treated wastewater.

Crotamiton seemed to be relatively persistent in WWTPs, but with concentrations in wastewater always below 0.15 μg/L. Consequently, CTMT is not a suitable marker of domestic wastewater in surface and groundwater as its concentrations were too low, at least in Switzerland.

For the investigated lakes, CBZ, DiOH-CBZ, AAA, and BTri quantitatively reflected the actual anthropogenic burden, considering domestic wastewater as the main source. A comparison of mean per capita loads in treated wastewater and lake water indicated some dissipation of all substances in surface waters, especially of AAA. In cases where water residence times are high, slow dissipation may therefore result in an underestimation of the wastewater contamination. However, in water bodies with short water residence times such as rivers and streams, dissipation will be negligible for these compounds. When using BTri as a marker for domestic wastewater, care should be taken to identify and characterize possible additional sources in the investigated water body.

For the detection and quantification of groundwater contamination with domestic wastewater, results of the present study indicate that CBZ and PMD are suitable markers. This is supported by previously reported data that suggested that both substances are quite persistent in or on the way to groundwater [5,22,23,25,40]. Benzotriazole and DiOH-CBZ were detected in groundwater as well, but seemed to be partially eliminated through a soil or subsurface passage or in groundwater. Nevertheless, detection of these substances in varying concentrations relative to the more persistent compounds may yield further information on the pathway of wastewater to groundwater provided that the elimination processes and their respective rates are known.

Overall, using a set of hydrophilic, persistent marker substances (vs a single substance) is recommended to detect and quantify domestic wastewater in surface and groundwater. Presence of several marker substances in a particular natural water sample certainly is a more reliable indicator for contamination with wastewater compared to the presence of a single substance, both, qualitatively and quantitatively. Furthermore, concentration ratios of different marker substances may provide additional information on source and entry pathways of contamination and environmental processes.

SUPPORTING INFORMATION

Table S1. Structure and properties of compounds investigated.

Table S2. Sampling date, wastewater throughput, and population serviced in WWTPs, Canton of Zurich, Switzerland.

Table S3. Volume, water residence time, water through-flow, and population in the catchment area of the lakes studied.

Table S4. Characterization of groundwater sampling sites.

Table S5. MRM transitions, limits of quantification, and reproducibility of compounds and internal standard investigated.

All found at DOI: 10.1897/08–606.S1 (90 KB PDF).

Acknowledgements

The present study was funded by the Swiss Federal Office for the Environment (FOEN), Bern, Switzerland. We thank A. Bächli, C. Balsiger, M.E. Balmer, H.-R. Buser, B. Hitzfeld, A. Weber, and the personnel of the WWTPs.

Ancillary