Occurrence and risk assessment of pesticides and pharmaceuticals in viticulture impacted watersheds from Northwest Spain

Chemosphere 341 (2023) 140098 Available online 6 September 2023 0045-6535/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/bync/4.0/). Occurrence and risk assessment of pesticides and pharmaceuticals in viticulture impacted watersheds from Northwest Spain V. Fern´ andez-Fern´ andez a , M. Ramil a , R. Cela b , I. Rodríguez a , * a Department of Analytical Chemistry, Nutrition and Food Sciences, IAQBUS - Institute of Research on Chemical and Biological Analysis, Universidade de Santiago de Compostela, R/Constantino Candeira SN, 15782, Santiago de Compostela, Spain b Mestrelab Research Center (CIM), Av. Barcelona 7, 15706, Santiago de Compostela, Spain HIGHLIGHTS GRAPHICAL ABSTRACT •Automated, sensitive determination of 73 pollutants in continental waters •Seasonal and spatial pulses of pesticides in viticulture impacted areas •Pharmaceutical residues affected by flowrate of water courses •Cardiovascular and psychiatric drugs pointed as markers of municipal wastewater •Olmesartan identified as potential hazardous based on risk quotients ARTICLE INFO Handling editor: Myrto Petreas Keywords: Pharmaceuticals Pesticides Fresh water Occurrence Risk assessment Liquid chromatography tandem mass spectrometry ABSTRACT An automated analytical methodology was developed, validated and applied to monitor 73 organic pollutants (pesticides and pharmaceuticals) in surface and groundwater samples obtained in watersheds from an intensive viticulture, rural region, in the Northwest of Spain. Filtered samples were concentrated using a reusable solidphase extraction sorbent, on-line combined with liquid chromatography tandem mass spectrometry (LC-MS/ MS). The analytical procedure achieved limits of quantification between 1 ng L −1 and 10 ng L −1 , with a throughput of 2 samples hour −1 , providing accurate recoveries for more than 90% of the 73 selected compounds, using calibration solutions prepared in ultrapure water (in presence of methanol and formic acid) as neat solvent. The distribution and the concentrations of pesticides in small streams impacted by discharges of treated municipal wastewaters were different in rural and residential areas. On the other hand, pharmaceuticals showed a similar distribution in both streams. In surface waters from viticulture impacted watersheds, with a limited influence of municipal wastewaters, pulses of pesticides were noticed, with values above 100 ng L −1 for several fungicides. Cardiovascular pharmaceuticals, psychiatric drugs and/or their transformation products were also * Corresponding author. E-mail address: [email protected] (I. Rodríguez). Contents lists available at ScienceDirect Chemosphere journal homepage: www.elsevier.com/locate/chemosphere https://doi.org/10.1016/j.chemosphere.2023.140098 Received 3 August 2023; Received in revised form 4 September 2023; Accepted 5 September 2023 Chemosphere 341 (2023) 140098 2 ubiquitous in these samples, with low, but relatively stable concentrations among sampling campaigns. Within the suite of investigated compounds, maximum pesticide residues remained below their predicted-non effect concentration (PNEC) in all samples. On the other hand, the environmental concentrations of the cardiovascular drug olmesartan stayed systematically above its PNEC in fresh water samples. 1. Introduction Scarcity and pollution of water resources are two relevant challenges to be faced in order to guarantee a sustainable development. Inland waters are particularly susceptible to contamination with anthropogenic chemicals from different sources. Application of pesticides in intensive agriculture areas might lead to hot spot contamination problems, either due to direct transport processes (i.e. run-off from crops during rain events and soil erosion), and/or to the misuse of these compounds. In addition, indirect sources, such as soil percolation and atmospheric drift, might contribute to the spread of pesticides in ground and surface water (Bexfield et al., 2021; Smalling et al., 2015; Suciu et al., 2020; Vera-- Candioti et al., 2021). Viticulture is recognized as one of the agriculture practices using the highest rates of pesticides (particularly fungicides) per hectare. Pollution of soil and groundwater in intensive viticulture areas has been highlighted in several previous studies (Manjarres-L´ opez et al., 2021; Suciu et al., 2020; Vallejo et al., 2019). Municipal wastewaters, sometimes combined with those from local industries, represent a relevant vector of pharmaceuticals, personal care, household products, and high production volume chemicals (HPVC), including certain pesticides, in the aquatic environment. Many of these compounds are incompletely removed at sewage treatment plants (STPs), ending in surface watercourses and reservoirs, and even reaching groundwaters (Loos et al., 2010; Meffe and de Bustamante, 2014). In dispersed population areas, domestic wastewaters receive limited treatment. This fact, added to leaks from sewers and septic tanks, might contribute additionally to pollute surface and groundwater with anthropogenic compounds. Monitoring the presence of organic pollutants in surface and groundwaters is