Seasonal variation of pesticides in surface water and drinking water wells in the annual cycle in western Poland, and potential health risk assessment

Drinking water wells on a riverbank filtration sites are exposed to contamination from farmlands (like pesticides) that had migrated from the contaminated river. In this study, pesticide contamination of the Warta River and riverbank filtration water at the Mosina-Krajkowo well field (Poland) were examined during the annual cycle. Among the 164 pesticides analysed, 25 were identified. The highest concentrations occurred in the river water and decreased along the flow path from the river to wells. Only the most persistent substances were detected at the farthest points. During the study, seasonal changes in pesticide concentrations and differences in the types of occurring substances were observed. Most substances and the highest concentrations were detected in May 2018, while the lowest number and the lowest concentrations were detected in February 2018. Spring is the period of increased exposure of water to pollution, which is correlated with increased pesticides use and increased rainfall. Seven toxic and persistent pesticides were found with the highest concentrations in water: isoproturon, nicosulfuron, imidacloprid, terbuthylazine, chlorotoluron, S-metalachlor, and prometryn. Pesticides are widely used in the study area; therefore, a potential health risk assessment was performed. The hazard quotient (HQ) values did not exceed one, which indicated a less significant health risk.

supplies drinking water to a large agglomeration (900,000 inhabitants) of the Poznań city in Poland. The selected site is located on the bank of the Warta River in Krajkowo. There are favourable hydrogeological conditions, because of the sediments of two groundwater bodies overlap, that is, the Warszawa-Berlin ice-marginal valley aquifer (shallow) and the Wielkopolska buried valley aquifer (deep). The total thickness of the water-bearing sediments is 40 m. The shallow aquifer is composed of coarse sands in the deeper section and fine sands in the upper section, and the deep aquifer comprises coarse sands and gravels beneath the fine sands. These aquifers are locally separated by a glacial till 37 . The well field is located near the forest and agricultural areas. It consists of 56 vertical wells located on the higher terrace, 400-1000 m away from the river; 29 vertical wells located in the flood plain, 60-80 m from the riverbank; 11 vertical wells supplied by artificial ponds; and the horizontal well (HW).
Sampling and sample analysis. The   The data from the first three of the seven sampling campaigns were previously published 15,27 . Samples were collected at eight sampling points: one in the river, two at observation wells, and five at production wells: HW and four vertical wells (Fig. 1). The sampling points were selected to represent different types of wells and different distances from the river (Table 1). Each production well was continuously pumped before sampling, and the observation wells were pumped using a portable pump (MP-1 type, Grundfos). The water in the well column was flushed a minimum of three times before sampling. Water was sampled in 1000 mL high-density polyethylene (HDPE) bottles. The bottles were filled to prevent degassing. The water samples were transported to the laboratory of the Institute of Plant Protection, National Research Institute in Poznań (Department of Pesticide Residue Research) on the same day for analysis.
Immediately after arrival at the laboratory, the samples were filtered, and 500 ml of water sample was taken for pesticide extraction. All selected pesticides were isolated from water samples by solid phase extraction (SPE, graphitised carbon black), followed by reverse-phase ultra-performance liquid chromatography with quadrupole mass detection (RP-UPLC-MS/MS) 38,39 . The studies included more than 160 active substances in pesticidess from different pesticide activities. The identified pesticides during the annual cycle included herbicides, fungicides, and insecticides, and the quantification limits for concentrations of individual compounds were from 0.005 to 0.02 µg/L. All methods were carried out in accordance with relevant guidelines and regulations.
Health risk assessment. Based on published standards an assessment of non-carcinogenic risk to human health was conducted for children (6 years old) and adults (70 years old) and the risks were calculated using hazard quotients (HQs) 3,24,36 The HQ values were estimated using Eq. (1) with the assumption that the river water and RBF water were untreated. An HQ value > 1 indicates that significant risk could occur, and HQ < 1 indicates no significant risk 3,24 . where, CDI is the protracted daily intake of pesticides through ingestion (mg/kg/day) estimated using Eq. (2), and ADI is the acceptable daily intake (mg/kg/day) 3,40 . The ADI values come from and are approved by the United States Environmental Protection Agency (US EPA) and Food and Agriculture Organization (FAO) of the United Nations reports.
where, C is the concentration of a pesticide in a river or RBF water (mg/L), IR is the water ingestion rate (0.87 L/ day for a 6 year-old child, 1.41 L/day for a 70-year-old an adult), EF is the exposure frequency (365 days/year), ED is the exposure duration (6 years for a child, 70 years for an adult), BW is the body weight (20 kg for a child and 70 kg for an adult), and AT is the average time (lifespan) (2190 days for child and 25,550 days for an adult) ( Table 2) 3,24 .

