Health Risk Assessment in Agricultural Soil Potentially Contaminated by Geogenic Thallium: Influence of Plant Species on Metal Mobility in Soil-Plant System

In agricultural soils, thallium (Tl) of geogenic origin may represent a potential risk for human health, mainly via ingestion of food crops. In this work, a pot experiment was carried out to evaluate (1) the bioavailability of Tl and other potentially toxic elements (PTEs) in an agricultural soil with naturally occurring Tl; (2) the uptake and accumulation of PTEs in Lactuca sativa L. var. acephala, Diplotaxis tenuifolia L. DC and Silene latifolia Poir; (3) the health risks arising from plant and soil ingestion by different subpopulations and dermal contact of soil by farmers. In soil, only Tl and Pb pseudototal contents were above Italian screening values. Nevertheless, the promptly bioavailable contents of all PTEs were always below internationally recognized trigger values. Plants affected PTE bioavailability in soil by their rhizodepositions and accumulated PTEs in their shoots. Acceptable risks (hazard index < 1) arose from dietary intake of both L. sativa L., D. tenuifolia L. and dermal contact of soil by farmers. Significant health risks can derive from the intake of S. latifolia Poir. (accumulating high Tl concentrations), in particular by children (HI = 74). In conclusion, an adequate management and crop selection are needed to profitably exploit soils with geogenic Tl for agricultural purposes.


Introduction
Thallium (Tl) is a nonessential and highly-toxic element for living organisms, generally showing a greater acute and chronic toxicity than other harmful elements, such as As, Cd, Hg and Pb [1,2]. Serious Tl poisoning may lead to gastrointestinal dysfunction, ascending paralysis and mental disorders, whilst polyneuritis may occur with chronic poisoning [3][4][5]. The lethal dose of Tl for an adult human is only 8-10 mg kg −1 ; therefore, it is classified as one of the 13 priority metal pollutants by the United States Environmental Protection Agency and also listed in the European Union (EU) as a dangerous pollutant [6][7][8]. The main routes of human exposure to Tl are the ingestion of contaminated food and drinking water [9]. Once ingested, Tl quickly reaches the principal human organs and eventually accumulates in the nervous system [10].
Although Tl is a rare element, ranging from 0.1 to 2.0 mg kg −1 in terrestrial environments (average content of 0.5 mg kg −1 in the continental crust) [11,12], it can occur at higher extent in various sulphide the 0-20 cm soil layer ( Figure S1) and then mixed thoroughly to form a composite sample which was successively used for the pot experiment with three plant species.
A portion of discrete samples and composite sample was dried at 50 • C until constant weight, homogenized and sieved at 2 mm, in order to evaluate the main physicochemical properties.
The study area is in the Piana Campana graben, to the NW of the Somma-Vesuvius complex and NE of the Phlegrean Fields ( Figure S1), the main active volcanoes of Campania. In this volcanic area, there was an accumulation of pyroclastic deposits [35]. Therefore, the soils of this territory are the result of a different extent of andosolization, which produced soils that can be classified as Pachi-Vitric Andosols and Hypereutric Cambisols [36,37].

Experimental Setup
The pot experiment was carried out in an unheated polyethylene greenhouse sited at the Department of Agricultural Sciences, University of Naples Federico II, southern Italy (40 • 49 N, 14 • 15 E). Seeds of lettuce (Lactuca sativa L. var. acephala) and perennial wall rocket (Diplotaxis tenuifolia L. DC) where bought from a local supplier, while white campion seeds (Silene latifolia Poir.) were collected in 2017 from spontaneous plants growing near the roads or in open fields of the Parco Gussone (Department of Agricultural Sciences, University of Naples Federico II, southern Italy). Then, seeds were surface-sterilized before storing by immersion in 5% (V/V) sodium hypochlorite for 10 min and washed three times in sterile water [38]. Soil from the composite sample was transferred to pots (four replicates per plant species; Ø 16 cm; V 2 L). Then, seeds of white campion (0.4 g per pot), lettuce (0.2 g per pot) and perennial wall rocket (0.1 g per pot) were sown at the end of October 2018 in the pots ( Figure S1). White campion and lettuce were then thinned to five plants per pot after germination. Pots were watered regularly to keep soil moisture close to field capacity and all plants were harvested three times in the same pot (from March to June 2019) during the experiment that lasted for a total of 33 weeks.

