Improved fully automated method for the determination of medium to highly polar pesticides in surface and groundwater and application in two distinct agriculture-impacted areas
Maria Vittoria Barbieri, Luis Simon Monllor-Alcaraz, Cristina Postigo, Miren Lopez de Alda
11 IMPROVED FULLY AUTOMATED METHOD FOR THE DETERMINATION OF MEDIUM TO HIGHLY POLAR PESTICIDES IN SURFACE AND GROUNDWATER AND APPLICATION IN TWO DISTINCT AGRICULTURE-IMPACTED AREAS.
Maria Vittoria Barbieri a , Luis Simón Monllor-Alcaraz a , Cristina Postigo a* , Miren López de Alda a* a Water, Environmental and Food Chemistry Unit (ENFOCHEM), Department of Environmental Chemistry, Institute of Environmental Assessment and Water Research (IDAEA-CSIC), Barcelona, Spain *Corresponding authors:
Cristina Postigo ( [email protected] Miren López de Alda ( [email protected] Institute of Environmental Assessment and Water Research (IDAEA-CSIC) Department of Environmental Chemistry C/ Jordi Girona 18-26, 08034 Barcelona, Spain. Tel: +34-934-006-100, Fax: +34-932-045-904
ABSTRACT
Water is an essential resource for all living organisms. The continuous and increasing use of pesticides in agricultural and urban activities results in the pollution of water resources and represents an environmental risk. To control and reduce pesticide pollution, reliable multi-residue methods for the detection of these compounds in water are needed. In this context, the present work aimed at providing an analytical method for the simultaneous determination of trace levels of 51 target pesticides in water and applying it to the investigation of target pesticides in two agriculture-impacted areas of interest. The method developed, based on an isotopic dilution approach and on-line solid-phase extraction-liquid chromatography-tandem mass spectrometry, is fast, simple, and to a large extent automated, and allows the analysis of most of the target compounds in compliance with European regulations. Further application of the method to the analysis of selected water samples collected at the lowest stretches of the two largest river basins of Catalonia (NE Spain), Llobregat and Ter, revealed the presence of a wide suite of pesticides, and some of them at concentrations above the water quality standards (irgarol and dichlorvos) or the acceptable method detection limits (methiocarb, imidacloprid, and thiacloprid), in the Llobregat, and much cleaner waters in the Ter River basin. Risk assessment of the pesticide concentrations measured in the Llobregat indicated high risk due to the presence of irgarol, dichlorvos, methiocarb, azinphos ethyl, imidacloprid, and diflufenican (hazard quotient (HQ) values>10), and an only moderate potential risk in the Ter River associated to the occurrence of bentazone and irgarol (HQ>1).
Keywords: polar pesticides, transformation products, occurrence, on-line solid-phase extraction, liquid chromatography-tandem mass spectrometry, risk assessment Introduction
The extensive, and sometimes excessive, use of pesticides has led to significant undesired effects on the environmental and human health (Han et al., 2018; Islam et al., 2018; Mostafalou and Abdollahi, 2017). As a result, developed countries have withdrawn from the market the most toxic and persistent pesticides, such as the organochlorine ones, and have promoted the use of comparatively more polar substances, expected to degrade more rapidly and to be less toxic to non-target organisms (Sander et al., 2017). However, their use in large amounts has still resulted in their accumulation in the environment. Nowadays, over 500 active ingredients are available in the European market, with an estimated total turnover of around 400,000 tons (Eurostat, 2020). Spain is ranked as the first country in Europe with the largest pesticide consumption (according to pesticide sales (Eurostat, 2020), 72 M kg on average in the period 2011-2018). Water plays an important role in the environmental fate of pesticides, because it transports these substances from agricultural to other areas, by flushing them away into the rivers via rainfall or irrigation runoff, or leaching them through the soil into groundwater bodies (Beitz et al., 2012). Pesticides applied in urban areas may also reach the aquatic environment after incomplete removal in wastewater treatment plants (WWTPs) (Köck-Schulmeyer et al., 2013; Rousis et al., 2017). Once applied or released in the environment, these compounds can be transformed by different natural processes (photodegradation, hydrolysis, oxidation, biotransformation). These processes hardly mineralize the parent compounds and, consequently, a wide spectrum of transformation products (TPs) are formed, some of them being even more toxic than the corresponding parent compound (Andreu and Picó, 2004; Richardson and Ternes, 2014). To reduce contamination by pesticides and TPs in the environment and minimize their impact on aquatic organisms and human health, the European Commission has established guidelines that influence the selection and application of pesticides, as well as maximum allowable concentrations (MAC) in both surface water and groundwater. The Groundwater Directive (EC, 2006a) sets maximum limits of 0.1 µg/L for individual pesticides and TPs and 0.5 µg/L for total pesticides in groundwater to preserve its quality, in line with the limits set in the Drinking Water Directive for waters intended for human consumption (EC, 1998). In surface water, the Directive 2013/39/EU (EC, 2013) establishes MACs for up to 45 priority substances, including 24 pesticides or biocides, in inland and other surface waters as well as in biota. Furthermore, five neonicotinoid pesticides, the carbamate methiocarb, and the semicarbamazone metaflumizone are currently included in the Watch List of substances for Union-wide monitoring in the field of water policy (European Decision 2018/840 (EC, 2018)). To meet European requirements, it is necessary to use analytical techniques that allow the monitoring of pesticide residues in surface and groundwater with high selectivity and sensitivity, such as liquid chromatography (LC) coupled to mass spectrometry (MS or MS ) (Hernández et al., 2005; Picó and Barceló, 2015). However, to measure the low concentrations at which some pesticides are present and toxic in water, a sample enrichment step is still required. Solid phase extraction (SPE) is nowadays considered the method of choice for this purpose (Pérez-Fernández et al., 2017). SPE can be performed in off-line mode, but the fully automated on-line approach has become increasingly attractive as it requires minimum intervention of the operator, which results in reduced sample processing time and improved reproducibility and accuracy of the results. Moreover, it grants high sample throughput and high sensitivity using low sample amounts, which becomes very relevant in case of required sample shipment and/or small storage space (Rossi and Zhang, 2000; Singer et al., 2010). In spite of these advantages, only few methodologies published for the analysis of polar pesticides in water are fully automated (Camilleri et al., 2015; Hurtado-Sánchez et al., 2013; Mann et al., 2016; Quintana et al., 2019; Rubirola et al., 2017; Singer et al., 2010). In this context, and with the final aim of advancing knowledge on the analysis and monitoring of regulated and non-regulated medium to highly polar pesticides in water bodies, the present work focused on developing and validating a fast and simple analytical methodology based on on-line SPE-LC-MS/MS to determine 51 pesticides in environmental waters at pg/L or ng/L levels in a single run. The selected pesticides belong to different chemical classes and the list includes pesticides of major concern included in the priority substances list or the Watch List, some of their transformation products, and pesticides commonly applied in Spain, and particularly, in Catalonia. To the best of our knowledge, this work provides for the first time validation figures for the analysis of 10 pesticides and TPs (i.e., azinphos ethyl, azinphos-methyl oxon, dichlorvos, diflufenican, fenthion oxon, fenthion oxon sulfone, fenthion oxon sulfoxide, fenthion sulfone, fenthion sulfoxide, and oxadiazon) in water samples with the use of an on-line SPE-LC-MS/MS approach. As a part of the validation process and to fulfil the overall aim of our work, the developed method was applied to the analysis of surface water and groundwater samples collected in two agriculture-impacted areas of Catalonia (NE Spain), to evaluate the occurrence and fate of the target pesticides and compliance of these water bodies with the current EQS. The results obtained were also used to assess the potential environmental risk that the pesticides found may pose for aquatic organisms in these areas. Materials and methods
Standards and solvents
High purity (>96%) standards of the 51 pesticides selected as target analytes and stable isotope-labeled (SIL) analogs for 45 of them were purchased from Fluka (Honeywell Specialty Chemicals Seelze GmbH, Germany), Toronto Research Chemicals (North York, ON, Canada), Cambridge Isotope Laboratories (Tewksbury, MA, USA), Sigma Aldrich (Merck KGaA, Darmstadt, Germany) or Dr. Ehrenstorfer (LGC Standards, Teddington, UK). The list of the target analytes and their main physical-chemical properties is provided in Table 1. Stock standard solutions of the individual analytes and SIL standards were prepared in methanol (MeOH), except in the case of simazine and its SIL analog that was prepared in dimethyl sulfoxide. All stock individual solutions (1000 µg/mL) were stored in amber glass bottles in the dark at -20 °C. Working standard solutions containing all analytes were prepared by appropriate dilution of the stock individual standard solutions at different concentrations (0.5 to 2000 ng/mL) in MeOH. A MeOH-based solution containing the mixture of SIL standards at a concentration of 1000 ng/mL was also prepared. These mixtures were used to prepare the aqueous standard solutions that defined calibration curves and in the validation studies. Pesticides-grade solvents MeOH, acetonitrile (ACN), and LC-grade water were supplied by Merck (Darmstadt, Germany).
