The aims of this work are to evaluate the presence of antibiotics in surface waters in a French water basin, where the presence of livestock is relatively important, and to understand the behaviour of antibiotics in drinking water treatment plants (DWTPs). Two sampling sites were chosen because of their livestock density and the presence of DWTPs in areas where urban activities are different. A large range of veterinary and human antibiotics were analysed in raw and treated water from the French Seine-Normandy Basin, based on the development of two analytical methodologies using solid-phase extraction and high-performance liquid chromatography-mass spectrometry. Clorsulon (an anthelmintic), fluoroquinolones, macrolides, sulfonamides (such as sulfamethoxazole, sulfadiazine), tetracyclines and trimethoprim were detected in raw surface water. Regarding the efficiency of drinking water treatment, an ozone/granular activated carbon combination proved to be effective in removing most antibiotics except danofloxacin and enrofloxacin which have an ionisable character and insufficient ozonation kinetic constant. Chlorination proved to be ineffective in removing antibiotics passing through the previous stages.

INTRODUCTION

The presence of pharmaceuticals in water resources and especially in drinking water is generating growing concerns not only from consumers, but also from water producers. Indeed, due to their usages, administration route and metabolism, Boxall et al. (2003) demonstrated that most antibiotics and anti-microbial agents, according to their physical properties (adsorption, hydrophobicity, biodegradability, photolysis sensitivity, etc.), have a high probability of reaching the environment. Once absorbed (incorporated into food or by injection), they are digested by the organism, in varying degrees (from a few per cent to 100% for those administered orally) and then excreted in the form of a metabolite and a parent compound. Excreta can be discharged directly into the environment (outdoor breeding, pisciculture), in wastewater treatment plants (pets) or stored before spreading (in the case of intensive breeding, spreading of manure) (Boxall et al. 2004). Veterinary drugs can persist in the soil for a few days to several years. Their half-life is influenced by several factors: temperature, pH and presence of manure. Boxall et al. (2003) reported that tetracyclines and quinolones are the most persistent in the environment. Acute effects on human health by these compounds from ingested drinking water are lower than other exposure pathways (Sanders 2010). However, one of the main health concerns related to the massive use of antibiotics in veterinary medicine is that micro-organisms become resistant to these compounds (Pallares et al. 2003; Gao et al. 2012). The existence of transfer phenomena of the resistance between human, animal and environmental micro-organisms has now been well established (Granier 2010). International authorities, among which the WHO and the World Organisation for Animal Health (formerly the Office International des Epizooties – OIE), recommend a more cautious use of antibiotics. In this context, it is clear that the detection of veterinary antibiotics in drinking water would strongly raise concerns by consumers, especially as veterinary antibiotic production represents more than 5,000 tonnes a year in the European Union (2004 value; Kools et al. 2008). This production rate is twice as high as that of human pharmaceuticals. In this context, the objective of the work here is to evaluate the presence of veterinary antibiotics in some French water resources and examine their behaviour during different treatment processes in two different drinking water treatment plants (DWTPs).

MATERIAL AND METHODS

Samples were taken from two DWTPs during two seasons in 2011–2012: the first campaign was performed in December and the second in April. The first plant located on the River Seine, called ‘Seine upstream 2’ on the map of sampling sites (Figure S1, available in the online version of this paper), comprises several treatment lines: the sampled line includes a settling unit followed by filtration using a first stage granular activated carbon (GAC) process, ozonation and second stage GAC filtration and UV disinfection. After mixing with water from other lines (same treatment and flow rate), water is chlorinated before being sent to the distribution system. The second plant located on the River Louette (see location in the Supplementary Information, available in the online version of this paper) includes a flotation unit followed by GAC filtration and water is chlorinated after being mixed with borehole water. The Seine River was selected as it is the water supply for the majority of the 12 million people living in the Paris area and its vast basin covers 30% of the French population. It is influenced by both agricultural and animal breeding activities. Livestock in the Seine-Normandy basin mainly consists of cattle with more than 2.1 million livestock units (LU) followed by pigs with around 900 thousand LU. Some figures on the number of cattle in French Seine-Normandy Basin are given in the Supplementary Information (Table S1, available in the online version of this paper). In contrast, the Louette River is a very small river located in an area practically devoid of animal breeding (less than 1,000 LU). In addition, other samples were collected along the Seine River (with and without the presence of DWTPs) as well as in a number of smaller rivers in Normandy. Two groundwater samples were taken along the Normandy coast, as well as groundwater samples at a treatment site located in the vicinity of Rouen, Normandy (results not shown). To summarise, 27 different types of surface water (including water entering the DWTP described in this study), five different types of groundwater and 10 types of treated water were sampled during both campaigns (see Figure S1 for location and Table S2 for the detailed description of treatment lines, Supplementary Information, available in the online version of this paper).