Plant Sampling, Processing and Analysis
Each harvest of shoot biomass was carried out when stem length was at least 15 cm. All the fresh shoot samples were thoroughly washed with tap water and then deionized water, gently blotted, weighed (fresh weight), oven-dried at 60 • C to constant dry weight and, later, finely ground in an electric mill. Shoot dried samples were firstly digested by concentrated HNO 3 and secondly by aqua regia in a microwave oven, then the concentration of PTEs in the extracts were detected by inductively coupled plasma mass spectrometry (ICP-MS) (analytical package VG101-EXT) at Acme Analytical Laboratories Ltd. (Vancouver, Canada) within a time span of 25 days from receiving the samples to final delivery of analytical results. Certified reference material (oriental tobacco leaves CTA-OTL-1) and lab standard materials (STD CDV-1 and STD V16) were used to monitor the data quality of analyses, with metal recoveries around ±8% of the certified values. PTE concentrations in the plants were compared to PTE thresholds indicated in EC Reg. 1881/2006 for vegetables (Cd: 0.2 mg kg −1 f.w.; Pb: 0.3 mg kg −1 f.w.) [39]. In addition, the range values of PTEs found in food plants growing on nonpolluted sites (Cd: 0.029-0.4 mg kg −1 d.w.; Cu: 6-8 mg kg −1 d.w.; Pb: 0.7-3.6 mg kg −1 d.w.; Tl: 0.02-0.125 kg −1 d.w.; Zn: 44-73 kg −1 d.w.) by Kabata-Pendias, [12], were used as reference values.
To evaluate the accumulation of PTEs, the bioaccumulation coefficient (BAC) was calculated as follows: where C shoots and C soil are the PTE concentrations (mg kg −1 ) in shoots and soil, respectively [28,29].

Soil Sampling at Pot Scale and Physicochemical Characterization
At the end of the experiment, the soils in the pots were dried at 50 • C until constant weight, homogenized and sieved at 2 mm before analysis. Afterwards, all the sieved and dried soil samples (before and after plant cultivation) were characterized for the following physicochemical properties: pH-H 2 O (1:2.5 soil:water solution ratio, pH meter GLP 22, Crison), electrical conductivity (1:5 soil:water solution ratio, Conductimeter basic 30, Crison), organic carbon [40], extraction of promptly bioavailable fraction of PTEs by ammonium nitrate (described in the Section 2.4.2).
In the discrete and composite soil samples collected before plant cultivation, the texture (pipette and sieving method, after pretreatment with H 2 O 2 to oxidize organic matter and particle dispersion by sodium hexametaphosphate), total nitrogen (Kjeldahl method), available phosphorous (Olsen method), exchangeable potassium (extraction with 0.5 M ammonium acetate and potassium evaluation by spectrophotometer) and pseudototal content of PTEs (described in the Section 2.4.1) were also determined.

Determination of Pseudototal Content of PTEs
Pseudototal PTE contents were quantified in 1 g of each pulverized soil sample by aqua regia digestion [41] and ICP-MS at Acme Analytical Laboratories Ltd. (Vancouver, Canada), within a time span of 25 days from receiving the samples to final delivery of analytical results. Certified reference material ERM-CC141 Loam Soil and BCR-141R Calcareous Loam Soil (European Reference Materials, ERM ® ) and lab standard materials (STD DS10 and STD OREAS45EA) were used to monitor the data quality of soil analyses, with metal recoveries around ±10% of the certified values.

Extraction of Promptly Bioavailable Fraction of Metals by Ammonium Nitrate
Ten grams of 2 mm-sieved soil were extracted with 25 mL of 1 M NH 4 NO 3 solution (m/V = 1:2.5) into a 50 mL FalconTM tube, by shaking for 120 min at 20 ± 2 • C with a horizontal shaker [42]. The extract was separated from the solid residue by centrifugation at 1000× g for 10 min and the supernatant was filtered by 25 mL syringe equipped with a 0.45 µm disposable in-line membrane filter, acidified by 65% HNO 3 (final acid concentration: 1%) and then stored at 4 • C in a refrigerator prior to being analysed. The promptly bioavailable metal contents in the filtrates were measured by ICP-MS at Acme Analytical Laboratories Ltd. (Vancouver, BC, Canada) within a time span of 25 days from receiving the samples to final delivery of analytical results. The detection limits of Cd, Cu, Pb, Tl and Zn were respectively 0.005, 0.010, 0.020, 0.001 and 0.025 mg kg −1 .