Table 1.
Target pesticides, main physical-chemical properties, and current legislative status. 1
Analyte
Chemical class
Formula ǂ MM (g mol -1 ) ǂ Solubility (mg L -1 ) ǂ K oc (mL g -1 ) ǂ K ow logP ǂ GUS ǂ DT50 ǂ (days) Legislative status ǂ Currently used in Spain ǂ EQS Ω (µg/L) Method LODs ԑ (ng/L) 2,4-D Alkylchlorophenoxy C H Cl O ✔ ✔ Acetamiprid ԑ Neonicotinoid C H ClN ✔ ✔ Alachlor × Chloroacetamide C H ClNO Atrazine × Triazine C H ClN Azinphos-ethyl
Organophosphate C H N O PS Azinphos-methyl
Organophosphate C H N O PS Azinphos-methyl-oxon
Metabolite C H N O PS 301.26 * 0.77* - - -
Bentazone
Benzothiazinone C H N O S 240.30 7112 55 -0.46 1.95 80 ✔ ✔ Bromoxynil
Hydroxybenzonitrile Br C H (OH)CN 276.90 38000 302 0.27 1.71 13 ✔ ✔ Chlorfenvinphos × Organophosphate C H Cl O P 359.60 145 680 3.80 1.72 7 X 0.3
Chlorpyrifos × Organophosphate C H Cl NO PS 350.58 1.05 5509 4.70 0.58 5 ✔ ✔ Chlortoluron
Phenylurea C H ClN O 212.68
74 196 2.50 ✔ ✔ Cyanazine
Triazine C H ClN Clothianidin ԑ Neonicotinoid C H ClN O S 249.68 340 123 0.90 3.74 40.3 X ✔ Deisopropylatrazine
Metabolite C H ClN Desethylatrazine
Metabolite C H ClN Diazinon
Organophosphate C H N O PS 304.35 60 609 3.69 1.51 4.3 X
Dichlorvos × Organophosphate C H Cl O P 220.98 18000 50 1.90 0.69 - X 7 x 10 -4 Diflufenican
Carboxamide C H F N O ✔ ✔ Dimethoate
Organophosphate C H NO PS ✔ ∞ Diuron × Phenylurea C H Cl N O 233.09 35.6 680 2.87 2.65 8.8 ✔ ✔ Fenitrothion
Organophosphate C H NO PS 277.23 19 2000 3.32 0.48 1.1 X ✔ Fenitrothion oxon
Metabolite C H NO P 261.17* 301 * 21* 1.69* - - -
Fenthion
Organophosphate C H O PS Fenthion oxon
Metabolite C H O PS 262.26* * 2.31* - - -
Fenthion oxon sulfone
Metabolite C H O PS 294.03*
Fenthion oxon sulfoxide
Metabolite C H O PS 278.26* 1222* 11* 0.15* - - -
Fenthion sulfone
Metabolite C H O PS - - - Fenthion sulfoxide
Metabolite C H O PS
183 1.92* - - -
Fluroxypyr
Pyridine compound C H Cl FN O ✔ ✔ Imidacloprid ԑ Neonicotinoid C H ClN O ✔ ✔ Irgarol × Triazine C H N S 253.37 7 1569 3.95 - - X 0.016
Isoproturon × Phenyluera C H N O 206.28 70.2 251* 2.5 2.61 40 X ✔ Linuron
Phenyluera C H Cl N O ✔ Malaoxon
Metabolite C H O PS 314.29* 7500* - - -
Malathion
Organophosphate C H O PS ✔ ✔ MCPA
Organophosphate C H ClO -0.81 2.98 13.5 ✔ ✔ Mecoprop
Aryloxyalkanoic acid C H ClO ✔ Methiocarb ԑ Carbamate C H NO S 225.31 27 182* 3.18 1.82 1.6 ✔ ✔ Metolachlor
Chloroacetamide C H ClNO Molinate
Thiocarbamate C H NOS 187.30 1100 190 2.86 1.89 4 X ✔ Pendimethalin
Dinitroaniline C H N O ✔ ✔ Propanil
Anilide C H Cl NO 218.08 95 149 2.29 -0.51 1.2 X ✔ Quinoxyfen × Quinoline C H Cl FNO 308.13 0.05 23° 4.66 -0.8 5 ✔ ✔ Simazine × Triazine C H ClN ✔ Terbuthylazine
Triazine C H ClN ✔ ✔ Terbutryn × Triazine C H N S 241.36 25 2432 3.66 2.21 27 X 0.34
Thiacloprid ԑ Neonicotinoid C H ClN S 252.72 184 615° 1.26 1.1 1000 ✔ Thiamethoxam ԑ Neonicotinoid C H ClN O S 291.71 4100 56 -0.13 3.58 30.6 X ✔ Thifensulfuron methyl
Sulfonylurea C H N O S ✔ ✔ Triallate
Thiocarbamate C H Cl NOS 304.7 4.1 3034 4.06 0.61 104 ✔ ✔ × Compound included in the list of priority substances. EC Directive 2013/39/EU of the European Parliament and of the Council of 12 August 2013 amending Directives ԑ Compound included in the European Watch List and corresponding maximum acceptable method detection limit (ng/L). EC Commission Implementing Decision (EU) Parliament and of the Council and repealing Commission Implementing Decision (EU) 2015/495 (notified under document C(2018) 3362). Retrieved from: https://goo.gl/nR4ezg. ǂ The PPDB, Pesticide Properties Database. http://sitem.herts.ac.uk/aeru/footprint/index2.htm. - Lewis, K.A., Tzilivakis, J., Warner, D. and Green, A. (2016). An international database for pesticide risk assessments and management. Human and Ecological Risk Assessment: An International Journal, 22(4), 1050-1064. * Data estimated using the US Environmental Protection Agency EPISuite TM ° ^ Calculated using the mathematical formula: GUS = log10 (half-life) x [4 - log10 (Koc)]. Ω Environmental Quality Standards (EQS) for priority substances in surface waters. ∞ Commission implementing regulation (EU) 2019/1090 of 26 June 2019 concerning the non-renewal of approval of the active substance dimethoate. Member States shall withdraw authorizations for plant protection products containing dimethoate as active substance by 17 January 2020 at the latest.
16 MM: molecular mass; Solubility: solubility in water at 20 o C; K oc : organic carbon partition coefficient; K ow : octanol-water partition coefficient; GUS: 17 leaching potential index; DT50: biodegradability, water phase only, expressed as half-life in days; Legislative status: ✔ approved, X not approved. 18 19 20 21 22 23 0 24 25 On-line solid-phase extraction
26 On-line SPE of the water samples was performed with a commercial Prospekt-2 27 system (Spark Holland, Emmen, The Netherlands) connected in series with the LC-MS/MS 28 instrument. Before automated on-line SPE and analysis, the water sample was fortified at a 29 concentration of 200 ng/L with the mixture of SIL compounds that will be used as surrogate 30 standards and centrifuged at g-force of 2500 xg (3500 rpm) and room temperature for 10 31 min to remove suspended particles (centrifuge 5810 R, Eppendorf Ibérica, Spain). Then, 5 mL 32 of the sample, calibration solution and/or blank was delivered using a 2 mL high-pressure 33 syringe onto a previously conditioned CHROspe cartridge Polymer DVB (divinylbenzene 34 polymer, 10 mm x 2 mm i.d., 25-35 µm particle size) (Axel Semrau GmbH & Co. KG, 35 Srockhövel, Germany) at a flow rate of 1 mL/min. Conditioning of the cartridge was also 36 performed via the high pressure dispenser (HPD) unit with 1 mL of ACN and 1 mL of LC-grade 37 water (5 mL/min). Upon sample loading, the cartridge was washed with 1 mL of LC-grade 38 water to complete sample transfer and remove highly polar components of the matrix and 39 the analytes were eluted with the LC mobile phase onto the LC analytical column. The system 40 configuration allows the preconcentration of the next sample in a batch while the LC-MS 41 analysis of the previously extracted sample takes place. The entire system was controlled 42 through SparkLink Version 3.10 (Spark Holland). 43 44
LC-MS/MS analysis
45 1 LC-MS/MS analyses were performed using a 1525 binary HPLC pump connected in 46 series with the Prospekt-2 system and a TQD triple-quadrupole mass spectrometer equipped 47 with an electrospray (ESI) interface (Waters, Milford, MA, USA). 48 LC separation was carried out with a Purospher ® STAR RP-18 end-capped column 49 (100 mm x 2 mm i.d., 5 µm particle size) preceded by a guard column (4 mm x 4 mm i.d., 5 50 µm) of the same packing material (Merck, Darmstadt, Germany), and a linear gradient of ACN 51 and water as mobile phase at a flow rate of 0.2 mL/min. The gradient started with an ACN 52 composition of 10% that was increased to 50% in 5 min, to 80% in the next 20 min, and 100% 53 in the following 6 min. Then, the chromatographic column was reequilibrated with the 54 mobile phase initial conditions in the following 9 min. In total, the analysis time, including the 55 sample extraction step, was 40 min. 56 MS/MS detection was performed in the selected reaction monitoring (SRM) mode, 57 recording one SRM transition per SIL compound and two SRM transitions per target analyte, 58 with the first one and more abundant being used for quantification and the second one for 59 confirmation. A total of 146 SRM transitions was acquired in six separate retention windows, 60 to maximize the sensitivity of the MS instrument (Figure 1). The ESI interface was operated in 61 both positive (PI) and negative (NI) ionization modes according to the preferential ionization 62 mode of the target analytes (43 were analyzed in PI and 8 in NI). Table 2 summarizes the 63 optimum SRM transitions and ionization conditions for each selected analyte. Other specific 64 optimized MS conditions were as follows: capillarity voltage, 3.5 kV; extractor voltage, 3 V; RF 65 lens voltage, 1.8 V; source temperature, 150 °C; desolvation temperature, 450 °C. Nitrogen 66 was used as cone gas (flow, 30 L/Hr) and desolvation gas (flow, 680 L/Hr); and argon was 67 used as collision gas (flow, 0.19 mL/min). MassLynx 4.1 software from Waters was used to 68 perform instrument control, data acquisition, and quantification. 69 2
Table 2.
