Occurrence and human health implications of chemical contaminants in vegetables grown in peri-urban agriculture

Recent studies have proven that vegetables cultivated in peri-urban areas are exposed to a greater concentration of organic microcontaminants (OMCs) and trace elements (TEs) than those grown in rural areas. In this study, the occurrence and human health risk of chemical contaminants (16 TEs and 33 OMCs) in edible parts of lettuce, tomato, cauliflower, and broad beans from two farm fields in the peri-urban area of the city of Barcelona and one rural site outside the peri-urban area were assessed. The concentration of TEs and OMCs ranged from non-detectable to 17.4mg/kg and from non-detectable to 256μg/kg, respectively. Tomato fruits showed the highest concentration of TEs and OMCs. Principal component analysis indicated that the occurrence of chemical contaminants in vegetables depended on the commodity rather than the location (peri-urban vs rural). Risk assessment using hazardous quotient (HQ) and threshold of toxicological concern (TTC) approaches showed that the risk for the consumption of target vegetables in the peri-urban area was low and similar to that observed for the rural site. Total HQ values for TEs were always below 1, and a minimum consumption of 150g/day for children and 380g/day for adults is required to reach the TTC due to the presence of pesticides. Further studies are needed to estimate the combined effect of TEs and OMCs on human health.


Introduction
The consumption of vegetables grown with peri-urban agriculture has grown exponentially due to the increase in urban demand (Olsson et al., 2016). Nevertheless, recent studies have highlighted that vegetables cultivated in peri-urban areas are generally exposed to a higher concentration of chemical contaminants than those grown in rural areas (Christou et al., 2017b;Säumel et al., 2012). Many studies have demonstrated that industrial activities, dense traffic flows, and the reuse of treated wastewater for irrigation are important sources of contamination of food crops (Christou et al., 2017a;Kalavrouziotis et al., 2008;Khan et al., 2008;Nabulo et al., 2006). Therefore, concerns about the chemical contamination of vegetables grown on peri-urban farms have risen due to the potential human health risk implications of consuming contaminated vegetables (Augustsson et al., 2018;Huang et al., 2018). The main health risks associated with peri-urban horticulture include the contamination of vegetables with agrochemical residues and trace elements (TEs) such as heavy metals (Birley and Lock, 1999). Nevertheless, it is important to notice that soil amending may result in a higher vegetable exposure to heavy metals (Antisari et al., 2015). Conversely, the use of reclaimed water for crop irrigation might end with a reduction of heavy metals uptake in crops caused by antagonist effects (Kalavrouziotis and Koukoulakis, 2010), but it can also increase crop exposition to organic microcontaminants (OMCs).
The detection of OMCs in the edible parts of plants has recently raised concerns regarding human exposure to these contaminants (Kalavrouziotis et al., 2012;Khan et al., 2008;Pan et al., 2014). In this regard, it has been reported that the presence of OMCs such as lamotrigine and 10,11-epoxycarbamazepine in sweet potatoes and carrots could have adverse effects on human health (Malchi et al., 2014;Prosser and Sibley, 2015). In the case of TEs such as heavy metals, although the potential risk associated with their presence in peri-urban vegetables is minimal, it may exceed the target hazard quotient for vulnerable populations such as children or pregnant woman (Hough et al., 2004;Nabulo et al., 2010). Nevertheless, few field studies exist on the occurrence of OMCs in vegetables (Malchi et al., 2014;Wu et al., 2014), and no risk assessment study has been done on the co-occurrence of metals, pesticides, and OMCs in peri-urban horticulture.
Estimation of the dietary exposure to chemical contaminants requires data on food consumption and the occurrence of each contaminant, which must then be compared with the pertinent health-based guidance value for the chemical of concern (WHO, 2008). Risk assessment can be based on either deterministic or probabilistic approaches. Deterministic approaches use a single value, such as a mean or percentile, to describe model variables. Their main drawback is thus the lack of insight they offer into the range of possible exposures and the proportion of the population that remains at risk. In contrast, probabilistic approaches, which are increasingly widely used (Quijano et al., 2017), account for the variability in food consumption and consumer body weight (BW), as well as the variability in the occurrence of the contaminant. Thereby, they address uncertainty and variability (Mondal and Polya, 2008), taking all possible scenarios into account (EFSA, n.d.;FAO, 2009).
Although risk assessment methodologies have mainly focused on individual chemicals and a single exposure route (oral, inhalational, or dermal), many of these chemicals act on the same target receptor by similar pathways. Therefore, the cocktail effect has been deemed an important factor in assessing the interaction between chemicals (WHO, 2017). This effect can take the form of an independent action, a dose addition (whereby one compound in the mixture is replaced by another) or an interaction (either synergistic or antagonistic compared to the predicted values based on dose addition) (Borchers et al., 2010). A conservative risk assessment approach consists of the sum of the hazard quotients (HQs), assuming that the effect of exposure to the individual chemical contaminants is additive and there are no synergistic or antagonistic effects (Prosser and Sibley, 2015).
In previous studies, we found that vegetables cultivated in peri-urban areas are exposed to a higher concentration of chemical contaminants than those from rural sites, but the occurrence of these contaminants in lettuce vegetables was similar . The present study aims to expand on those results and assess the occurrence of these contaminants (16 TEs and 33 OMCs) in several vegetables (lettuce, tomato, cauliflower, and broad beans), as well as to assess the potential risk to human health associated with the consumption of vegetables grown on peri-urban farms. This work was carried out on two farm plots located in the peri-urban area of Barcelona (NE Spain) and a rural plot located in a pristine area. The hypothesis was that the higher contamination exposure in peri-urban agriculture could result in a greater occurrence of contaminants in eatable vegetables and, consequently, in a greater human health risk.