Health Risk Assessment
PTEs may cause both carcinogenic and noncarcinogenic risks through different pathways. According to Italian law, the main risk for human health in agricultural lands is through crop ingestion [43]; however, agricultural workers' exposure through ingestion and dermal contact of soil were evaluated in a health risk assessment [44]. The noncarcinogenic risks were assessed by calculating the hazard quotient (HQ) for each exposure pathway and each PTE [17,45,46]: where ADD is the average daily dose (mg kg −1 day −1 ) and RfD is the chronic reference dose of each PTE (mg kg −1 day −1 ). Oral reference dose (RfDo) values used in the dietary and soil ingestion health risk assessment are shown in Table S1 [47]. Reference dose for dermal absorption (RfDd) was calculated for each PTE from oral reference dose using equation [48]: where RfDd is the adjusted reference dose for dermal exposure (mg kg −1 day −1 ) and ABS GI is the gastrointestinal absorption factor (dimensionless).
The cumulative risk of all PTEs for each exposure pathway was evaluated by calculating the Hazard Index (HI) by using the equation: HQ or HI values >1 indicate significant risk to human health [45,46]. The ADD was calculated for the dietary intake (ADD dietary ) for different age groups (children, 3-9 years; adolescents, 10-17 years; adults, 18-64 years; elderly, 65-74 years) while ADDs for soil ingestion (ADD ing ) and dermal contact with soil (ADD dermal ) were assessed for adult farm workers as follows [43,45,49,50]: where: • C plant (mg g −1 ) and C soil (mg kg −1 ) are the PTEs concentrations in plants and soil, respectively. For each plant, max PTE concentrations found in the shoot at the three harvests were used in this study; • IngR plant (g day −1 ) and IngR soil (mg day −1 ) are the plants and soil ingestion rate per day, respectively; • SA is the exposed skin surface area (cm 2 ); • AF is the skin adherence factor (mg cm −2 day −1 ); • ABS is the dermal absorption factor (dimensionless); • EF is the exposure frequency (day year −1 ); • ED is the exposure duration (years); • BW is the body weight (kg); • AT is the period over witch exposure is averaged (days). For non-carcinogenic risk assessment AT = ED × 365 days while for carcinogenic risk assessment AT = 70 × 365 assuming that carcinogenic effects may occur also when exposure ends.
Carcinogenic risks were evaluated by equation: where CR is the cancer risk (dimensionless) and SF is the cancer slope factor (mg kg −1 day −1 ) −1 . Oral slope factor (SFo) values are shown in Table S1 [47], while the dermal slope factor (SFd) was evaluated by the following equation [48]: where SFd is the dermal slope factor, SFo is the oral slope factor and ABS GI is the gastrointestinal absorption factor (dimensionless). The acceptable or tolerable risk is within the range of 1 × 10 −6 -1 × 10 −4 and values lower than 1 × 10 −6 are considered negligible. Carcinogenic risk values higher than 1 × 10 −4 are likely to be harmful to humans [51].
All the parameters used in the health risk assessment equations are shown in Tables S2 and S3.

Statistical Analysis
All data were subjected to analysis of variance (ANOVA) via a general linear model and mean values were separated according to a least significant difference (LSD) test with p < 0.05, using the software IBM SPSS Statistics 25. The Kolmogorov-Smirnov and Levene tests were used to verify normality of distribution and homogeneity of variance, respectively. Logarithmic transformation was applied, when needed, to variables that did not accomplish normality. Pearson correlation analysis was carried out between soil and plant PTE concentrations by XLStat (Addinsoft, Paris, France) to evaluate general tendencies.