Results
Occurrence of pesticides in surface and riverbank filtration (RBF) water samples. Among the 164 analysed pesticides, 22 were detected (LOQ > 0.005 μg/L) in the Warta River (lotic surface water) with pesticide concentrations ranging 0.031-0.472 μg/L. Generally, the concentrations in the Warta River were higher than those at the other sampling points (Table 3, Supplementary file). The exception was in June 2018, when a higher concentration of the sum of pesticides (0.197 μg/L) was observed in the observation well, 168b/2, than that in the Warta River (0.111 μg/L). However, in the case of RBF water samples, 21 active substances were detected. Pesticides were detected at each sampling point at various concentrations (Table 3, Supplementary file). The highest concentrations and the lowest reduction in pesticides were observed in 168b/2 (maximum 0.197 μg/L) and HW (0.191 μg/L) (two points located nearest to the river). However, at each subsequent point, the concentrations decreased compared to that in the river water. Similar concentrations were detected in the two analysed wells located in the flood plain during each sampling campaign. Lower concentrations occurred in 78b/1 s, whereas no pesticides or very low concentrations were detected in the farthest sampling point (50A) located at the higher terrace. The concentration reduction increased as the distance between the river and well increased. Notably, only persistent pesticides were detected at the most distant points, for example, the herbicides chlorotoluron and isoproturon 42 .
The concentrations of pesticides in the surface and groundwater changed with the study period (Fig. 2). The highest pesticide concentration in surface water occurred in May 2018 (0.472 μg/L). The lowest concentrations were recorded in February (minimum 0.031 μg/L). In the remaining months, the concentrations of pesticides in river water were not variable (ranging 0.100-0.171 μg/L). The situation was slightly different for RBF water. The highest pesticide concentrations were observed in June 2018 (Fig. 2). Similar to the Warta River, the lowest concentrations occurred in February 2018. Generally, the concentration of pesticides was lowest in the winter months. Significantly lower concentrations of pesticides in RBF water were observed in August 2018 (0.009-0.035 μg/L) than in August 2017 (0.019-0.112 μg/L). This may be due to the lower rainfall that occurred in July 2018 (monthly total of 10 mm) than in July 2017 (monthly total of 93 mm) (Fig. 2).
In August 2018, fewer compounds (8 pesticides) were found in water than in August 2017 (13 pesticides       www.nature.com/scientificreports/ River it was 0.031 µg/L and in RBF water 0-0.024 µg/L, respectively). The most frequently detected pesticide in the highest concentration both in surface water and RBF water was nicosulfuron. Chlorotoluron also often appeared in RBF water in high concentrations. Generally, in the Warta River and at the closest points to the river, primarily the same substances were present. Most of the same substances occurring in the river also occurred in the HW. However, a similar trend was observed in the observation wells (168b/2 and 177b/1). At the other sampling points, the number of substances present was significantly reduced. Imidacloprid, nicosulfuron, isoproturon, and chlorotoluron were dominant in the drinking water wells (19L and 1AL). Notably, at the points further from the river, particularly 78b/1s and 50A, some substances were detected which did not occur in the river or at HW during the sampling campaigns in the given series, for example, isoproturon and chlorotoluron. Moreover, priority substances in the field of water policy not allowed for use in today's agriculture were detected in water samples; prometryn and simazine were banned within the European Union in 2007, and isoproturon in 2017.
Seven pesticides occurring in the surface and RBF water had the highest concentrations: imidacloprid, isoproturon, nicosulfuron, terbuthylazine, chlorotoluron, S-metalachlor, and prometryn (Fig. 5). Among these compounds, nicosulfuron stands out the most in the figure, exhibiting the highest concentration at six sampling points.
Health risk assessment. The non-carcinogenic human risks for a single pesticide posed by the ingestion of surface and RBF water were calculated as HQs. The HQ values are summarised in Table 5. HQs were calculated for azoxystrobin, boscalid, fluopicolide, fenpyroximate, propiconazole, tebuconazole, imidacloprid, thiamethoxam, and mesotrione, owing to the availability of acceptable daily intake values. The values were similar and ranged from 1E−05 to 1E−07. In general, the highest values were obtained for tebuconazole (1E−05) and the lowest for propiconazole (1E−07). The risk for children was higher than that for adults.