On-line SPE-LC-MS/MS conditions for the analysis of the 51 investigated pesticides and SIL analogs. Analyte
Retention time (min)
SRMs (m/z) Precursor ion>product ion
Cone (V)
Collision Energy (eV)
SRM ratio (SRM1/SRM2)
Negative ESI mode
Bromoxynil 7.5 276>81/276>79 40 20/30 16.6
Bromoxynil C Bentazone 7.5 239>132/239>197 30 25/20 2.8
Bentazone d Fluroxypyr α MCPA 7.9 199>141/201>143 25 10/10 2.7
MCPA d Mecoprop 7.9 213>141/213>71 25 10/10 10.2
Mecoprop d Propanil 14.9 216>160/218>162 25 20/20 1.4
Propanil d Fenitrothion 19.8 262>152/262>122 25 20/30 9.6
Fenitrothion d Positive ESI mode
Thifensulfuron methyl 7.3 388>167/388>141 25 15/20 6.4
Thifensulfuron methyl d Desethylatrazine (DEA) 7.4 188>146/188>79 30 15/25 11.5
Desethylatrazine d β Deisopropilatrazine d Thiamethoxam
Thiamethoxam d Clothianidin d Imidacloprid d Acetamiprid d Dimethoate d γ Thiacloprid d Dichlorvos d Simazine d Cyanazine d δ Fenthion oxon sulfone d Fenthion sulfoxide d Fenitrothion oxon d Chlortoluron d Isoproturon d Atrazine d Diuron 12.8 233>72/235>72 25 15/15 1.5
Diuron d Fenthion oxon d Fenthion sulfone d Terbuthylazine d Methiocarb d Linuron d Azinphos methyl d ε Terbutryn d Irgarol d Metolachlor d Alachlor d Malathion d Chlorfenvinphos d Azinphos ethyl d Diazinon d Diflufenican d Oxadiazon d ζ Pendimethalin d Chlorpyrifos d Triallate C α Compound quantified using mecoprop d as surrogate standard. β Compound quantified using thiamethoxam d as surrogate standard. γ Compound quantified using fenthion sulfoxide d as surrogate standard. δ Compound quantified using chlortoluron d as surrogate standard. ε Compound quantified using linuron d as surrogate standard. ζ Compound quantified using chlorpyrifos d as surrogate standard.
77 78 79 4 Figure 1.
Total Ion Current (TIC) chromatograms obtained from the analysis of a fortified 82 groundwater sample (100 ng/L) showing the acquisition of the 146 SRM transitions set for 83 determination of the 51 target pesticides and their 45 SIL analogs in six different acquisition 84 windows along the analytical run. 85 86
Method performance
87 The analytical method was validated in terms of linearity, accuracy, precision, 88 sensitivity, and matrix effects, in both surface water and groundwater. The validation in 89 groundwater was carried out using a pooled sample of groundwater from various aquifers 90 located in Catalonia (NE Spain). In the case of surface water, a pooled sample of water from 91 three Catalonian rivers, namely, Segre, Llobregat, and Tordera, was used. 92 Eleven calibration solutions within the concentration range 0.5-2000 ng/L, 93 constructed after appropriate dilution of the working standard solutions in LC-grade water 94 5 were used to evaluate method linearity. Quantification was done using an isotope dilution 95 approach, i.e., considering the ratio between the peak area of each analyte and that of its 96 corresponding SIL analog, except in the case of six compounds, for which SIL analogs were 97 not available. These compounds were quantified using SIL standards presenting similar 98 structure, retention time, and/or recoveries. Method linearity was expressed with the 99 coefficient of determination (r ) of the weighted linear regression model obtained for each 100 analyte. 1/x was used as a weighting factor to reduce the influence of the high 101 concentration data points in the model. 102 Accuracy and precision of the method in LC-grade water, surface, and groundwater 103 were appraised with the analyte recovery and its repeatability after n=5 replicated analyses 104 of each matrix fortified at three different concentration levels (10 ng/L, 100 ng/L, and 1000 105 ng/L). Background concentration levels of each target pesticide in each matrix were taken 106 into account in the calculations. 107 The method sensitivity was evaluated through the calculation of limits of detection 108 (LOD), limits of quantification (LOQ), and limits of determination (LODet). LOD and LOQ were 109 experimentally estimated from the analysis of the water matrices fortified at the lowest level 110 (10 ng/L) as the analyte concentration giving a signal to noise ratio of 3 in the case of LOD 111 and 10 in the case of LOQ. LODet coincides with the minimum concentration of a compound 112 that can be quantified (LOQ of SRM1) and confirmed (LOD of SRM2). 113 To evaluate the matrix effects produced by co-extracted matrix components, analyte 114 peak areas obtained after on-line SPE-LC-MS/MS analysis of surface water and groundwater 115 fortified at 100 ng/L were compared with those obtained after on-line SPE-LC-MS/MS 116 analysis of LC-grade water fortified at equal concentration. Negative matrix effect values 117 occur when the analyte signal in LC-grade water is higher than in fortified surface or 118 groundwater, and indicate ionization suppression effects. On the contrary, positive matrix 119 6 effect values occur when the analyte signal is higher in surface and groundwater than in LC-120 grade water, and indicate signal enhancement effects. 121 Sampling locations and water collection
123 The presence of the target pesticides was investigated in the last stretches of the two 124 largest river basins of Catalonia (NE Spain), i.e., the Llobregat and the Ter River basins. 125 Surface water of these two basins is used to supply drinking water to about 4.5 million 126 people in Barcelona and its metropolitan area (Postigo et al., 2018). The Llobregat River is 127 located in an area with an important concentration of industries (e.g., tannery, food 128 products, textile, pulp, and paper industries), and high population density, and thus, with an 129 important demand of water. This river is highly impacted by domestic and industrial 130 wastewater discharges (> 30 wastewater treatment plants) and surface runoff from 131 agricultural areas (González et al., 2012). On the contrary, the Ter River basin is characterized 132 by a low population density and intense agricultural activities (e.g., crops of rice, corn, alfalfa, 133 and apple trees, among others). It also receives the impact of some metallurgic, pulp mill, 134 textile, and tannery industries (Céspedes et al., 2006). 135 The sampling campaigns were conducted in February 2017 in the Llobregat River and 136 in June 2018 in the Ter River. A total of 11 surface water samples were collected from the 137 Llobregat River, and the same number (6 surface water and 5 groundwater samples) from 138 the Ter River (Figure 2). Grab sampling was done in all surface water locations. Groundwater 139 samples were collected after pumping each well for few minutes (10-20 min) to remove 140 stagnant water, at the minimum flow rate possible, and steady conditions of physical-141 chemical parameters (i.e. temperature, pH, and conductivity). All samples were collected in 142 amber polyethylene terephthalate (PET) bottles and transported under cool conditions to the 143 laboratory, where they were stored upon arrival at -20 °C in the dark until analysis. 144 7 145 146
Figure 2 . Sampling locations in each of the river basins investigated. Figure courtesy of J. Montaner from IRTA (Ter) and V. Sola from CUADLL (Llobregat). 147 8
Risk assessment
The potential environmental risk associated with the pesticides found in the investigated samples was assessed using the hazard quotient (HQ) approach (EPA, 1997), following the equation: HQ=MEC/PNEC. This approach compares the measured environmental concentration (MEC) for each compound with its predicted no-effect concentration (PNEC), i.e., the concentration at which no toxic effects are expected to occur. To assess the worst-case scenario, the maximum pesticide concentration measured in the various investigated samples (MEC max Results and discussion
