An Efficient LC–HRMS-Based Approach to Evaluate Pesticide Contamination in Water Bodies with Measurement Uncertainty Considerations

Abstract

Over recent decades, the global occurrence of pesticide residues in aquatic environments has been a pivotal issue; however, their trace-level concentrations necessitate the establishment of ultra-sensitive and reliable analytical approaches. To this end, the present study describes the optimization and validation of an LC-HRMS-based method for the accurate determination of 18 pesticides in river and sea water, accompanied by a measurement uncertainty estimation. This method was applied to analyze 17 real samples from agriculture and aquaculture-impacted areas in Greece and Albania. Different solid-phase extraction (SPE) protocols were tested. For the analysis, cutting-edge Orbitrap MS technology and MS/MS fragmentation, along with the use of matrix-matched calibration curves, provided unprecedented accuracy (<5 ppm) and sensitivity for the confirmation of positive detections. Regarding method performance, exceptional linearity was obtained; the limits of quantification ranged from 1.7 ng L⁻¹ to 90 ng L⁻¹, recoveries varied from 61% to 96% in river water, while slightly higher recoveries (60–111%) were observed in seawater. In all cases, repeatability and intra-laboratory reproducibility were below 15%. The measurement expanded uncertainty (U′, k = 2) was estimated considering precision and bias. MU% values were lower than 50% in all cases, as recommended in SANTE guidelines and applied to the quantified results. The matrix effect study exhibited negative values (<20%) for all compounds. Application to real samples showed a low pesticide contamination load that should not be underestimated.

1. Introduction

Water pollution is a major global environmental issue that severely affects ecosystem sustainability and public health [1]. Pesticides and plant-protection products, widely employed in agriculture to protect crops, are among the principal contributors to pollution loads [2]. These chemicals frequently contaminate watersheds because of runoff from agricultural land into rivers and seas [3]. Simultaneously, aquaculture practices further intensify the strain in aquatic environments [4], exacerbating the risk of contamination [5,6]. Given that pesticides are integral to global agricultural productivity, they also pose a significant risk of environmental contamination through atmospheric transportation, even at considerable distances from their point of application [2,6].

A substantial proportion of legacy pesticides are included in the European Union’s “priority list” of compounds owing to their detrimental effects on the environment and human health [5]. Their widespread presence in aquatic ecosystems, particularly in areas with intensive agricultural and aquaculture activities, has elicited considerable concern from scientific and regulatory authorities [7,8,9]. Consequently, there is an urgent need for continuous monitoring and accurate assessment of their concentrations. These compounds have the potential to destabilize ecosystems and represent constant hazards to biodiversity due to their bioaccumulation and persistence [10,11,12]. Moreover, extended exposure to these pollutants has adverse effects on public health, with certain compounds identified as carcinogenic or harmful [13,14].

The expansion of aquaculture has resulted in the occurrence of a multitude of contaminants such as antibiotics, microplastics, organophosphate esters and pesticides in the aquatic environment. Specifically, herbicides, fungicides, and insecticides have been widely used to mitigate undesirable vegetation, mosses, harmful algae, fungal infections, and parasitic infections in the aquaculture [15]. Yet, pesticides used in agriculture may end up into the aquaculture because of run-off. However, limited research has been conducted on the prevalence of these substances in aquatic systems in Greece and Albania [6,16,17,18]. Both countries exhibit significant agricultural and aquaculture activities, rendering these areas susceptible to pollution from associated compounds [19,20]. Moreover, they have consistently been among the leading producers of fish farming products in the European Union over the past decade [20]. In 2022, Greece ranked fourth in the EU in terms of the total output of farmed aquatic organisms measured in tons of live weight. Greece contributed approximately 13.0% of the EU production, which rendered it the leading aquaculture in terms of revenue among EU countries. The estimated value of Greece’s aquaculture output in 2022 is EUR 844 million, accounting for 17.4% of the total earnings in the EU [20]. In contrast, Albania has significantly increased all forms of aquaculture activity, leading to an expansion in aquaculture production. In 2022, the aquaculture sector surpassed 50% of the country’s overall fishery production [19,20].

Based on the above data, and although the vulnerability of such ecosystems to pesticide prevalence is well recognized, comprehensive research on these contaminants within these regions remains insufficient. Additionally, there is a notable gap in the literature regarding the application of advanced analytical techniques, including liquid chromatography in conjunction with high-resolution mass spectrometry (HRMS) to accurately identify and quantify these substances [6].