Based on The Hazardous Substances Data Bank (HSDB) the carcinogenic risk was also assessed 50 . Propiconazol, tebuconazole are possible human carcinogenic. Azoxystrobin and mesotrione are classified as not likely to be carcinogenic to humans. Boscalid has suggestive evidence of carcinogenicity, but is not sufficient to assess humans carcinogenic potential. No evidence for carcinogenicity was observed in rats administered fluopicolide in food for 24 months. Imidacloprid and thiamethoxam are non-carcinogenicity for humans.

Discussion
Water pollution caused by pesticides is notably high in the RBF well field, where the wells are supplied by the infiltration of surface water [25][26][27] . Thus far, previous studies have proven that the highest concentration of pesticides occurs in the surface water, and the contaminants decrease along the flow path to the wells. Research conducted on the Mosina-Krajkowo RBF well field has shown a high but incomplete reduction of contaminants 15,27,33 . Overall, a continuation of the aforementioned research in this study confirmed these findings. The concentrations of pesticides were reduced with increasing distance from the river. The highest concentrations were observed at surface water and HW. This trend changed in June 2018, when higher concentrations were observed in the observation wells. This could be due to the very high concentrations of pesticides in the river in the previous month (May 2018). Polluted water was supplied to wells in June 2018. The relationship between pesticide concentration and rainfall was also revealed, primarily in August 2017 and 2018. Previous research has focused on the occurrence and reduction of pesticides as a group of pollutants 27 . This study focused on the individual substances; pesticides were not considered as a group, but the behaviours of individual substances were analysed.
The study was conducted during the annual cycle and indicated the period of the year when water was most susceptible to pollution. The highest concentration of pesticidess in surface water could be combined with the period of increased chemical crop protection. Farmlands are often located close to rivers, which causes pesticides to rapidly migrate to the surface water 51,52 . In the study area, plant protection is conducted mainly between March and October, with particularly high protection measures adopted during spring. This was confirmed by the high concentration of pesticides in May and June. The lowest concentrations of pesticides in February could Notably, the present study included RBF water, where pollution is detected after a long time compared to surface waters. The delay is due to the specificity of the RBF well field. River water supplying the wells infiltrates from the river into the aquifer, which lasts from a few days to several weeks 33 . The highest concentrations of pesticides in both river and RBF water were in spring. However, varied concentrations were not the only differences. Individual substances also occurred seasonally. Depending on the sampling campaign, other substances were detected and reached the highest concentrations at the selected sampling points. It could be associated with the period of use and the persistence of the considered substances. In June and July 2018, nicosulfuron was detected at the highest concentrations. It was also found in other months (August and November 2017, May and August 2018), but at lower levels. Similar behaviour was observed for imidacloprid. It occurred at higher concentrations in June, July, and August 2018, and at lower concentrations in August, November, and May 2017. The same findings were obtained in the studies carried out in the Cachapoal River basin in Central Chile, where some of the pesticides were present only in the winter season (terbuthylazine, simazine, atrazine, and DIA), and some only at the end of the summer season (chlorpyrifos, and DEA) 54 .
During the spring and summer months, more substances were detected. This was particularly significant in river water. In February, two substances were most frequently present in the samples. Chlorotoluron and isoproturon occurred both in the river and in wells (even those farthest from the river). These two substances were also present in the other sampling campaigns, particularly at the farthest points (78b/1s, 50A).
The most common substances used for plant protection in this research area were herbicides. This is consistent with the worldwide consumption (the amount used in mass units) of pesticides, at 47.5% herbicides, 29.5% insecticides, and 17.5% fungicides 52 . Herbicides are also prevalent in other European countries, for example, Greece 36 . Herbicides are pesticides used to control weeds in crops. The most common pesticidess present in the highest concentration in collected samples was a sulfonylurea herbicide, nicosulfuron, which is widely used for weed control in corn 55 . The next was chlorotoluron, a phenylurea herbicide used to control grass weeds in cereal or fruit production. Chlorotoluron is also used to protect crops, primarily during late autumn and early spring. Prolonged use of chlorotoluron may cause its accumulation in the environment 56 . S-metalachlor is commonly Table 5. HQ values indicating non-carcinogenic human risks. ADI acceptable daily intake; C the concentration of a pesticide in a river or RBF water, HQ hazard quotients.

Pesticide
Sampling point ADI (mg/kg/day)  57 . The obtained results are in contrast with the findings of the study conducted in the Choluteca River Basin of Honduras, where the pesticides present in water are weakly correlated with the current pesticide applications 58 .