Once received in the laboratory, prior to extraction, water samples were filtered via 0.7 μm glass fibre filters. To preserve all the targeted samples, these were kept at −20 °C if analyses were not performed directly after filtration (Capdeville 2011). In the rare cases where precipitation was observed after freezing, samples were re-filtered at 0.7 μm before extraction.

Two methodologies, including SPE followed by analysis using high-performance liquid chromatography coupled with tandem mass spectrometry (HPLC-MS/MS), were applied to the samples in order to quantify 31 compounds (four sulfonamides, one diaminopyrimidine, six quinolones, four tetracyclines, five macrolides, two beta-lactamases, two anthelmintics, two anticoccidials, four antiparasitics, one antifungal). One iodinated X-ray contrast agent and four pesticides were selected as indicators of urban wastewaters and agricultural activities. Owing to the range of physico-chemical properties of these compounds, the development of two analytical methodologies was necessary, especially for pH and extraction water volume.

Solid-phase extraction conditions are described in detail in Table 1: methodology 1 was applied for the extraction of 22 compounds and methodology 2 for the extraction of 14 analytes (Table 3).

Table 1

Conditions of extraction by solid phase extraction

  Method 1 Method 2 
Sample volume 500 mL 250 mL 
pH 4 (diluted HCl) 7 (diluted HCl and/or NH4OH) 
EDTA 500 mg/L 
Internal standards Yes Yes 
Type of cartridge HLB, 6 cc, 500 mg HLB, 6 cc, 500 mg 
Conditioning 10 mL of acetonitrile 6 mL of methanol 
 10 mL of water 6 mL of water 
Drop-off flow rate 3 mL/min 3 mL/min 
Rinsing 6 mL water/methanol (70/30, v/v) + 2% NH4OH 
Elution 7 mL of acetonitrile 6 mL water/methanol (10/90, v/v) + 2% NH4OH 
  Method 1 Method 2 
Sample volume 500 mL 250 mL 
pH 4 (diluted HCl) 7 (diluted HCl and/or NH4OH) 
EDTA 500 mg/L 
Internal standards Yes Yes 
Type of cartridge HLB, 6 cc, 500 mg HLB, 6 cc, 500 mg 
Conditioning 10 mL of acetonitrile 6 mL of methanol 
 10 mL of water 6 mL of water 
Drop-off flow rate 3 mL/min 3 mL/min 
Rinsing 6 mL water/methanol (70/30, v/v) + 2% NH4OH 
Elution 7 mL of acetonitrile 6 mL water/methanol (10/90, v/v) + 2% NH4OH 

After drying under a nitrogen stream, extracts were analysed using combined HPLC with a triple quadrupole Quantum mass spectrometer (ThermoElectrons). The extracts from the first SPE method were analysed in positive ionisation mode following the conditions of method 1 in Table 2 and those from the second SPE method were analysed in positive (method 2a) and negative (method 2b) modes.

Table 2

Conditions of HPLC-MS/MS analyses

HPLC conditions
 
 Method 1
 
Method 2a
 
Method 2b
 
Injected volume 10 μL 10 μL 10 μL 
Mobile phase flow rate 200 μL/min 200 μL/min 200 μL/min 
Type of column T3 column (2.1 × 150 mm, 3 μm, 100 Å, Waters®Xbridge C18 (2.1 × 150 mm, 3.5 μm, 130 Å, Waters®Xbridge C18 (2.1 × 150 mm, 3.5 μm, 130 Å, Waters®
Column temperature 30 °C 30 °C 30 °C 
Solvent A water + 0.1% formic acid water + 0.1% formic acid + 1.6 mmol/L ammonium acetate water + 0.1% formic acid + 1.6 mmol/L ammonium acetate 
Solvent B methanol methanol methanol 
Gradient Time (min) A % B % Time (min) A % B % Time (min) A % B % 
100 95 45 55 
100 95 45 55 
27 100 27 100 12 100 
30 100 37 100 18 100 
40 100 38 95 19 45 55 
   45 95 25 45 55 
MS/MS conditions 
Type of ionisation ESI ESI ESI 
Polarity positive positive negative 
Acquisition type SRM SRM SRM 
Spray voltage 3,500 V 3,000 V 2,700 V 
Sheath gas pressure 30 psi 30 psi 30 psi 
Aux. gas pressure 5 psi 0 psi 10 psi 
Capillary temperature 300 °C 300 °C 300 °C 
Source CID − 5 − 5 to −1 
Calibration internal internal internal 
HPLC conditions
 
