Evaluation of the Health Risk and Distribution Characteristics of Pesticides in Shallow Groundwater, South Korea

: In this work, a method of simultaneously analyzing pesticide concentration and assessing its risks was developed. Assessments were conducted to evaluate the distribution characteristics and risks to human health of pesticides in shallow groundwater in agricultural areas. We developed multi-residue analytical methods using liquid chromatography (LC-MS/MS) and gas chromatography–tandem mass spectrometry (GC-MS/MS) to analyze 57 pesticides in groundwater. In addition, risk assessments were performed by setting scenarios considering the routes of pesticide infiltration into groundwater. For the simultaneous analysis of 57 pesticides, the liquid–liquid extraction method was applied twice using dichloromethane under acidic and alkaline conditions. The extract was concentrated and analyzed using LC-MS/MS (41 pesticides) and GC-MS/MS (16 pesticides). The precision and accuracy ranges of the analytical methods were 0.1~12.9% (within ± 15%) and 80.3~113.6% (within ± 20%), respectively. The limit of quantification was found to range from 0.0004 to 0.0677 µ g/L. In total, 57 pesticides were monitored in 200 groundwater wells from 2019 to 2020. Twenty-six pesticides, including metolachlor and imidacloprid, were detected, with an average concentration of 0.0008 µ g/L in groundwater. The pesticide types and detection levels differed depending on the survey period and surrounding land. When the risks associated with alachlor, metolachlor, and carbofuran were assessed, their health risks when found in groundwater were evaluated to be negligible (non-carcinogenic risk: less than 10 − 3 , carcinogenic risk: less than 10 − 6 ).


Introduction
In South Korea, groundwater has been developed and used mostly for residential, industrial, and agricultural purposes.It has high utilization potential in the future as only approximately 22% of its development capacity has been used: 2.98 billion m 3 of groundwater has been used per year from the 1.69 million groundwater utilization facilities.The results of groundwater quality evaluations over the last ten years (2011~2020) show that the quality is relatively good; the non-conformity rate is 2.8% (1.1~5.1%) on average.However, concerns regarding the contamination of shallow groundwater due to agricultural activities (pesticides and livestock manure) have been raised.Since two organophosphorus pesticides (diazinon 20 µg/L and parathion 60 µg/L) have been set and managed as nondrinking groundwater reference items in South Korea, groundwater quality monitoring for more diverse pesticides in groundwater is needed (Ministry of Environment (ME), 2022) [1].
The European Environment Agency (EEA) has set an upper limit of 0.1 µg/L for a single pesticide and 0.5 µg/L as the limit for total pesticide concentrations in groundwater [2].Recently, the EEA published pesticide status survey results for surface water (29 countries) and groundwater (22 countries) for 30 EU member countries for samples taken between 2013 and 2020 [3].Imidacloprid and metazachlor were mainly detected in surface water, whereas atrazine and its metabolites were mainly found in groundwater.Among all survey points, the proportion of points at which 0.1 µg/L was exceeded for at least one pesticide was reported as 4~11% for groundwater.In addition, according to ETC/ICM technical paper data [4], groundwater monitoring for 228 pesticides was conducted at 13,863 points for 8 years.
In California, USA, pesticide monitoring and risk assessment for groundwater wells are performed every year by the California EPA Department of Pesticide Regulation, the State Water Resources Control Board, and the United States Geological Survey [5].The state recently tested for 196 pesticides in a total of 5195 groundwater wells [6].One or more pesticides were detected in 476 wells, and a total of 39 pesticides were detected.The detection level was different by pesticide, but the concentration range observed was between 1 and 10 µg/L.The detected pesticides included dibromochloropropane, ethylene dichloride, and simazine, as well as atrazine and its decomposition products (2,4-diamino-6-chloro-s-triazine).In South Korea, the status of commonly used pesticides must also be identified through groundwater quality monitoring for the active management of shallow groundwater in rural areas.
In this study, we selected 57 pesticides based on pesticide usage in South Korea and studied analysis methods that could be used to analyze them simultaneously as much as possible.Next, 57 pesticides were investigated in the groundwater monitoring network of the MOE to study the detection characteristics of pesticides in shallow groundwater.In addition, a human risk assessment was conducted considering the routes of exposure to pesticides in shallow groundwater.