Method optimization
The analytical method developed is based on a methodology previously described for the analysis of 22 pesticides in environmental waters (Köck-Schulmeyer et al., 2014; Köck-Schulmeyer et al., 2013; Postigo et al., 2010). One of the analytical improvements incorporated in this methodology is the expansion of the list of the targeted pesticides with 29 additional medium to highly polar pesticides, including 8 TPs, i.e., acetamiprid, azinphos ethyl, azinphos-methyl, azinphos-methyl oxon, bromoxynil, chlorfenvinphos, chlorpyrifos, clothianidin, 9 dichlorvos, diflufenican, fenitrothion oxon, fenthion oxon, fenthion oxon sulfone, fenthion oxon sulfoxide, fenthion sulfone, fenthion sulfoxide, fluroxypyr, imidacloprid, irgarol, malaoxon, methiocarb, oxadiazon, pendimethalin, quinoxyfen, terbutryn, thiacloprid, thiamethoxam, thifensulfuron methyl, and triallate, and SIL analogs for 24 of them. These pesticides and TPs were selected considering their feasibility for LC-MS analysis, their current use in Spain, and their inclusion in the EU legislation (as a priority, Watch List or banned substances) (EC, 2013; EC, 2018) (Table 1). The current method, unlike the previous one, removes suspended particles by centrifugation instead of filtration. Moreover, this step is conducted after surrogate standard addition to account for pesticides present in the whole matrix and reduce the loss of analytes due to adsorption onto the filters. Compared to the previous method, the analysis time is reduced to half due to the determination of all targeted pesticides and TPs in a single analytical run. This was possible thanks to the use of a generic sorbent for the simultaneous preconcentration of all analytes, and the switch of polarity ionization during MS acquisition. The optimization of the MS/MS conditions for the detection of the new analytes included in the methodology was performed by on-column injection of individual standard solutions of each compound. Full scan acquisition was used to select the molecular ion and the best ionization mode and optimum declustering potential for its detection, and product ion scan acquisition allowed obtaining the optimum collision energies to register the two most abundant and selective fragment ions (SRM transitions) in each case. MS/MS conditions for each pesticide and SIL analog are provided in Table 2. Up to six time-acquisition windows were established to maximize the acquisition time for each SRM transition and hence improve method sensitivity. An example of the extracted ion chromatograms of the target compounds obtained after on-line SPE-LC-MS/MS analysis of a surface water sample fortified with the targeted pesticides at a concentration of 100 ng/L (500 ng/L for those compounds with LOD above 100 ng/L) is provided in Figure 3. 0
Figure 3.
Extracted ion chromatograms (XIC) of the target pesticides after on-line SPE-LC-MS/MS analysis of a surface water sample fortified at a concentration of 100 ng/L (or 500 ng/L in the case of those compounds marked with *). 1 1
Figure 3. (continued). 2 3 2
Method validation
4 Tables 3-5 and Figure 4 summarize the method performance in groundwater, surface 5 water, and LC-grade water, in terms of linearity, recovery, repeatability, sensitivity, and 6 matrix effects at the three concentration levels investigated (10 ng/L, 100 ng/L, and 1000 7 ng/L). 8 The linearity of the method expanded between 0.5 ng/L and 2000 ng/L for most of 9 the compounds (1000 ng/L was the upper linearity range in the case of clothianidin, 10 dichlorvos, diuron, fenitrothion oxon, fenthion sulfone, fenthion oxon sulfone, malathion, 11 mecoprop, molinate, simazine, and thifensulfuron methyl). The weighted linear regression 12 models presented a coefficient of determination (r ) higher than 0.99 for all compounds 13 except for fenitrothion (0.981) and fenthion sulfone (0.983) (Table 3). 14 15 3 Table 3.
Method performance in terms of linearity, recovery, repeatability (RSD, relative standard deviation), and sensitivity (limits of detection (LOD) and 16 limits of determination (LODet)) for the target pesticides in surface water and groundwater. Analyte Linearity (r ) Groundwater Surface water
Accuracy and precision (100 ng/L) Sensitivity Accuracy and precision (100 ng/L) Sensitivity Analyte recovery ± RSD(%) LOD ng/L LODet ng/L Analyte recovery ± RSD(%) LOD ng/L LODet ng/L 2,4-D
Acetamiprid
Alachlor
Atrazine
Azinphos ethyl
Azinphos methyl
Azinphos methyl oxon
Bentazone
Bromoxynil
Chlorfenvinphos
Chlorpyrifos
Chlortoluron
Clothianidin
Cyanazine
DEA
DIA
Diazinon
Dichlorvos
Diflufenican
Dimethoate
Diuron
Fenitrothion
Fenitrothion oxon
Fenthion oxon
Fenthion oxon sulfone Fenthion oxon sulfoxide
Fenthion sulfone
Fenthion sulfoxide
Fluroxypyr
Imidacloprid
Irgarol
Isoproturon
Linuron
Malaoxon
Malathion
MCPA
Mecoprop
Methiocarb
Metolachlor
Molinate
Oxadiazon
Pendimethalin
Propanil
Quinoxyfen
Simazine
Terbuthylazine
Terbutryn
Thiacloprid
Thiamethoxam
Thifensulfuron methyl
Triallate
18 19 20 5
Table 4.
Recovery and repeatability (RSD, relative standard deviation) obtained from the 21 replicate (n=5) analysis of groundwater and surface water fortified with the target analytes at 22 concentrations levels of 10 and 1000 ng/L . Analyte
Groundwater Surface water Analyte recovery ± RSD (%) Analyte recovery ± RSD (%) 10 ng/L 1000 ng/L 10 ng/L 1000 ng/L 2,4-D
91 ± 15 95 ± 9 126 ± 7 127 ± 4
Acetamiprid
80 ± 19 101 ± 14 93 ± 5 80 ± 5
Alachlor
82 ± 11 80 ± 16 125 ± 5 98 ± 6
Atrazine
118 ± 10 119 ± 4 89 ± 5 94 ± 8
Azinphos ethyl
112 ± 17 92 ± 13 118 ± 17 85 ± 8
Azinphos methyl
114 ± 20 112 ± 18 113 ± 9 100 ± 10
Azinphos methyl oxon
121 ± 3 90 ± 5 BLOD 83 ± 23
Bentazone
107 ± 15 86 ± 16 104 ± 5 81 ± 8
Bromoxynil
82 ± 6 82 ± 18 121 ± 4 90 ± 5
Chlorfenvinphos
113 ± 17 90 ± 4 120 ± 14 93 ± 6
Chlorpyrifos
116 ± 11 89 ± 4 105 ± 7 122 ± 3
Chlortoluron
123 ± 16 88 ± 14 114 ± 11 87 ± 18
Clothianidin
BLOD 99 ± 7 BLOD 90 ± 12
Cyanazine
127 ± 24 120 ± 20 126 ± 4 120 ± 13
DEA
81 ± 6 93 ± 4 90 ± 6 102 ± 8
DIA
BLOD 104 ± 8 125 ± 3 112 ± 6
Diazinon
106 ± 10 104 ± 9 125 ± 18 106 ± 4
Dichlorvos
BLOD 87 ± 3 123 ± 7 95 ± 5
Diflufenican
BLOD 84 ± 20 BLOD 80 ± 17
Dimethoate
BLOD 84 ± 5 BLOD 110 ± 9
Diuron
111 ± 16 121 ± 4 85 ± 17 86 ± 11
Fenitrothion
BLOD 98 ± 3 BLOD 81 ± 5
Fenitrothion oxon
82 ± 17 88 ± 5 118 ± 8 91 ± 3
Fenthion oxon
107 ± 4 124 ± 20 102 ± 8 115 ± 6
Fenthion oxon sulfone
83 ± 5 106 ± 8 BLOD 94 ± 17
Fenthion oxon sulfoxide
99 ± 12 110 ± 16 98 ± 7 89 ± 8
Fenthion sulfone
116 ± 4 120 ± 4 BLOD 83 ± 11
Fenthion sulfoxide
84 ± 3 104 ± 8 95 ± 5 107 ± 11
Fluroxypyr
BLOD 97 ± 6 BLOD 79 ± 20
Imidacloprid
84 ± 4 83 ± 4 112 ± 15 118 ± 5
Irgarol
110 ± 15 122 ± 16 116 ± 10 102 ± 11
Isoproturon
104 ± 5 83 ± 5 107 ± 7 119 ± 13
Linuron
92 ± 4 106 ± 12 109 ± 6 86 ± 20
Malaoxon
120 ± 11 124 ± 10 125 ± 12 121 ± 13
Malathion
BLOD 80 ± 20 BLOD 88 ± 5
MCPA
115 ± 19 86 ± 13 93 ± 20 113 ± 13
Mecoprop
99 ± 16 86 ± 18 105 ± 14 106 ± 15
Methiocarb
111 ± 15 116 ± 19 88 ± 11 104 ± 12
Metolachlor
122 ± 10 114 ± 13 122 ± 19 118 ± 4
Molinate
BLOD 81 ± 8 BLOD 88 ± 7 Oxadiazon
BLOD 109 ± 20 BLOD 104 ± 7
Pendimethalin
BLOD 99 ± 4 BLOD 121 ± 4
Propanil
120 ± 4 85 ± 17 103 ± 19 110 ± 9
Quinoxyfen
95 ± 3 88 ± 21 73 ± 5 106 ± 19
Simazine
123 ± 12 86 ± 20 96 ± 4 79 ± 19
Terbuthylazine
115 ± 19 111 ± 13 111 ± 7 83 ± 4
Terbutryn
99 ± 3 100 ± 3 117 ± 13 95 ± 8
Thiacloprid
112 ± 15 99 ± 6 122 ± 3 81 ± 18
Thiamethoxam
BLOD 80 ± 6 88 ± 12 98 ± 12
Thifensulfuron methyl
108 ± 14 84 ± 19 115 ± 3 91 ± 11
Triallate
75 ± 19 113 ± 19 106 ± 20 87 ± 14
BLOD: Below limit of detection 24 25 26
Table 5.