 Method 1
 
Method 2a
 
Method 2b
 
Injected volume 10 μL 10 μL 10 μL 
Mobile phase flow rate 200 μL/min 200 μL/min 200 μL/min 
Type of column T3 column (2.1 × 150 mm, 3 μm, 100 Å, Waters®Xbridge C18 (2.1 × 150 mm, 3.5 μm, 130 Å, Waters®Xbridge C18 (2.1 × 150 mm, 3.5 μm, 130 Å, Waters®
Column temperature 30 °C 30 °C 30 °C 
Solvent A water + 0.1% formic acid water + 0.1% formic acid + 1.6 mmol/L ammonium acetate water + 0.1% formic acid + 1.6 mmol/L ammonium acetate 
Solvent B methanol methanol methanol 
Gradient Time (min) A % B % Time (min) A % B % Time (min) A % B % 
100 95 45 55 
100 95 45 55 
27 100 27 100 12 100 
30 100 37 100 18 100 
40 100 38 95 19 45 55 
   45 95 25 45 55 
MS/MS conditions 
Type of ionisation ESI ESI ESI 
Polarity positive positive negative 
Acquisition type SRM SRM SRM 
Spray voltage 3,500 V 3,000 V 2,700 V 
Sheath gas pressure 30 psi 30 psi 30 psi 
Aux. gas pressure 5 psi 0 psi 10 psi 
Capillary temperature 300 °C 300 °C 300 °C 
Source CID − 5 − 5 to −1 
Calibration internal internal internal 

Compound detection was performed in selected reaction monitoring (SRM) mode and for each analyte of interest, two ion transitions were chosen for the quantification of the compound and the confirmation of its identification (see Table S3 in Supplementary Information, available in the online version of this paper).

Several types of control measures were implemented throughout the procedures in order to guarantee the quality of the results. Firstly, internal marked standards were added to all samples as well as to the standard solutions in order to monitor losses during extraction and injection. The concentration levels obtained for the target compounds were corrected by the overall result of the procedure. Then, to take into account the possible sources of contamination coming from bottling, reagents and the laboratory, a blank extraction was performed per sample series. Finally, water from each site was spiked to 100 ng/L with each analyte and the results obtained were used to correct the results and quantification limits.

Recoveries for both methodologies vary from 25 to 100%, in the water spiked with all the compounds of interest. Recoveries for internal standards are in the range of 50–95%. Limits of quantification are between 2 and 50 ng/L for groundwater, 2 and 360 ng/L for surface water and treated water. All the details on the validation of analytical methodologies are given in the Supplementary Information (Table S3).

RESULTS AND DISCUSSION

Results from the first DWTP during the first sampling period

Table 3 gives the quantification values for antibiotics and pesticides along DWTP n°1 found during the first sampling campaign (December). Note that flumequine concentration levels are not available due to extraction problems on this set of samples. The influence of urban wastewater at this site can be considered as negligible. Indeed, Bruchet & Mandra (1994) estimated that when the Seine river flow is around 300 m3/s, urban wastewater represents 5% of the river flow downstream to Paris and its major urban and industrial wastewater plants. As the present site studied is located upstream of the Paris area, the contribution of wastewater to the river flow is <1%. Concentration levels for quinolones, sulfonamides and a few macrolides in raw water demonstrate the impact of veterinary sources on the River Seine. The highest diversity of antibiotics was found upstream of Paris. While the lower basin (Normandy) is dominated by cattle breeding, the upper basin is characterised by mixed farming with a greater diversity of animal species including poultry, goats, sheep, pigs and cattle. Even if the flow rate is high (150 m3/s) compared to those measured in other sampling points of this project (such as Normandy rivers), a variety of antibiotics are found in the River Seine, showing that dilution does not necessarily reduce concentration levels to under the limit of quantification. It is surprising to observe that concentration levels of sulfonamides in the River Seine are in the same order of magnitude as those reported by Shelver et al. (2010) in surface and groundwaters in the immediate vicinity of pig farms (sulfadimethoxine: 2–32 ng/L, sulfamethazine: 2–5 ng/L and sulfamethoxazole: 20–43 ng/L). Concentration levels of oxytetracycline (6 ng/L) are similar to those found in two rivers in Luxembourg (Pailler et al. 2009). However, concentration levels of fluoroquinolones in this study are higher than those reported by Tamtam et al. (2009) for surface waters impacted by livestock activities (concentration levels lower than 20 ng/L).