An essential component of analytical method validation is the comprehensive evaluation of measurement uncertainty, which plays a crucial role in ensuring the reliability of results [21,22,23,24]. This procedure is important for the analytical workflow because measurement uncertainty estimation facilitates the accurate interpretation of results [22,23]. Consequently, regulatory decision-making in environmental monitoring campaigns will improve dramatically due to the 95% probability of the expected concentration levels [22,23].

Furthermore, the integration of solid-phase extraction (SPE) with high-resolution technologies, such as UHPLC–Orbitrap HRMS, offers several advantages for the detection of low concentrations of pesticides in environmental water samples [25,26,27,28]. SPE can achieve high preconcentration factors and sample purification, thereby enhancing the sensitivity of the method for detecting such compounds [25,28]. This capability is particularly important for the analysis of complex environmental matrices, such as aquaculture water, where the presence of co-eluted compounds can affect the chromatographic profile of the analytes [25,26,29]. Simultaneously, UHPLC-Orbitrap HRMS provides unparalleled resolution and precision, enabling rapid and consistent sample analysis [6,30]. Consequently, the combination of these technologies ensures exceptional method detection and quantification levels, rendering this approach ideal for monitoring environmental contaminants in sensitive ecosystems.

The overall goal of the present study was to validate a simple, efficient SPE–LC–HRMS-based protocol for the successful and accurate determination of contemporary pesticides in freshwater and seawater and to address the measurement uncertainty. To this end, the specific objectives of the study were as follows: (a) to optimize critical factors involved in sample preparation, including extraction cartridges, elution solvents, and sample pH, to enhance recovery and reduce interference, (b) to establish a robust HRMS method for the accurate quantification of targeted pesticides in water samples, (c) to perform a comprehensive validation according to regulatory guidelines considering a range of pesticide concentrations relevant to environmental monitoring, (d) to quantify measurement uncertainty including all contributing factors, and (e) to ensure the applicability of the method by analyzing a diverse set of water samples and assessing at the same time the prevalence and concentrations of the selected pesticides in the sampling area for the first time.

2. Materials and Methods

2.1. Chemicals, Reagents and Standard Solutions

Methanol (MeOH) and acetonitrile (ACN) were of high purity, suitable for LC–MS analysis, while dichloromethane (DCM) was over 99.5% pure, all supplied by Fisher Scientific (Leicestershire, UK). Acetone with purity greater than 99.5% was obtained from Honeywell (Morris Plains, NJ, USA). Ultra-pure water with conductivity < 0.055 μS/cm, provided by an Evoqua Ultrapure Water System Ultra Clear 20 TWF EDI (Evoqua Water Technologies, Pittsburgh, PA, USA), was used in all extraction steps, except for the preparation of the chromatographic eluent system, where water of LC–MS purity purchased from Fisher Scientific (Leicestershire, UK) was used. Formic acid (FA) and acetic acid (HAc) of 98–100% purity from Merck (Darmstadt, Germany) were used for the adjustment of pH value. Ammonium formate of 98–100% purity, 25% ammonia solution, and sodium hydroxide (NaOH) of 98% purity were also purchased from Merck.

The standard compounds were of high purity (>97%) and were supplied in solid form by Sigma Aldrich (Darmstadt, Germany). The selected pesticides, including herbicides, insecticides, and fungicides, demonstrated mostly medium to high polarity and were thiamethoxam, dimethoate, acetamiprid, thiacloprid, fluometuron, chlorantraniliprole, metalaxyl, diuron, boscalid, myclobutanil, linuron, S-metolachlor, prometryn, tebupirimfos, chlorpyrifos, fluazifop-p-butyl, imazalil, and fenpyroximate. A KERN ACS 80-4N (Balingen, Germany) four-decimal precision analytical balance was used to weigh standard compounds. Eppendorf (Vienna, Austria) automatic adjustable volume pipettes (10–100 μL, 100–1000 μL) and glass microsyringes (5 μL, 10 μL, 50 µL, 100 µL, 250 µL, and 500 µL) from Innovative Labor Systeme (Stützerbach, Germany) were used to collect and transfer different sample volumes.