Recovery and repeatability (RSD, relative standard deviation) obtained from the 27 replicate (n=5) analysis of LC-grade water fortified with the target analytes at concentration 28 levels of 10, 100 and 1000 ng/L, and limits of detection (LOD) and determination (LODet) 29 achieved. 30
Analyte Analyte recovery ± RSD (%) Sensitivity 10 ng/L 100 ng/L 1000 ng/L LOD ng/L LODet ng/L 2,4-D
79 ± 14 81 ± 12 88 ± 5 6.1 20
Acetamiprid
106 ± 9 81 ± 19 95 ± 9 0.16 0.53
Alachlor
93 ± 15 105 ± 1 97 ± 3 1.2 3.8
Atrazine
102 ± 5 123 ± 14 92 ± 2 0.14 0.88
Azinphos ethyl
113 ± 14 85 ± 6 98 ± 12 0.42 1.4
Azinphos methyl
81 ± 6 92 ± 13 121 ± 13 0.38 1.3
Azinphos methyl oxon
126 ± 10 100 ± 7 108 ± 11 3.1 10
Bentazone
76 ± 20 88 ± 20 113 ± 10 4.3 14
Bromoxynil
111 ± 5 99 ± 5 121 ± 8 2.6 8.6
Chlorfenvinphos
112 ± 11 106 ± 4 112 ± 15 0.24 0.80
Chlorpyrifos
123 ± 18 120 ± 10 104 ± 18 0.44 1.5
Chlortoluron
125 ± 20 98 ± 5 107 ± 14 0.13 0.42
Clothianidin
113 ± 11 100 ± 4 80 ± 5 2.3 7.5
Cyanazine
115 ± 7 112 ± 5 124 ± 3 0.081 0.28
DEA
90 ± 6 100 ± 26 102 ± 8 2.3 7.9
DIA
105 ± 21 120 ± 5 116 ± 13 4.4 15
Diazinon
82 ± 3 103 ± 5 125 ± 6 0.042 0.16
Dichlorvos
94 ± 20 120 ± 15 113 ± 14 5.4 18
Diflufenican
121 ± 11 96 ± 13 100 ± 19 1.2 4.0
Dimethoate
120 ± 9 117 ± 17 84 ± 19 0.76 2.6
Diuron
109 ± 4 127 ± 12 124 ± 3 0.13 0.43
Fenitrothion
106 ± 12 123 ± 5 120 ± 17 2.6 8.8
Fenitrothion oxon
120 ± 4 85 ± 8 112 ± 6 0.79 2.6
Fenthion oxon
110 ± 12 119 ± 3 122 ± 7 0.17 0.59
Fenthion oxon sulfone
99 ± 13 125 ± 4 112 ± 20 2.8 9.4 Fenthion oxon sulfoxide
110 ± 4 98 ± 5 109 ± 6 0.13 0.43
Fenthion sulfone
85 ± 20 109 ± 15 120 ± 6 4.2 14
Fenthion sulfoxide
89 ± 5 93 ± 20 106 ± 16 0.41 1.4
Fluroxypyr
BLOD 103 ± 18 102 ± 14 29 95
Imidacloprid
124 ± 20 103 ± 12 81 ± 18 0.87 2.9
Irgarol
89 ± 7 121 ± 16 87 ± 21 0.85 2.8
Isoproturon
91 ± 24 98 ± 16 90 ± 3 0.15 0.50
Linuron
122 ± 3 108 ± 8 114 ± 8 0.58 1.9
Malaoxon
90 ± 19 117 ± 16 123 ± 14 0.15 0.50
Malathion
83 ± 5 82 ± 12 82 ± 12 3.4 12
MCPA
118 ± 6 101 ± 20 82 ± 7 5.5 19
Mecoprop
92 ± 17 109 ± 4 81 ± 13 1.1 3.6
Methiocarb
110 ± 16 123 ± 11 108 ± 12 0.41 1.4
Metolachlor
112 ± 3 108 ± 7 114 ± 4 0.086 0.32
Molinate
93 ± 16 82 ± 17 122 ± 8 1.1 3.6
Oxadiazon
100 ± 4 82 ± 19 88 ± 19 1.3 4.5
Pendimethalin
BLOD 94 ± 4 93 ± 7 17 55
Propanil
122 ± 15 112 ± 10 112 ± 17 0.90 3.0
Quinoxyfen
109 ± 12 86 ± 3 92 ± 7 1.1 3.6
Simazine
113 ± 12 96 ± 20 104 ± 13 0.31 1.1
Terbuthylazine
122 ± 13 117 ± 6 115 ± 9 0.14 0.48
Terbutryn
92 ± 11 106 ± 4 120 ± 4 0.19 0.66
Thiacloprid
97 ± 3 110 ± 18 80 ± 14 0.059 0.21
Thiamethoxam
119 ± 22 120 ± 20 89 ± 1 1.8 6.0
Thifensulfuron methyl
83 ± 1 118 ± 12 80 ± 19 0.022 0.06
Triallate
114 ± 20 110 ± 14 118 ± 4 3.8 13
BLOD: Below limit of detection 31 32
33 Analyte recoveries observed in each of the investigated matrices were in general in 34 good agreement at the three concentration levels (Tables 3-5). Analyte losses during 35 extraction and variations in analyte ionization due to matrix effects were well compensated 36 with the use of SIL standards, as indicated by the recoveries obtained, always between 80% 37 and 120%, except in a few cases that slightly deviated from this range. Likewise, relative 38 standard deviations (RSD) nearly always below 20%, or very close, indicated good 39 repeatability, as corresponds to automated methodologies with minimal sample 40 manipulation. 41 The average LODs and LODets obtained in surface water were between 0.3 and 19 42 ng/L and between 0.8 and 40 ng/L, respectively, for most of the compounds (86%), while in 43 8 groundwater these limits ranged between 0.1 and 28 ng/L and from 0.4 to 63 ng/L for all 44 compounds, respectively. 45 The extent of matrix effects in both surface and groundwater is shown in Figure 4. 46 Significant matrix effects (±20% variation of the signal) were observed for 90% of the 47 compounds in both matrices. MS signal enhancement was observed only for bentazone 48 (+54%) in groundwater and for fenitrothion (+94%), DIA (+37%), and MCPA (+37%) in surface 49 water. In all other cases, matrix effects occurred in the form of signal ionization suppression, 50 with values up to -100%. 51 52
Figure 4.