Table 3

Quantification of veterinary antibiotics and human tracers in DWTP n°1 during the first sampling campaign (ng/L)

    Raw water Settling + GAC filtration GAC + Ozone UV disinfection Treated water 
Sulfonamides Sulfadiazine1 <2 <2 <2 <2 
Sulfamethazine1 <2 <2 <2 
Sulfamethoxazole1 11 <2 <2 <2 <2 
Sulfadimethoxine1 <2 <2 <2 
Diaminopyrimidine Trimethoprim1 <2 <2 <2 <2 
Quinolones Ciprofloxacin1 96 <60 <60 <60 <60 
Enrofloxacin1 46 34 28 <25 <25 
Danofloxacin1 64 27 29 24 23 
Marbofloxacin2a <4 <4 <4 <4 <4 
Flumequine2a NA NA NA NA NA 
Oxolinic acid2a <4 <4 
Tetracyclines Tetracycline1 <10 <10 <10 <10 <10 
Oxytetracycline1 <4 <4 NA 
Doxycycline1 <20 <20 <20 <20 <20 
Chlortetracycline1 <20 <20 <20 <20 NA 
Macrolides Lincomycin1 <4 <4 <4 <4 NA 
Clindamycin1 <2 <2 
Tylosin1 <10 <10 <10 <10 <10 
Erythromycin1 14 <4 <4 
Roxithromycin1 14 <4 
Beta-lactamases Cefazolin2a <40 <40 <40 <40 <40 
Ceftiofur2a <8 <8 <8 <8 <8 
Anthelmintics Albendazole2a <4 <4 <4 <4 NA 
Oxfendazole2a <4 <4 <4 <4 <4 
Anticoccidials Amprolium2a <20 <20 <20 <20 <20 
Sulfaquinoxaline2a <4 <4 <4 <4 <4 
Antiparasitics Levamisole2a <4 <4 <4 <4 <4 
Metrifonate2a <4 <4 <4 <4 <4 
Dicyclanil2a <4 <4 <4 <4 <4 
Clorsulon2b 30 <20 <20 <20 <20 
Antifungal Parconazole2a <8 <8 <8 <8 <8 
Iodised chemical contrast agent Iopromide1 68 38 21 24 <20 
Pesticides Simazine1 <2 
Atrazine1 30 11 11 
Isoproturon1 10 <2 <2 <2 <2 
Diuron1 <2 <2 <2 
    Raw water Settling + GAC filtration GAC + Ozone UV disinfection Treated water 
Sulfonamides Sulfadiazine1 <2 <2 <2 <2 
Sulfamethazine1 <2 <2 <2 
Sulfamethoxazole1 11 <2 <2 <2 <2 
Sulfadimethoxine1 <2 <2 <2 
Diaminopyrimidine Trimethoprim1 <2 <2 <2 <2 
Quinolones Ciprofloxacin1 96 <60 <60 <60 <60 
Enrofloxacin1 46 34 28 <25 <25 
Danofloxacin1 64 27 29 24 23 
Marbofloxacin2a <4 <4 <4 <4 <4 
Flumequine2a NA NA NA NA NA 
Oxolinic acid2a <4 <4 
Tetracyclines Tetracycline1 <10 <10 <10 <10 <10 
Oxytetracycline1 <4 <4 NA 
Doxycycline1 <20 <20 <20 <20 <20 
Chlortetracycline1 <20 <20 <20 <20 NA 
Macrolides Lincomycin1 <4 <4 <4 <4 NA 
Clindamycin1 <2 <2 
Tylosin1 <10 <10 <10 <10 <10 
Erythromycin1 14 <4 <4 
Roxithromycin1 14 <4 
Beta-lactamases Cefazolin2a <40 <40 <40 <40 <40 
Ceftiofur2a <8 <8 <8 <8 <8 
Anthelmintics Albendazole2a <4 <4 <4 <4 NA 
Oxfendazole2a <4 <4 <4 <4 <4 
Anticoccidials Amprolium2a <20 <20 <20 <20 <20 
Sulfaquinoxaline2a <4 <4 <4 <4 <4 
Antiparasitics Levamisole2a <4 <4 <4 <4 <4 
Metrifonate2a <4 <4 <4 <4 <4 
Dicyclanil2a <4 <4 <4 <4 <4 
Clorsulon2b 30 <20 <20 <20 <20 
Antifungal Parconazole2a <8 <8 <8 <8 <8 
Iodised chemical contrast agent Iopromide1 68 38 21 24 <20 
Pesticides Simazine1 <2 
Atrazine1 30 11 11 
Isoproturon1 10 <2 <2 <2 <2 
Diuron1 <2 <2 <2 

NA = not available;1 2a 2b = numbers indicate the analytical method.

Contributions from agricultural activities are also confirmed by the presence of pesticides, such as triazines and phenylureas. But, when comparing their respective levels, medication now seems to dominate over pesticides; an observation already made during the Amperes Study on the River Seine downstream of Paris (Bruchet et al. 2011). Sulfamethoxazole and erythromycin, both from human and veterinary medicine, are also quantified and reflect the contribution of urban effluent, as confirmed by the iopromide concentration level (68 ng/L). This can be compared to findings in the Armistiq research project (Besnault et al. 2015), where intensive treatment processes were evaluated for the elimination of micropollutants in wastewater. Concentration levels of sulfamethoxazole and erythromycin in wastewater were between 35–365 ng/L and 14–219 ng/L, respectively. These results show that concentration levels measured in this study are lower than those observed at the entrance of WWTP, thereby confirming a dilution effect in the River Seine.