thus required to understand the distribution of these compounds in the environment, to identify markers of anthropogenic pollution, and to assess their potential eco-toxicological risks. Liquid chromatography tandem mass spectrometry (LC-MS/MS) is regarded as the gold-standard for the determination of minor pollutants in the aquatic environment. Usually, LC-MS/MS analysis follows a previous concentration step. Solid-phase extraction (SPE) is the most resorted sample concentration technique, given the variety of available sorbents, and the possibility to process simultaneously several samples with the aid of a multi-position vacuum manifold (Montes et al., 2023). A step forward in the development of more sustainable, higher throughput, analytical methodologies is automation. In this regard, the SPE process can be performed with a dedicated device, working as an extra autosampler coupled to the LC-MS system, which mimics the different steps involved in the off-line SPE extraction and clean-up processes, based on disposable sorbents. These systems show a high versatility in terms of available sorbents and use of sequential elution and/or clean-up steps (Postigo et al., 2010; Rubirola et al., 2017; Trenholm et al., 2009). Another automation option involves the use of an additional LC pump and an extra valve connected to a high-pressure SPE cartridge. In this configuration, the own autosampler of the LC instrument withdraws the sample, which is thereafter pushed into the SPE sorbent by the auxiliary LC pump. Compounds are further recovered from the sorbent using the same mobile phases employed in the LC separation step (Hurtado-S´ anchez et al., 2013; Quintana et al., 2019). This second configuration is less demanding in terms of dedicated instrumentation; moreover, it reduces operational costs since it employs reusable cartridges (Hong et al., 2019). This research has been focussed on two major aims. The first one was to develop and to validate a fully automated procedure (based on reusable SPE sorbents on-line connected to LC-MS/MS) for the sensitive determination of trace organic pollutants, related with agriculture activities and discharges of municipal sewage. A suite of pharmaceuticals, selected attending to their limited biodegradability during treatment of municipal wastewaters (L´ opez-García et al., 2018); previous reports of their presence in treated wastewater (Styszko et al., 2021), and/or surface water samples obtained in the Northwest of Spain (Castro et al., 2019; Fern´ andez-Fern´ andez et al., 2022), was chosen as markers of municipal sewage in the considered watersheds. Other markers of municipal wastewater initially included in the study were the stimulant caffeine, the organophosphorus flame retardant tris(2-chloroisopropyl) phosphate (TCPP) and the UV-filter benzophenone-3. The selection of agriculture pesticides comprised mainly fungicides and insecticides. Both are intensively employed in vineyards surrounding two of the watersheds considered in the study. Moreover, some of them have been reported in agriculture soils obtained in the same geographic area (P´ erez-May´ an et al., 2020), and in surface and groundwater from vineyard areas from other regions in Spain (Manjarres-L´ opez et al., 2021; Vallejo et al., 2019). The second aim of the study was to assess the relevance of both sources of pollutants, and to investigate the temporal fluctuations of selected compounds (depending on seasonal uses and variations in the flowrate of water courses), in two watersheds from an intensive viticulture, dispersed population, area in the Northwest of Spain. To this end, four sampling campaigns were programmed. Finally, found concentrations were employed to identify those compounds representing an environmental hazard on the basis of measured levels and predicted non-effect concentrations (PNEC) in fresh water. 2. Material and methods 2.1. Solvents, standards and sorbents Methanol (MeOH) and acetonitrile (ACN), both LC-MS grade quality, were purchased from Merck (Darmstadt, Germany). Formic acid (FA) was obtained from same supplier. Ultrapure deionized water (18.2 MΩ cm −1 ) was produced using a Geni U system (Rephile, Shanghai, China). Standards for a selection of pharmaceuticals, including several excretion metabolites, pesticides, caffeine, TCPP and benzophenone-3 were purchased from Sigma-Aldrich (St. Louis, Missouri, USA). The initial selection of targets includes 75 compounds. Their names, CAS numbers, log D values and pKa values are provided as supplementary information, Table S1. It is worth noting that, some of the selected compounds (i.e. azolic species, venlafaxine and its metabolite, azoxystrobin and trimethoprim) have been included in the last update of the European Watch List of compounds to monitor in the aquatic environment, published in 2022 (European Commission, 2022). The log D values of target compounds, at pH 7, ranged from - 2, case of valsartan acid and atenolol, to above 5 units, for some pesticides and pharmaceuticals (i.e. quinoxyfen and sertaconazole), Table S1. The suite of analytes includes compounds existing in neutral, protonated (many of the basic pharmaceuticals) and deprotonated (i.e. irbersatan and the metalaxyl transformation product CGA 62826) forms at