Multiresidue analysis of pesticides in three Indian soils: method development and validation using gas chromatography tandem mass spectrometry

Abstract The paper reports a multiresidue method that was validated on 220 multi-class pesticides in three major Indian soils, namely, (i) new alluvial soil (NAS); (ii) red lateritic soil (RS) and (iii) black soil (BS) from three different regions. An ethyl acetate-based extraction method with a freezing-out cleanup step was employed for sample preparation, followed by gas chromatography-tandem mass spectrometric analysis. The method that was initially optimized on BS worked satisfactorily for the other two soil matrices. At the spiking level of 10 µg/kg (LOQ), the recoveries were satisfactory (within 70–120%) with precision-RSDs, ≤20% (n = 6) for 85, 88.6, and 89% of compounds in BS, RS, and NAS respectively. At 20 µg/kg, the method performance was satisfactory in each soil for all pesticides. When this validated method was applied to analyse 25 field samples, 6 pesticides were detected in them. In each case, precision (RSD) was <20%. The method sensitivity, accuracy and precision complied with the SANTE/2020/12830 guidelines. The method can be applied for environmental monitoring and risk assessment purposes, thus aiding in regulating pesticide usage in agricultural fields. The limitations and future scope of the study are also discussed. Highlights A multiresidue method is reported for simultaneous analysis of multi-class pesticides in diverse soils The method was validated on 220 pesticides in new alluvial, red lateritic and black soils Sample preparation involved extraction with ethyl acetate and cleanup by a freezing step The residues were estimated by gas chromatography tandem mass spectrometry (GC-MS/MS) The method accuracy and precision complied with the EU’s SANTE guidelines