Water Quality Survey Point Selection and Survey Cycles
Two hundred wells in the nationwide non-drinking groundwater monitoring network were tested twice a year from 2019 to 2020.Target points were selected to create a uniform distribution of survey points by region and surrounding pollution sources (e.g., industrial activities, urban areas, and background points).For agricultural areas, wells to be surveyed were selected based on their proximity to rice paddies, fields, and cultivation facilities.

Groundwater Sampling and Field Measurement Item Analysis
Groundwater sampling was performed using the sample collection and preservation method (ES 04130.1e) in the water pollution test standard, and the water temperature, pH, electrical conductivity (EC), oxidation-reduction potential (ORP), and dissolved oxygen (DO) were measured in the field [7].Separate sampling was carried out for the analysis of cations and anions in 125 mL polyethylene bottles after filtration using a membrane filter (0.45 µm, mixed cellulose ester, Advantec, Tokyo, Japan).Cations were stored by adjusting the pH to 2 or lower through the addition of HNO 3 .Pesticide samples were collected in 1 L brown glass bottles, and the bottles were sealed to prevent contact with the atmosphere.All samples were stored and transported in a 4 • C icebox.

Methods to Analyze Pesticides in Groundwater (Liquid Chromatography and Gas Chromatography-Tandem Mass Spectrometry)
We filtered 1 L of groundwater through a 0.45 µm PTFE filter (Advantec, Chiyoda-ku, Tokyo, Japan), and 500 mL was accurately obtained in a separating funnel.Then, 1 + 1 HCl was added to adjust the pH to a range of 3~4, 20 g of NaCl was added, and the sample was mixed.Next, 50 mL of DCM was added twice to perform the first extraction.After readjusting the pH to 10 by adding 5 M NaOH to the water layer, 50 mL of DCM was applied twice for the secondary extraction of residual pesticides from the sample.Then, 200 mL of DCM extract was concentrated into approximately 5 mL using a rotary evaporator (N-1300, EYELA, Koishikawa Bunkyo-ku, Tokyo, Japan).The sample was transferred into a concentration vial and completely dried using a nitrogen concentrator (MGS-3100, EYELA).After complete drying, 0.5 mL of ACN was added for dissolution and 0.25 mL of the sample was taken.Then, 0.25 mL of 100 mM ammonium formate buffer was added to make a volume of 0.5 mL, and this solution was analyzed using LC-MS/MS (Agilent 6495 TQ, Santa Clara, CA, USA).Table 1 shows the LC-MS/MS analysis conditions to simultaneously analyze 41 pesticides.We added 0.25 mL of ACN into the remaining solution of 0.25 mL to make 0.5 mL, and this solution was analyzed using GC-MS/MS (Agilent 7000C TQ, Santa Clara, CA, USA).Table 2 shows the GC-MS/MS analysis conditions to simultaneously analyze 16 items.As shown in Figure 1, health risks were assessed by setting the exposure scenarios for drinking (ingestion) and non-drinking water (dermal contact and inhalation) [13,14].The exposure scenarios of dermal contact and inhalation routes were set for showers and indoor/outdoor agricultural activities.The reference concentration for the risk assessment was set to the 95th percentile concentration, a high-end risk, which is known to affect approximately 68% of the receptors within the total distribution through exposure [15,16].The 95th percentile was derived using crystal ball ver.11.0 (Oracle, Redwood Shores, CA, USA) based on the precision survey results, and the physicochemical properties and toxicity reference values for each item were investigated (Tables 3 and 4).Table 5 lists the contents of the formulas and the related factors for each route obtained during risk assessment.
[ 19,20] Total non-cancer risk HQ = LADD Inges.+LADD Dermal.+LADD Inhal.R f D [17] Table 6 presents the physiological exposure coefficients and representative values applied in this study.In the case of agricultural activities, the reference data for the usage, usage hours, and usage frequency of agricultural canals as well as the skin exposure area (body surface area) were established considering 10 activities with the longest indoor/outdoor labor hours [21].When the indoor agricultural labor VF was calculated, the standard value of the agricultural house size in Table 6 was considered.When outdoor agricultural labor VF was calculated, the body exposure of the pollutants was calculated using a box model, as shown in Figure 2.