Soil Physicochemical Properties, Pseudototal and Promptly Bioavailable Contents of PTEs
The main physicochemical properties and pseudototal content of PTEs of the composite soil sample used in the pot experiment, obtained by mixing discrete soil samples collected from the agricultural soil, are reported in Table 1.
The soil was sandy loam and showed a slightly acid to subneutral pH (6.43), which made the bioavailability of the main nutrients such as phosphate optimal for plant growth. The electrical conductivity was quite low, hence there was no accumulation of soluble salts which could potentially inhibit plant growth. The soil showed a medium-low content in organic matter (OM), typical of soils exploited for agricultural purposes in Mediterranean countries. The content in total N was moderate, in a good equilibrium with the organic C content; the C/N ratio was 9.57, a value which indicated that the primary inputs of organic matter from crop residues were properly mineralised by soil microbiota. The medium-high availability of P and the good cation exchange capacity (CEC) adequately saturated by exchangeable K revealed a good soil chemical fertility and availability of the major plant nutrients. The good fertility of the volcanic soils of metropolitan area of Naples was also highlighted in other studies [36,37].
The pseudototal contents of Cd, Cu and Zn were below the screening values set by the Italian Ministerial Decree 46 for agricultural soils [43], but also lower than the baseline values of volcanic soils in the metropolitan area of Naples proposed by Cicchella et al. [52] and De Vivo et al. [53]. These findings can be observed in all the eight discrete samples collected from the agricultural soil (Table S4) and, as a consequence, in the composite soil sample used in the pot experiment (Table 1). On the other hand, the pseudototal content of Tl was twofold higher (Table 1 and Table S4) than the screening value for agricultural soils (1 mg kg −1 , [43]); however, it was within the range of Tl baseline values (1.7-2.7 mg kg −1 ) found by Cicchella et al. [52] and De Vivo et al. [53] in the volcanic soils originating from the Somma-Vesuvius complex. Therefore, according to these comprehensive geological studies, the Tl content in the studied farmland should be basically of geogenic nature. Similarly, the pseudototal content of Pb exceeded of~20% the screening value for agricultural soils (100 mg kg −1 , [43]), but it was within the medium-high baseline values of the metropolitan area of Naples [52,53], due to both geogenic factors (e.g., hydrothermal fluids related to Vesuvius fractures) and anthropogenic contribution (fall-out and accumulation of Pb from tetraethyl lead, added to gasoline as an anti-knock agent until the 1990s).
The bioavailable content of PTEs in the soil, along with pH, EC and OC measures, before (T0) and after the growing cycle of lettuce (T1-L), perennial wall rocket (T1-R) and white campion (T1-S) plants, are shown in Table 2.  After plant growing cycles, we found a statistically significant increase of both soil pH and EC values, probably due to the enhanced biological activity triggered by the release of rhizodepositions by plant roots. Nevertheless, the content in organic C after plant growing cycles did not showed statistically significant differences in comparison to the initial medium-low content.
The promptly bioavailable fractions of all the PTEs in the soil before plant growth were lower than the trigger values settled by Germany and Slovak Republic to indicate a potential risk for growth and food safety of crops (Table 2) [33,34,54]. This means that all the PTEs are basically poorly bioavailable in the soil, even those showing pseudototal contents higher than screening values for agricultural soils [43], such as Tl (the most concerning metal in the site) and Pb. In terms of percentage, the promptly bioavailable fractions of Cd, Cu, Pb, Tl and Zn were the 2.1, 0.2, <0.1, 3.4 and 0.5%, respectively, of the pseudototal contents. These outcomes can be justified by the presumed geogenic nature of all the PTEs in the soil [52,53], which can be likely occluded or interlayered in the soil mineral lattice. Additionally, the soil physicochemical properties and the subneutral pH may contribute to minimise the solubility and bioavailability of PTEs in the soil, which tends to precipitate as metal (hydr)oxides or form stable and sparingly soluble organomineral complexes [31,55].
The promptly bioavailable contents of Cu and Tl in the soil increased after plant growing cycles, whereas the opposite was observed for Zn bioavailable fraction ( Table 2). A significant reduction of bioavailable Zn (from 43 to 92%) was also observed by Rocco et al. [56] in two soils potentially polluted by Cu and Zn of Campania (southern Italy), after one year of phytoremediation with Eucalyptus camedulensis L. and Populus nigra L. The increased bioavailability of Tl after plant growth might be related to the enhanced bioavailability of K triggered by rhizodepositions and soil microbiota at the soil-root interface, since the geochemistry of two elements in the soil is very similar [57]. Moreover, since Tl seems not to be strongly complexed by humic acids in soils [58], in the OM-rich rhizospheric soil, Tl can move from soil to pore waters and thus become bioavailable to plants [8,59]. No statistically significant variation from the start point was found for bioavailable Pb after plant growing cycles, as well as for bioavailable Cd whose concentrations after plant growth were below the instrumental detection limit (0.005 mg kg −1 ). The perennial wall rocket was able to rise 4.6-fold the initial Cu bioavailable fraction after three harvest cycles, but also white campion and lettuce plants produced a significant increase of bioavailable Cu (3.3-fold and 2.5-fold higher than the start point, respectively). This enhanced Cu bioavailability after plant growth may be due to the strong affinity of the metal for dissolved organic matter (DOM). Rhizodepositions and the consequent increase in microbial activity usually contribute to raise the concentration of DOM in the rhizospheric soil which often leads to a concomitant increase of Cu promptly bioavailable content [60]. Likewise, the promptly bioavailable content of Tl was raised of 57% after three harvest cycles of lettuce and perennial wall rocket and of 14% after three harvest cycles of white campion. On the other hand, the initial bioavailable content of Zn was lowered 49, 57 and 72% by white campion, lettuce and perennial wall rocket plants, respectively.