neutral pH. Out of the initial selection of 75 target species (Table S1), 73 compounds could be determined at trace levels with the proposed method. Isotopically labelled analogues (deuterated or 13 C species) of some of the analytes were purchased either from Sigma-Aldrich, or from Toronto Research Chemicals (North York, Canada), Table S2. For those compounds whose labelled analogues were not V. Fern´ andez-Fern´ andez et al. Chemosphere 341 (2023) 140098 3 available, selection of surrogate standards (SSs) was made based on similar structures, and/or proximity of their chromatographic retention times. Two different sorbents were evaluated for on-line SPE of water samples. Reversed-phase PLRP-S (12.5 mm ×2.1 mm) cartridges, containing polystyrene-divinyl benzene (PS-DVD) as sorbent (particle size 15–20 μ m), were acquired from Agilent (Wilmington, DE, USA). Mixedmode, reversed-phase and weak-cation exchange, Strata-X-CW (20 mm ×2 mm) cartridges, containing a polymeric weak cationic exchanger (25 μ m), were provided by Phenomenex. 2.2. Samples Samples of surface water were collected in three different watersheds in the Northwest of Spain. The first one was selected as representative of a small river running through a peri-urban, residential area, receiving the treated wastewater from a municipal STP with a capacity of 14,000 habitant equivalents (h.e.). The STP is equipped with an activated sludge unit and a final UV treatment (catchment 1). Sampling points were fixed 3 Km upstream (S1) and downstream (S2) the STP discharge point. In this watershed, detached houses are surrounded by gardens and self-consumption orchards. The 2nd catchment corresponds to a medium size river (maximum and minimum flowrates during the sampling period were 11.96 m 3 s −1 and 1.47 m 3 s −1 ), running through a hilly, rural, intensive viticulture area. Sampling points were set in one of its tributary streams, which is also affected by discharges of municipal treated wastewater (the SPT has a capacity of 11,950 h.e., and it is equipped with biological and UV treatment units), (S3), and at different sites along the main river (S4 to S6). In the 3rd catchment, sampling points were set in four sites (S7, S8, S10 and S12) of a large water reservoir (total capacity 60 million of m 3 ), three small streams (S9, S14 and S15) connected to the reservoir, and two springs of groundwater (S11 and S13). As in case of the 2nd watershed, streams and springs were located in a hilly rural area, devoted to vineyards within the Denomination of Origin Ribeiro (Galicia, Spain). The river feeding the water reservoir receives the treated wastewater from a 100.000 inhabitants city located 15 Km upstream sampling point S12. Table S3 summarizes the codes, the exact position and the type of water in each sampling point. A map with the position of sampling sites, including also the nearby STPs, is shown in Fig. 1. Grab water samples were taken in polypropylene flasks, transported to the laboratory and frozen within the next 6 h. Analysis of frozen samples (−20 ◦C) was carried out within the next seven days. Four different sampling campaigns were performed between the beginning of March 2022 and the end of July 2022. This interval covers the end of winter, before application of fungicides and insecticides to vineyards, and the worst scenario at the end of July, with low flowrates in all watercourses (the exception was the reservoir with a dammed water volume in the range from 52 to 55 million of m 3 through the study), and intensive application of pesticides to vineyards. 2.3. Sample preparation Before analysis, samples were filtered (0.22 μ m pore size, PTFE hydrophilic syringe filters) and made to the mark in 5 mL, volumetric glass flasks containing 0.75 mL of MeOH and 0.02 mL of FA. Calibration standard solutions were prepared in the same way, using ultrapure water with same contents of MeOH and FA as water samples, in the range of concentrations from 1 ng L −1 to 500 ng L −1 , with SSs at 200 ng L −1 . 2.4. On-line SPE and determination conditions On-line concentration of water samples and determination of target compounds were carried out using an Agilent 1290 Infinity II, LC instrument, connected to a triple quadrupole (QqQ) mass analyzer (MS), Fig. 1. Map of sampling points in the three watersheds, including the position of the nearby municipal sewage treatment plants (STPs). V. Fern´ andez-Fern´ andez et al. Chemosphere 341 (2023) 140098 4 Agilent 6495, equipped with a jet stream-type electrospray ionization source (ESI). Compounds were separated using a Zorbax Eclipse XDB- C18 column (100 mm ×2.1 mm, 3.5 μ m), also purchased from Agilent, operated at a flowrate of 0.3 mL min −1 , and maintained at 30 ◦C. Ultrapure water (phase A) and ACN (phase B), both 0.1% in FA, were combined as mobile phases as follows: 0–4 min (15% B), 16–17 min (85% B), 17.1–19 min (100% B), and 19.1–25 min (15% B). Nitrogen (N 2 ) was employed as nebulizing and drying gas in the ESI source at values of 20 PSI (137.9 KPa) and 14 L min −1 , at a temperature of 200 ◦C. N 2 was also used as sheath gas at 11 L min −1 (250 ◦C). The voltages of the ESI needle, the nozzle and the fragmentor were set at 3000 V, 1500 V and 166 V, respectively. The m/z values corresponding to precursor