Introduction
Soil, a non-renewable natural component of our ecology, is the foundation of successful agriculture and the primary source of nutrients for the entire human population. The application of pesticides in modern agricultural practices has been unavoidable to increase crop productivity and protect agricultural produces from various diseases and pest infestations. In India, various classes of pesticides including organophosphorus (OP), synthetic pyrethroids (SP) etc. are used in agriculture. Many of these pesticides can accumulate in the soil for long periods of time before being transferred to crops. [1] According to Ashesh et al. [2] though organochlorines (OC) are banned in agriculture, India has the highest level of OC pesticide pollution in the world. Besides, pesticides are likely to get exposed to chemical as well as microbial degradation, producing metabolites of variable toxicological significance. According to Kumari et al. [3] the insecticide pyriproxyfen and six of its metabolites, including 4-hydroxy pyriproxyfen, persisted in soil with half-lives ranging from 2.6 to 30 days. They slowed down the sucrase and dehydrogenase activities and affected the overall soil health. As pesticides are hazardous to human and animal health and also to the ecosystem, [1,4,5] their widespread use has become a global concern with regard to food and environmental safety. Hence, a risk assessment of pesticide residues in soil is of high importance.
As a heterogeneous complex matrix, soil comprises organic (e.g. humus 10-15%, lipids, carbohydrates, lignin, flavonoids, pigments, resins and fulvic acids) and inorganic (e.g. variable percentage of sand, silt and clay) components. [6] In recent years, researchers have revealed that the physico-chemical properties of soil and the nature of pesticides [þ their isomers and toxic metabolites (e.g. tetrahydrophthalimide for captan)] determine the persistence, fate, environmental behavior and pollution risk, once a compound is introduced into the soil. [7][8][9] Pesticides are retained in soil due to their polarity and ionic activities, making residue analysis difficult, and thus appropriate methods for their analysis are warranted.
In India, there is a rising scientific interest to investigate pesticide contamination in soil. For example, Mondal et al. [10] studied the Hooghly River basin in Eastern India, where 31 pesticides were determined by GC-MS/MS. Using a modified QuEChERS workflow, they found non-agricultural OC pesticides in every sample of river water and sediment. In the Cauvery delta region of South India, Menon et al. [11] examined the fate of neonicotinoid pesticide residues in paddy soil and water, and found that neonicotinoids are less persistent in the water-soil systems of the delta region due to photolysis and rapid microbial degradation. Elsewhere, Joseph et al. [12] used the QuEChERS method coupled with the analysis by GC with electron capture detection (ECD) to investigate 17 persistent OC pesticides in Cardamom Hill Reserve soils. They concluded that although the OC residue intakes in this region were low for humans, but the levels in soil were harmful to plants, invertebrates and small birds. In a recent review, [13] the authors commented that in the absence of a strong legal framework and farmers' lack of awareness, inappropriate pesticide uses in India contributed to soil and sediment contamination, as well as health problems for human and aquatic life. For this, there is an interest in industrial and governmental sectors to develop analytical methods to monitor pesticide residues in soil to protect human health and the environment.
Analytical methods for multi-class pesticides based on mass spectrometry (MS) have recently garnered the interest of numerous researchers worldwide, as these techniques are simple and reliable in terms of confirmatory identification and quantification. Over the past three decades, GC-MS has become a standard method for identifying and quantifying the contaminant residues in complex matrices, and its use for multiresidue pesticide analysis has become increasingly prevalent. Previous studies targeted a variety of analytes including OCs and SPs, [14] OCs [15] and OPs, OCs, SPs, carbamates and triazoles. [16] The instruments used included LC-MS [14] and GC coupled with MS, ECD and nitrogenphosphorus detector. [6,17] Different analytical steps including pressurised liquid extraction (PLE) followed by GC-MS/MS [17] and PLE followed by GC-time of flight MS [18] were used. These studies had a limitation of having a long GC run time for analysis. e.g. 41.88 min by Łozowicka et al. [6] and 47 min by Fernandes et al. [19] The non-MS detectors like ECD and NPD also lacked selectivity and did not provide confirmatory identifications. All of these indicate that researchers are continually seeking to develop more improved methods.
Several researchers have studied conventional and modern sample preparation methods such as liquid-solid extraction, [20] PLE, [21] microwave assisted solvent extraction, [22] supercritical fluid extraction [23] and modified QuEChERS [15,24] for the determination of pesticide residues in soils. Almost for two decades, researchers have studied the application of the QuEChERS method developed by Anastassiades et al. [25] For instance Asensio-Ramos, [5] developed a GC-MS/MS method and estimated only a few pesticides (majorly OPs with one thiadiazine) while using the QuEChERS method in three different soils (forestal, ornamental and agricultural, pH 4. 2-5.9). A recent study [6] described a QuEChERS-based method in a type of Polish soil (pH-6.6, organic matter-1.45%, clay content-2.43%) for analysing a wide range of compounds (216 pesticides), although it required a long chromatographic run time of 41.88 min. In another study, Leyva-Morales et al. [17] reported a PLE based method (using dichloromethane:acetone, 50:50, v/v) with the final analysis by multiple techniques, including GC-ECD, GC with a pulsed flame photometric detector and GC with a thermionic specific detector. Thus, the method involved three separate chromatographic analyses, turning the method quite complex and difficult to implement in routine situations. Furthermore, no mass spectrometry-based confirmation was used in it. All of these limitations inspired us to develop a selective but also a time-and cost-effective method.
In India, there are three major types of agricultural soils, namely new alluvial soil (NAS), red lateritic soil (RS) and Western Maharashtra plain zoneblack soil (BS). [26] The average contribution of NAS to India's agricultural output is 45.6%, followed by BS (15%) and RS (10.6%), which make up approximately 71% of India's agricultural land. [27] As information is scarcely available regarding multiresidue analysis of pesticides in NAS, RS and BS, the aim of this study was to optimise and validate a method targeting determination of a wide variety of pesticides in these soils. The performance of the method was verified by determining 220 GC-amenable pesticides from various chemical classes. The combination of an optimised sample preparation procedure with selective estimation of multi-class pesticides by GC-MS/MS in these typical agricultural soils of India adds value to the existing literature in this field.

Chemical reagents and materials
The certified reference standards of pesticides (!98% purity) were purchased from Sigma-Aldrich (Steinheim, Germany) and Dr Ehrenstorfer GmbH (Augsburg, Germany). The HPLC grade ethyl acetate was obtained from J. T. Baker (Center Valley, CA, USA). Acetic acid and anhydrous sodium sulfate were purchased from Merck (Bengaluru, India). The dispersive solid phase extraction (dSPE) sorbents [e.g., primary secondary amine (PSA) and C 18 (end-capped)] were procured from Agilent Technologies (Santa Clara, CA, USA). The polytetrafluoroethylene (PTFE) syringe filters (0.22 mm) were sourced from Chromatopak Analytical Instrumentation Pvt. Ltd. (Mumbai, India). The HPLC grade water was generated using a Sartorius water purification system (Gottingen, Germany).

Selection of target analytes
A total of 220 pesticides (GC amenable) (Appendix Table  A1) from various chemical classes that are widely used in agriculture were chosen. [28] The list included a range of compounds that reveal diverse physico-chemical properties including OPs, SPs, OCs and heterocyclics, as well as their metabolites. [29] Among them, some are reported as soil pollutants, [6] while certain others are banned or of restricteduse in the country, thus necessitating a regular monitoring. [30] Preparation of standards The stock solutions of the individual pesticide standards (1000 mg/mL) were prepared in volumetric flasks (certified class 'A') by dissolving 10 (±0.1) mg of each analyte in 10 mL ethyl acetate The solutions were stored in the dark at À20 ± 2 C. By appropriate dilution, a working standard mixture (10 mg/mL) was prepared by mixing the appropriate quantities of individual solutions with volume makeup. An intermediate standard solution of 1 mg/mL was prepared and stored at À20 ± 2 C in the dark. The stability of the standards (degradation limit, within ±10%) was weekly monitored as per the clause F9 of SANTE/11312/2021 [31] guidelines. The calibration standards (2, 5, 10, 20 and 50 ng/mL) were prepared by serially diluting the intermediate standard with ethyl acetate. For matrix-matched standards of the same concentrations, the blank matrix extracts were used for dilution.