Stock solutions of each compound were prepared separately at a concentration of 2000 mg L⁻¹ in methanol. The solutions were checked for quality by mass spectrometry, transferred to 8 mL glass-sealed vials, and stored at −20 °C. Diluted standard working solutions of the selected compounds and their mixtures were prepared prior to analysis at various concentrations, with appropriate dilutions in methanol and water/methanol (90:10 v/v).

A Millipore filtration device (Darmstadt, Germany) and Daihan Labtech (Wilmington, NC, USA) magnetic stirrer were used for water filtration and stirring, respectively. A Thermo Fisher Scientific (Rockwood, TN, USA) HyperSep Glass Block 60104-243 extraction device connected to a vacuum pump was used for solid phase extraction (SPE). For the optimization and application of the solid-phase extraction method, Oasis HLB (divinylbenzene/N-vinylpyrrolidone co-polymer cartridges, 200 mg, 6 mL) extraction cartridges from Waters Corporation (Milford, MA, USA) were used. For preconcentration of the extracts, a Techne Dri-Block DB-3D (Staffordshire, UK) heating evaporator under a gentle stream of nitrogen was used. In all cases, prior to injection into the LC–MS system, the samples were filtered through 0.22 μm pore diameter polytetrafluoroethylene (PTFE)-filled syringe filters (Millipore, Cork, Ireland).

A Niskin-type sampler was used to collect water samples, and glass fiber filters GF/B, 0.7 mm from Whatman (Maidstone, UK) were used before the extraction. A Consort C932 pH meter (Turnhout, Belgium) and WTW LF 325-B conductometer (Weilheim, Germany) were used to measure the physicochemical characteristics of the water samples.

2.2. Instrumentation

An ultra-high-performance liquid chromatography (UHPLC) coupled with an LTQ/Orbitrap FT mass detector system was used for the separation and determination of the selected pesticide compounds. The system included an autosampler (Accela AS autosampler model 2.1.1), an automatic sample flow pump (Accela quaternary gradient U-HPLC-pump model 1.05.0900), and a hybrid mass analyzer, LTQ-Orbitrap XL 2.5.5 SP1, Thermo Fisher Scientific (Bremen, Germany). The linear ion trap (LTQ) section of the hybrid mass analyzer was equipped with an ion maximum electrospray ionization (ESI) source operated in the positive or negative ion mode. Instrument control and further processing of the accurate mass spectrum (mass range m/z 50–2000) with high resolution (60,000 FWHM) was performed using the Xcalibur software program Xcalibur v. 2.2 of the company Thermo Electron (San Jose, CA, USA). Selected analytes were separated on a SpeedCore PFP (pentafluorophenyl) reversed-phase analytical column (50 mm length × 2.1 mm internal diameter, 2.6 μm particle size) from Fortis Technologies (Cheshire, UK). Mobile phase consisted of (i) water and (ii) methanol, both acidified with 0.1% formic acid.

2.3. Sample Preparation

Oasis HLB extraction cartridges (200 mg, 6 mL) were placed in a solid-phase extraction apparatus connected to a vacuum pump. The activation of the sorbent was performed by the successive addition of 5 mL of methanol and 5 mL of LC–MS water at a flow rate of ~1 mL min⁻¹. Immediately after activation and before the adsorbent material was dried, 250 mL of the aqueous sample was extracted by passing it through the cartridge under vacuum at a flow rate of ~2 mL min⁻¹. At the end of the extraction and before drying, the cartridges were washed with 5 mL ultrapure water and left under vacuum for 30 min to completely remove moisture. This was followed by elution with 2 × 5 mL methanol under vacuum at a flow rate of 2 mL min⁻¹. Afterwards, the eluent was dried under a moderate stream of nitrogen at a temperature of 30 °C and reconstituted in 0.5 mL of 90/10 (v/v) water/methanol with 0.1% formic acid. A PTFE filter μm pore size of 0.22 μm was used to filter the final extract, which was subjected to LC-LTQ/Orbitrap MS analysis.

2.4. Sampling Area and Real Samples

The suggested analytical approach was utilized in a “case study” conducted in Greece and Albania. The samples were obtained from aquatic environments that are affected by aquaculture activity, notably fish farms or mussel farms, in addition to other human activities. In Greece, specific locations were chosen from three regions within the water bodies of Epirus. These regions include the Arachthos catchment area, the Louros watershed where freshwater fish farm units are located, and the Sayada Strip, which is renowned for seawater aquaculture. Within Albania, specific locations were chosen within the Butrint Lagoon near mussel farming facilities.