Sampling site description
The study was carried out in the Llobregat River delta and its low valley (NE Spain) (Fig. 1). Two sampling sites, with a surface area >0.1 ha, were sampled in this protected farmland spanning 3300 ha in the metropolitan area of the city of Barcelona (Paül and McKenzie, 2013). A rural site outside the peri-urban area of influence, located at 400 m asl, was also selected for comparison.
The rural plot (P1) and the two peri-urban farm plots (P2-P3) were chosen for the disparity between them in terms of exposure to atmospheric and irrigation water pollution derived from industrial, urban, and agricultural activities. The rural plot (P1) is irrigated with well and rainwater using a drip irrigation system. In contrast, the peri-urban sampling sites (P2-P3) are watered by means of furrow irrigation with water from the Llobregat River, which contains discharges from 80 wastewater treatment plants (WWTPs) and is distributed through open-air channels (Marcé et al., 2012). The irrigation water used in P2 was also supplied by the effluent-dominated Infanta Channel creek, which collects effluents from 10 WWTPs. More detailed information is described elsewhere (Margenat et al., 2017).

Sampling regime
Vegetables were selected on the basis of their edible parts, i.e., the leaf (lettuce), the fruit (tomatoes and broad beans), or the flower (cauliflower). The specific types of vegetables sampled (lettuce, tomato, cauliflower, and broad beans) were chosen because they are some of the most heavily consumed vegetables in the Barcelona metropolitan area. The vegetables were harvested when they reached their commercial size, and only the edible part was used for the analysis.
The lettuce (Lactuca sativa L. cv. Batavia) was planted in March 2016 and harvested in May 2016. The tomato seedlings (Lycopersicon esculentum Mill) were cultivated in summer 2016 and picked in September 2016. The cauliflower (Brassica oleracea L. cv Skywalker) and broad beans (Vicia faba L.) were seeded in winter and harvested in March 2017. All vegetables were planted at all the studied sites, except the broad beans, which were not planted in P1.
Each farm plot was divided into 5 sections, and 5 to 15 vegetable specimens were collected per section. A quarter of every head of lettuce and cauliflower was mixed and comminuted together with the aid of liquid nitrogen and a porcelain mortar, while the tomatoes and broad beans were mixed and comminuted in their entirety, per section. Therefore, 5 composite samples were collected per plot and vegetable. The samples were stored at −20°C until they could be analyzed.