Sample selection, collection and pretreatment
For the purpose of the study, NAS, RS and BS were chosen. These were collected from three different geographic locations: (i) NAS from the University Research Station, Bidhan Chandra Krishi Viswavidyalaya (BCKV) (Mohanpur, Nadia District, West Bengal), (ii) RS from the Regional Research Station, BCKV (Jhargram, Midnapore District, West Bengal) and (iii) black soil (BS) from Pune, Maharashtra State. These soils were collected from three zones: NAS from the new alluvial soil zone, RS from the red and laterite zone and BS from the Western Maharashtra plain zone. The physicochemical properties of these soils are mentioned in Table 1. These soils were preferred over others due to their texturally diverse physicochemical properties and location in prime crop-growing regions. These soil samples were collected from fields that had not been treated with pesticides within the past three years.
The sample collection, preservation and storage were based on the US EPA Method 1699 (https://www.google. com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja &uact=8&ved=2ahUKEwj-5vbK2Kv8AhVy_DgGHcBECxcQ FnoECA0QAw&url=https%3A%2F%2Fwww.epa.gov%2F sites %2Fdefault%2Ffiles%2F2015-10%2Fdocuments%2Fmethod_ 1699_2007.pdf&usg=AOvVaw3QTnfAcQ6NUn3-FN2oHDyb). Following the standard sampling protocol, the samples were collected from the upper layer (between 15 and 20 cm in depth) and air-dried, ground, passed through a 2 mm sieve. They were subsampled by the usual method of quartering and stored in zip-lock bags at À20 C. The samples' physicochemical properties were evaluated: soil texture by the hydrometer method; [31] pH by the soil water suspension method [after adding deionised water in 1:2.5 (g/mL) ratio] and soil organic carbon content by the Walkley and Black wet oxidation method. [32,33] Instrumental parameters for GC-MS/MS analysis The sample analysis was performed using a GC (model GC 2010, Shimadzu Corporation, Kyoto, Japan) equipped with an AOC-20i auto-injector and hyphenated to a triple quadrupole mass spectrometer (TQ8040, Shimadzu Corporation, Kyoto, Japan). Analytical separation of the target analytes was performed using an Rxi V R -5Sil MS (30 m Â 0.25 mm, 0.25-lm film thickness) column from Restek Corporation (Bellefonte, PA, USA). In the splitless mode, 1 lL of volume was injected into a gooseneck liner. At a constant flow rate of 1.2 mL/min, ultrapure grade helium 5.0 (99.999% purity) was used as the carrier gas, while argon was used as the collision gas. The oven temperature was programmed with an initial temperature of 90 C (1 min hold), ramped at 35 C/min to 130 C (0 min hold), then at 10 C/min to 240 C (1 min hold) and finally at 15 C/min up to 290 C (4 min hold). This resulted in a total run time of 21.31 min. The temperatures of the transfer line and the electron ionisation (EI) were kept constant at 290 and 230 C, respectively. The voltage of the detector was set to 0.70 kV. The solvent delay time was 1.5 min. The data acquisition, which started at 2 min, was separately performed in the scan mode (to evaluate matrix effects) and multiple reaction monitoring (MRM) mode. The compound identifications relied on the correct selection of the MRM transitions using the identification criteria specified in the SANTE/11312/2021 [31] guidelines, which included detection of a minimum of 2 product ions and additionally, ion ratio variations (samples to the calibration standards in the same batch of analysis) within the tolerance limit of 30%. To control the instrument, acquire sample batches and process data, the GC-MS/MS LabSolutions V R software (Version 4.45 SU2, Japan) was used.