Regarding the effectiveness of DWTP processes, good removal has been observed for three fluoroquinolones, sulfonamides, trimethoprim, oxytetracycline, erythromycin, roxithromycin and clorsulon during clarification (coagulation/settling/filtration). The benefits of clarification for removing ionisable drugs have already been recorded in previous work (Tabe et al. 2009). However, an ozone/GAC combination does not provide further reduction. For example, when considering danofloxacin, whose Log Kow is equal to 0.4 at a pH of 7, it is normal that it is not easily adsorbed by carbon. Also, its ozonation constant by molecular ozone (around 103 mole−1 s−1) is insufficient for a good removal rate to be achieved under this plant's ozone treatment conditions (residual of 0.3 mg/L for 20 min).

UV disinfection (40 mJ/cm2) is also ineffective on danofloxacin, which is found at low levels in chlorinated, distributed water (23 ng/L). This dosage used for disinfection is 10 to 20 times lower than dosages required for the oxidation of micropollutants. Thus, the apparent decrease of several antibiotics during UV disinfection (enrofloxacin, sulfadimethoxine, etc.) is probably observed because these chemicals are practically at their limit of quantification before this stage. Therefore, it was not possible to draw conclusions for compounds whose concentrations are too close to their limit of quantification. Consequently, the slight differences observed (roxithromycin, oxolinic acid, atrazine and simazine) between UV disinfected water and discharged water, mixed with that of other treatment lines (same treatment processes) probably reflect analytical uncertainties rather than a difference in effectiveness between the treatment process lines. To conclude, the concentration levels of this complicated mix of antibiotics found in raw water are systematically below their limits of quantification, meaning that this sophisticated combination of processes has a high impact on these micropollutants, except for traces of danofloxacin. The specific behaviour of danofloxacin can be explained by its physico-chemical properties. This compound is ionised at pH neutral and will therefore not be adsorbed by activated carbon filtration. In addition, its reactivity to molecular ozone is known to be low (Dodd et al. 2006).

Results from the first DWTP during the second sampling period

Figure 1 shows the changes into the concentration levels of some compounds quantified during the second sampling campaign, carried out during a rainy and cold period. The flow rate was higher than that of the first sampling period (280 m3/s compared to 150 m3/s) but this does not affect the concentration levels found in raw water for most antibiotics (in contrast to the observed decrease in atrazine and iopromide). Therefore factors other than the flow rate (use of antibiotics, spreading of manure, rainfall events, etc.) govern the concentration levels. In particular, manure spreading is carried out more during spring and the important surface run-off during the sampling period may cause a possible higher level than expected of antibiotic concentrations in raw water.
Figure 1

Concentration levels of antibiotics, atrazine and iopromide along the first DWTP during the second sampling period (ng/L).

Figure 1

Concentration levels of antibiotics, atrazine and iopromide along the first DWTP during the second sampling period (ng/L).

Only six antibiotics were found in raw water: sulfamethoxazole, marbofloxacin, tylosin, roxithromycin and clorsulon, the concentration levels of which are in the same range as those found in the first sampling period (from 14 to 28 ng/L) (Table S4, Supplementary Information, available in the online version of this paper). The three quinolones quantified during the first campaign between 46 and 96 ng/L in raw water, were not present during this second sampling period. By contrast, the concentration of amprolium is particularly high (130 ng/L). It is used purely for veterinary purposes as an anticoccidial for poultry. The reason for this high level is not clear. Owing to its physico-chemical properties (alkaline molecule having two nitrogen heterocycles and a free amine function), it is not removed by the settling-first stage GAC filtration process but it is easily attacked by ozone and totally transformed during this stage. As in the previous campaign, clarification (settling/GAC filtration) has shown its effectiveness against the identified antibiotics. The remaining trace levels of four compounds are then eliminated during ozone treatment. Finally, no antibiotics were detected after the second stage GAC treatment, even if iopromide, a poorly adsorbable compound, were still present. This difference with the previous sampling campaign is probably due to a difference in the GAC filter's condition (higher level of saturation). More generally, regarding the entire range of compounds studied, only danofloxacin succeeds in crossing part of the very comprehensive treatment processes in the first plant studied, which combines two levels of GAC filtration, ozonation and UV disinfection.