Analytical procedures
The chemicals and reagents used for the analytical methodologies are described in Section 1 of the supporting material (SM).

Extraction of trace elements
A 1 g of vegetable tissue, previously dried, was extracted using HNO 3 /HCl 4 digestion at 95°C for 3 h (EPA, 1994). The samples were then transferred to a 100-mL volumetric flask and centrifuged. A 10 mL aliquot was then diluted with 40 mL Milli-Q water prior to analysis in ICP-MS and ICP-OES. Further information is provided in Section 2.1 of the SM.

Extraction of organic microcontaminants
Briefly, 2 g of fresh sample (cauliflowers, tomato and broad beans), previously spiked with 12.5 ng of a surrogate mix (Section 1 SM), were transferred to a glass centrifuge tube containing 4g Na 2 SO 4 and 1 g of NaCl. Then, 10 mL of ACN were added to the tubes, and they were sonicated 15 min and centrifuged at 3000 rpm. The supernatant was collected and transferred to another tube, and 75 mg of C18, 75 mg of PSA, and 1350 mg of Na 2 SO 4 were added. The tube was sonicated for 15 min and centrifuged as above.
Finally, the supernatant was evaporated under a gentle stream of nitrogen to approximately 1 mL, and 50 mL of water were added. The samples were percolated through SPE cartridges (100 mg/6 mL), preconditioned with 2 mL methanol and 2 mL of MilliQ water, and eluted with ethyl acetate. The final extracts were evaporated to ca. 100 μL and spiked with 37 ng of TPhA. A 50 μL aliquot was directly analyzed by GC-MS/MS, and another 50 μL aliquot was derivatized by adding 10 μL of TMSH in the chromatograph injector port.
Extraction of carbamazepine and 10,11-epoxycarbamazepine from vegetable samples was carried out by sonication followed by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) (Tadić et al., n.d.). For this purpose, 0.5 g of fresh-weight vegetable was spiked with 50 ng of carbamazepine-13 C and left to equilibrate for 30 min. The samples were then sonicated with 10 mL of MeOH for 15 min and centrifuged 15 min at 3000 rpm. The extraction was per Margenat et al. Environment International xxx (2018) xxx-xxx formed twice, and the extracts were combined and reduced to ca. 1 mL with nitrogen gas and reconstituted with 10 mL of LiChrosolv water. The samples were subsequently percolated through SPE cartridges (STRATA X, 100 mg/6 mL), previously conditioned with 1 mL of MeOH and water, respectively. The cartridges were washed with water/methanol (95:5, v/v) and eluted with 2 mL of a mixture of MeOH/ethyl acetate (1:1, v/ v). The final extracts were reduced almost to dryness, resuspended in 1 mL of water, and filtered (0.22 μm) prior to LC-MS/MS analysis. LODs and LOQs were calculated for each analyte as three and ten times the signal from the baseline noise (S/N ratio) of 3 blank samples, respectively. Further details on the GC-MS/MS and LC-MS/MS and analytical quality parameters (LOD, LOQ, and recoveries) are provided elsewhere (Sections 2.2 and 2.3 and Tables 1-3 SM).
It is important to notice that the extraction method used for the determination of OMCs in lettuce, as well as the analytical quality parameters are likewise reported elsewhere .