2.5. Method Validation

Method validation was conducted according to the SANTE guidelines [23]. The acceptability criteria for the method performance included recovery (R%), precision expressed as repeatability RSD_r and within-laboratory reproducibility RSD_wR, linearity (R²), matrix effect (ME%), method detection and quantification limits (MDLs and MQLs), and combined measurement uncertainty (MU%). For the evaluation of %, a calibration curve was prepared in a water extract, and then it was compared to one prepared in the solvent mixture, while a blank extract was used to represent the signal of the non-spiked extract of the matrix. Calculations were based on the following equation (Equation (1)):

$$ME \left(\%\right)=\left(\left({\displaystyle \frac{Slope of calibration curve in matrix}{Slope of calibration curve in solvent}}\right)-1\right)\times 100$$

(1)

2.6. Measurement Uncertainty

The value of the expanded measurement uncertainty (U′) was determined by applying a coverage factor of k = 2 (at a confidence level of 95%) to the combined measurement uncertainty (u′), as indicated in Equation (2). The combined measurement uncertainty (u′) occured as a contributionof the uncertainty component for the bias (u′(bias)) and the uncertainty component for the precision (u′(precision)), as represented by Equation (3).

Residue results do not need to be adjusted for method bias if the mean bias is less than 20%, and the default expanded measurement uncertainty of 50% is not exceeded. If the bias exceeds 20%, it is possible to mathematically adjust the result by employing a recovery factor. In this scenario, the initial result acquired from the analysis of the relevant pesticides was multiplied by the recovery factor, which was calculated as 100% divided by the recovery percentage.

The initial estimation did not account for recovery correction, as outlined in Equation (4), whereas the subsequent estimation integrated recovery correction, as shown in Equation (5). Equation (6) was employed to assess the u′(precision), which was determined by computing the within-laboratory reproducibility RSD_wR of the pesticides.

$${\mathbf{U}}^{\mathbf{\prime }}=\mathbf{k}\times {\mathbf{u}}^{\mathbf{\prime }}$$

(2)

$${\mathbf{u}}^{\mathbf{\prime }}=\sqrt{{{\mathbf{u}}^{\mathbf{\prime }}\left(\mathbf{b}\mathbf{i}\mathbf{a}\mathbf{s}\right)}^2+{{\mathbf{u}}^{\mathbf{\prime }}\left(\mathbf{p}\mathbf{r}\mathbf{e}\mathbf{c}\mathbf{i}\mathbf{s}\mathbf{i}\mathbf{o}\mathbf{n}\right)}^2}$$

(3)

$${\mathbf{u}}^{\mathbf{\prime }}\left(\mathbf{b}\mathbf{i}\mathbf{a}\mathbf{s}\right)=\sqrt{{\mathbf{m}\mathbf{e}\mathbf{a}\mathbf{n}}_{\mathbf{b}\mathbf{i}\mathbf{a}\mathbf{s}}^2+{\mathbf{S}\mathbf{D}.\mathbf{P}}_{\mathbf{b}\mathbf{i}\mathbf{a}\mathbf{s}}^2}$$

(4)

$${\mathbf{u}}^{\mathbf{\prime }}\left(\mathbf{b}\mathbf{i}\mathbf{a}\mathbf{s}\right)={\displaystyle \frac{{\mathbf{R}\mathbf{S}\mathbf{D}}_{\mathbf{w}\mathbf{R}}}{\sqrt{\mathbf{N}}}}$$

(5)

$${{\mathbf{u}}^{\mathbf{\prime }}\left(\mathbf{p}\mathbf{r}\mathbf{e}\mathbf{c}\mathbf{i}\mathbf{s}\mathbf{i}\mathbf{o}\mathbf{n}\right)=\mathbf{R}\mathbf{S}\mathbf{D}}_{\mathbf{w}\mathbf{R}}$$

(6)

The mean bias error is the arithmetic mean of the bias, SD.P_bias denotes the standard deviation of bias in the population, RSD_wR signifies the reproducibility of recovery within the laboratory, and N denotes the number of recovery tests conducted. Typically, the expanded measurement uncertainty (MU) should be lower than the 50% threshold commonly used in the SANTE guidelines.