Quality Control Results by Pesticide Item
In this study (2019-2020), we established methods to analyze 57 residual pesticides in groundwater using GC-MS/MS (16 pesticides) and LC-MS/MS (41 pesticides).As methods to verify the effectiveness, the linearity of the calibration curve, the limit of quantification (LOQ), accuracy, precision, and method blank (MB) were analyzed.
The calibration curve of the 16 GC-MS/MS residual pesticides was prepared from 0.0001 to 0.1 μg/L, and its linearity was confirmed because the R 2 value was 0.995 or higher.The relative standard deviation was calculated through seven repeated analyses with 0.01 μg/L of pesticide using the established analytical methods.It was multiplied by 3.14 to calculate the detection limit and by 10 to calculate the LOQ.Consequently, the LOQ ranged from 0.0014 to 0.0677 μg/L for the entire research period.Due to improved analytical methods, the LOD and LOQ were relatively lower in 2020 (Table 7).Accuracy was repeatedly analyzed by preparing four samples for each concentration in the same manner.Accuracy ranged from 87.6 to 101.9% (less than ±20%).Precision was also analyzed in the same way as accuracy and was found to range from 0.5 to 5.1% (less than ±10%) for the research period (Table 8).During the monitoring period, the field blank sample was confirmed as non-detectable, and the average RFD (%) of the double sample was confirmed to be 7.1%.

Quality Control Results by Pesticide Item
In this study (2019-2020), we established methods to analyze 57 residual pesticides in groundwater using GC-MS/MS (16 pesticides) and LC-MS/MS (41 pesticides).As methods to verify the effectiveness, the linearity of the calibration curve, the limit of quantification (LOQ), accuracy, precision, and method blank (MB) were analyzed.
The calibration curve of the 16 GC-MS/MS residual pesticides was prepared from 0.0001 to 0.1 µg/L, and its linearity was confirmed because the R 2 value was 0.995 or higher.The relative standard deviation was calculated through seven repeated analyses with 0.01 µg/L of pesticide using the established analytical methods.It was multiplied by 3.14 to calculate the detection limit and by 10 to calculate the LOQ.Consequently, the LOQ ranged from 0.0014 to 0.0677 µg/L for the entire research period.Due to improved analytical methods, the LOD and LOQ were relatively lower in 2020 (Table 7).Accuracy was repeatedly analyzed by preparing four samples for each concentration in the same manner.Accuracy ranged from 87.6 to 101.9% (less than ±20%).Precision was also analyzed in the same way as accuracy and was found to range from 0.5 to 5.1% (less than ±10%) for the research period (Table 8).During the monitoring period, the field blank sample was confirmed as non-detectable, and the average RFD (%) of the double sample was confirmed to be 7.1%.
The calibration curve of the 41 LC-MS/MS residual pesticides was also prepared from 0.0001 to 0.1 µg/L in the same way as GC-MS/MS, and its linearity was confirmed because the R 2 value was 0.995 or higher.The LOQ, accuracy, and precision were also analyzed using the same method as those for GC-MS/MS to verify effectiveness.The LOQ ranged from 0.0005 to 0.0545 µg/L.Acetamiprid showed the lowest LOQ, while fludioxonil exhibited the highest LOQ (Table 9).
The accuracy ranged from 80.3% to 113.6% and satisfied the requirement of less than ±20%; however, the range was somewhat wider than that observed for GC-MS/MS.Precision ranged from 0.1% to 12.9% and satisfied the requirement for less than ±15% for all concentrations; however, the range was also wider than that observed for GC-MS/MS, similar to that for accuracy.In 2019, novaluron showed higher precision (0.34~12.9%) than the other components at low concentrations and benzobicyclon exhibited the highest precision (0.51~7.8%) at high concentrations, indicating differences depending on the concentration.In 2020, precision was lower than that in 2019, with 0.2~5.8% at low concentrations and 0.4~2.9% at high concentrations (Table 10).MB was analyzed using purified water that is used for glass cleaning, and the 57 residual pesticides were not detected during the research period.
results.When groundwater types were analyzed based on the major cation/anion analysis results, the Ca-HCO3 type indicating uncontaminated shallow groundwater was found to represent 39%, the Ca-(Cl+NO3 − ) type indicating the influence of artificial pollution sources was found to represent 39%, the Na-HCO3 type indicating groundwater-media reactions was found to represent 15%, and the Na-(Cl+NO3 − ) type indicating the effects of seawater was found to represent 7% (Figure 3) [9][10][11].