Sample size optimisation
To determine the effect of the sample size on the sensitivity, accuracy and precision, three different sample sizes (2, 5 and 10 g) were selected. The samples were spiked at 100 mg/kg of the test analytes, mixed with 10 mL of distilled water and extracted with 10 mL of ethyl acetate in separate batches (n ¼ 6).

dSPE cleanup optimisation
As mentioned earlier, the main fraction of the soil matrix is humus, which comprises mainly the organic acids and their conjugate bases. In addition, 20% of humus occurs as lipids. [34,35] Because of their high solubility in ethyl acetate, these lipids might get co-extracted with the target pesticide residues. The dSPE cleanup effect for 1 mL extract was evaluated using the following sorbent combinations: Cleanup 1 (50 mg PSA, 150 mg Na 2 SO 4 ), Cleanup 2 (25 mg PSA, 25 mg C 18 and 150 mg Na 2 SO 4 ) and Cleanup 3 (25 mg PSA, 50 mg C 18 and 150 mg Na 2 SO 4 ). Each tube was vigorously vortexed for 1 min before being centrifuged at 5600 Âg for 5 min at 4 C. The combined effect of the dSPE cleanup and freezing was also evaluated.

Freezing
To optimise the freezing temperature, a portion of the ethyl acetate layer (2 mL) of each tube (corresponding to the optimised sample size) was pipetted out and kept in a 15 mL centrifuge tube at À20 C (for 30 min and 60 min) and À80 C (for 10 min and 20 min). The results were evaluated in terms of recoveries and matrix effects.

Shaking versus sonication
The effects of shaking and sonication were separately evaluated for spiked as well as incurred residues. The samples were kept on a shaker at a speed of 250 times/min for 10, 20 and 30 min. In a parallel experiment, the samples were sonicated for 10, 20 and 30 min. The recovery (%) results were comparatively evaluated. The optimised sample preparation method for the recovery studies For the recovery experiments, a sample (5 g) was placed in a 50 mL polypropylene centrifuge tube. An appropriate volume of standard solution containing the target analytes was added, mixed thoroughly and then vortexed for 30 s. Prior to extraction, the mixture was kept standing for 20 min at room temperature. Afterwards, 10 mL of distilled water was added to hydrate the matrix and hand shaken for 30 s. To it, 10 mL of ethyl acetate was added, followed by 10 g anhydrous Na 2 SO 4 . With no further delay, the resulting mixture was vigorously hand-shaken for 30 s and vortexed for 2 min. The tubes were shaken at the rate of 250 times/min on a mechanical shaker for 20 min, followed by centrifugation at 2800 Âg for 10 min. The clear supernatant (ethyl acetate phase) was taken into a 15 mL centrifuge tube and held at À80 C for 20 min. After the phase separation, 2 mL of the ethyl acetate layer was collected into an Eppendorf tube and then centrifuged at 11200 Âg for 5 min. The extract was filtered through a 0.22 mm polytetrafluoroethylene (PTFE) syringe filter, collected in an autosampler vial and injected into GC-MS/MS for detection and quantification.

Method validation
The method validation parameters such as accuracy, precision (Appendix Table A2), linearity, specificity, ruggedness and the matrix effect (ME) were evaluated as per the criteria described in the SANTE guidelines, SANTE/2020/12830 [36] and SANTE/11312/2021. [37] The criteria of analytical quality control are summarised in a Appendix Note. These analytical performance parameters were evaluated by analysing all soil matrices, each spiked at three different concentrations (10, 20 and 50 mg/kg). Each experiment was performed in six replicates.
The linearity of the calibration curves was studied using the concentrations of 2, 5, 10, 20 and 50 mg/kg. The limit of quantification (LOQ) for each pesticide was determined as the lowest spiked concentration that could be quantified with satisfactory recoveries (70-120%) and precision (an RSD of less than 20%). The accuracy was estimated through recovery experiments in six replicates (spiked at three-levels such as LOQ, 2 Â LOQ and 5 Â LOQ). The results were calculated using matrix-matched calibration standards that were injected at the start and end of each batch. The precision was expressed as within-the-laboratory repeatability in terms of RSD. By involving three different analysts, the recovery experiments were conducted on three different days, indicating intra-laboratory reproducibility.
The matrix effect was estimated by using the following formula: The matrix effect (ME) was evaluated as a percentage of signal suppression (À) or enhancement (þ). The ME was determined by comparing the peak areas of the matrixmatched standards (peak areas of the post-extraction spike) with the corresponding peak areas of standards in solvent at 0.02 mg/mL. According to Ferrer et al. [38] the targeted pesticides could be classified into three categories of ME as follows: soft (< ±20%), medium (±20-50%) and strong (more than ±50%).
The method's accuracy and precision were checked at the LOQ and higher levels of fortification as per the clause G3 of the SANTE/11312/2021 [31] guideline. In all cases, six separate analyses were performed. The recoveries and RSDs (precision) were estimated from the results. The precision for incurred residues in farm soils was also evaluated in the same way. The ruggedness of the method was confirmed by analysing a quality control (QC) sample (prepared by spiking each of the blank matrices at 50 mg/kg and processing by using the optimised method) in every batch.