Hazard quotient (HQ) for trace elements
The potential risk to human health resulting from consumption of TEs in vegetables was assessed using the hazard quotient (HQ) approach, which was calculated as follows (Dziubanek et al., 2017): (1) where RfD (reference dose) is the maximum tolerable daily intake (μg/ kg bw/day) of a specific metal (EPA, 2015; WHO, n.d.) that does not result in any harmful health effect, and EDI is the estimated daily intake (μg/kg bw/day). The EDI was calculated as follows: ( 2) where DI denotes the daily intake of edible parts of vegetables in g per day (the DI for each of the vegetables and population classes is given in Table 4-SM), C M is the 95th percentile value for the concentration of each TE in the vegetable tissue (μg/g fresh weight (fw)), and BW is the body weight (kg) of the target individual. The DI of fresh vegetables in Spain used in the calculations was taken from the EFSA's Comprehensive Food Consumption Database. For these data, all varieties of lettuce were considered, as well as tomatoes and tomato by-products, cauliflower, and all types of beans. An EDI value for a specific TE in excess of the corresponding RfD (i.e., HQ > 1) implies a potential risk to consumers. RfD values are obtained from chronic oral exposure studies (IRIS, 2018). Finally, the total hazard quotient (THQ) of each sampling site was calculated as the sum of all the HQs for all the chemicals to which an individual might be exposed. The values for selected metals Margenat et al. Environment International xxx (2018) xxx-xxx are shown in Table 3. The RfD value for Cr was calculated considering the Cr 3+ form, as it is the main chromium species present in lettuce (Asfaw et al., 2017).

Threshold of toxicological concern for organic microcontaminants
The human health risk associated with the intake of the selected OMCs through the consumption of the vegetable crops was assessed with the threshold of toxicological concern (TTC) approach using Toxtree software (Toxtree V2.6.13), which is based on the decision trees of Cramer and Kroes (Cramer et al., 1976;Kroes et al., 2000). This method is suitable for evaluating chemical compounds found at low concentrations in food products without toxicity data (Kroes et al., 2000). As a result, the OMCs were classified into three categories (I, II, III), corresponding to increasing toxicity or their potential genotoxicity and carcinogenicity. Class I contains substances with simple chemical structures with efficient modes of metabolism, suggesting a low order of oral toxicity. Class II consists of intermediate compounds and substances. Class III includes those that allow for no strong initial presumption of safety or may even suggest significant toxicity.
The selected TTC values were 30, 9, and 1.5 μg/kg bw/day for Classes I, II, and III, respectively (Munro et al., 1996). For the genotoxic substances, a TTC value of 0.0025 mg/kg bw/day was selected (Kroes et al., 2004). TTC values were calculated based on analysis of the chronic toxicity data of chemicals in three structural classes identified according to the Cramer decision tree (Cramer et al., 1976). The intake of OMCs above TTC values could pose a potential risk of exposure and requires a specific toxicity analysis of the targeted OMCs. In this study, the average body weights of a Catalan male adult (20-65 years) and child (4-9 years) were used, i.e. 70 and 24 kg, respectively (Generalitat de Catalunya, 2015). The daily consumption (DC, kg/day) by an adult or child to reach the TTC was calculated as follows: ( 3) where C OMC is the 95th percentile value for the concentration of each OMCs in the vegetable tissue (μg/kg fw).

Data analysis
The experimental results were statistically evaluated using the SPSS v. 22 package (Chicago, IL, US). All data sets were checked for normal distribution using the Kolmogorov-Smirnov test to ensure that parametric statistics were applicable. The overall comparison of the occurrence of chemical contaminants between farm plots (rural vs peri-urban) was performed with a paired-sample t-test (dependent samples), whereas the comparison of the concentration of each chemical contaminant between plots was analyzed by independent samples t-test. Statistical significance was defined as p ≤ 0.05. Principal component analysis (PCA) was performed on the concentration of TEs and OMCs in vegetables by using a correlation matrix. Table 1 shows the occurrence of TEs in vegetable crops from different sampling sites. Different vegetable species and cultivars differ in their ability to uptake, accumulate, and tolerate heavy metals. The concentration of TEs in vegetables ranged from non-detectable to 17 mg/ kg fw. Zn, B, and Mn were the most abundant, each of them with a concentration higher than 1 mg/kg fw. No statistical differences were generally observed between the rural and peri-urban sites (paired t-test taking into account all compound differences between sites, p-value > 0.05). Nevertheless, the concentrations of As, Cd, Pb, Mo, and Hg were greater in vegetables harvested in the peri-urban area (p-value < 0.05; broad beans were not included due to the lack of data for the rural site), except for Ba, Co, Mn and Sb which showed higher abundance in vegetables from rural agriculture. These findings are consistent with a previous study conducted in the same area for lettuce, in which Cd and Pb showed greater concentrations in the peri-urban site than in the rural one . They were also similar to those of other studies carried out in urban gardens (0.014 mg/kg fw Cd in fruit and 0.028 mg/kg fw Cd in leafy vegetables) (McBride et al., 2014).

