Occurrence of Selected Known or Suspected Endocrine-Disrupting Pesticides in Portuguese Surface Waters Using SPME-GC-IT/MS

: A multiresidue analytical methodology based on a solid-phase microextraction (SPME), fol-lowed by gas chromatography-ion trap mass spectrometry (GC–IT/MS), has been developed for trace analysis of 20 known or suspected endocrine-disrupting pesticides. The SPME conditions are optimized considering several key parameters to obtain the maximum sensitivity. After the optimization, the method validation is performed, and the limits of detection (ranged from 2–150 ng/L) and the coefﬁcient of determination (above 0.990) of studied compounds are determined for all the analytes. A robust sampling of twenty sampling points of surface water samples from the north and center of Portugal is performed, and the validated methodology is applied. In total, 20 compounds from four chemical families (13 organochlorine, 1 organophosphorus, 2 dicarboximide, and 4 pyrethroids) are studied, and the pesticides most frequently detected are eight organochlorine pesticides ( α -, β -HCHs, lindane, HCB, o, p (cid:48) -DDT, p , p (cid:48) -DDE, p , p (cid:48) -DDD, α -endosulfan), cypermethrin, and vinclozolin.


Introduction
The term endocrine-disruptors have been generally used to call compounds with characteristics in common, namely, with the capability to mimic the actions of hormones interfering with the behavior of endocrine systems in various living species [1]. The endocrine-disruptors compounds (EDCs) can be different chemical groups, such as pesticides, pharmaceuticals, and personal care products, polybrominated diphenyl ethers, among others [2]. Most of these are released to the environment by anthropogenic activities and may coexist and accumulate in crops and human bodies through food chains, posing risks to human health and the ecosystem [3,4]. Endocrine-disrupting pesticides (EDPs) are the largest group of EDCs. Despite all the benefits related to pesticide application (e.g., control of pests and diseases), they still can contaminate different environmental compartments being the aquatic environment the most vulnerable because of massive global use and its consequent degradation products [5,6]. The industry introduced several synthetic pesticides, such as organochlorine pesticides (OCs) (1940s), organophosphorus (OPs) (1960s), carbamates (1970s), and pyrethroids (PYR) (1980s). OCs, many of them classified as EDPs, were banned or restricted after the 1970s in most countries of the world, but continuously detected until today, due to their characteristics, such as high lipophilicity, bioaccumulation, high persistence, and potential of long-range transport [7]. The other chemical families of pesticides (e.g., OPs, carbamates, and PYR) are placed on the market The Douro is one of the longest rivers in the Iberian Peninsula, sharing its 930 km with Spain and Portugal [40]. The Douro River samples were representative of an urban area. Sousa, Cabrum, and Tâmega Rivers are tributaries of Douro River. The Leça River has its estuary near the port of Leixões. Chemical and bacteriological quality data led to classify the Leça River water quality as "bad" or "very bad" [41]. The Leça River is from the urban and industrial areas. The Minho River extends over 300 km in a southwest direction in the northwest corner of the Iberian Peninsula. Its terminal 70 km constitutes the natural border between Portugal and Spain, including the main estuarine axis of approximately 40 km (upstream limit of springtides) [42,43]. The Lima River estuary is an urban-industrialized estuary; the input of agricultural runoff impacts this system and urban and industrial sewage, which discharges nutrients and other substances that are transported from domestic, industrial, and agricultural areas into the estuarine area [44]. The Cávado River is in north Portugal. It runs 135 km from Gouveia to its mouth next to the city of Esposende. The Ave River basin is in a very populated region of the country. River water is intensively used for agricultural and industrial purposes, namely, textile industries [45]. Ria de Aveiro coastal lagoon is an estuarine system on the northwest coast of Portugal [46]. It is a shallow system that occupies an area of 43 km 2 at low tide. The Moscoso Stream and Caima River are tributaries from Ria de Aveiro estuarine zone. The Caima River drains a catchment area of about 196 km 2 , in the north-center region of Portugal [47]. These samples were from the urban, agricultural, and industrial areas.

Conditioning and Cleaning Procedure
The SPME (100 µm Polydimethylsiloxane (PDMS) fiber) procedure was performed with a manual fiber holder assembly supplied by Supelco (Darmstadt, Germany). A magnetic stirring and heater unit (AGIMATIC-N, Mundilab) was used during the SPME procedure. According to factory recommendations, the PDMS fiber used in the present study was conditioned at 250 • C in a current of helium for one hour in the GC-MS injector (Supelco ® ). Routinely, the conditioning was achieved by exposing the fibers to the injector port at 260 • C for 30 min before use. Fiber blanks were performed to ensure that no interferences from the fibers were present in GC chromatograms. After thermic desorption of the analytes in the GC, the fiber remained in the injector for several minutes for further cleaning at a high temperature. The liner in GC-MS was also replaced and cleaned every week to prevent the accumulation of analytes, and thus, unwanted peaks in the chromatograms.