and product ions of each compound, with retention times and collision energies for each transition, are summarized in Table 1 (except benzophenone-3 and TCPP). Those corresponding to SSs are given as supplementary information, Table S2. Compounds were determined in the dynamic multiple reaction mode (MRM), using a window of ±0.5 min around their retention times, Table 1 and Table S2. The connection between the analytical LC-ESI-MS/MS instrument and the on-line sample preparation unit was made with a two-positions, 10-port, valve, with the SPE cartridge connected between two ports of this valve. A scheme of the device, showing connections of analytical and auxiliary pumps with the 10-port valve, has been provided elsewhere (P´ erez-May´ an et al., 2021). The volume of sample, or calibration standard solutions, loaded in the on-line SPE cartridge was 0.4 mL. Table S4 summarizes the main events (injection of the sample, loading in the SPE cartridge, desorption and conditioning) involved in the on-line SPE procedure, with solvents and flowrates employed during washing and conditioning of the SPE sorbent. All the parameters related to on-line SPE, LC separation and ESI-MS/MS determination conditions were controlled using MassHunter software. The MassHunter Quantatitative package was employed for data managing and compounds quantification in spiked and non-spiked water samples. 2.5. Performance of the method and application to water samples The linearity and the accuracy of the analytical procedure (on-line SPE followed LC-ESI-MS/MS), accounting for the performance of sample concentration and ionization steps, were investigated with responses obtained for the quantification (Q1) transition of each compound, corrected with that measured for the assigned SS. Accuracy was tested using filtered river water samples spiked at three different levels: 10 ng L −1 , 50 ng L −1 and 250 ng L −1 . Identification of compounds in spiked and non-spiked samples was based on retention times and qualification (Q2) to Q1 product ions response ratios within 0.1 min, and ±30% of the average values obtained for calibration standards, respectively. During analysis of each batch of samples (including eight levels of calibration standards, and 15 samples processed in duplicate) a minimum of two procedural blanks (ultrapure water with addition of SS), two calibration standards, to verify the accuracy of calibration, and one spiked environmental sample were processed. Only those compounds not detected in procedural blanks above their LOQs, and showing relative recoveries in the range from 70% to 120% were considered for quantification. 3. Results and discussion 3.1. Method optimization Performance of on-line SPE methods depends on the efficiency of compounds retention and desorption steps. Reversed-phase (PLRP-S) and mixed-mode (reversed-phase and weak cationic exchange) polymers were tested as sorbents, whilst combinations of MeOH: H 2 O or ACN: H 2 O (in both cases containing a 0.1% of FA) were evaluated as mobile phases for desorption of the SPE sorbent and LC separation purposes. The mixed-mode cartridge led to wider peaks for basic compounds than the reversed-phase one, whilst similar chromatographic profiles were noticed for slightly basic species (i.e. azoles) and neutrals with both sorbents, see Fig. S1. Likely, this effect is associated with electrostatic interactions between strong bases and the mixedmode sorbent, which produce broad, tailing desorption bands. Consequently, the reversed-phase sorbent was selected for method development. Elution efficiency, assessed through chromatographic peak profiles, was also affected by the organic component of the mobile phase in the LC column. Considering the gradient described in section 2.3, better peak shapes, and slightly shorter retention times, were observed using ACN than with MeOH, figure not shown. Likely, these chromatographic differences are due to higher elution strength of the former solvent, leading to a narrow band of compounds transferred from the SPE cartridge to the head of the analytical column. Cross-contamination problems between consecutive chromatographic runs varied as function of the considered compound and the sample pre-treatment. On-line SPE of spiked samples at their natural pH (ca. 6.5–7 units for surface, and 5.8 in case of ultrapure water aliquots) reflected the existence of important carry-over problems. Memory effects were noticed not only for highly lipophilic species, such as sertaconazole (log D 5.60, Table S1), but also for medium polarity compounds, as it is the case of sertraline and its excretion metabolite norsertraline (log D values of 2.7 and 2.8, respectively, Table S1). Addition of methanol to water samples (15%, under final conditions), and particularly, acidification using FA (0.02 mL were added to 5 mL samples) minimized those problems, Table S5. Both factors increased the solubility of lipophilic (case of MeOH) and basic compounds (present in the protonated form at acidic pH) preventing sorption losses in the connections between the autosampler of the LC-MS/MS instrument with the 10-port valve coupled to the SPE sorbent. Another positive effect of acidification was an enhancement in the responses of certain compounds, either with