•
Pesticide status survey at the monitoring network points Figure 4 shows the results of a survey on the distribution status of residual pesticides at 200 points in the nationwide groundwater quality monitoring network for 2 years (twice/year).

•
Pesticide status survey at the monitoring network points Figure 4 shows the results of a survey on the distribution status of residual pesticides at 200 points in the nationwide groundwater quality monitoring network for 2 years (twice/year).In the survey results, 26 pesticides, including metolachlor, imidacloprid, and alachlor, were detected as being above the LOQ.Metolachlor exhibited the highest detection frequency (10.13%) followed by imidacloprid (8.50%), alachlor (8.13%), tricyclazole (5.88%), isoprothiolane (5.50%), and carbofuran (3.13%).Metolachlor exhibited the highest average concentration (0.0123 μg/L) followed by alachlor (0.0077 μg/L) and imidacloprid (0.0065 μg/L).The distribution status by pesticide class in groundwater is shown in Figure 5.A total of 11 pesticide classes were detected, including the amide class (e.g., alachlor and metolachlor), the azole class (e.g., tricyclazole and oxadiazon), and the neonicotinoid class (e.g., imidacloprid).When the pesticide detection status by measurement network type was compared, relatively higher detection levels were confirmed in the contamination and agricultural areas compared to the background measurement network.In the survey results, 26 pesticides, including metolachlor, imidacloprid, and alachlor, were detected as being above the LOQ.Metolachlor exhibited the highest detection frequency (10.13%) followed by imidacloprid (8.50%), alachlor (8.13%), tricyclazole (5.88%), isoprothiolane (5.50%), and carbofuran (3.13%).Metolachlor exhibited the highest average concentration (0.0123 µg/L) followed by alachlor (0.0077 µg/L) and imidacloprid (0.0065 µg/L).The distribution status by pesticide class in groundwater is shown in Figure 5.A total of 11 pesticide classes were detected, including the amide class (e.g., alachlor and metolachlor), the azole class (e.g., tricyclazole and oxadiazon), and the neonicotinoid class (e.g., imidacloprid).When the pesticide detection status by measurement network type was compared, relatively higher detection levels were confirmed in the contamination and agricultural areas compared to the background measurement network.