Sample Enrichment
Standard water samples were prepared by spiking ultrapure water with a mixture of pesticides (5000 ng/L of the twenty pesticides) to validate the SPME step. Atrazine D-5 was used as an internal standard at 5000 ng/L. The comparison between studies with ultrapure water and a surface water sample (residue-free) was conducted to evaluate the SPME efficiency.

Sample Enrichment
Standard water samples were prepared by spiking ultrapure water with a mixture of pesticides (5000 ng/L of the twenty pesticides) to validate the SPME step. Atrazine D-5

Gas Chromatography
Gas chromatograph-ion trap mass spectrometer (Thermo Trace-Ultra GC from Thermo Fisher Scientific (Waltham, MA, USA)) equipped with a fused-silica capillary column ZB-XLB (30 m × 0.25 mm ID, 0.25 µm film thickness, Phenomenex) was used for the separation of the study pesticides, using helium 99.99% as carrier gas at a 1.3 mL/min flow rate. Injector mode was split/spitless injector in the split mode (split/ratio 10) at 260 • C during the chromatographic run. The oven temperature was as followed: Initial oven temperature was held at 60 • C for 1 min, programmed with a gradient of 20 • C/min up to 200 • C where it stays for 1 min and then an increase of 5 • C/min up to 245 • C where it stays for 32 min. The mass detector conditions were: Transfer line temperature 250 • C, ion source temperature 250 • C, and ionization mode electron impact at 70 eV. The retention times were determined in the full scan mode and validation study in the selected ion monitoring (SIM) mode. In SIM mode, the main ions (Q1, Q2, and Q3) were selected for each analyte (Table 1). A matrix-matched calibration curve (using a residue-free surface water sample) was performed using nine concentration levels (10, 50, 100, 500, 1000, 1500, 2000, 2500, and 3000 ng/L). The addition of the IS, atrazine D-5, was performed at 1000 ng/L. The limits of detection (LOD) and quantification (LOQ) of the method were evaluated considering a signal-to-noise ratio of 3 and 10, respectively. Each concentration was measured twice. Precision was evaluated by calculating the relative standard deviation (RSD) from 3 replicates.

Ionic Strength Adjustments
The effect of the amount of salt influenced the extraction efficiency of some analytes present in water samples. All the chemical families presented a good performance without the use of NaCl (Figure 2A). A few compounds (HCH isomers and diazinon) showed higher areas with the addition of 0.2% NaCl. This may be related to the increase in the ionic strength of the matrix and consequently decreases the solubility of these compounds in water. However, the results showed that the salt aggregate does not help in most of the studied compounds. The literature has recently reported the presence of several priority pesticides, most of them classified as EDPs in aquatic environments [24,[34][35][36]38,48,49]. Table 3 summarizes the results obtained in the present study with 20 surface water samples in terms of frequency of detection, minimum, maximum, and mean values detected. As the values obtained are different in terms of range, the graphs below show the data distribution by samples and each compound. All the data, including the value below LOQ (estimated by the analytical value), were counted for the data analysis.    Figure 2B showed the best results for the complete desorption of most pesticides with the desorption temperature at 260 • C. Bifenthrin and iprodione presented complete desorption with a temperature of 250 • C.

Organic Solvent
The effect of organic solvent was evaluated, and the addition of 2.5% methanol improved the extraction efficiency of the target compounds. The effect of this parameter is well demonstrated in Figure 2C.

Extraction Temperature and Time
In terms of extraction time, the results showed that as the time increased up to 45 min, the extraction efficiency improved. However, for longer times (60 min), the performance of the methodology decreased ( Figure 2D,E). The results obtained for the extraction temperature study showed similar behavior to the extraction time. As the temperature was increased, better areas were obtained. Only the dicarboximide pesticides had an opposite behavior at 60 • C.
The best conditions were selected, and the extractions were performed in glass vials by immersing the SPME fiber (PDMS) in 10 mL of water sample with 2.5% methanol,

Method Validation
After the optimization of the methodology, the validation study was performed. The proposed method measured different families of pesticides (OCs, PYR, OPs, and dicarboximide). Also, the methodology showed good performance, and the analytical parameters obtained at optimized conditions were calculated and presented in Table 2. Good linear regressions (R 2 ≥ 0.9904 in all cases) were obtained over the tested analytical range. The method achieved LODs and LOQs between 0.2-150 ng/L and 0.6-450 ng/L, respectively. The precision was evaluated, and the relative standard deviation (RSD) was lower than 15% is acceptable for all the studied analytes. The validation parameters obtained are in the same order as those reported for other authors applying SPME [19].