basic (i.e. sertraline) or acid properties (valsartan acid). Fig. 2 compares the chromatographic peaks for a selection of compounds corresponding to acidified and non-acidified aliquots of the same river water matrix, spiked at 50 ng L −1 . Responses for some species, particularly neutrals as it is the case of benalaxyl, were hardly affected by sample pH (Fig. 2A); however, in other cases they were enhanced significantly, either as a result of a better retention in the SPE sorbent (case of acidic compounds, Fig. 2B), or due to an improved ionization efficiency of bases, Fig. 2C. A limitation found during development of the SPE on-line method was the presence of some compounds in procedural blanks. This problem affected the two HPVC, initially included in the selection of targets: benzophenone-3 and TCPP, Table S1. Likely, the source of contamination were mobile phases, and particularly ultrapure water employed to condition the SPE sorbent, and to push samples and standards into the on-line cartridge. Using direct injection of sample extracts (i.e. obtained by off-line SPE) a delay column, installed between the LC pump and the injection valve, permits to discriminate the peak coming from the sample from the contribution of the mobile phase, which is shifted to slightly longer retention times (Fern´ andez-Fern´ andez et al., 2022). However, with the on-line combination of SPE and LC-MS/MS employed in this research (using the mobile phase of the analytical column to desorb the SPE cartridge) any compound present in the aqueous phase (ultrapure water) used to load the sample in the SPE sorbent, and/or to rinse this sorbent, is retained by the cartridge and further transferred to the analytical column. This limitation prevented the determination of benzophenone-3 and TCPP at concentrations below 50 ng L −1 in environmental water samples. Obviously, their residues in ultrapure water are far below this level; however, as the volume of ultrapure water passed through the SPE cartridge between analysis is significantly larger than the concentrated volume of sample (c.a. 14 mL versus 0.4 mL, see Table S4), the contamination problem was enhanced. V. Fern´ andez-Fern´ andez et al. Chemosphere 341 (2023) 140098 5 Table 1 Name, uses, retention times, MS/MS determination conditions and linearity of the SPE on-line LC-MS/MS procedure. Compound Uses Retention time (min) Precursor ion Q1 (CE) Q2 (CE) Ratio (Q2/ Q1) IS Linearity (R 2 .1–500 ng L −1 ) Acetamiprid Pesticide 7.64 223.1 126.0 (27) 56.1 (12) 0.52 Acetamiprid-d3 0.999 Ametoctradin Pesticide 11.71 276.2 176.1 (48) 70.0 (32) 0.15 Myclobutanil-d 4 0.99 Amisulpride Pharmaceutical 5.01 370.2 242.0 (20) 196.0 (40) 0.75 Flecainide-d 4 0.999 Amitriptyline Pharmaceutical 5.87 278.2 233.1 (16) 91.1 (36) 0.96 Flecainide-d 4 0.999 Atenolol Pharmaceutical 4.85 267.0 190.0 (16) 144.9 (25) 1.33 Tramadol- 13 C-d 3 0.991 Atorvastatin Pharmaceutical 12.37 559.3 440.2 (20) 276.1 (40) 0.1 Myclobutanil-d 4 0.998 Azoxystrobin Pesticide 12.34 404.1 372.1 (8) 329.1 (32) 0.33 Myclobutanil-d 4 0.998 Benalaxyl Pesticide 13.87 326.2 148.1 (27) 91.1 (48) 0.75 Myclobutanil-d 4 0.999 Boscalid Pesticide 12.52 343.0 307.1 (16) 139.9 (27) 0.2 Myclobutanil-d 4 0.99 Caffeine Stimulant 5.63 195.3 137.9 (22) 110.1 (25) 0.32 Carbendazim-d 3 0.997 Candesartan Pharmaceutical 10.06 441.1 235.1 (20) 192.1 (32) 0.9 Irbesartan-d 4 0.997 Carbamazepine Pharmaceutical 9.24 237.1 194.1 (21) 193.1 (41) 0.31 Flecainide-d 4 0.999 Carbendazim Pesticide 5.22 192.1 160.1 (16) 132.1 (32) 0.2 Carbendazim-d 3 0.98 Citalopram Pharmaceutical 8.36 325.2 109 (28) 262.1 (16) 0.32 Flecainide-d 4 0.999 Climbazole Pharmaceutical 9.20 293.1 197.1 (16) 141.0 (24) 0.2 Climbazole-d 4 0.999 Clomipramine Pharmaceutical 9.71 315.2 86.1 (20) 58.1(56) 0.75 Clotrimazole-d 5 0.993 Cloperastine Pharmaceutical 9.75 330.2 201.0 (16) 166.1 (40) 0.58 Clotrimazole-d 5 0.998 Clothianidin Pesticide 7.08 250.0 169.1 (8) 131.9 (8) 0.58 Clothianidin-d 3 0.999 Clotrimazole Pharmaceutical 9.64 277.1 164.9 (28) 239.0 (60) 0.45 Clotrimazole-d 5 0.99 Clozapine Pharmaceutical 7.27 327.1 270.1 (24) 191.9 (52) 0.7 Imazalil-d 5 0.994 Cyflufenamid Pesticide 14.94 413.1 295.1 (10) 203.0 (48) 0.99 Myclobutanil-d 4 0.99 Cyprodinil Pesticide 10.94 226.1 92.9 (40) 108.1 (24) 0.43 Cyprodinil-d 5 0.999 Diclofenac Pharmaceutical 12.64 296.0 214.0 (40) 250.0 (10) 0.51 Myclobutanil-d 4 0.998 Difenoconazole Pesticide 13.87 406.0 251.1 (25) 111.1 (60) 0.25 Myclobutanil-d 4 0.992 Dimethomorph Pesticide 11.23; 11.52 388.1 301.1 (20) 165.1 (32) 0.41 Dimethomorph-d 6 0.998 Fenamidone Pesticide 12.33 312.1 92.2 (28) 236.1 (8) 0.3 Miconazole-d 5 0.98 Flecainide Pharmaceutical 8.46 415.1 398.0 (24) 301.0 (40) 0.61 Flecainide-d 4 0.999 Fluconazole Pharmaceutical 6.80 307.1 219.9 (20) 70.0 (44) 0.51 Tramadol 13 C-d 3 0.999 Fluopicolide Pesticide 12.74 392.9 172.9 (20) 365.0 (12) 0.08 Myclobutanil-d 4 0.999 Haloperidol Pharmaceutical 8.63 376.1 123.0 (44) 165.1 (24) 0.92 Flecainide-d 4 0.998 Imazalil Pesticide 8.91 297.1 69.0 (16) 255.0 (12) 0.27 Imazalil-d 5 0.99 Imidacloprid Pesticide 7.26 256.1 208.9 (12) 175.1 (12) 0.93 Imidacloprid-d 4 0.999 Iprovalicarb Pesticide 12.08 321.2 119.1 (16) 202.9 (0) 0.1 Myclobutanil-d 4 0.999 Irbesartan Pharmaceutical 9.48 429.2 207.1 (24) 