Occurrence of trace elements
The total concentration of TEs per vegetable was as follows (calculated as the average of the total concentration of TEs in the 3 plots): 26 mg/kg fw (broad beans), 19 mg/kg fw (tomato fruits), 9.0 mg/kg fw (lettuce), and 8.9 mg/kg fw (cauliflower). Zn, Cu, B, and Mn, were the most abundant TEs in broad beans (14 mg/kg fw for Zn, 4.0 mg/kg fw for Cu, 2.9 mg/kg fw for B, and 3.0 mg/kg fw for Mn, on average) and tomato fruits (8.0 mg/kg fw for Zn, 3.1 mg/kg fw for Cu, 4.1 mg/kg fw for B, and 2.5 mg/kg fw for Mn, on average), whereas Cu and B were lower than 0.7 and 2.0 mg/kg fw respectively in lettuce and cauliflower. Mn concentration was similar in all of the studied vegetables. These results are in keeping with those of other studies conducted in the Basque Country (Spain) (Trebolazabala et al., 2017), which showed that Zn and Cu were detected in tomato fruits in a similar range of concentrations (12.43-26.7 mg/kg fw for Zn and 3.563-16 mg/kg fw for Cu). Similarly, in a study carried out in Turkey (Bagdatlioglu et al., 2010), Cu was detected at a higher concentration in tomato fruits than in lettuce. Furthermore, in the same study broad beans showed maximum levels for both Cu and Zn. Another study conducted in Bangladesh observed that the concentration of Cu was higher in tomato fruits than in the rest of vegetables studied (brinjal, bean, carrot, green chilli, onion, and potato) (Shaheen et al., 2016).
Finally, it should be noted that Li, Cd, Sb, and As were only detected in lettuce, due to their higher accumulation in the edible parts of this crop compared to root and fruit vegetables (Singh, 2012). The values of Cd and Pb were compliant with the maximum levels for these TEs in foodstuffs set out in Directive 1881/2006/EC of 19 December 2006 (0.2 mg/kg and 0.05 mg/kg fw for Cd in lettuce and the other vegetables, respectively, and 0.3 mg/kg and 0.10 mg/kg fw for Pb in lettuce and cauliflower and in tomatoes and broad beans, respectively).

Occurrence of organic microcontaminants
Only 10 of the 33 OMCs assessed in the vegetables were detected in at least one site. Overall, carbamazepine was the most frequently detected OMC (75%) ( Table 2). This is consistent with the demonstrated ubiquity of this compound in surface water samples, as well as its high plant uptake in greenhouse and field-grown experiments. Due to its neutral molecular form in a wide range of pH values and its log Kow (2.45), meaning it is rapidly uptaken by plants and accumulated at higher concentrations (Franklin et al., 2016;Goldstein et al., 2014;Riemenschneider et al., 2016). The concentration of OMCs in vegetables ranged from non-detectable to 256 μg/kg fw (dimethomorph). Dimethomorph (110-256 μg/kg fw for tomato fruits in the peri-urban site P2), methyl paraben (106-193 μg/kg fw for lettuce in P1), bisphenol F (32-92 μg/kg for tomato fruits in P3), and TCEP (97-124 μg/kg fw for tomato fruits in P3) were among the OMCs detected at the highest concentrations. This is consistent with the fact that direct application of fungicides, WWTP effluents (methylparaben), and chemicals released by plastic pipelines (bisphenol F) have been demonstrated to be the main sources of pollution in lettuce . Similarly, dimethomorph has already been detected in tomato fruits due to its direct application in agriculture (Walorczyk, 2013). Although the paired t-test did not reveal statistical differences between the concen