•
Groundwater quality survey results according to agricultural area cultivation purpose The results of the survey conducted twice a year at 50 points in the agricultural monitoring network (12 rice paddies, 19 fields, and 19 facility cultivation sites, for cultivation purposes) indicated that 18 pesticide items were detected (Figure 6).In the agricultural area, imidacloprid exhibited the highest detection frequency (18%) followed by metolachlor (13%), isoprothiolane (11%), alachlor (10%), and tricyclazole (10%).
Water 2024, 16, x FOR PEER REVIEW 13 of 16 • Groundwater quality survey results according to agricultural area cultivation purpose The results of the survey conducted twice a year at 50 points in the agricultural monitoring network (12 rice paddies, 19 fields, and 19 facility cultivation sites, for cultivation purposes) indicated that 18 pesticide items were detected (Figure 6).In the agricultural area, imidacloprid exhibited the highest detection frequency (18%) followed by metolachlor (13%), isoprothiolane (11%), alachlor (10%), and tricyclazole (10%).In the pesticide detection frequency investigation results according to the cultivation purpose, imidacloprid, which is a representative neonicotinoid pesticide, showed relatively higher detection frequencies in rice paddies, fields, and facility cultivation.It kills pests by blocking neurotransmitter receptors and has been widely used in rice paddies [26].Imidacloprid is less likely to be leached into groundwater because it has high soil absorptivity, with its Freundlich adsorption constant (Kf) ranging from 1.7 to 2.6 and its soil organic carbon adsorption coefficient (Koc) ranging from 228 to 249 [27].In the pesticide detection frequency investigation results according to the cultivation purpose, imidacloprid, which is a representative neonicotinoid pesticide, showed relatively Water 2024, 16, 584 13 of 16 higher detection frequencies in rice paddies, fields, and facility cultivation.It kills pests by blocking neurotransmitter receptors and has been widely used in rice paddies [26].Imidacloprid is less likely to be leached into groundwater because it has high soil absorptivity, with its Freundlich adsorption constant (Kf) ranging from 1.7 to 2.6 and its soil organic carbon adsorption coefficient (Koc) ranging from 228 to 249 [27].
Etridiazole and isoprothiolane showed relatively higher detection frequencies in groundwater around fields.Etridiazole is a sterilizing agent that can inhibit the oxidative decomposition of lipids, and it is known as a mobile compound with moderate persistence and high volatility [28].It is less likely to be released into groundwater but can remain in a water environment over an extended period because of its stable characteristics against hydrolysis and aquatic photolysis.Etridiazole was detected only at two points, as suggested above, but a maximum of 1.185 µg/L was detected at one point, indicating that continuous monitoring is required at this point.Isoprothiolane, a sterilizing agent that inhibits phospholipid biosynthesis and methyltransferase, is used in large quantities for rice cultivation.It has a high water solubility of 48 mg/L, and thus, it is highly likely to be introduced into the water environment, including groundwater [29].When pesticides were investigated in major rivers in South Korea in 2011, isoprothiolane exhibited a relatively high detection frequency compared to other pesticides because it was sprayed in the form of granules with large input per unit area and was released into rivers during rainfall or drainage while staying in rice paddies [30].
Alachlor, metolachlor, and tricyclazole showed relatively high detection frequencies in groundwater around facility cultivation sites.Alachlor and metolachlor are herbicides that have a mechanism of inhibiting the synthesis of long-chain fatty acids.Alachlor is the second most commonly used herbicide in the USA and is mainly used for corn cultivation.It has high mobility in water environments and exhibits high water solubility of 170.3 mg/L and moderate environmental persistence.Therefore, alachlor and its decomposition products are highly likely to be detected in groundwater [31].In South Korea, alachlor is mainly used to remove weeds during the cultivation of potatoes and beans, and it was detected at a concentration of 0.62 µg/L in the Yeongsan River water system (river water) and at up to 4.2 µg/L in a rice paddy district (paddy water) [32].Metolachlor has been classified as a human carcinogen by the US EPA, and its average half-life in water is known to be 200 days [33].It is highly likely to be released into groundwater when used as a pesticide because its chemical properties are similar to those of alachlor.Tricyclazole, a pesticide used as a sterilizing agent in rice cultivation, is known to be stable against hydrolysis under liquid conditions.Its half-life in soil ranges from 58 to 795 days, and its toxicity to living organisms in the environment is considered low.It is less likely to be released into groundwater due to its high organic carbon adsorption coefficient of 533~1199 [34].In the results of the survey on the status of pesticides in domestic rivers conducted by Hwang et al. (2019), however, tricyclazole exhibited a higher detection frequency than other pesticides.They stated that this was because the pesticides scattered during spray and those left in rice paddies were introduced into rivers by rainfall [35].A detailed investigation indicated that tricyclazole also exhibited a higher detection frequency than other pesticides as in the river water survey results presumably due to the same reason.