195.0 (24) 0.18 Irbesartan-d 4 0.999 Lamotrigine Pharmaceutical 6.39 256.0 43.1(40) 210.8 (32) 0.22 Lamotrigine- 13 C 3 0.999 Mandipropamid Pesticide 12.51 412.1 328.1 (8) 125.0 (40) 0.69 Myclobutanil-d 4 0.997 CGA 62826 Pesticide 8.94 266.1 220.1 (12) 192.1 (15) 0.67 Metalaxyl- 13 C 6 0.999 Metalaxyl Pesticide 10.38 280.2 220.1 (10) 192.1 (12) 0.36 Metalaxyl- 13 C 6 0.999 (continued on next page) V. Fern´ andez-Fern´ andez et al. Chemosphere 341 (2023) 140098 6 3.2. Analytical performance The determination coefficients (R 2 ) of calibration graphs for standards in ultrapure water (from 1 ng L −1 to 500 ng L −1 ) stayed above 0.99 for most compounds, Table 1. For highly lipophilic species, i.e. sertaconazole, linearity was lost at concentrations above 250 ng L −1 , with a reduction in the slope of calibration curves for higher levels. Efficiency of on-line SPE and potential variations in the yield of ESI ionization for environmental water samples versus standards in ultrapure water were evaluated together, using samples fortified at two different concentrations levels: 50 ng L −1 and 250 ng L −1 . Differences in peak areas (without SS correction) for spiked (n =3) and non-spiked (n =2) aliquots of river water were normalized to those measured for standards in ultrapure water. Obtained values are provided as supplementary information, Table S6. Fig. S2 summarizes the number of compounds with normalized relative responses in three different ranges: Table 1 (continued) Compound Uses Retention time (min) Precursor ion Q1 (CE) Q2 (CE) Ratio (Q2/ Q1) IS Linearity (R 2 .1–500 ng L −1 ) Metconazole Pesticide 13.07 320.1 70.0 (28) 124.9 (52) 0.09 Tebuconazole-d 9 0.991 Methiocarb Pesticide 11.76 226.1 169.0 (4) 121.1 (12) 0.48 Myclobutanil-d 4 0.999 Metoxyfenozide Pesticide 12.84 369.1 149 (15) 313.2 (8) 0.62 Myclobutanil-d 4 0.999 Metrafenone Pesticide 15.12 409.1 209.1 (8) 226.9 (16) 0.68 Myclobutanil-d 4 0.994 Mirtazapine Pharmaceutical 6.18 266.2 194.9 (28) 72.0 (20) 0.73 Miconazole-d 5 0.998 Myclobutanil Pesticide 12.24 289.1 70.1 (16) 125.1 (32) 0.25 Myclobutanil-d 4 0.999 N-desmethyl citalopram Pharmaceutical 8.23 311.22 108.9 (28) 262.0 (16) 0.49 Flecainide-d 4 0.998 Norsertraline Pharmaceutical 9.31 275.0 158.8 (20) 129.1 (30) 0.1 Norsertraline- 13 C 6 0.996 O-desmethyl venlafaxine Pharmaceutical 6.11 264.2 58.0 (17) Venlafaxine-d 6 0.999 Olmesartan Pharmaceutical 7.67 447.2 207 (24) 195 (20) 0.09 Irbesartan-d 4 0.998 Penconazole Pesticide 12.95 284.1 70.1 (15) 159.0 (30) 0.45 Myclobutanil-d 4 0.996 Prochloraz Pesticide 11.44 376.0 308.0 (4) 70.0 (24) 0.85 Myclobutanil-d 4 0.999 Propiconazole Pesticide 13.31 342.1 159.0 (32) 69.1(16) 0.8 Myclobutanil-d 4 0.999 Propranolol Pharmaceutical 7.85 260.2 116.1 (20) 183.1 (20) 0.6 Tramadol- 13 C-d 3 0.999 Pyraclostrobin Pesticide 14.46 388.1 193.8 (8) 163.1 (20) 0.78 Myclobutanil-d 4 0.994 Pyrimethanil Pesticide 9.32 200.1 82.1 (25) 106.9 (20) 0.71 Pyrimethanil-d 5 0.999 Quinoxyfen Pesticide 15.23 308.0 196.9 (36) 161.9 (45) 0.48 Myclobutanil-d 4 0.990 Sertaconazole Pharmaceutical 11.07 437.0, 439.0 180.9 (36) 180.9 (36) 0.63 Myclobutanil-d 4 0.990 a Sertraline Pharmaceutical 9.48 306.1 158.9 (36) 275.0 (12) 0.83 Norsertraline- 13 C 6 0.994 Tebuconazole Pesticide 12.60 308.1 70.0 (40) 124.9 (47) 0.08 Tebuconazole-d 9 0.997 Tebufenoxide Pesticide 13.54 353.2 133 (16) 297.1 (0) 0.42 Tebuconazole-d 9 0.992 Terbutryn Pesticide 9.70 242.0 185.9 (20) 68.0 (60) 0.32 Imazalil-d 5 0.999 Tetraconazole Pesticide 12.57 372.0 70.0 (24) 158.8 (32) 1 Myclobutanil-d 4 0.999 Thiabendazole Pesticide 5.40 202.0 175.0 (28) 131.1 (40) 0.73 Tramadol- 13 C-d 3 0.999 Thiacloprid Pesticide 8.42 253.0 126.0 (16) 90.0 (40) 0.36 Imidacloprid-d 4 0.999 Thiamethoxam Pesticide 6.51 292.0 211.1 (8) 132.0 (24) 0.34 Thiamethoxam-d 4 0.998 Thiophanate-methyl Pesticide 9.48 343.0 151.0 (20) 93.1 (56) 0.25 Thiophanate-methyl- d 6 0.992 Tramadol Pharmaceutical 6.51 264.2 58.1(15) Tramadol- 13 C-d 3 0.999 Trazodone Pharmaceutical 7.60 372.2 176.1 (24) 147.9 (40) 0.7 Tramadol- 13 C-d 3 0.999 Triadimenol Pesticide 11.43 296.1 70.1 (8) 99.1 (8) 0.04 Myclobutanil-d 4 0.997 Trifloxystrobin Pesticide 15.1 409.1 186.1 (12) 145.0 (52) 0.74 Myclobutanil-d 4 0.996 Trimethoprim Pharmaceutical 5.66 291.4 230.2 (25) 122.8 (25) 0.59 Lamotrigine- 13 C 3 0.999 Valsartan acid Pharmaceutical 8.28 267.1 206.1 (20) 178.1 (36) 0.52 Valsartan acid-d 4 0.999 Venlafaxine Pharmaceutical 7.33 278.0 58.1 (25) 260.0 (9) 0.27 Venlafaxine-d 6 0.999 Zoxamide Pesticide 14.29 336.0 187.0 (16) 159.0 (44) 0.71 Myclobutanil-d 4 0.999 a R 2 corresponding to the range from 5 to 250 ng L −1 . V. Fern´ andez-Fern´ andez et al. Chemosphere 341 (2023) 140098 7 below 70%, 70%–120% and above 120%. Depending on the considered concentration level, from 65 to 70 compounds (out of 73 targets) displayed relative responses between 70% and 120%. Accuracy and precision were investigated using surface water samples spiked at three different concentrations: 10 ng L −1 , 50 ng L −1 and 250 ng L −1 . At the lowest addition level, assays were carried out using water from a river flowing through an area without known discharges of urban wastewater and scarcely impacted by agriculture activities. For medium and high addition levels, samples collected at