Field sample analysis
The method's performance was evaluated on 25 field samples collected from agricultural farms in Maharashtra and West Bengal states of India (Appendix Tables A3 and A4). Each sample (2 kg) was mixed thoroughly and analysed by the optimised method. The analysis of each positive sample was repeated six times. For these, six random portions (5 g each) were drawn from each positive sample and analysed separately by the optimised method. Each GC-MS/MS batch included the matrix blanks (soil extracts with no pesticides), spiked samples and matrix-matched calibration standards (2, 5, 10, 20 and 50 mg/kg) bracketing the field samples. This was as per the clause C5 of SANTE/11312/2021 [31] guidelines.

Sample size optimisation
Almost all of the analytes provided an RSD of <20% for a 2 g sample. The ME for 85% of the test analytes was less than 20%. But, when using a 2 g sample size, we noted that the residues got diluted by five times, and confirmatory/quantitative analysis of incurred residues at low-levels (10-20 mg/kg) was difficult. This limitation was solved when a 5 g sample size was chosen. For sample sizes of 5 and 10 g, the RSDs were satisfactory (<20%) for most analytes. But, when compared to a 5 g sample size, the ME for approximately 22% of the compounds (e.g. dichlorvos, atrazine, fenitrothion, profenofos, propargite, p,p 0 -DDT and pretilachlor) was higher for a 10 g sample. As a result, 5 g was chosen to be the best sample size for further study. Figure  1(a,b) presents the results of the sample size optimisation for certain representative (early, mid and late eluting) pesticides from different chemical classes.
Sample preparation method (optimisation of cleanup for the 5 g sample size) On direct analysis of the sample extract, 13% of the analytes showed recoveries higher than 130%. A significant ME (>50%), was noted for 26% of the analytes. This indicated that removal of interfering co-extractives from the matrices was necessary to minimise MEs. Freezing-out was an effective way to clean all matrix extracts. By putting the extracts at À80 C for 20 min, a large extent of matrix-derived compounds was reduced as reflected in a cleaner GC-MS full scan spectrum (Appendix Figure A1). In our earlier studies, freezing at À80 C provided satisfactory cleanup effects in palm oil, [39] and also in sesame seeds. [40] The cleanup step's effectiveness was demonstrated by the satisfactory recovery of 86.7% of tested pesticides. Due to higher matrix effects, the ion ratio (the qualitative to quantitative MRMs, %) deviations (from the matrix-matched standards) of some compounds (e.g. dichlorvos and parathion) did not comply with the acceptance criteria (<30%), resulting in their false negative detection. Such false detection issues were resolved by performing the freezing step (Appendix Figure A3(a,b)). Additionally, the peak shape of several compounds (e.g. chlorothalonil, oxyfluorfen, chlorfenapyr and fipronil) was improved with the freezing treatment (Appendix Figure  A4(a-d)).

dSPE cleanup optimisation
The Cleanup 1 strategy included only PSA, which increased the pH of the BS extract to 7.8. Satisfactory results were obtained for only 71% of the target analytes. The RSDs were mostly <20%. A lower recovery for a few herbicides (e.g. fluzifop-p-butyl [57% (±9%)], clomazone [61% (±7%)] and propanil [57% (±10%)]) might be due to their reaction with the dSPE sorbents, also observed in previous studies. [41,42] Hence, the amount of PSA was reduced and combined with C 18 for Cleanup 2. This provided satisfactory results for 84.8% of the analytes. The results for Cleanup 2 and Cleanup 3 were comparable. As Cleanup 3 (with an increased amount of C 18 ) had no additional effect, Cleanup 2 was chosen as the most cost-effective and the optimum sorbent combination for the dSPE cleanup.

Combined effect of freezing and dSPE cleanup
When the effectiveness of Cleanup 2 was evaluated, $83.3% of the analytes demonstrated a recovery in the range of 70-120%. A significant number of compounds, for example, chlorfenapyr, tetrachlorvinphos, triazophos and tricyclazole, suffered from poor recoveries (less than 30%) when the dSPE sorbents were used for cleanup. The results showed that freezing at À80 C for 20 min (without a dSPE cleanup) recovered the highest number of compounds (Figure 2(a)) within the acceptable range of 70-120%. The recovery data of some compounds including allidochlor, dichlobenil, chlorthalonil, biphenyl, ethion, mevinphos, paclobutrazole, phorate, profenofos, tau-fluvalinate, trans-chlordane, tricyclazole and 2,3,5,6 tetrachloro aniline supporting this fact are presented in Figure 2(b). Freezing outperformed dSPE because it achieved roughly the same recovery ($85 versus $83% of the compounds) and RSD while being less expensive in terms of sorbents (not required) and manual labor (no weighing of sorbents). Freezing in conjunction with dSPE had minor effect on the recoveries when compared to freezing alone. This aligned with the findings of Rutkowska et al. [43] who reported similar results while analysing pesticide residues in dry herb matrices by GC-MS/MS. In LC-MS/MS analysis also, Zaidon et al. [44] and Caldas et al. [45] reported satisfactory multiresidue analysis of pesticides in paddy soil using QuEChERS extraction. In another study, Zhou et al. [46] noted a satisfactory recovery for the herbicide, pyrazosulfuran ethyl in certain Chinese soils, where the sample preparation workflow involved acetonitrile extraction followed by direct HPLC analysis without any dSPE cleanup. Thus, the findings of this study mark a significant improvement in the scope and performance of soil analysis for pesticide residues.