Results from the second DWTP during the first sampling period

Concentrations of quantified veterinary antibiotics at the DWTP n°2 are presented in Table 4. Its situation is quite different from that of DWTP n°1 because of the low impact of urban or livestock activities. However, nine veterinary antibiotics were quantified in the River Louette, including danofloxacin and enrofloxacin, which are specifically employed in veterinary medicine. Indeed, there is a riding stable a few kilometres upstream to this DWTP. Therefore, due to the very low river flow rate (<0.5 m3/s), the impact of this activity can be clearly measured in the raw water. Regarding the effectiveness of water treatments, flotation hardly affects the level of antibiotics and it would seem that the GAC filter releases several compounds, such as trimethoprim, tetracyclines or sulfamethoxazole. Sloughing effects from GAC filters have been well established for decades (Newcombe 1994).

Table 4

Quantification of veterinary antibiotics and human tracers in DWTP n°2 during the first sampling campaign (ng/L)

    Raw water Floated water GAC filtered water  Treated water 
Sulfonamides Sulfamethazine <2 <2 
Sulfamethoxazole <2 10 <2 
Sulfadimethoxine <2 <2 NC 
Diaminopyrimidine Trimethoprim 11 14 
Fluoroquinolones Ciprofloxacin 75 <60 <60 <60 
Enrofloxacin 32 26 <25 <25 
Danofloxacin 48 30 27 23 
Tetracyclines Tetracycline <10 <10 17 61 
Oxytetracycline <4 NC 
Doxycycline 39 <20 <20 59 
Chlortetracycline <20 <20 45 <20 
Macrolides Clindamycin <2 <2 <2 
Erythromycin <4 <4 
Roxithromycin 13 <4 <4 
Antiparasitic Clorsulon <4 <4 
Pesticides Simazine 14 14 
Atrazine 79 82 35 35 
Isoproturon <2 <2 
Diuron <2 <2 
    Raw water Floated water GAC filtered water  Treated water 
Sulfonamides Sulfamethazine <2 <2 
Sulfamethoxazole <2 10 <2 
Sulfadimethoxine <2 <2 NC 
Diaminopyrimidine Trimethoprim 11 14 
Fluoroquinolones Ciprofloxacin 75 <60 <60 <60 
Enrofloxacin 32 26 <25 <25 
Danofloxacin 48 30 27 23 
Tetracyclines Tetracycline <10 <10 17 61 
Oxytetracycline <4 NC 
Doxycycline 39 <20 <20 59 
Chlortetracycline <20 <20 45 <20 
Macrolides Clindamycin <2 <2 <2 
Erythromycin <4 <4 
Roxithromycin 13 <4 <4 
Antiparasitic Clorsulon <4 <4 
Pesticides Simazine 14 14 
Atrazine 79 82 35 35 
Isoproturon <2 <2 
Diuron <2 <2 

Concentrations of doxycycline or tetracycline in treated water are significantly higher than those measured at the GAC outlet. This could be explained by contamination of the borehole water mixed with the treatment line water. Unfortunately, separate sampling of the plant water before it is mixed with water from two boreholes cannot be performed at the plant.

Finally, the total level of antibiotics measured in the drinking water produced amounts to 0.17 μg/L. Note that danofloxacin and atrazine again pass through this DWTP process. The presence of pesticides and some human antibiotics (such as roxithromycin) indicates the impact of agricultural and human activities in the water catchment area of this small river.

Global view of the results obtained from the various sites in the Seine-Normandy Basin

The results obtained from these two previous sites, as well as the other sites sampled in the Seine-Normandy Basin (results not shown in detail), are summarised in Table 5. For each antibiotic, the number of positive samples in the raw and treated waters and the concentration range (in brackets) are specified.

Table 5

Summary of results: number of positive samples and concentration range in brackets (ng/L)