point S15 (see Table S3) were used. Recoveries were calculated for all compounds at the three addition levels, but caffeine at 10 ng L −1 . The obtained data are shown in Table 2. At 10 ng L −1 and 50 ng L −1 , 68 out of 73 compounds showed relative recoveries in the range from 70% to 120%. At 250 ng L −1 , only caffeine showed a relative recovery outside of the above range (68% ±5%). The standard deviations associated to above recoveries varied in the interval between 0.2% and 14%. The limits of quantification (LOQs) of the procedure were estimated from responses obtained for the lowest level spiked water sample, except in case of caffeine. LOQs remained equal to, or lower than 10 ng L −1 , with a value of 1 ng L −1 attained for 35 pollutants. In case of caffeine the LOQ was set at 25 ng L −1 . These LOQs are in the range of values reported for off-line SPE, providing a 500-fold concentration factor, using the same model of LC-MS/MS instrument (Montes et al., 2023). Advantages versus the off-line option are a lower cost (samples are concentrated in reusable SPE cartridges) and a significant reduction of the sample preparation time. Compared with previous applications of SPE on-line combined with LC-MS/MS, based also on reusable sorbents, this research permitted the simultaneous determination of pesticides and pharmaceuticals, whilst the former studies dealt only either with pesticides (Hurtado-S´ anchez et al., 2013; Quintana et al., 2019), or with pharmaceuticals (Hong et al., 2019). Moreover, it doubles the number of analytes covered in a similar scope study (dealing with pesticides and pharmaceuticals) using disposable sorbents for the on-line combination of SPE and LC-MS/MS (Rubirola et al., 2017). It is also worth noting that, for those compounds compiled in the recent revision of the European Watch List, the LOQs achieved in this research remain between one and two orders of magnitude below the maximum acceptable procedural LOQs (European Commission, 2022). 3.3. Application to surface and spring water samples Table S7 summarizes the concentrations of pollutants, found above their LOQs, in the set of samples obtained from the three catchments. Compounds were grouped in pharmaceuticals and pesticides, and sorted alphabetically within each group. Caffeine, maintained out of both groups, was considered as a marker of wastewater contamination of Fig. 2. LC-MS/MS chromatograms for the quantification (Q1) transition of selected compounds in acidified (left), and non-acidified (right), aliquots of the same river water. Addition level 50 ng L −1 . Asymmetry factors (As) are included in the figure. V. Fern´ andez-Fern´ andez et al. Chemosphere 341 (2023) 140098 8 Table 2 Accuracy of the method for samples of surface water spiked at different concentration levels and procedural LOQs. N =3 replicates of spiked and non-spiked water aliquots. Compound 10 ng L −1 50 ng L −1 250 ng L −1 LOQ Mean SD Mean SD Mean SD (ng L −1 ) Acetamiprid 108% 7% 87% 3% 99% 1% 1 Ametoctradin 98% 11% 87% 3% 99.8% 0.5% 10 Amisulpride 93% 3% 84% 1% 99.7% 0.3% 1 Amitriptyline 102% 4% 87% 1% 99% 1% 1 Atenolol 102% 7% 81% 1% 100% 4% 5 Atorvastatin 101% 10% 99% 4% 101% 2% 5 Azoxystrobin 107% 10% 116% 2% 100% 1% 1 Benalaxyl 108% 9% 100% 2% 101% 2% 1 Boscalid 111% 6% 119% 5% 103% 3% 10 Caffeine n.e. n.e. 141% 2% 68% 5% 25 Candesartan 109% 6% 80% 3% 100% 3% 10 Carbamazepine 106% 2% 93% 1% 99.9% 0.2% 1 Carbendazim 136% 6% 98% 1% 101% 1% 1 Citalopram 103% 4% 84% 1% 98% 2% 1 Climbazole 108% 8% 88.4% 0.2% 100% 1% 1 Clomipramine 112% 4% 89% 1% 99% 2% 1 Cloperastine 104% 4% 93% 1% 99% 1% 1 Clothianidin 99% 8% 91% 4% 98% 2% 5 Clotrimazole 102% 4% 86% 1% 99% 1% 1 Clozapine 82% 3% 83% 2% 99% 1% 1 Cyflufenamid 114% 14% 103% 2% 100% 1% 1 Cyprodinil 112% 8% 98% 1% 100% 1% 2 Diclofenac 113% 6% 105% 5% 96% 4% 10 Difenoconazole 100% 6% 95% 4% 99.8% 0.5% 5 Dimethomorph 103% 5% 86% 2% 100% 1% 2 Fenamidone 98% 5% 194% 3% 98% 3% 1 Flecainide 97% 7% 87% 1% 99% 1% 1 Fluconazole 115% 8% 93% 5% 99% 2% 5 Fluopicolide 92% 8% 86% 1% 102% 3% 5 Haloperidol 105% 4% 90% 1% 102% 2% 2 Imazalil 103% 8% 90% 4% 100.4% 0.3% 5 Imidacloprid 110% 3% 88% 2% 99% 1% 5 Iprovalicarb 94% 7% 90% 3% 102% 2% 2 Irbesartan 101% 7% 86% 2% 95% 2% 2 Lamotrogine 101% 8% 85% 1% 101% 2% 5 Mandipropamid 106% 6% 132% 1% 102% 2% 5 CGA 62826 171% 7% 211% 3% 100% 2% 5 Metalaxyl 97% 7% 90% 2% 100% 1% 1 Metconazole 102% 9% 80% 2% 99% 2% 5 Methiocarb 82% 7% 77% 2% 101% 2% 5 Metoxyfenozide 108% 1% 101% 1% 100% 1% 1 Metrafenone 106% 3% 105% 1% 101% 1% 1 Mirtazapine 74% 9% 83% 1% 98% 2% 1 Myclobutanil 83% 9% 93% 2% 99% 1% 1 N-desmethyl citalopram 86% 4% 81% 1% 100% 1% 1 Norsertraline 110% 10% 89% 2% 102% 2% 10 O-desmethyl venlafaxine 108% 7% 84% 1% 96% 2% 1 Olmesartan 144% 4% 86% 3% 94% 1% 2 Penconazole 99% 5% 93% 1% 102% 3% 1 Prochloraz 108% 5% 93% 1% 103% 3% 5 Propiconazole 106% 6% 95% 2% 99.8% 0.4% 5 Propranolol 100% 6% 80% 1% 100.1% 0.3% 2 Pyraclostrobin 96% 5% 91% 2% 100.2% 0.2% 1 Pyrimethanil 99% 8% 89% 2% 101% 1% 2 Quinoxyfen 103% 8% 93% 1% 101% 1% 1 Sertaconazole 80% 34% 85% 3% 98% 2% 5 Sertraline 95% 7% 85% 1% 99.9% 0.1% 1 Tebuconazole 104% 8% 82% 3% 100% 2% 5 Tebufenoxide 97% 5% 77% 1% 101% 1% 1 Terbutryn 102% 7% 86% 1% 99.5% 0.5% 10 Tetraconazole 102% 7% 106% 3% 100% 2% 2 Thiabendazole 103% 