Monitoring of 20 Known or Suspected Endocrine-Disrupting Pesticides in Portuguese Surface Water Samples
The national and international authorities and the scientific community have shown interest and concern on the topic of water pollution and its adverse implications for humans and the ecosystem.
The literature has recently reported the presence of several priority pesticides, most of them classified as EDPs in aquatic environments [24,[34][35][36]38,48,49]. Table 3 summarizes the results obtained in the present study with 20 surface water samples in terms of frequency of detection, minimum, maximum, and mean values detected. As the values obtained are different in terms of range, the graphs below show the data distribution by samples and each compound. All the data, including the value below LOQ (estimated by the analytical value), were counted for the data analysis. Regarding the 20 spot surface water samples, only the diazinon and iprodione were not detected in the analyzed samples. The αand β-endosulfan were the most detected EDPs (75%), and the methoxychlor (5%) and deltamethrin (5%) were the least detected. Although the samples were collected in different seasons, no significant differences were observed. Even so, pyrethroids were more detected in the summer samples.

Organochlorine Pesticides in Surface Water Samples
Several decades after the OCs were banned, they are frequently detected in different environmental samples, due to their persistence and bioaccumulation. These compounds are in continuous movement between the different environmental compartments, such as water, soil, sediments, air, and/or bioconcentrate, in the living organisms [50].  The highest individual mean value was found for HCHs isomers in the Douro River (Mean ∑HCH = 2150 ng/L) and Cávado River (Mean ∑HCH = 350 ng/L). Indeed, the cumulative values were always above the maximum allowable concentration-Environmental Quality Standards (MAC-EQS) established by the Directive 98/83/EC and 2013/39/EU [15,16]. HCHs isomers were also detected in the Minho and Sousa Rivers, but below the quantification limit (<LOQ). The organochlorine compound, HCB, presented concentrations 18. The endosulfan isomers studied were (α, β), and the MAC-EQS = 0.01 was considered for endosulfan. All rivers, except the Lima River, showed the presence of endosulfan isomers. Finally, methoxychlor was observed at a concentration of 33 ng/L in the Tâmega River. In general, the rivers with the greatest OCs contamination were the Douro, Cávado, and the Sousa Rivers, as shown in Figure 3. Values above MAC-EQS were found in the seven river samples. Figure 4 presents the chromatogram obtained from the Douro and Cávado Rivers samples, demonstrating the confirmation by GC-MS of the presence of OCs.
the Douro, Cávado, and the Sousa Rivers, as shown in Figure 3. Values above MAC-EQS were found in the seven river samples. Figure 4 presents the chromatogram obtained from the Douro and Cávado Rivers samples, demonstrating the confirmation by GC-MS of the presence of OCs.   Comparing present data with Cruzeiro et al., [37] in Portuguese surface water samples, the 4,4 -DDT and 4,4 -DDD were detected in 100% of samples analyzed (N = 20), while in the present work, it was 40%. Indeed, the values obtained were above the restrictive European legislation (98/83/EC and 2013/39/EU). The same compound, 4,4'-DDT was quantified in another Portuguese study in concentrations 21.5 fold above the maximum (10 ng/L) level acceptable for inland and surface waters [38]. Although these authors reported HCB levels below its method quantification limit, the opposite occurred in the present work, where these values exceeded the LOQ. In 2017, 4,4 -DDT and HCB were also detected in the Douro River estuary, in 8 and 19 samples, respectively, over 2013739/EU levels [35]. Some studies reported pollution by these OCs in other samples in Portugal, such as, sediments samples from the Douro River [30,49,51] fruits [26], soil [28,52], and human samples [53]. This study contributes to the continuous evaluation of these compounds and emphasizes the relevance of continuous monitoring studies. Even though the usage of OCs was banned in Portugal, the recorded levels can be a consequence of their persistence or their misuse or illegal use. However, with this work, we can say that its incidence seems to be decreasing, at least for DDT.