Plant Biomass and PTE Uptake in Relation to Soil Promptly Bioavailable Contents
The dried biomass and PTE concentrations in shoots of the three plant species, at the first, second and third harvests, are shown in Table 3.
The dried biomass of the three plants was not significantly different between the first and second harvests, while a reduction of 51% for lettuce and of 37% for white campion at the third harvest occurred in comparison to the second harvest (Table 3). Lettuce requires a sufficient supply of essential macronutrients (particularly nitrogen) generally by mineral fertilizers, which can be integrated with plant biostimulants [61,62]. Consequently, the biomass reduction showed by lettuce was probably related to low agronomical inputs. In contrast, perennial wall rocket showed no biomass reduction at the third harvest, showing a good adaptability to continuous harvests, as observed by Caruso et al. [21] reporting a stable biomass production up to five consecutive harvests. Cadmium concentration in all plant shoots (Table 3) were above the reference ranges observed in food plants grown in nonpolluted sites [12], despite the low bioavailability of Cd in soil ( Table 2). White campion also showed Cd concentration (fresh weight) higher than the threshold for vegetables at the third harvest (Table S5- [39]), suggesting a potential risk for human health linked to the ingestion of this plant. On average, lettuce and white campion showed the highest Cd concentration, twofold higher than perennial wall rocket (Table 4). The BAF of Cd was higher than other PTEs in most of the studied plants (Table 5), indicating its high mobility in the soil and transfer into plant tissues [50]. Furthermore, lettuce and white campion showed a BAF above unity (Table 5), thus suggesting a high Cd accumulation ability of the two crops in the studied soils [63]. Leafy vegetables such as lettuce showed a high tendency to accumulate Cd in previous studies [64,65], while white campion reported high Cd accumulation in polluted soils [28]. Copper concentration in plant tissues was higher than reference ranges measured in food plants grown on nonpolluted sites (Table 3- [12]). Additionally, perennial wall rocket showed the highest Cu concentration in shoots as compared to the other studied plants (Table 4), in accordance to the high bioavailable Cu content measured in the cultivation soil after plant growing cycles (above the trigger value settled by Germany and Slovak Republic; Table 2 [33,34,54]). Cu is a micronutrient for plants and this can justify, at least in part, the high Cu concentration in plant shoots. Indeed, many studied demonstrated as root exudates are able to mobilise Cu in the soil [66,67].
Lead concentration in plant shoots (ranging from 0.57 to 0.66 mg kg −1 ) was below the reference ranges proposed by Kabata-Pendias [12] for food plants (Table 3; Table 4). The low concentration in plants and the low mobility of this PTE in soil (Table 2) suggested a low risk of Pb transfer into the food chain. However, white campion showed Pb concentration in shoots (fresh weight) higher than the threshold for vegetables at the third harvest (Table S5- [39]), suggesting a potential risk for human health linked to the ingestion of this plant.
Thallium concentration in perennial wall rocket and white campion was higher than reference ranges proposed by Kabata-Pendias [12] for food plants. This result was consistent with the high bioavailable Tl content in the soil, above the threshold for food safety (Table 2- [33,34,54]). D'Orazio et al. [8] reported that plants belonging to Brassicacea family as perennial wall rocket may accumulate high concentrations of Tl. On the other hand, the low Tl concentration in lettuce plants was similar to that of Bunzl et al. [68] in plants grown in noncontaminated soils, to indicate that lettuce plants are sensitive to Tl and do not take easily it up, even when its bioavailability is enhanced in the rhizosphere soil. White campion showed the highest Tl concentration in shoots as compared to the other studied plants and a BAF > 1 ( Table 4, Table 5), which indicates a high Tl accumulation efficiency of this plant. Accordingly, many authors demonstrated that white campion highly accumulates Tl in its shoots and can show a Tl BAF up to 11 in soils polluted by anthropogenic Tl [27,28,69]. Our findings thus reveals that white campion can take up and accumulate high Tl concentrations also in soil where Tl derives from geological sources.
Zinc concentration in lettuce shoots at the third harvest was higher than reference ranges proposed by Kabata-Pendias [12] for food plants ( Table 3). The concentrations of Zn found in our lettuces were analogous to those observed by Antoniadis et al. [19], who grew the plants in soils showing a similar Zn mobility. Zn is an essential micronutrient for plants, moreover it is generally quite mobile and bioavailable in the soil [67,70], hence it is conceivable that plants as lettuce can take up and assimilate high concentrations of this metal from the rhizosphere soil.
In general, PTE concentrations in plant shoots tended to increase in the second and third harvests, except for Tl in lettuce and white campion ( Table 3). The promptly bioavailable fractions of Cu and Tl in the soil raised from the time T0 to the end of the experiment, after plant growth and three consecutive harvests, while on the contrary those of Cd and Zn decreased over time (Table 2). This behaviour for Cd and Zn suggested that the uptake of the promptly bioavailable fractions of these two metals by plants was intense and faster than the progressive conversion of less bioavailable pools into easily available pools in the soil [71]. On the contrary, the promptly bioavailable contents of Cu and Tl in the soil were enhanced over time probably because the less available pools were easily turned into easily available pools by root exudates and soil biota [72].
These trends were also ascertained by correlating the metal promptly bioavailable contents in the soil and PTE concentrations in plants (Table 6).
A statistically significant positive soil vs plant correlation was observed for Cu and Tl (only for perennial wall rocket), while a statistically significant negative soil vs plant correlation was noticed for Zn (Table 6). A similar positive correlation between NH 4 NO 3 -extractable fraction in soil and total content in plant was reported for Cu and Tl in previous studies, to indicate that the fluctuations of promptly bioavailable contents of metals in the soil are generally related to the amounts taken up and assimilated by the plants [73,74].