9% 75.7% 0.5% 103% 7% 1 Thiacloprid 87% 7% 73% 1% 100% 2% 1 Thiamethoxam 103% 9% 83% 3% 101% 2% 3 Thiophanate-methyl 66% 9% 61% 6% 101% 3% 5 Tramadol 61% 9% 86% 0% 95% 1% 1 Trazodone 104% 7% 93% 1% 99% 1% 1 Triadimenol 113% 7% 92% 1% 101% 3% 5 Trifloxystrobin 103% 3% 94% 2% 100% 1% 1 Trimethoprim 119% 7% 92% 1% 101% 3% 2 Valsartan acid 103% 9% 85% 4% 96% 1% 5 Venlafaxine 97% 4% 85% 2% 99% 2% 1 Zoxamide 109% 9% 99% 2% 101% 2% 1 V. Fern´ andez-Fern´ andez et al. Chemosphere 341 (2023) 140098 9 springs and streams. As regards the initial selection of pharmaceuticals and transformation products, just haloperidol, clotrimazole and sertaconazole remained undetected in all sampling sites. In case pesticides, 15 compounds (ametoctradin, clothianidin, cyflufenamid, fenamidone, iprovalicard, metconazole, methiocarb, metrafenone, prochloraz, pyraclostrobin, quinoxyfen, tebufenozide, thiacloprid, thiophanate-methyl and trifloxystrobin) were never detected. The highest residues of pharmaceuticals, pesticides and, in most cases caffeine, were measured at sampling points S2 and S3, located in streams from residential and rural areas affected by direct discharges of municipal wastewater from STPs, with similar capacities (expressed as h.e.) and water treatment facilities. Fig. 3A and B depict total concentrations of pharmaceuticals and pesticides in both points, through the four sampling campaigns. Sampling point S3 showed higher residues of pharmaceuticals and pesticides than S2. Ratios of concentrations between the rural (site S3) and the residential (site S2) streams for pharmaceuticals and pesticides are shown in Fig. 3C. In the four sampling campaigns, higher ratios were found for pesticides compared to pharmaceuticals, with a clear trend to increase from winter to summer. This trend agrees with the seasonal application of fungicides and insecticides in vineyards, located in the rural area, during spring and summer. The range of pesticides detected at sampling point S3 includes compounds with medium to high soil leaching potential, such as carbendazim and azoxystrobin (groundwater ubiquity score, GUS, values of 2.21 and 3.10, respectively), as well as some insecticides not authorized since several years ago, but with a high stability in agriculture soils, such as imidacloprid and thiamethoxam (https://sitem.herts.ac.uk/aeru/ ppdb). Likely, some of these compounds have persisted in vineyard soils since previous years, being transported to the watercourse through run-off and soil erosion. Residues of pesticides at sampling point S3 not only increased from the 1st to the 4th sampling campaign (Fig. 3C); moreover, their relative concentrations varied significantly. In particular, the sum of metalaxyl and its soil transformation product (CGA 62826) increased in the campaign of July, representing around 40% of the total concentration of pesticides at this point, Fig. 3D. Thus, delayed release of stable pesticides from vineyard soils (case of imidacloprid), and direct inputs (spray drifts and run-off transport) of more labile compounds (such as metalaxyl and its transformation product CGA62826) during, or immediately after, application contribute to the levels of these compounds measured in sampling point S3. Maximum residues of metalaxyl plus its environmental degradation product, azoxystrobin and carbendazim at sampling point S3 were 251.2 ng L −1 , 127.8 ng L −1 and 146.9 ng L −1 respectively, Table S7. These values are within the range of concentrations reported for several fungicides in a national scale survey carried out in USA (Battaglin et al., 2011), but they are lower than those measured in surface water impacted by agriculture activities in other areas of Spain, such as the Llobregat river in Catalonia, with residues of carbendazim above 1000 ng L −1 (Quintana et al., 2019). Maximum pesticide residues at sampling point S3 are also lower than those found in springs and wells from other vineyard areas in Spain (Manjarres-L´ opez et al., 2021), and those reported for a vineyard affected catchment in France, with maximum dissolved concentrations of tebuconazole and dimethomorph above 1000 ng L −1 (Rabiet et al., 2010). Fig. 4A summarizes the normalized concentrations of different groups of pharmaceuticals at rural (S3) and residential (S2) streams, both impacted by discharges of STPs. Conversely to seasonal variations in the distribution of pesticides observed at site S3 (Fig. 3D), no spatialtemporal changes were noticed among relative concentrations of the different kinds of pharmaceuticals. Found residues were dominated by the group of cardiovascular drugs, followed by the sum of psychiatric pharmaceuticals, and the pain-relief agent tramadol Fig. 4A. Top concentration compounds within the two groups of pharmaceuticals were O-desmethyl venlafaxine followed by lamotrigine (psychiatric drugs) n.e., not evaluated. Fig. 3. Overview of pharmaceuticals and pesticides in surface waters impacted by municipal treated wastewater in residential (S2) and rural (S3) areas. A and B, sum of concentrations for pharmaceuticals and pesticides, respectively. C, ratios of total concentrations between rural and residential areas. D, normalized concentrations of pesticides at sampling point S3. V. Fern´ andez-Fern´ andez et al.