Matrix effects observed at a concentration level of 100 ng/L in groundwater (a) and 53 surface water (b). 54 55 9 In the last ten years, of all the methodologies published in the peer-reviewed 56 literature for the determination of non-polar and polar pesticides in water samples, only a 57 few of them are fully automated (Camilleri et al., 2015; Hurtado-Sánchez et al., 2013; Mann 58 et al., 2016; Quintana et al., 2019; Rubirola et al., 2017; Singer et al., 2010) (Table 6). Among 59 them, the present method and the recent one described by Quintana et al. (2019) are the 60 only capable of determining more than 50 pesticides in water samples, covering a wide 61 spectrum of medium to highly polar compounds. A special feature of the methodology here 62 presented as compared to the others is the large proportion of SIL analogs used for 63 quantification (88%). The use of SIL compounds for almost all targeted analytes indeed 64 requires an initial investment of economic resources for their acquisition; however, it is 65 essential to correct for matrix effects and ensure the production of reliable results in any 66 water matrix. 67 68 0
Table 6.
On-line SPE LC-MS/MS methodologies published in the peer-reviewed literature in the last ten years for the simultaneous determination of 69 medium to highly polar pesticides in groundwater and surface water . Number of pesticides Analyte overlap Water matrix Pre-extraction step Quantification method Accuracy (analyte recovery, %) Precision (RSD, %) Sensitivity (LOQ, ng/L) Matrix effects Reference 51
GW/SW centrifugation isotope dilution (88% of SIL) . 80-127 <20 GW: 0.4-63 SW: 0.8-40 (86% of analytes) ± 100 This study
30 SW/GW/DW centrifugation isotope dilution (22% of SIL) -40 to 42 <40 5-25 -161 to 100 (Quintana et al., 2019)
13 SW/DW/EWW filtration isotope dilution (86% of SIL) <10 relative bias a <10 SW: 0.3 – 2.1 ± 100 (Rubirola et al., 2017)
8 DW/SW/GW filtration isotope dilution (22% of SIL) 72-198 <40 GW: 8-62 SW: 10-64 not provided (Mann et al., 2016)
4 SW acidification (2.5‰ formic acid) not provided 86-114 <30 0.1-10 ± 20 (Camilleri et al., 2015)
10 SW filtration Standard addition 74-129 <14 0.3-33 not provided (Hurtado-Sánchez et al., 2013)
9 SW/WW filtration Isotope dilution (60% of SIL) 71-103 <20 SW: 3-100 ± 100 (Singer et al., 2010) 71 GW, groundwater; SW, surface water; SIL, stable isotope-labeled analogs; DW, drinking water; EWW, effluent wastewater; WW, wastewater. 72 a Relative bias (%) = ((theoretical concentration − experimental concentration)/theoretical concentration) × 100
73 74 75 76 1 Overall, on-line methods allow lowering the LODs to a higher extent than off-line analytical approaches because the complete sample (5 mL in our case) is transferred into the LC-MS systems. In comparison with other automated analytical methods available in the literature, the LODets obtained with the methodology developed, between 0.4 and 63 ng/L in groundwater and between 0.8 and 40 ng/L in surface water (for 86% of the compounds), are overall comparable or lower than those previously reported by other authors, for instance: LOQs from 0.3 to 33 ng/L reported by Hurtado-Sánchez et al. for 10 pesticides in surface water (Hurtado-Sánchez et al., 2013), from 3 to 100 ng/L reported by Singer et al. for 20 pesticides also in surface water (Singer et al., 2010), and LOQs values above 8 ng/L in groundwater and 10 ng/L in surface water as reported by Mann et al. (Mann et al., 2016). In this respect, it may be worth mentioning that the LODets in our method incorporate the confirmation by the SRM2, and thus could be higher than the LOQ of the SRM1 if the LOD of the SRM2 is above that value. Despite this, our automated approach provides the best sensitivity for the analysis of azinphos-methyl, bromoxynil, clothianidin, quinoxyfen, terbuthylazine, terbutryn, and thifensulfuron methyl, and to the best of our knowledge, this is the first time that an on-line SPE-LC MS/MS-based method is validated for the analysis of azinphos ethyl, azinpho- methyl oxon, dichlorvos, diflufenican, fenthion oxon, fenthion oxon sulfone, fenthion oxon sulfoxide, fenthion sulfone, fenthion sulfoxide, and oxadiazon in surface and groundwater samples. Moreover, the sensitivity of the presented methodology allows its application to monitor the target priority substances in surface waters below their respective lowest EQS (EC, 2013): alachlor, LOD=6 ng/L vs.
EQS =
300 ng/L; atrazine, 2 ng/L vs.
600 ng/L; chlorfenvinphos, 0.6 ng/L vs.
100 ng/L; chlorpyrifos, 0.6 ng/L vs.
30 ng/L; diuron, 0.6 ng/L vs.
200 ng/L; irgarol, 1 ng/L vs. vs.
300 ng/L; quinoxyfen, 5 ng/L vs.
15 ng/L; simazine, 5 ng/L vs. vs. vs.
2 ng/L) and four of the five neonicotinoids included in our study (imidacloprid 4 ng/L, thiacloprid 0.3 ng/L, acetamiprid 1 ng/L, thiamethoxam 1.8 ng/L, and as an exception clothianidin 18 ng/L, in all cases vs Occurrence in water samples
The developed methodology was applied to the analysis of 22 water samples collected in two agriculture-impacted areas of Catalonia with different predominant crops and pressures. The results obtained (lowest and highest concentrations, average concentrations, and detection frequencies) are summarized in Table 7, while individual concentrations of the pesticides found in the investigated samples are provided as SI in Tables 8 and 9. Of the 51 investigated compounds, 28 were detected in the Llobregat basin. The pesticide pattern observed in the studied area (Figures 2 and 5), which includes small tributaries (Pt 1 and Pt 2) and their confluence into the main river (Pt 3), as well as irrigation and drainage channels that give service to surrounding farms (Pt 4-11) and in some points, receive the input of WWTP effluents (Pt 6 and Pt 9), was characterized by the generalized presence of diuron and terbutryn throughout the investigated stretch, with punctually high concentrations of other compounds in certain sites, such as bromoxynil in Pt 4 and linuron in Pt 7 and 10. A variety of compounds were present in some locations, viz. , Pt 7 and Pt 10, in line with the variety of small exploitations dedicated to the cultivation of different crops (Figure 5). In this profile, it was also notable the presence of 2,4-D in the two sites most directly affected by the input of WWTP effluents (Pt 6 and Pt 9), which reflects a likely poor removal of this compound in the WWTPs. 4
Figure 5.
Cumulative levels of the most abundant (>100 ng/L in at least one sample) and/or frequently detected (>36%) pesticides found in the lower basin of the Llobregat River. Alachlor, atrazine, cyanazine, fenthion sulfoxide, fenthion oxon, irgarol, isoproturon, malaoxon, malathion, metolachlor, molinate, propanyl, simazine, terbuthylazine, and thiacloprid, detected at lower concentrations in fewer samples, are not represented in the figure.
Figure 6.
Cumulative levels of the targeted pesticides measured in the Ter River water samples.
Pt 1 Pt 2 Pt 3 Pt 4 Pt 5 Pt 6 Pt 7 Pt 8 Pt 9 Pt 10 Pt 110200400600800100012001400160018002000 Sampling locations C on c en t r a t i on [ ng / L ] Chlorfenvinphos Diuron Bromoxynil Linuron Terbutryn 2,4-D Diflufenican Chlortoluron Imidacloprid Azinphos ethyl Methiocarb Diazinon Mecoprop Dichlorvos
Pt 1 Pt 3 Pt 4 Pt 5b Pt 7 Pt 8 Pt 2 Pt 5a Pt 6 Pt 9020406080100120
Sampling locations
Groundwater C on c en t r a t i on [ ng / L ] Terbutryn Metolachlor MCPA Diuron Diazinon Irgarol Bentazone
Surface water Table 7.