U N C O R R E C T E D P R O O F
A. Margenat et al. Environment International xxx (2018) xxx-xxx trations of OMCs in vegetables from the rural and peri-urban areas (p-value > 0.05), individual analyses showed that the concentrations of dimethomorph, TCEP, and MPB were different between the two areas (p-value < 0.05). The higher concentrations of dimethomorph and TCEP in the peri-urban area can be explained due to the application of fungicide and the wet/dry deposition of the fire-retardant compound on the vegetable surface, respectively. The occurrence of fire-retardant compounds in the atmosphere in areas close to cities has already been demonstrated (Ren et al., 2016), mostly in the particle phase (gas phase <5%), and it has been found that rainfall can lead to a scavenging effect, promoting the wet deposition of these compounds. The total concentration of OMCs per vegetable was as follows (calculated as the average of the total concentration of OMCs in the 3 plots): tomato fruits (163 μg/kg fw), lettuce (104 μg/kg fw), broad beans (19 μg/kg fw), and cauliflower (5 μg/kg fw). The highest average concentrations were found in tomato fruits for TCEP (64 μg/kg fw) and bisphenol F (33 μg/kg fw). Finally, it is important to note that the concentration of pesticides (dimethomorph, carbendazim, and indoxacarb) in vegetables complied with the maximum residue levels of pesticides in or on food and feed of plant and animal origin under Council Directive 91/414/EC (EC, 2006) (Table 5-SM).

Sources and distribution of chemical contaminants (PCA)
Principal component analysis (PCA) was performed on the whole data set to gain further insight into the sources and distribution behavior of the various parameters assessed in the irrigation waters (Table 6-SM). The PCA reduced the 23 measured variables to 6 principal components with eigenvalues > 1, which explained 75% of the total variability observed. Components explaining small data variance (i.e., <10%) were not retained and were assumed to be mostly due to background and noise contributions. Therefore, only the first three principal components, accounting for 55% of the total variability, were studied. The first principal component (PC1), which accounted for 23% of the variance, had high positive loading values (>0.6) for B, Cu, and Mo, but negative loadings for Co, Ba, and MPB. This negative correlation between parameters indicates that while the positive variables were not accumulated in plants, the negative ones were. This fit perfectly with the lettuce plants. The second component explained 18% of the variance and had positive loadings (>0.6) for Cr, Mn, Co, Ni, Mo, carba mazepine, and MPB. This component correlated with the chemicals that were highly abundant in lettuce and broad beans. The third component accounted for 13% of the variance and had positive loadings (>0.6) for B, Rb, TCEP, and BPF. It correlated with the chemicals found to be most abundant in tomato fruits. Fig. 2 shows the score plots for PC1 vs PC2 (PC1 vs PC3 is provided in Fig. 1-SM). Both plots grouped samples in 4 groups depending on the vegetable (lettuce, tomato, cauliflower, or broad bean). Hence, the abundance of chemicals in the various samples collected depended on the vegetable rather than the location.