  Surface water Groundwater Treated 
Clorsulon 16 (6–107) 2 (4) 
Roxithromycin 15 (10–313) 3 (11–41) 
Oxolinic acid 10 (4–20) 1 (11) 4 (5–13) 
Tylosin 11 (5–196) 1 (3) 
Erythromycin 11 (4–45) 2 (2–6) 
Sulfamethoxazole 9 (2–108) 1 (2) 
Marbofloxacin 6 (7–21) 2 (14–67) 4 (8–23) 
Danofloxacin 6 (48–500) 2 (23) 
Ciprofloxacin 4 (75–481) 1 (10) 
Trimethoprim 5 (2–90) 2 (3–14) 
Sulfadiazine 2 (7–8) 1 (18) 2 (2–30) 
Clindamycine 5 (3–5) 
Levamisol 3 (7–9) 1 (6) 
Enrofloxacin 3 (32–46) 
Amprolium 2 (129–132) 1 (179) 
Doxycycline 2 (20–39) 1 (59) 
Sulfamethazine 2 (4–10) 1 (16) 
Tetracycline 2 (16–37) 1 (61) 
Sulfadimethoxine 1 (6) 1 (39) 
Oxytetracycline 2 (8–53) 
Flumequine 2 (12–18) 
Chlortetracycline 1 (47) 
Sulfaquinoxaline 1 (6) 
Albendazole 1 (10) 
Lincomycin 
Oxfendazole 
Ceftiofur 
Parconazole 
Dicyclanil 
Metrifonate 
Cefazolin 
  Surface water Groundwater Treated 
Clorsulon 16 (6–107) 2 (4) 
Roxithromycin 15 (10–313) 3 (11–41) 
Oxolinic acid 10 (4–20) 1 (11) 4 (5–13) 
Tylosin 11 (5–196) 1 (3) 
Erythromycin 11 (4–45) 2 (2–6) 
Sulfamethoxazole 9 (2–108) 1 (2) 
Marbofloxacin 6 (7–21) 2 (14–67) 4 (8–23) 
Danofloxacin 6 (48–500) 2 (23) 
Ciprofloxacin 4 (75–481) 1 (10) 
Trimethoprim 5 (2–90) 2 (3–14) 
Sulfadiazine 2 (7–8) 1 (18) 2 (2–30) 
Clindamycine 5 (3–5) 
Levamisol 3 (7–9) 1 (6) 
Enrofloxacin 3 (32–46) 
Amprolium 2 (129–132) 1 (179) 
Doxycycline 2 (20–39) 1 (59) 
Sulfamethazine 2 (4–10) 1 (16) 
Tetracycline 2 (16–37) 1 (61) 
Sulfadimethoxine 1 (6) 1 (39) 
Oxytetracycline 2 (8–53) 
Flumequine 2 (12–18) 
Chlortetracycline 1 (47) 
Sulfaquinoxaline 1 (6) 
Albendazole 1 (10) 
Lincomycin 
Oxfendazole 
Ceftiofur 
Parconazole 
Dicyclanil 
Metrifonate 
Cefazolin 

In total, 27 surface water samples, five groundwater samples and 10 treated water samples were analysed. For surface waters, clorsulon was detected with an occurrence rate of 16/27 = 59%. The presence of this compound at a high occurrence rate suggests the potential presence, in lower concentrations, of ivermectin, an ecotoxic compound at ng/L levels, administered simultaneously (ANMV 2011).

In general, 20 molecules were quantified at least twice in surface water and have an occurrence rate of higher than 5%. It is therefore worth retaining these for follow-up or future study. Nearly half of the substances examined were detected in treated water, with oxolinic acid in first place (40% occurrence rate). Logically, the highest concentration levels in antibiotics (fluoroquinolones during the first campaign) were found in small rivers in zones where livestock density is highest (results not shown). The majority of compounds and the most widespread, appear to be clorsulon (an antihelmintic from the benzene-sulfonamide family), fluoroquinolones (oxolinic acid, danofloxacin, ciprofloxacin and enrofloxacin), macrolides (erythromycin, roxithromycin, tylosin, clindamycin), certain sulfonamides (in particular sulfamethoxazole and sulfadiazine) and trimethoprim (various). Some antibiotics can be detected in high, but isolated, concentration levels (amprolium). Tetracyclines were found at relatively low frequencies at concentration levels of between 8 and 53 ng/L at some sites. According to an ANSES report, this group of antibiotics is, by far, the one presenting the highest tonnage (150 tonnes against 4 tonnes for fluoroquinolones as example) (ANSES 2010). Veterinary antibiotics were detected in groundwater located in a zone with a low livestock density. This was the case in both measurement campaigns.

An evaluation of risks related to antibiotics measured in quantities higher than the LOQ in treated water was carried out according to recommendations by the French ANSES (Agence nationale de sécurité sanitaire de l'alimentation, de l'environnement et du travail) (ANSES 2013). This covered seven antibiotics, all used for veterinary purposes. Two of these are also prescribed for humans. A comparison between the guide value derived from the reference toxicological value and the maximum concentration level measured in treated water shows a high safety margin, always significantly higher than 1 (between 66 and 18,750). For each antibiotic, the health risk for persons drinking treated water containing residues can be considered as insignificant or acceptable.

CONCLUSIONS

The quantification of a large range of veterinary and human antibiotics in surface water and along two DWTPs demonstrates the omnipresence of these compounds in the environment, coming from animal breeding. Sulfonamides, quinolones, macrolides and the antiparasitic clorsulon are present in both raw water samples studied at concentration levels below 100 ng/L. Variations in flow rate between two seasons in a large river such as the River Seine show that the impact of dilution and/or season is low on antibiotic concentration levels but other factors such as manure spreading, use of antibiotics and rainfall events may also explain the difference in concentration levels. However, in a small river such as the River Louette (flow rate lower than 0.5 m3/s), the impact of a single horse riding stable is measurable with, for example, significant concentration levels of quinolones of between 50 and 75 ng/L.