3. Results and Discussion

The target pesticides were identified as [M + H]⁺ adduct ions, on the condition that the experimental accurate mass matches the theoretical one within a tolerance of ±5 ppm, as well as the retention time of the analytes in the sample align with that in the standards within a ±0.2 min tolerance in the same analytical batch. For additional verification, MS fragmentation was utilized, requiring at least one fragment ion. The UHPLC-Orbitrap MS/MS data are summarized in Table S1. To better correlate mass accuracy with matrix effects, a mixture of all target pesticides at a concentration of 50 μg L⁻¹ in the matrix was injected three times, and the average mass deviation, or mass error (Δ), was calculated. As shown in Table S1, at a resolution of 60,000, the mass error was consistently less than 3 ppm. The fact that accurate mass measurements were unaffected by complex matrix components demonstrates that the LTQ/Orbitrap MS is effective in separating compounds from potential interferences.

3.1. Sample Preparation Optimization

In order to determine the optimal extraction conditions for pesticides in water, three extraction protocols, namely HLB1, HLB2, HLB3, (as shown in Table S2) were evaluated in triplicate (n = 3). In all cases, Oasis HLB extraction cartridges (200 mg, 6 mL) were utilized due to their superior efficacy in extracting polar pesticides [31,32,33], given that they are specifically designed with a hydrophilic-lipophilic balance to ensure optimal results. The efficiency of the three procedures was examined under various extraction conditions.

In the HLB1 and HLB3 protocols, the pH of the sample was not modified prior to extraction. However, in the HLB2, the sample was acidified to achieve a final pH of 2.5–3. The HLB1 and HLB3 protocols involved activation and equilibration of the adsorbent by sequentially adding 5 mL of methanol and 5 mL of LC–MS grade water. In contrast, the HLB2 procedure utilized acidified LC–MS water with a pH range of 2–2.5 for this stage. In all three situations, the volume of the sample was 250 mL, while the flow rate remained constant at approximately 1–2 mL min⁻¹.

Regarding the washing of the cartridges once the full sample volume was loaded and prior to elution, 5 mL of LC–MS grade water was used in all three tests. During the elution process, 2 × 5 mL of methanol (MeOH) was used for HLB1 and HLB2. In HLB3, a mixture of solvents was utilized: 3 mL of dichloromethane, 3 mL of hexane, and 3 mL of acetone, in that specific order, collected in the same tube. The eluate was dried using a moderate stream of nitrogen and then dissolved again in 500 µL of the initial mobile phase (90% H₂O/MeOH and 0.1% formic acid (v/v)).

Figure 1 and Table S3 depict the recovery rates of pesticides obtained by the three tested extraction protocols. For most of the selected pesticides, the recoveries were largely within acceptable ranges for all the three protocols. Namely, yielded recoveries ranged from 61% to 96% for “HLB1”. However, for “HLB2”, the recoveries ranged between 37% and 87%; some of them were calculated below the acceptable limits. Despite the varying pK_a values of the selected pesticides, typical acidic pesticides were not included in the target list; hence, acidified extraction conditions did not favor significantly the obtained recoveries. Recoveries obtained by “HLB3”, which employed a combination of elution solvents, varied between 49% and 95%. HLB2” and “HLB3” methods resulted in a significant proportion of compounds with recoveries that were deemed unacceptable, rendering them inappropriate for a multi-residual analysis.

3.2. Method Validation

For validation purposes, water samples from pristine areas were collected and screened prior to spiking to ensure the absence of target analytes. The MDLs and MQLs, calculated as the concentrations corresponding to signal to noise ratio (S/N) of 3 and 10, respectively, varied from 0.5 ng L⁻¹ to 30 ng L⁻¹ and from 1.7 ng L⁻¹ to 90 ng L⁻¹. Except for fluometuron, diuron, chlorpyrifos, tebupyrimfos, boscalid, and myclobutanil, the quantification limits were significantly lower than 10 ng L⁻¹.

The linearity of the method was verified by constructing a seven-point calibration curve using spiked samples at a concentration range from the lowest quantification limit (MQL) to approximately 100 times the MQL. The curve exhibited a linear trend, and the calculated coefficient of determination (r²) was greater than 0.99, indicating excellent linearity.

Table 1. MDL, MQL, linearity range and correlation (r²) of the method.