Although insecticides are the least frequently occurring group of pesticides, imidacloprid was observed to have high concentrations. These results are of concern because imidacloprid is one of the most toxic insecticides for bees 59 . In addition, neonicotinoid insecticides are toxic to aquatic invertebrates 60,61 . This toxic and persistent neonicotinoid insecticide is also observed at high concentrations in the area of intense corn and soybean production in the Midwestern United States 62 and agricultural regions of southern Ontario in Canada 63 . Sultana et al. 63 suggested further research to determine whether imidacloprid pollution is a global problem. The literature review and research conducted have proven this. Additionally, banned pesticides (isoproturon, prometryn, and simazine) were detected in the surface and RBF water. However high stability and durability, as well as low concentrations of these substances suggest that these is residues from past use.
Pesticides are toxic and persist in the environment. Some of them are not reduced or only partially eliminated, during the natural processes (RBF) and through treatment systems 64,65 . Water from the RBF well field is used for consumption; therefore, risk assessment studies are necessary. None of the HQs (for children or adults) for a single pesticide exceeded one, which implied that no significant health risk occurred due to the potential daily ingestion of river or RBF water. Similar results have been achieved in research on rivers in northern Greece, wherein the potential non-carcinogenic risk is low 36 . Research results from Japan indicate that diazinon and fenitrothion pose a high risk 3 . In the study conducted on the Mosina-Krajkowo well field, these substances were not present. Research conducted in drinking water in Ethiopia showed that there was no acute risk; however, chronic human health risks were observed 23 . The occurrence of hazards in different locations necessitates the constant monitoring of water quality and potential risk assessment. Equally important is research into the cumulative health risk of consuming water containing a mixture of pesticides. However, standards are needed to provide for the interaction between pesticides.

Conclusions
Research conducted in the annual cycle showed the presence of pesticides in the Warta River and RBF well field water, where the river is the main source of water. Of the 164 analysed pesticides, 25 were detected at the sampling points. Overall, the highest concentrations were observed in the Warta River. The concentrations in the wells were lower, which was the effect of the processes that occurred during riverbank filtration (RBF). Water pollution caused by pesticides shows seasonal variation. The highest pesticide concentrations occurred during the period of using pesticides in agriculture.
The pesticides found in collected water samples included herbicides, insecticides, and fungicides. Approximately 60% of the detected substances were herbicides, because of their popularity and spraying techniques. The occurrence of individual pesticides was also analysed. The greatest variation in pesticides, similar to the highest concentrations, was observed in the period of increased pesticide use by farmers. In the winter month (February 2018), only the most persistent pesticides were detected (chlorotoluron, imidacloprid, isoproturon, and prometryn). Most pesticides were detected in June 2018 (17), May 2018 (15) and November 2017. At the highest concentrations in surface and RBF water, seven pesticides occurred: imidacloprid, isoproturon, nicosulfuron, terbuthylazine, chlorotoluron, S-metalachlor, and prometryn. Nicosulfuron, chlorotoluron, and S-metalachlor are used for weed control, for example, in corn or cereal, which are common in the study area. Isoproturon, prometryn, and simazine have been banned in the European Union.
During the research, a very toxic substance, imidacloprid, was detected at high concentrations. This result confirmed that this insecticide is a global problem. Literature data from different parts of the world describe the presence of this substance at high concentrations.
The non-carcinogenic human risks posed by ingestion of surface and RBF water were calculated as the HQs for a single substances: azoxystrobin, boscalid, fluopicolide, fenpyroximate, propiconazole, tebuconazole, imidacloprid, thiamethoxam, and mesotrione. These values did not exceed one, which indicated a low risk. Tebuconazole exhibited the highest values of HQs, and propiconazole had the lowest. The potential risk for children was slightly higher than that for adults. The risks calculated for individual substances for surface and RBF water were similar; however, both were low. A further step should be to identify the risk of multi-pesticide consumption and calculate the cumulative risk.
Conducted research performed for RBF site, where groundwater quality is strongly depending on surface (source) water quality, covering the entire year allowed to show the variability of pesticide concentrations in the river and wells in relation to the periods of their use in agriculture and the hydrological situation (rainfall). It was also determined to what extend various pesticides are removed during underground flow of water from the river to wells. The presence of banned pesticides has been also identified. The presented research indicate the importance of pesticide monitoring at RBF sites and can be used to determine the principles of pesticides monitoring and the needs for further engineering water treatment in order to protect public health.