Risk assessment for trace element exposure
The estimated daily intake (EDI) of TEs was compound and vegetable dependent. On average, Zn was the compound with the highest  Margenat et al. Environment International xxx (2018) xxx-xxx intake (9.0 × 10 −3 and 1.0 × 10 −2 mg/day for adults and children, respectively), whereas tomatoes had the highest concentration of TEs on average per farm plot (4.3-7.2 mg/kg fw). Oral reference doses (RfDs) were used to assess the human health risk (Table 3), except for Co, Hg, Li, and Rb, for which the oral RfD was not available. Therefore, the risk assessment was evaluated for 12 TEs. HQs ranged from 4.8 × 10 −8 to 0.2 for As in broad beans and Pb in tomato fruits (P3), respectively. Of these 12 TEs, on average, Pb posed the greatest health risk to adults and children, followed by Zn, As, Mn, B, Mo, Cd, Cu, Ni, Ba, As, and Cr. These findings are consistent with those of other studies carried out in China and Ethiopia (Chang et al., 2014;Dziubanek et al., 2017;Woldetsadik et al., 2017), where food crops irrigated with either river water or treated wastewater in rural and urban sites did not pose risks to human health.
Pb was also assessed with three Benchmark Dose Lower Confidence Limits (BMDLs) to evaluate the effects of lead exposure in humans: BMDL 10 for chronic kidney disease development at 0.63 μg/kg bw/day, BMDL 01 for systolic blood pressure effects at 1.5 μg/kg bw/day, and BMDL 01 for developmental neurotoxicity (which would apply to fetuses and infants) at 0.5 μg/kg bw/day, as determined by the WHO (1986) and EFSA (2010). The EDI values found for Pb in the present study due to the ingestion of these vegetables (Table 7-SM) were in the range of 5.2 · 10 −5 to 5.2 · 10 −1 μg/kg bw/day. These values were lower than all three of the aforementioned BMDLs, except for tomato fruits in Plot 3. This is in agreement with the high lead content found in the soil from this plot (164 mg/kg dw) in a previous study .
The THQs obtained for all age groups were <1 for all vegetables and sites studied. Tomato fruits showed the highest THQ on average (0.40 and 0.57 on average for adults and children, respectively), followed by lettuce (0.23 and 0.30 for adults and children, respectively), cauliflower (0.02 and 0.003 for adults and children, respectively), and broad beans (0.0001 and 0.005 for adults and children, respectively). The fact that the highest THQ was observed for tomatoes is probably due to the high consumption of this fruit in Spain compared to the other studied vegetables (see the Material and methods section). Nevertheless, no statistical differences were observed in the individual and total HQs between vegetables grown in the peri-urban area and rural site (paired t-test, p > 0.05). Although for some TEs such as Pb in tomatoes and Zn in lettuce the HQs were greater in the peri-urban agriculture, statistical analysis, comparing rural and peri-urban vegetables (including all vegetables), resulted in no statistical differences between both areas for any of the studied TEs. The total sums of the THQs (for adults) for each farm plot, taking into consideration all TEs and the consumption of all the studied vegetables, were as follows: 0.82 (P3), 0.59 (P2), and 0.54 (P1). Farm plots irrigated with unplanned reclaimed water (de facto reuse) showed the highest risk (P2-3), although it was still <1 for adults; hence no risk was assumed (Zeng et al., 2018). These findings are consistent with previous studies, which have found that farm plots irrigated with treated wastewater result in a higher risk due to the occurrence of TEs than control plots irrigated with groundwater (Khan et al., 2008).
Although the THQs obtained were lower than 1 in all sites, it should be taken into account that the assessment included only the consumption of the analyzed vegetables. The values could be higher if other vegetables are considered. Therefore, future studies should evaluate the risk associated with other vegetables consumed per person and day in a real-life scenario.