Shaking versus sonication
When compared to sonication, shaking provided a better recovery of the compounds. As shown in Figure 3, 20 min of shaking time offered 10-15% higher recoveries in comparison to 10 min of shaking. But, when the shaking time was increased from 20 to 30 min, no significant changes in recoveries were observed. Thus, a shaking time of 20 min was chosen.

Method validation
GC-MS/MS-multiresidue analysis of pesticides  Tables 2 and 3. Different parameters including linearity (expressed as R 2 ), LOQ, ME, accuracy (expressed as recovery), precision and ruggedness (expressed as the RSD) were used to validate this optimised method. The specificity of the method was determined by analysing the blank samples of BS, RS and NAS. At the retention time of the targeted pesticides, there was no background peak of the targeted MRMs above a signal-to-noise ratio of 3. A good linearity (R 2 ! 0.99) was obtained in the calibration curve for all compounds over a concentration range of 2-50 mg/kg. A slightly lower R 2 value ($0.98) was noted for fipronil (BS), hexachlorobenzene (NAS), cypermethrin (RS) and permethrin (RS). As shown in Appendix Figure A2, a considerable ME was primarily observed in BS for phenylureas (e.g. linuron), aromatic ethers (e.g. oxyfluorfen), polychlorobenzene (e.g. hexachlorobenzene), pentachloronitrobenzene (quintozene) and dinitrile (e.g. chlorothalonil) groups, the mid-eluting OPs (e.g. phorate sulfide, tolclofos methyl, chlorfenvinphos etc.) and late-eluting synthetic pyrethroids (e.g. cypermethrin).
According to the findings, 69% of the compounds had a lower ME (within ±20%) in BS, while 59 pesticides showed a medium ME (ranging from 21 to 49%). Among all, only 9 pesticides including tolcofos methyl, phorate sulfide, linuron, 4th peak of cypermethrin (eluting at 19.762 min), delta-HCH, endosulfan sulfate, etridazole, oxyfluorfen and parathion showed a stronger ME of above-50%. In RS, tolcofos methyl showed the highest ME value of 78.4%. In this soil, 83% of pesticides showed a lower ME (less than 70%), out of which 33% had ME values within ±21-49%. Four  Bold terms were recovered satisfactorily using the method.
pesticides, namely alpha-endosulfan, chlorothalonil, fenvalerate and tatradifon exhibited signal enhancements of above 50%. urovi c-Pej cev et al. [47] mentioned similar problem for chlorothalonil by QuEChERS. A signal suppression was observed only for (E)-chlorfenvinfos in BS as well as RS. In NAS, 90% compounds showed a nominal ME. A moderate ME was observed for 17 compounds, but a higher ME was noted for chlorpyrifos, alpha-endosulfan and oxyfluorfen. Appendix Figure A4 depicts it for selected pesticides. The ME was lowered by freezing in the study. A considerable ME of the same or different compounds was also previously observed in soil matrices. For instance, Fernandes et al. [19] reported high MEs for 12 pesticides (out of 36 target compounds), which included alpha-and beta-HCH, hexachlorobenzene, o,p-DDT, bupirimate, chlorpyrifos, fludioxonil, malathion, methiocarb and pendimethalin. Similarly, Asensio-Ramoset et al. [5] reported a significant ME for buprofezin, chlorpyrifos, chlorpyrifos-methyl, diazinon, dimethoate, ethoprofos, fenitrothion, malaoxon, malathion and phosmet. According to Łozowicka et al. [6] positive MEs (signal enhancements) were reported for alpha-endosulfan, beta-endosulfan, endosulfan sulfate, dichlorvos, methamidophos, oxyflurofen and oxamyl, whereas negative MEs (signal suppressions) were observed in the cases of bupirimate, dichlobenil, etaconazole, propham and trifloxystrobin. The use of matrix-matched calibration graphs (soil typespecific) brought the recoveries (accuracies) within the acceptable range of 70-120% in every case. Therefore, isotopically-labelled internal standards were not required for any of the compounds.