Minimum, and mean concentrations and detection frequency of the targeted pesticides in the investigated water samples. a Mean calculated considering values Concentration (ng/L) Detection frequency b (%) Min Max Mean a Llobregat River 130 200 30 18 Alachlor 18 24 3.8 18 Atrazine <6.7 21 5.6 55 Azinphos ethyl 10 110 17 27 Bromoxynil <22 1520 150 27 Chlorfenvinphos <2.9 67 12 55 Chlortoluron 18 67 13 36 Cyanazine 29 29 2.6 9 Diazinon Dichlorvos <20 130 16 45 Diflufenican 130 150 25 18 Diuron 24 500 170 100 Fenthion oxon 37 37 3.4 9 Fenthion sulfoxide 32 32 2.9 9 Imidacloprid <10 190 19 27 Irgarol <6.6 41 7.6 45 Isoproturon <7.1 25 4.8 27 Linuron 130 520 100 27 Malaoxon 24 24 4.4 18 Malathion <17 32 6 27 Methiocarb 50 130 17 18 Metolachlor Molinate 27 33 5.5 18 Propanil 19 19 1.7 9 Simazine 16 20 3.3 18 Terbuthylazine Terbutryn Thiacloprid <0.79 31 4.3 27 Ter River Bentazone 110 110 9.8 9 Diazinon Diuron 14 14 1.3 9 Irgarol MCPA 18 18 1.6 9 Metolachlor 15 24 3.6 18 Terbutryn Table 8. Concentrations (ng/L) of the individual pesticides and cumulative pesticide concentrations (TOTAL) measured in the water samples collected in the Llobregat River. n.d.: not detected PESTICIDES Pt 1 Pt 2 Pt 3 Pt 4 Pt 5 Pt 6 Pt 7 Pt 8 Pt 9 Pt 10 Pt 11 2,4-D n.d. n.d. n.d. n.d. n.d. 197 n.d. n.d. 133 n.d. n.d. Alachlor n.d. n.d. n.d. n.d. n.d. n.d. 18 n.d. n.d. 24 n.d. Atrazine n.d. 13 <6.7 n.d. n.d. <6.7 17 n.d. n.d. 21 <6.7 Azinphos ethyl n.d. 10 n.d. n.d. n.d. n.d. 69 n.d. n.d. 106 n.d. Bromoxynil n.d. 74 <22 1520 n.d. n.d. n.d. n.d. n.d. n.d. n.d. Chlorfenvinphos n.d. 4.8 n.d. n.d. 4.8 <2.9 48 n.d. n.d. 67 <2.9 Chlortoluron n.d. 67 n.d. n.d. 30 n.d. 18 n.d. n.d. 27 n.d. Cyanazine n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 29 n.d. Diazinon 20 2.9 n.d. n.d. n.d. n.d. n.d. n.d. n.d. 71 18 Dichlorvos n.d. <20 <20 n.d. <20 n.d. n.d. <20 n.d. 130 n.d. Diflufenican n.d. n.d. n.d. n.d. n.d. n.d. 134 n.d. n.d. 145 n.d. Diuron 502 24 249 245 258 91 42 118 61 63 235 Fenthion oxon n.d. n.d. n.d. n.d. n.d. n.d. 37 n.d. n.d. n.d. n.d. Fenthion sulfoxide n.d. n.d. n.d. n.d. n.d. n.d. 32 n.d. n.d. n.d. n.d. Imidacloprid n.d. n.d. n.d. n.d. n.d. n.d. <10 n.d. n.d. 194 <10 Irgarol n.d. n.d. n.d. <6.6 <6.6 n.d. 33 <6.6 n.d. 41 n.d. Isoproturon n.d. n.d. n.d. n.d. n.d. n.d. 24 <7.1 n.d. 25 n.d. Linuron n.d. n.d. n.d. n.d. n.d. n.d. 524 n.d. 132 484 n.d. Malaoxon n.d. n.d. n.d. n.d. n.d. n.d. 24 n.d. n.d. 24 n.d. Malathion n.d. n.d. n.d. <17 n.d. n.d. 25 n.d. n.d. 32 n.d. Methiocarb n.d. n.d. n.d. n.d. n.d. n.d. 50 n.d. n.d. 131 n.d. Metolachlor Molinate n.d. n.d. n.d. n.d. n.d. n.d. 33 n.d. n.d. 27 n.d. Propanil n.d. n.d. n.d. n.d. n.d. n.d. 19 n.d. n.d. n.d. n.d. Simazine 16 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 20 n.d. Terbuthylazine n.d. n.d. n.d. n.d. n.d. n.d. 24 n.d. 3.5 30 n.d. Terbutryn 121 8.9 63 63 71 160 45 127 24 55 n.d. Thiacloprid <0.79 n.d. n.d. n.d. n.d. n.d. 16 n.d. n.d. 31 n.d. TOTAL 664 205 311 1830 364 448 1260 244 354 1800 252 7 1 Table 9. Concentrations (ng/L) of the individual pesticides and cumulative pesticide 2 concentrations (TOTAL) measured in the water samples collected in the Ter River. 3 n.d.: not detected 4 Total concentration calculated considering only values >LOQ 5 As shown in Table 7, the herbicides diuron and terbutryn were the most ubiquitous 6 pesticides, occurring in 100% and 91% of the samples analyzed, respectively. Diuron is an 7 effective herbicide used to treat invasive vegetation on both agricultural and non-agricultural 8 sites. It is also useful in removing mildew and killing algae. Thus, such a widespread 9 occurrence may be associated with its use in agriculture but also in industrial and urban 10 environments. The ubiquitous presence of diuron in this river has been already reported in 11 previous studies (Köck-Schulmeyer et al., 2012; Masiá et al., 2015). On the other hand, 12 terbutryn presence may be attributed to its release from the river sediments (Barbieri et al., 13 2019; Masiá et al., 2015) or nearby soils, where it may be accumulated, because the use of 14 this herbicide/algaecide as a plant protection product has been banned for nearly a decade in 15 the EU (EC, 2002). The sorption of terbutryn onto solid particles during its use in the past is 16 supported by its low water solubility (25 mg/L) and its moderately high octanol-water 17 partition coefficient (log Kow=3.7) (Table 1). 18 Diuron was also one of the targeted pesticides that presented the highest 19 concentrations (up to 500 ng/L), only surpassed by bromoxynil (1520 ng/L) and linuron (520 20 PESTICIDES Pt 1 Pt 2 Pt 3 Pt 4 Pt 5a Pt 5b Pt 6 Pt 7 Pt 8 Pt 9 Pt 10 Bentazone n.d. n.d. n.d. n.d. n.d. n.d. n.d. 108 n.d. n.d. n.d. Diazinon n.d. n.d. n.d. n.d. n.d. n.d. n.d. 2.3 4.6 n.d. n.d. Diuron n.d. n.d. n.d. n.d. n.d. n.d. 14 n.d. n.d. n.d. n.d. Irgarol n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 5.4 n.d. MCPA n.d. n.d. 18 n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. Metolachlor n.d. n.d. 15 n.d. n.d. n.d. n.d. n.d. n.d. 24 n.d. Terbutryn n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 4.1 5.3 n.d. TOTAL n.d. n.d. 33 n.d. n.d. n.d. 14 110 8.7 34 n.d. 8 ng/L). Bromoxynil and linuron are both herbicides used to control annual broadleaf weeds on 21 crop and non-crop sites. 22 In addition to diuron, bromoxynil and linuron, various other pesticides, namely, 2,4-23 D, azinphos-ethyl, dichlorvos, diflufenican, imidacloprid, methiocarb, and terbutryn, were 24 found at concentrations above the limit of 100 ng/L set for individual pesticides in water 25 intended for human consumption (EC, 1998). This is a concern since the Llobregat River 26 water is an important source of drinking water for the city of Barcelona and its metropolitan 27 area. Total pesticide concentrations in the Llobregat River waters ranged between 205 and 28 1830 ng/L, being above the limit set for total pesticides of 500 ng/L in four out of the eleven 29 investigated locations (Pt 1, Pt 4, Pt 7, and Pt 10) (Figures 2 and 5). In Pt 1 and Pt 4, the 30 exceedance is basically due to the large presence of one specific pesticide (diuron and 31 bromoxynil, respectively), whereas in the other sites (Pt 7 and Pt 10) the exceedance is due 32 to the concurrent presence of a variety of pesticides at low concentrations. Overall, the main 33 groups contributing to total pesticide levels were triazines, organophosphorus, ureas, and 34 neonicotinoids (Figure 5). 35 Many of the pesticides detected are priority substances in the field of water policy 36 (EC, 2013) or are included in the EU Watch List (EC, 2018). EQS exceedances (EQS provided in 37 Table 1) were only observed in the case of the antifouling agent irgarol in two locations (33 38 and 41 ng/L vs its EQS of 16 ng/L) and the insecticide dichlorvos in five locations (from 20 39 ng/L to 130 ng/L, far above its EQS of 0.7 ng/L). The use of irgarol and dichlorvos is currently 40 prohibited in the EU (EC, 2006b; EC, 2009; EC, 2016). Both substances were also found in the 41 Llobregat River sediments (Barbieri et al., 2019) which may be the source of these pollutants 42 into the river water. While dichlorvos has not been previously investigated in the Llobregat 43 River waters, irgarol was previously reported to occur in a tributary of the Llobregat River at 44 a maximum concentration of 5 ng/L (Quintana et al., 2019). 45 9 Methiocarb and the neonicotinoids imidacloprid and thiacloprid were found at 46 concentrations (up to 130, 190 and 31 ng/L, respectively) much higher than the maximum 47 acceptable method LODs established in the Watch List for the monitoring of these 48 substances (2 ng/L for methiocarb and 8.3 ng/L for the neonicotinoids). Given that the 49 acceptable method LODs provided by the implementing decision 2018/840 (EC, 2018) 50 coincide with the substance-specific PNEC in water, the measured concentrations could 51 affect aquatic organisms. 