Compound	MDL (ng L−1)	MQL (ng L−1)	Linearity Range (ng L−1)	r2
Diuron	3.0	10.0	MQL–1000	0.9997
Fluazifop-p-butyl	2.0	6.0	MQL–1000	0.9987
Fluometuron	30.0	90.0	MQL–1000	0.9987
Linuron	8.0	23.0	MQL–1000	0.9994
Prometryn	0.5	1.7	MQL–500	0.9992
S-metolachlor	0.7	2.3	MQL–500	0.9990
Chlorantraniliprole	1.0	3.3	MQL–1000	0.9989
Chlorpyrifos	8.4	25.6	MQL–1000	0.9991
Dimethoate	2.0	6.0	MQL–500	0.9994
Tebupirimfos	9.0	30.0	MQL–1000	0.9997
Thiacloprid	0.9	2.5	MQL–500	0.9991
Thiamethoxam	3.0	10.0	MQL–1000	0.9970
Boscalid	5.0	16.0	MQL–500	0.9991
Imazalil	2.0	6.0	MQL–500	0.9994
Metalaxyl	0.8	2.3	MQL–500	0.9984
Myclobutanil	4.0	12.0	MQL–1000	0.9987
Acetamiprid	0.9	2.5	MQL–500	0.9997
Fenpyroximate	2.0	6.0	MQL–500	0.9990

provides information on the detection and quantification limits, the linear range, and the correlation coefficient (r²) for the compounds.

Method accuracy was investigated by means of trueness (recovery) and precision [34]. The selected spiking concentration levels for the compounds were 10 ng L⁻¹, 50 ng L⁻¹, and 200 ng L⁻¹. For each concentration level, five samples (n = 5) were spiked and analyzed within the same day to determine method repeatability expressed as RSD_r, while intra-laboratory reproducibility (RSD_wR) was assessed by three successive days of analysis (n = 3 × 5 = 15).

The mean recovery values (n = 5) for the pesticides in river waters varied from 59% (chlorantraniliprole) to 104% (fluazifop-p-butyl) at the low spiking level, from 61% (thiacloprid, boscalid) to 96% (imazalil) at the medium level, and from 68% (dimethoate) to 93% (fluazifop-p-butyl) at the high level. In the case of sea water, the recoveries were marginally elevated, primarily as a result of greater ionic strength and matrix effect [35]. The recoveries ranged from 62% (chlorantraniliprole) to 111% (fluazifop-p-butyl) at the low level, from 60% (thiacloprid) to 95.4% (imazalil) at the medium level, and from 68% (dimethoate) to 97.2% (fluazifop-p-butyl) at the highest level. Figure S1 illustrates the number of pesticides in rivers and sea waters that exhibit recoveries ranging from 70% to 120%.

All pesticides exhibited %RSD_r < 13% in both river and sea waters. The intra-laboratory repeatability, expressed as the relative standard deviation (RSD_wR%), was below 15%. It should be noted that the maximum acceptable threshold for all instances is 20%, and no pesticide exceeded this limit.

Table 2. Recoveries (%R), repeatability (%RSD_r) and intra-laboratory reproducibility (RSD_WR%) of the method in water.