Method performance
At the lowest fortification level of 10 mg/kg, 85% of the analytes demonstrated satisfactory method performance (recoveries ranged between 70 and 120% with a < 20% RSD). At 10 mg/kg, relatively poor recoveries (<70%) were recorded for p,p 0 -DDT, p,p 0 -DDD, p,p 0 -DDE, o,p 0 -DDE, o,p 0 -DDT, hexachloro-exo-epoxide, cypermethrin-1, aldrin, transfluthrin, trans-nonachlor and fluchloralin. Since precision in these results was satisfactory (RSD, <10%) the method performance complied with the acceptance criteria of SANTE/11312/2021 [31] (clause G6) guidelines. However at 20 mg/kg, all of them exhibited satisfactory recovery (70-120%) and precision results. Compounds such as allethrin, bioallethrin, cycloate, endrin, flucythrinate, paraoxon-methyl, fenothrin and tetramethrin were not detected in BS at any of the three spiking levels. Only a handful of compounds, for instance, fenitrothion, propoxur, tau-fluvalinate and azinophos ethyl, showed recoveries of more than 120%, although their RSDs were <10%. At the fortification levels of 20 and 50 mg/kg, all analytes complied with the method performance criteria (Tables 2 and 3). Thus, the method LOQ was less than 0.05 mg/kg (50 mg/kg) in all cases, in compliance with the SANTE's Guidance Document SANTE/2020/12830. [36] Since maximum residue limits (MRL) are only applicable to raw agricultural commodities and not available for soil, it was not possible to compare the LOQ values with any MRLs.
Certain compounds including aldrin, tolyfluanid, dichlofluanid, dieldrin, fipronil, propargite, profenofos, chlorothalonil, p,p'-DDT, lambda-cyhalothrin, pyriproxyfen, diazinon, oxadiazon and edifenphos were recovered satisfactorily using the method (highlighted in bold, Tables 2 and 3). In contrast, these compounds were recovered poorly in a previous study by, [48] In a similar vein, a poor recovery was noted for these compounds in our earlier study on animal feed. [49] The results here, however, showed non-detection of allethrin, flucythrinate, phenothrin, tetramethrin, cycloate, endrin and paraoxon methyl in BS. Some compounds did not meet the identification criteria of the ion ratio matching (within ±30%), especially at lower concentrations ( 10 mg/kg), where isobaric interferences from co-eluting matrix compounds could have altered the value and resulted in nondetection. Similar observations were also mentioned in Walorczyk et al. 2012 (Figure 4(a)). Another reason for false negatives in the case of some compounds (e.g. paraoxon methyl, cycloate and endrin) might be because of their recovery loss in the sample preparation stage. For cycloate (a thiocarbamate ester), alkaline hydrolysis might be the reason of degradation of residues and relatively poor recoveries in BS. [50] Whereas, in case of paraoxon methyl, the degradation of residues could have been due to the oxidation as a result of the interactions with the mixed mineral oxides present in soil matrices, also suggested by Zimmermann et al. [51] As mentioned earlier, the method was at first optimised in BS before its performance was evaluated in the remaining two soils. Figure 4(b,c) illustrate the compounds that went undetected in NAS and RS respectively. For example, boromophos ethyl, chlorobenzilate, propanil and nitrofen were not detected in NAS, while bromofenvinphos, cis-permethrin, etofenprox, phenothrin and tetramethrin remained undetected in RS. In the NAS matrix, 89% of the tested pesticides satisfied the method validation criteria with the associated recoveries within 70-120% and RSDs 20% at the LOQ of 10 mg/kg. Among all, only prochloraz and ethalfluralin suffered from a poor recovery of <60%, with an associated RSD 20%. However, their recoveries were satisfactory at 20 mg/kg level.
At both spiking levels (10 and 20 mg/kg), the recoveries of the pesticides in RS were satisfactory (70-120%) except for beta-endosulfan (an OC) and EPN (a phosphonic ester). At 10 mg/kg, in RS, 88.63% of the target pesticides satisfied the method validation criteria. The recoveries were in the range of 70-120% for the majority of pesticide-matrix combinations. In all the tested soil matrices (BS, RS and NAS), a similar group of problematic compounds was observed with recovery values ranging from 60 to 130% with a consistently low RSD of 20% (Tables 2 and 3, highlighted in bold, Appendix Figure A5).