Evaluation Results for the Health Risks of Pesticides
Table 11 shows the results of the risk assessment by human exposure and exposure route based on the 95th percentile concentration for each item derived from the pesticide background survey results.
The risk assessment results showed that the non-carcinogenic risk was less than 0.001 for all items.The carcinogenic risk of metolachlor, which is known to be carcinogenic, was also evaluated to be less than 10 −6 in all exposure routes.The health risks caused by pesticide exposure in groundwater were found to be negligible (non-carcinogenic risk: less than 10 −3 ; carcinogenic risk: less than 10 −6 ).The background survey target points, however, are mostly less likely to be directly exposed to pesticides.Therefore, it will be desirable to continuously perform detailed investigations for groundwater around farmland where pesticides are used, perform risk assessment after the accumulation of survey results over a certain level, evaluate the results, and present management measures.

Conclusions
In this study, we identified the distribution status and background concentrations of pesticides in South Korea by developing and using multi-residue analytical methods for pesticides in groundwater.In addition, the detection characteristics according to land use as well as the behavioral characteristics and health risks of the items mainly detected were evaluated to contribute to the preparation of management measures for pesticides in groundwater in South Korea.The following conclusions were drawn:

•
Multi-residue analytical methods (LC-MS/MS 41 items and GC-MS/MS 16 items) were developed for 57 pesticides.The precision and accuracy ranges of the analytical methods were 0.2~12.9%(within ±15%) and 80.3~113.6%(within ±20%), respectively, and the LOQ was found to be in the range from 0.0004 to 0.0677 µg/L.• The pesticide distribution status survey results for groundwater showed that the detected concentrations were less than the minimum concentration (diazinon, 20 µg/L) in the domestic groundwater pesticide standard at all points for all items.Some items (GUS > 2.8) that are highly likely to be released into groundwater (e.g., alachlor and metolachlor), however, tended to be detected regardless of land use, confirming that it is necessary to develop countermeasures for them.• When the health risks of the items mainly detected were assessed, carcinogenic and non-carcinogenic risks were found to be less than 10 −6 and 10 −1 for all items, confirming that their health risks are negligible.

HCO 3 −
was analyzed through 0.05 N HCl titration in the field.The cation/anion analysis results were used for the classification of groundwater types using the Piper diagram.The Ca 2+ -HCO 3 − type represents shallow groundwater and the Ca 2+ -(Cl − + NO 3 − ) type represents artificial contamination.The Na + -HCO 3 − type indicates the reaction between shallow groundwater and underground media, and the Na + -(Cl − + NO 3 − ) type indicates the influence of seawater [9-12].

Figure 1 .
Figure 1.Human exposure scenarios based on groundwater use.

Figure 1 .Table 3 .
Figure 1.Human exposure scenarios based on groundwater use.Table 3. Physical and chemical properties of the three pesticides.

Figure 2 .
Figure 2. Box model of outdoor agricultural labor VF output.

Figure 2 .
Figure 2. Box model of outdoor agricultural labor VF output.

Figure 3 .
Figure 3. Groundwater type at the monitored sites.

Figure 3 .
Figure 3. Groundwater type at the monitored sites.

Water 2024 ,
16, x FOR PEER REVIEW 12 of 16

Figure 5 .
Figure 5. Distribution status of the pesticide classes in groundwater.Figure 5. Distribution status of the pesticide classes in groundwater.

Figure 5 .
Figure 5. Distribution status of the pesticide classes in groundwater.Figure 5. Distribution status of the pesticide classes in groundwater.

Figure 6 .
Figure 6.Results of the detailed investigation of pesticides in groundwater.

Figure 6 .
Figure 6.Results of the detailed investigation of pesticides in groundwater.

Table 3 .
Physical and chemical properties of the three pesticides.

Table 4 .
Human toxicity reference dose and carcinogenic potential of the three pesticides.

Table 4 .
Human toxicity reference dose and carcinogenic potential of the three pesticides.

Table 5 .
Risk assessment formula considering ingestion, dermal contact, and inhalation.

Table 6 .
Exposure coefficients and representative values related to the groundwater multi-pathway for the human exposure algorithm.

Table 11 .
Risk and exposure assessment results of pesticides in groundwater monitored from 2019 to 2020.