52 As for TPs, malaoxon was detected in two samples at similar concentration levels 53 than its parent compound, malathion (24 ng/L in the case of malaoxon in both samples vs 25 54 and 32 ng/L of malathion). The presence of malaoxon is of concern, considering that it is 60 55 times more toxic than its parent compound (Jensen and Whatling, 2010). Moreover, the 56 occurrence of malathion in drinking water sources is also worrisome as it may convert into 57 malaoxon during chlorine-based disinfection of water (Ohno et al., 2008) if it survives to the 58 water treatment train. Two TPs of the currently banned organophosphate insecticide 59 fenthion, namely, fenthion oxon and fenthion sulfoxide, were also detected in one of the 60 sampling locations (Pt.7, Figure 2) at concentrations of 37 ng/L and 32 ng/L, respectively, 61 while the parent compound was not detected in any sample. 62 In the Ter River, pesticide pollution was much less severe than in the Llobregat River 63 (Figure 6). In total, 7 pesticides (bentazone, diazinon, diuron, irgarol, MCPA, metolachlor, and 64 terbutryn) were detected in this area (Table 7). Total concentrations of pesticides in 65 groundwater were very low (up to 34 ng/L in Pt 9), being slightly higher in surface waters (up 66 to 110 ng/L in Pt 7). Thus, pesticide levels did not exceed in any case the limit of 500 ng/L set 67 in the European legislation for the sum of pesticides in groundwater (EC, 2006a) and waters 68 intended for human consumption (EC, 1998). However, bentazone was found in one of the 69 surface water samples (Pt 7) at a level (108 ng/L) higher than the standard of 100 ng/L set for 70 individual pesticides. Bentazone is extensively used as an herbicide in agriculture and 71 0 especially in rice fields, and likely to reach water bodies due to its high mobility in soils or via 72 runoff (high water solubility = 7112 mg/L and low log K ow = 0.46). The location where 73 bentazone was found corresponded indeed with a drainage channel of a rice cultivation area, 74 and was the most polluted among the investigated sites. 75 Regarding the occurrence of priority pesticides in the Ter surface waters, only three 76 were found (terbutryn, irgarol, and diuron), but none of them at concentrations above its 77 corresponding MAC (Tables 1 and 9). 78 Four compounds currently banned in Europe were also detected, including terbutryn, 79 irgarol, metolachlor, and diazinon (EC, 2002; EC, 2006b; EC, 2016). Their presence may be 80 due to improper use of the stock of these compounds or rather to their release by leaching 81 or runoff from soils or sediments, where the contaminants could have accumulated over 82 time. 83 Pesticide contamination in the Ter River has been scarcely investigated before. A 84 study conducted in 2001 in the same area revealed the presence of atrazine, DEA, and 85 metolachlor at levels below 100 ng/L in samples from the Ter River after water treatment in a 86 plant (Quintana et al., 2001). The use of these compounds is currently banned in Europe and 87 this could explain the absence of atrazine and its metabolite desethyl atrazine (DEA) in our 88 study, and the low levels of metolachlor (15 ng/L) found in surface water. 89 90 Environmental risk assessment 91 Hazard quotients (HQs) calculated for the various individual pesticides detected in 92 the samples based on their maximum concentrations measured are provided in Table 10. In 93 the case of the Llobregat River basin, six compounds, namely irgarol, dichlorvos, methiocarb, 94 azinphos ethyl, imidacloprid, and diflufenican presented HQ values above 10, in Pt 10, and 95 thus, they represent a potentially high risk for the aquatic organisms. This risk is associated 96 1 with their very low PNEC values ( Figure 7. Hazard Quotients (HQ) in the samples analysed, based on the individual HQs of the 125 pesticides measured in each sample. The HQs corresponding to those pesticides detected in 126 the samples but not specified in the legend have been grouped as “ Others ”. 127 128 3 Table 10. Hazard quotient (HQ) values calculated for the pesticides measured in water 129 samples of the Llobregat and Ter River basins. 130 Pesticides MEC a (µg/L) PNEC b (µg/L) HQ Llobregat River Bromoxynil 1.5 0.5 3.04 CFP 0.067 0.1 0.67 Chlortoluron 0.067 0.1 0.67 Cyanazine 0.029 0.19 0.15 Diazinon 0.071 0.01 7.08 Dichlorvos 0.13 0.0006 Diflufenican 0.15 0.009 Diuron 0.51 0.2 2.51 Fenthion oxon 0.037 0.2 0.18 Fenthion sulfoxide 0.032 - - Imidacloprid 0.19 0.0083 Irgarol 0.041 0.0035 Isoproturon 0.025 0.3 0.08 Linuron 0.52 0.1 5.24 Malaoxon 0.024 0.31 0.08 Malathion 0.032 0.006 5.30 Methiocarb 0.13 0.01 Metolachlor 0.028 0.2 0.14 Molinate 0.033 3.8 0.01 Propanil 0.019 0.2 0.09 Simazine 0.020 1 0.02 Terbuthylazine 0.030 0.06 0.51 Terbutryn 0.16 0.065 2.45 Thiacloprid 0.031 0.01 3.10 Ter River Bentazone 0.11 0.1 1.08 Diazinon 0.0046 0.01 0.46 Diuron 0.014 0.2 0.07 Irgarol 0.0054 0.0035 1.54 MCPA 0.018 0.5 0.04 Metolachlor 0.024 0.2 0.12 Terbutryn 0.0053 0.065 0.08 a MEC: maximum environmental concentration measured 131 b Conclusions 135 A fast and simple analytical methodology based on on-line SPE-LC-ESI-MS/MS has 136 been developed for the analysis of a wide range of medium to highly polar pesticides in 137 surface water and groundwater. Advanced aspects of the proposed method are its capability 138 to determine in a single run a high number of multi-class pesticides (51) using a low sample 139 volume (5 mL), in a considerably short analysis time (40 min) and with very high reliability of 140 results (due to the use of isotopically labeled analogs of 45 out of the 51 target compounds 141 for quantification by the isotope dilution method). For most of the initially targeted 142 compounds, the method shows satisfactory performance in terms of accuracy and 143 repeatability and provides enough sensitivity for their detection in ground and surface water 144 in compliance with the current legislation. 145 The application of the method to real samples showed a very different contamination 146 profile in the two investigated river basins. The average total pesticide concentration in the 147 Ter River samples was 17.6 times lower than in the Llobregat River samples, and only 7 148 pesticides were found in the Ter River versus 28 detected in the Llobregat River. The list of 149 pesticides found included priority and Watch List substances, and even pesticides currently 150 banned in Europe. The contamination pattern observed in the Llobregat River underlines the 151 significant contribution of the urban and industrial activities conducted in the metropolitan 152 area of Barcelona to pesticide pollution. 153 High risk for aquatic organisms was expected to be derived from the co-occurrence 154 of many pesticides in specific locations, where pesticides at high concentrations or with very 155 low PNEC values were present. These findings reveal that although less persistent than 156 organochlorine pesticides, medium to highly polar pesticides can be found in water at 157 potentially harmful levels. Further research is needed to understand the sources of these 158 compounds to control them, as well as to assess the real impact of pesticide co-occurrence 159 on the health of the aquatic ecosystems. 160 5 Acknowledgments 161 This work has received funding from the European Union’s Horizon 2020 Research and 162 Innovation Programme (WaterProtect project, No. 727450), the Spanish State Research 163 Agency (AEI) and the European Regional Development Fund (ERDF) (BECAS project, 164 CTM2016-75587-C2-2-R), the Spanish Ministry of Science and Innovation (Project CEX2018-165 000794-S), and the Government of Catalonia (2017 SGR 01404). Vinyet Solà from CUADLL 166 and Francesc Camps Saguè from IRTA are acknowledged for their help in the collection of 167 samples from the Llobregat River and the Ter River, respectively. 168 169