Comparison with the other methods described in bibliography
While 220 pesticides were analysed in three distinct Indian soil types in the current study, [6] another research reported a modified QuEChERS method for the analysis of 216 pesticides in a type of Polish soil. Other reports targeted a lesser number of pesticides (e.g. 123 by Tulve et al. [52] 58 by Yu et al. [16] and 34 by Nascimento et al. [53] ).
In India, over the past 15 years, the ethyl acetate-based method has been widely used in pesticide residue analysis, particularly in fruits and vegetables. [49,54,55] Compared to the QuEChERS method, the ethyl acetate extraction provided superior recoveries for the traditionally low-recovery compounds, e.g. OCs. On acetonitrile extraction (as performed in the QuEChERS method), the results pertaining to chlorobenzenes (RSD, 26-41%), chlorophenols (RSD, 21-35%) and HCH isomers (RSDs for alpha, beta, delta and gamma were 25, 31, 37 and 29% respectively) were less repeatable. These observations are also supported by an earlier study. [15] However, the ethyl acetate-based method here showed satisfactory and consistent recovery of the abovementioned compounds with RSDs of 20% (Tables 2  and 3) in all three soil types studied here. The absence of a dSPE cleanup step might have resulted in better recoveries in the study. This also establishes a significant advantage over the earlier methods (e.g. Łozowicka et al. [6] which always required a dSPE cleanup step for successful residue analyses performance. The chromatographic runtime of the current method was also significantly less (21.31 min) than the methods reported by Łozowicka et al. [6] (41.88 min) and Alam et al. (40.33 min). [56] Furthermore, when compared to the QuEChERS method, the ethyl acetate extraction here provided superior recoveries for many compounds. When some compounds were extracted with acetonitrile, the results pertaining to chlorobenzenes (RSD, 26-41%), chlorophenols (RSD, 21-35%) and HCH isomers (RSDs for alpha, beta, delta and gamma were 25, 31, 37 and 29% respectively) were less repeatable. An earlier study by Rouvi ere et al. [15] corroborated these findings. However, our method demonstrated satisfactory and consistent recovery of the compounds in all three soil types studied, with RSDs of 20% (Tables 2 and 3). Compared to Łozowicka et al. [6] our method thus yielded better recoveries and a higher throughput.

Method verification
When calculating against soil type-specific calibration curves, agreeable accuracies in results were obtained. To verify the robustness and effectiveness of the developed analytical method, the accuracy and precision were confirmed by an intra-laboratory study. The repeatability (intra-day precision) and reproducibility (inter-day precision) results are presented in Appendix Table A2. The RSD values were <20% for all the target analytes. These results fulfilled the regulatory compliance as per the SANTE guidelines. Additionally, 30 continuous injections of the matrixmatched standards at 10 and 50 mg/kg provided RSD values of less than 7% for all tested compounds, indicating good instrument repeatability (Appendix Table A2). The method was thoroughly validated in terms of confirmatory identification and quantification. Overall, the method demonstrated  Table 3. Method validation data in three soils at the spiking level of 10, 20 and 50 ng/g.

Black soil
New alluvial soil

Applications in field samples
In the farm soil samples, six different pesticides such as dichlorvos, biphenyl, diphenylamine, anthraquinone, tricylazole and chlorpyrifos were detected (Appendix Table A3). Shaking provided 15-20% higher residue recoveries of these incurred pesticides. The repeat extraction (the residual mass extracted again) did not result in any additional recovery of residues. This indicates a statisfactory extractability of the method. In all cases, the RSDs were low (<10%, n ¼ 6), establishing a satisfactory method precision.

Method throughput and cost
Four batches of extracted samples (48 number, involving 2 analysts) could be acquired through autosampler in a 24hour cycle of GC-MS/MS analysis. This throughput was almost double that of the method reported by Łozowicka et al. [6] as their GC runtime was almost 2-times longer. The average working day (for sample preparation) was 8-h long, but the use of autosampler made it possible to continue with GC-MS/MS injections round the clock, even beyond regular business hours. The method also appeared to be cost-effective. The consumable cost was around INR 650 ($8 USD) per sample, which is cheaper than most methods. For example, the method of Łozowicka et al. [6] required a dSPE cleanup, incurring an additional cost for sorbents and also a longer analysis time.

Conclusions
In this study, a simple, high throughput and cost-effective multiresidue method for the GC-MS/MS determination of multi-class pesticides has been reported, which was evaluated in three major Indian soil types for the first time. As demonstrated, the method performed satisfactorily in all soil matrices. The optimised procedure yielded clean extracts without the need for a dSPE cleanup, thereby reducing the consumable costs. The majority of the target compounds met the validation criteria, with recoveries ranging from 70 to 120% and RSDs 20% at LOQ and higher levels. Additionally, the chromatographic technique produced welldefined and well-shaped peaks in a shorter run time than previous reports. The sensitivity, accuracy and precision of the method complied with the SANTE's guidelines. Based on the successful performance validation, the method can be effectively used for the residue monitoring of pesticides in soil matrices across agricultural fields, assisting in pesticide residue management. The developed method could successfully analyse agricultural field samples, establishing its satisfactory extractability and precision in determining pesticide residues.