Pyrethroids and Dicarboximide Pesticides in Portuguese Surface Water Samples
In this study, two PYR (cypermethrin and fenvalerate) and one dicarboximide pesticides (vinclozolin) were detected. Figure 5 shows a plot with the mean concentration in ng/L reported by the sample. Cypermethrin was the most frequently detected, and the Cávado, Cabrum, and Minho Rivers presented higher values. Nevertheless, all the surface water samples showed concentration values above the admissible amounts MAC-EQS. Fenvalerate was detected only in two samples (Cávado and Lima Rivers) at levels of 290 and 100 ng/L, respectively. while in the present work, it was 40%. Indeed, the values obtained were above the restrictive European legislation (98/83/EC and 2013/39/EU). The same compound, 4,4'-DDT was quantified in another Portuguese study in concentrations 21.5 fold above the maximum (10 ng/L) level acceptable for inland and surface waters [38]. Although these authors reported HCB levels below its method quantification limit, the opposite occurred in the present work, where these values exceeded the LOQ. In 2017, 4,4′-DDT and HCB were also detected in the Douro River estuary, in 8 and 19 samples, respectively, over 2013739/EU levels [35]. Some studies reported pollution by these OCs in other samples in Portugal, such as, sediments samples from the Douro River [30,49,51] fruits [26], soil [28,52], and human samples [53]. This study contributes to the continuous evaluation of these compounds and emphasizes the relevance of continuous monitoring studies. Even though the usage of OCs was banned in Portugal, the recorded levels can be a consequence of their persistence or their misuse or illegal use. However, with this work, we can say that its incidence seems to be decreasing, at least for DDT.

Pyrethroids and Dicarboximide Pesticides in Portuguese Surface Water Samples
In this study, two PYR (cypermethrin and fenvalerate) and one dicarboximide pesticides (vinclozolin) were detected. Figure 5 shows a plot with the mean concentration in ng/L reported by the sample. Cypermethrin was the most frequently detected, and the Cávado, Cabrum, and Minho Rivers presented higher values. Nevertheless, all the surface water samples showed concentration values above the admissible amounts MAC-EQS. Fenvalerate was detected only in two samples (Cávado and Lima Rivers) at levels of 290 and 100 ng/L, respectively. Regarding other studies conducted in Portugal [38], our levels of cypermethrin were slightly lower (350 ng/L) in terms of environmental levels, compared with a concentration of 453 ng/L obtained in Ria Formosa lagoon in summer samples. The authors discussed the evidence that the higher concentration presented in samples from the summer period is related to the application of pesticides. In a study conducted in 2017 in the Douro River samples [35], nine samples (in a total of 24 sampling sites) presented concentrations over the limits established by EU, and an average concentration for cypermethrin of 35.9 ng/L. This study has already shown lower values than those detected in samples from the Douro Regarding other studies conducted in Portugal [38], our levels of cypermethrin were slightly lower (350 ng/L) in terms of environmental levels, compared with a concentration of 453 ng/L obtained in Ria Formosa lagoon in summer samples. The authors discussed the evidence that the higher concentration presented in samples from the summer period is related to the application of pesticides. In a study conducted in 2017 in the Douro River samples [35], nine samples (in a total of 24 sampling sites) presented concentrations over the limits established by EU, and an average concentration for cypermethrin of 35.9 ng/L. This study has already shown lower values than those detected in samples from the Douro River (230 ng/L) in the present study. The vinclozolin and iprodione were evaluated in the current work, but only vinclozolin was detected.
Although there is no legislation for this pesticide in surface waters, it was quantified in nine samples: The Douro River (41 ng/L, the Minho River (30 ng/L), the Tâmega River (33 ng/L), Ria Aveiro (32 µg/L), the Cabrum river (51 ng/L), the Moscoso Stream (25 ng/L), the Caima River (64 ng/L), and the Ave River (26 ng/L)). The vinclozolin is a fungicide frequently used to control diseases in vineyards [54]. The vineyards in the north region of Portugal are in the hills on the banks of the rivers. The use of this pesticide in agriculture systems may be one of the reasons for the frequent detection.

Conclusions
The proposed analytical method, SPME-GC-IT/MS, was firstly optimized for the simultaneous determination of 20 priority pesticides. In terms of analytical validation, the methodology showed good performance and fulfilled the requirements according to European analytical standards to determine the target compounds in water samples. We can conclude that this work stood out for the diversity of sampling sites in Portugal, despite being spot samples. In addition, the surface water samples from several sampling sites from the north end center of Portugal presented known and potential EDPs, with many of them above the established MAC-EQS. The following pesticides were detected in order of decreasing frequency of detection: αand β-endosulfan (75%) > cypermethrin (55%)> HCB, aldrin, dieldrin, endrin, and vinclozolin (45%) > p,p -DDE + o,p -DDT + p,p -DDD (40%) > HCH isomers (30%), > fenvalerate (10%) > methoxychlor and deltamethrin (5%).
The persistence of OCs was again proven, as well as the presence of other pesticides. Despite the legislation established for these samples, this work also concluded that it is still insufficient when the presence of other non-legislated compounds was confirmed.
This study reinforces the data regarding the presence of priority pesticides in the environment and concluded that detected levels are still alarming. The monitoring and control of pesticides in rivers are an important issue that should continue to be explored in many areas and countries, and although the SPME performs well, more automated extraction methods should be explored to decrease the time and the cost of the monitoring analysis.