Health Risk Assessment
The risk of non-carcinogenic toxicity to different age classes of humans due to dietary intake of PTEs from the studied crops are shown in Table 7.  In bold values >1.
Thallium was the element with the highest risk for human health, representing the 85%, 94% and 99% of the hazard index (HI), respectively, for lettuce, perennial wall rocket and white campion. This is a consequence of both high Tl concentration in plants and the lowest oral reference dose of Tl (0.001 mg kg −1 ), if compared to the other PTEs [75]. Thallium HQ was above 1 only for white campion ranging from 34.7 for elderly to 74.6 for children (Table 7) indicating a significant non-carcinogenic risk for human health. HQs found for Tl in our pot trial were similar to those of other studies in Tl-rich soils, this means that the possible intake of white campion may pose a serious risk for human health if grown in a soil rich in Tl of geogenic nature. In particular, Wang et al. [76] reported HQ up to 50 for children and 30 for adults in food plants grown on soils with total Tl concentrations up to 177 mg kg −1 . Similarly, Huang et al. [77], reported Tl HQ of 57.6 in soil with Tl concentration of 6.3 mg kg −1 . The noncarcinogenic hazard indexes (HI) related to the intake of lettuce and perennial wall rocket were below the safe level (<1). On the contrary, the HI for white campion was above 1, ranging from 34.8 for elderly to 74.9 for children (Table 7), thus indicating a potential noncarcinogenic risk due to the ingestion of this plant. Results of the health risk assessment for white campion also showed that children are at higher risk to PTE exposure than adults and elderly. This is essentially due to the high food ingestion of children in proportion to their body size and to the low ability to metabolize, detoxify and excrete toxins; in other words, a not-dangerous dose for an adult may be dangerous for a child [50,78].
Health risk assessment was also calculated for farm workers based on two exposure pathways: direct soil ingestion and soil dermal contact (Table 8). The risk linked to each PTE (HQ) and the total risk (HI) were below 1 for both the exposure pathways, thus there is no risk for farmers working on the studied farmland. The low risk is mostly related to the low exposure frequency of farm workers that in turn limits the exposure to PTEs [79]. As for dietary risk assessment, Tl was the most risky element for soil ingestion and dermal contact as well, representing 84% and 50%, respectively, of the hazard index (HI) ( Table 8). The high toxicity of this element, showing a very low oral and dermal RfD (0.001 µg kg −1 ), is the main responsible of this prominent contribution to the overall risk [47,48,75].
Only the carcinogenic risk of Pb was calculated, since the carcinogenic slope factors for Cd, Cu, Tl and Zn are not available [47]. The cancer risk of Pb through dietary intake of the three plants was always lower than 1 × 10 −6 , thus resulting negligible for all the classes of humans (Figure 1 [51]). Similarly, cancer risk of Pb for farm workers through ingestion and dermal contact of soil were insignificant, with values of 9.2 × 10 −7 and 4.1 × 10 −9 , respectively [51].