Risk assessment for OMC exposure
The health risk associated with the occurrence of OMCs in the different studied vegetables was assessed using the TTC approach. Only compounds identified above the LOQ were considered for the risk assessment study (10 of the 33 monitored compounds). The OMCs were classified as Class III (dimethomorph, surfynol 104, 2-MBT, BPF, CBZ, carbendazim, TCEP, and indoxacarb), Class I (MPB), and genotoxic (10,11-epoxycarbamazepine). Table 4 shows the DC required to reach TTC levels for each of the studied OMCs. The DC ranged from 0.38 kg/ day (dimethomorph in tomatoes, P2) to 7000 kg/day (dimethomorph in cauliflowers and broad beans, except for P2) for adults, and from 0.15 kg/day to 2400 kg/day for children. Hence, in the worst-case scenario, the daily consumption of tomato fruits grown in P2 required for a child or an adult to reach the TCC level is 150 g/day and 380 g/ day, respectively, which would not be a typical DI for tomato fruits and by-products (94.3 g/day and 196 g/day for children and adults, respectively, as shown in Table 4-SM). Similarly, the daily consumption of lettuce, tomato, and broad beans did not reach the TCC for any of the studied OMCs and therefore did not pose a risk to human health. Nevertheless, exposure to other compounds, such as 10,11-epoxycarbamazepine, should be taken into consideration in lettuce as the TTC can be reached with a consumption of 700 g/day or 300 g/day of lettuce for adults and children, respectively. These findings are consistent with those of Riemenschneider et al. (2016), who studied 28 OMCs (including 9 CBZ metabolites) in 10 vegetable species, including lettuce and tomato fruits irrigated with TWW. In that study, the daily consumption of lettuce required to reach the TCC was higher than 9 kg for all the studied OMCs, except 10,11-epoxycarbamazepine, for which it was 40 g for lettuce for adults; in contrast, it was higher than 300 kg for tomato fruits.
Since pesticides were the most important compounds in the risk assessment obtained with the TTC approach, and because they are already regulated through maximum residue levels (MRLs), the risk (HQ) for consumers was also assessed. This was done using the quotient between the EDI and the ADI, where the ADI was obtained from the MRL established by EFSA for pesticides in each vegetable and the daily intake (EC, 2005). Table 8-SM shows the HQs for pesticides in the different vegetables. As indicated by the TTC approach, dimethomorph posed the highest risk in tomatoes (HQ = 0.56), whereas the other pesticides showed HQs <0.04. The risk for pesticides is considered unacceptable when the HQ exceeds 1; thus the risk was not noticeable. These findings are consistent with other studies carried out in pesticides for several vegetables and fruits with HQ values always below 1 (Lemos et al., 2016;Li et al., 2017;Lozowicka, 2015). In general, the risk associated with pesticide application was found to be more relevant than that associated with other OMCs that can originate from irrigation or soil pollution.

Conclusions
Although the occurrence of some chemical contaminants, such as Pb or dimethomorph, was greater in vegetables grown in the peri-urban area than the rural one, the abundance of chemical contaminants depended on the vegetables evaluated (cauliflowers, tomatoes and lettuce). The following key conclusions can be drawn: -The concentrations of TEs ranged from non-detectable to 4.9 mg/kg fw in lettuce, from 2.3 × 10 −4 to 12 mg/kg fw in tomato fruits, from 1.1 × 10 −4 to 4.9 mg/kg fw in cauliflower, and from 1.5 × 10 −4 to 17 mg/kg fw in broad beans. In contrast, the concentration of OMCs ranged from non-detectable to 193 μg/kg fw in lettuce, from non-detectable to 256 μg/kg fw in tomato fruits, from non-detectable to 156 μg/kg fw in cauliflower, and from non-detectable to 206 μg/kg fw in broad beans.

U N C O R R E C T E D P R O O F
A. Margenat et al. Environment International xxx (2018) xxx-xxx Table 4 The daily consumption (kg/day) required to reach TTC levels for the selected OMCs in an adult (70 kg) and in a child (24 kg).  -THQs for TEs were below 1 in all studied vegetables and ranged from 0.0001 to 0.40 on average in adults and from 0.0005 to 0.57 in children. -The daily consumption of food crops required to reach the TTC values for OMCs was 150 g/day for children and 380 g/day for an adult. As the actual daily consumption of these vegetables is below these values, no risk due to the intake of these food crops is foreseen, although more toxicological data is needed.
The present study showed that the occurrence of target chemical contaminants in peri-urban agriculture did not affect the human health risk. However, further research is required, especially on the potential human health risk of the combined effects of OMCs and TEs.