Regarding the effectiveness of drinking water treatment processes, contradictory results were observed for clarification, a beneficial effect observed for a settling–GAC filtration combination and no effect with a flotation–GAC filtration combination. An ozone/GAC combination proved to be ineffective in removing certain antibiotics, which is logical if the properties of these molecules are considered (ionisable property for many of them, insufficient ozonation kinetic constants). In particular, danofloxacin and marbofloxacin appear to be very resistant to different treatment combinations. However ozone by itself proved to be extremely effective in destroying amprolium, a compound which has a free amine function. Chlorination proved to be ineffective in removing antibiotics having crossed through the previous stages.

The highest concentration of antibiotics detected in drinking water is 0.24 μg/L. Therefore no acute effects on human health are expected. Conversely, the frequent presence in the cold season of antibiotics in natural water, in particular fluoroquinolones, could constitute an environmental risk and, indirectly, a human one.

REFERENCES

REFERENCES
ANMV
2011
Characteristic Product Summary of Virbamec D for its veterinary use (RCP). http://www.ircp.anmv.anses.fr/ (accessed 19 April 2015)
.
ANSES
2010
FARM 2007–2008, Programme français de surveillance de l'antibiorésistance des bactéries d'origine animale (FARM 2007–2008, French program for antibiotic resistance monitoring of animals bacteria)
.
Report. Bialec, Nancy, France
.
ANSES
2013
Évaluation des risques sanitaires liés à la présence de résidus de médicaments dans les eaux destinées à la consommation humaine: méthode générale et application à la carbamazépine et la danofloxacine (Evaluation of health risks linked to the presence of medicines residues in drinking water: general method and application to carbamazepine and danofloxacin). Recommendation of ANSES, Seisin n°2009-SA-0210
.
Boxall
A. B. A.
Fogg
L. A.
Blackwell
P. A.
Kay
P.
Pemberton
E. J.
Croxford
A.
2003
Prioritisation of veterinary medicines in the UK environment
.
Toxicology Letters
142
(
3
),
207
218
.
Boxall
A. B. A.
Fogg
L. A.
Kay
P.
Blackwell
P. A.
Pemberton
E. J.
Croxford
A.
2004
Veterinary medicines in the environment
.
Reviews of Environmental Contamination and Toxicology
180
,
1
91
.
Bruchet
A.
Mandra
V.
1994
Influence of municipal wastewaters on the composition of background organics in the Seine River. Consequences for the production of drinking water
.
Presented at AWWA Annual Conference Sunday Seminar
.
New York City
,
USA
.
Bruchet
A.
Robert
S.
Esperanza
M.
Janex Habibi
M. L.
Miège
C.
Coquery
M.
Budzinski
H.
Le Ménach
K.
2011
Natural attenuation of priority and emerging contaminants during river bank filtration and artificial recharge
.
European Journal of Water Quality
42
(
2
),
123
133
.
Capdeville
M. J.
2011
Etude des cycles biogéochimiques des contaminants organiques dits ‘émergents’ dans les systèmes aquatiques (Study of the biogeochemical cycles of organic contaminants called « emerging » in aquatic systems)
.
Thesis
.
University of Bordeaux
,
France
.
Granier
S.
2010
Epidémiologie comparée des béta-lactamases (Comparative epidemiology of beta-lactamases). Communication in the meeting, Microbial Resistance in Animal Health, ANSES, 18 novembre 2010
.
Kools
S. A. E.
Moltmann
J. F.
Knacker
A.
2008
Estimating the use of veterinary medicines in the European Union
.
Regulatory Toxicology and Pharmacology
50
(
1
),
59
65
.
Sanders
P.
2010
Panorama des voies de transmission des bactéries et gènes de résistance aux antibiotiques: place de l'alimentation (Panorama transmission channels of bacteria and resistance genes to antibiotics instead of food). Communication in the meeting, Microbial Resistance in Animal Health, ANSES, 18 novembre 2010
.
Tabe
S.
Jamal
T.
Seth
R.
Yue
C.
Yang
P.
Zhao
X.
Schweitzer
L.
2009
PPCPs and EDCs – occurrence in the Detroit River and their removal by ozonation, Report Project #3071
.
Water Research Foundation
,
Denver, CO
,
USA
.
Tamtam
F.
Le Bot
B.
Eurin
J.
Mercier
F.
Desportes
A.
Chevreuil
M.
2009
Surface water contamination by antibiotic residues in rural area
.
European Journal of Water Quality
40
(
2
),
175
186
.