Conclusions
This study aimed to assess the potential risks for human health deriving from the agriculture use of a soil located in the volcanic metropolitan area of Naples (southern Italy) and showing pseudototal contents of Tl and Pb of geogenic nature above screening values for agricultural soils. The soil extraction by 1M NH4NO3 basically revealed that all the PTEs, including Tl and Pb, were poorly bioavailable in the soil. This behaviour is essentially due to the geogenic nature of all the PTEs in the soil, which are likely occluded or interlayered in the soil mineral lattice. Nevertheless, repeated croppings of L. sativa L., D. tenuifolia L. and S. latifolia Poir. raised the promptly bioavailable content of Cu and Tl in the soil above internationally recognized trigger values and accumulated relatively high concentrations of Cd, Cu, Tl and Zn in shoots. However, acceptable risks (HI < 1) were found, arising from the dietary intake of lettuce and perennial wall rocket, soil ingestion and dermal contact by farm workers. Only the ingestion of white campion might lead to significant health risks, in particular for children. No carcinogenic risks due to Pb exposure were found.
The findings of this work demand the proper management of soil used for agricultural purposes, in order to minimise the possible transfer of Tl and other PTEs to food crops and hence to consumers

Conclusions
This study aimed to assess the potential risks for human health deriving from the agriculture use of a soil located in the volcanic metropolitan area of Naples (southern Italy) and showing pseudototal contents of Tl and Pb of geogenic nature above screening values for agricultural soils. The soil extraction by 1M NH 4 NO 3 basically revealed that all the PTEs, including Tl and Pb, were poorly bioavailable in the soil. This behaviour is essentially due to the geogenic nature of all the PTEs in the soil, which are likely occluded or interlayered in the soil mineral lattice. Nevertheless, repeated croppings of L. sativa L., D. tenuifolia L. and S. latifolia Poir. raised the promptly bioavailable content of Cu and Tl in the soil above internationally recognized trigger values and accumulated relatively high concentrations of Cd, Cu, Tl and Zn in shoots. However, acceptable risks (HI < 1) were found, arising from the dietary intake of lettuce and perennial wall rocket, soil ingestion and dermal contact by farm workers. Only the ingestion of white campion might lead to significant health risks, in particular for children. No carcinogenic risks due to Pb exposure were found.
The findings of this work demand the proper management of soil used for agricultural purposes, in order to minimise the possible transfer of Tl and other PTEs to food crops and hence to consumers eating vegetables grown on site or farmers working out in the field. It is evident that the selection of food crops is crucial to cultivate on site, since different plant species can differently affect the mobility and bioavailability of Tl and other PTEs in the soil by their own rhizodepositions, taking up and accumulating the metals in the edible portions to a different extent.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4395/10/6/890/s1, Figure S1: Map of the studied site and sampling points (top) in the rural area of Afragola (Naples, Campania, Italy) and greenhouse pot experiment with L. sativa, D. tenuifolia and S. latifolia. Table S1: Oral reference doses (RfDo) related to non-carcinogenic health risk and oral slope factor (SFo; unitless) related to carcinogenic health risk. Table S2: Reference parameters used in health risk assessment on the consumption of vegetables by children, teenagers, adults and elderly. Table S3: Reference parameters used in health risk assessment on the ingestion and dermal contact of soil by farm workers. Table S4: Pseudo-total content of PTEs (mg kg −1 ) in the eight discrete soil samples collected from the studied agricultural soil. Table S5: PTEs concentrations (mg kg −1 f. w.) in shoots of L. sativa, D. tenuifolia and S. latifolia from the first, second and third harvest.