	River Water									Sea Water
Pesticide	10 ng L−1 *			50 ng L−1			200 ng L−1			10 ng L−1 *			50 ng L−1			200 ng L−1
	(%)R	(%)RSDr	(%)RSDR	(%)R	(%)RSDr	(%)RSDR	(%)R	(%)RSDr	(%)RSDR	(%)R	(%)RSDr	(%)RSDR	(%)R	(%)RSDr	(%)RSDR	(%)R	(%)RSDr	(%)RSDR
Acetamiprid	71	7	7.5	65	6.2	9.1	80	2.5	6.2	80	3.2	6.6	65.2	8.3	4.1	82.4	5.1	4.2
Boscalid	-	-	-	61	6.6	7	85	3.5	4.2	-	8.5	7.7	64.8	4.9	3.2	83.8	3.2	12.2
Chlorantraniliprole	59	8.5	6.7	81	2.5	5.1	87	8	11.5	62	10.2	7.2	82	12.2	6.6	86	8.3	2.9
Chlorpyrifos	-	-	-	78	5.1	8.3	75	8.7	14.9	-	6.2	6.2	79	11.5	12.2	76	6.2	4.9
Dimethoate	61	9.8	11.2	78	1.3	6.2	68	4.4	3.2	64	2.5	4.2	74.8	5.1	8.8	68	4.2	5.3
Diuron	69	7.2	8.1	73.1	4.1	5.2	79	2.4	2.9	71	3.2	6.2	72.9	2.4	8.5	79.8	7	8.5
Fenpyroximate	81	9.9	11.3	74.2	9.4	11	78	7.7	4.2	84	6.7	6.7	76.8	9	3.2	81	7	8.5
Fluazifop-p-butyl	104	9.6	9.5	82	3.2	6.1	93	6.5	8.6	111	4.4	4.9	84.3	4.2	6	97.2	8.5	8.3
Fluometuron	-	-	-	69	2.9	9	82	3.7	6.7	-	8.8	4.1	70.4	5.1	5.7	83.8	6.4	6.2
Imazalil	75	6.7	7.7	96	3.1	6.7	81	2.5	8.5	78	8.3	8.6	95.4	3.2	6.6	84	6	8.6
Linuron	-	-	-	87	1.1	5.2	80	1.3	4.9	-	2.5	3.2	88.7	4.2	6.1	86	6.2	8.2
Metalaxyl	84	2.4	8.4	75	1.3	2.4	82	12.2	7.7	86	4.2	6.7	76.8	4.2	3.1	82.9	5.7	8.8
Myclobutanil	-	-	-	73.1	8.3	12.1	78	11.5	8.4	-	8.5	3.2	74.9	5.1	6.2	78.8	4.9	5.1
Prometryn	78	5.1	7.7	78	2.6	6.2	86	2.3	3.2	76	4.9	5.1	80	8.5	9.9	89.1	4.2	11.5
S-metolachlor	72	4.2	8.6	69	2.9	4.2	83	2.4	12.2	73	3.2	1.3	70.9	3.2	4.4	87.5	8.5	11.9
Tebupirimfos	-		-	83	7.2	4.2	80	8.8	6.6	-	4.9	3.2	81.2	4.2	2.5	80	8.8	3.5
Thiacloprid	69	1.4	9.9	61	4.9	8.5	79	2.5	12.2	70.2	3.2	6.6	60	8.5	5.3	81	2.5	2.5
Thiamethoxam	77	3.9	4.4	62.5	3.2	4.9	75	5.3	11.5	75.9	4.2	9.2	63.7	12.2	8.8	77.2	4.9	6.6

* Compounds with MQL higher than 10 ng L⁻¹ are noted with (-).

presents the method’s recoveries, repeatability, and reproducibility for these compounds at three spiking levels.

The use of matrix matched calibration curves alongside isotopically labeled internal standards (if available) effectively reduces calculation errors caused by ionization [29]. The matrix effect was investigated in marine waters, given its heightened significance resulting from the elevated salinity levels. Figure 2 depicts the matrix effect in the presence of pesticides. The matrix effect generally exhibited negative values (<−20%), indicating moderate signal suppression for all studied pesticides while a stronger signal suppression was calculated for dimethoate (31% signal reduction).

3.3. Measurement Uncertainty

The expanded measurement uncertainty (MU%) for each pesticide compound was evaluated at three different concentration levels (10, 50, and 200 ng L⁻¹), yielding results below 50%, which indicates a high level of agreement with the SANTE requirements, as shown in

Table 3. Expanded measurement uncertainty (%MU) * for three spiking levels in river and sea water.

	River Water			Sea Water
	10 ng L−1 **	50 ng L−1	200 ng L−1	10 ng L−1	50 ng L−1	200 ng L−1
Acetamiprid	8.2	10.0	42.2	42.6	4.5	37.6
Boscalid	-	7.7	31.9	-	3.5	41.1
Chlorantraniliprole	7.3	39.7	38.2	7.9	45.4	33.1
Chlorpyrifos	-	48.1	16.3	-	13.4	5.4
Dimethoate	12.3	45.8	3.5	4.6	9.6	5.8
Diuron	8.9	5.7	42.7	6.8	9.3	46.0
Fenpyroximate	48.4	12.0	47.4	37.2	3.5	43.9
Fluazifop-p-butyl	28.2	38.5	25.7	25.6	34.6	24.4
Fluometuron	-	9.9	39.1	-	6.2	37.0
Imazalil	8.4	16.8	41.9	9.4	17.3	38.3
Linuron	-	28.1	41.3	-	27.0	34.7
Metalaxyl	36.5	2.6	46.1	32.2	47.6	40.1
Myclobutanil	-	13.3	9.2	-	6.8	44.7
Prometryn	47.7	46.0	29.1	5.6	47.8	32.8
S-metolachlor	9.4	4.6	42.1	1.4	4.8	38.5
Tebupirimfos	-	37.9	45.7	-	38.8	44.3
Thiacloprid	10.8	9.3	48.8	7.2	5.8	38.7
Thiamethoxam	47.5	5.4	12.6	10.1	9.6	48.5

* values in bold refer to the cases where recovery correction was applied; ** compounds with MQL higher than 10 ng L⁻¹ are noted with (-)

.

3.4. Investigation of the Selected Pesticides in River Water and Sea Water

Figure S2 provides the area and position of the sampling points, excluding those covered by a confidentiality agreement. Table S4 lists the primary physicochemical properties of the samples. The measurements for the water samples were conducted promptly after sampling, at the location of the collection.

Out of the 18 pesticides, 9 (50%) were found in 92% of the water samples (11 out of 12). Only two compounds were found at a concentration level higher than the quantification limit: imazalil and fenpyroximate. The compounds diuron, prometryn, S-metolachlor, tebupirimfos, metalaxyl, myclobutanil, and acetamiprid demonstrated occasionally some weak peaks (<MQL). However, dimethoate, thiacloprid, fluometuron, linuron, chlorpyrifos, and fluazifop-p-butyl were not detected in any of the samples.

The most frequently detected compounds at all sampling points were fenpyroximate (58%), myclobutanil (50%), and S-metolachlor (33%). Figure 3a displays positive detections in a descending order of rank. In terms of the categories of pesticide compounds, fungicides were the most detected pesticides (41%), followed by acaricides (25.8%), herbicides (24%) and insecticides (7%) (Figure 3b).

Regarding the presence of multiple pesticides in a water sample, Figure 4 demonstrates that at least one compound was detected in 91.6% of the samples. More specifically, in 16.7% of the samples, only one pesticide was detected. Additionally, in 75.0% of the samples, multiple compounds were identified (more than one), whereas 16.7% of the samples tested positive for four compounds, signifying the sampling locations with the highest pollution levels (R2 and R4). Notably, only 8.3% of the samples did not contain pesticides at all. Among the detected pesticides, the highest quantified concentration levels were reported for fenpyroximate at 77.0 ng/L ± 9.3 mg/kg (k = 2; 95%), as shown in Table S5.

Table S5 provides the cases of positive detections for each water system, specifically for the compounds that were detected, while Figure 4 depicts the contribution of pesticide compounds to the pollution load of each sampling point. Based on Table S5 and Figure 4, it can be observed that fenpyroximate was found in the water bodies of all three examined systems in Greece. Subsequently, myclobutanil was found in river waters, but its presence was not observed near fish farms. In Albania, S-metolachlor and metalaxyl were ubiquitous, yet below the limit of quantification in the three sampling stations, while imazalil was quantified to concentrations from 11.2 to 31.2 ng L⁻¹.

4. Conclusions

This study successfully established and validated an analytical method utilizing liquid chromatography coupled with high-resolution mass spectrometry for the quantification of 18 pesticides in water bodies adjacent to agricultural and aquaculture regions in Greece and Albania. The efficient method showed method quantification limits ranging from 1.7 ng L⁻¹ to 90 ng L⁻¹; recovery rates ranged from 61 to 96% in the river water for medium spiking level, while for seawater, they were slightly higher (60–111%) due to increased ionic strength. In all cases, the repeatability and intra-laboratory reproducibility was less than 15%, while linearity was excellent in all cases. Matrix effect study in seawater revealed strong signal suppression (<−20%) only for dimethoate. Relatively low concentrations were measured in the real samples. The most frequently detected compounds at all sampling points were fenpyroximate (58%), myclobutanil (50%), and S-metolachlor (33%).

By addressing a significant gap in existing literature, this study contributes novel data regarding the occurrence and concentration of these agrochemicals, thereby providing critical insights into the impact of agrochemical pollution on vulnerable aquatic ecosystems. Although the application of this method is exemplified through this specific case study, its broader applicability makes it a valuable model for pollutant monitoring in other regions that face comparable environmental challenges. These findings underscore the necessity for ongoing surveillance and assessment of pesticide residues to safeguard ecosystem health and to inform regulatory frameworks.