Introduction

A significant quantity of tailings and wastewater is produced during the mining exploitation and ore smelting processes which results in severe metal pollution in the surface water, groundwater, soils, and food crops (Amoakwah et al. 2020; Rodríguez-Estival et al. 2019). The metals in the mining environment pose a health risk to the population in the vicinity through different pathways like ingestion, dermal contact, or inhalation of dust and soil; ingestion and dermal contact of water; and consumption of food crops grown in contaminated soils or irrigated with polluted water (Doabi et al. 2018; Park and Choi 2013; Wu et al. 2015).

Consumption of metal-contaminated crops, particularly vegetables, from a mining dominated area form a major exposure pathway for the local consumers. Metal concentrations in plants grown in mine contaminated soils are considerably elevated compared to the reference soils (Cao et al. 2016). Nevertheless, in addition to soil, irrigation water and atmospheric deposition are impending sources of heavy metals in the vegetables (Bi et al. 2018; El-Radaideh and Al-Taani 2018; Li et al. 2012). However, in a similar environment, concentrations of metals in the vegetables are also dependent on plant categories and species.

Leafy vegetables accrue elevated metal concentrations compared to rootstalk and fruiting vegetables. Foliar deposition of metals through atmospheric dust emitted from anthropogenic sources form the governing pathway of metal uptake for the leafy vegetables (Luo et al. 2011; Noh and Jeong 2019; Xu et al. 2015). Also, the leafy vegetables have higher transportation rates of metals compared to other vegetable types (Hu et al. 2017). Crop-specific prediction models have been used by many researchers to estimate the uptake and accumulation of metals by the crops and thus to predict the dietary hazards of metals (Eid et al. 2018, 2020; Kumar et al. 2019; Legind and Trapp 2010). The empirical models can also be used for practical applications like soil remediation and health risk assessment (Kumar et al. 2020; Wang et al. 2015).

Singhbhum craton and shear zones are the highly mineralized areas of India harboring one of the biggest sources of high-grade iron ore and copper, respectively. These areas are under immense load of metal mining activities and subsequently their processing industries which together pave the way for metal contamination of the environs and consequently the food chains. There are a number of research works on metals in soil-vegetable systems but not many deals with the multifaceted system comprising of soil, dust, water, and leafy vegetables in a mining environment. In this complex system, human health risk needs to be quantified considering all the applicable pathways. Since there were multiple pathways of metals exposure, accounting for all of them is mandatory for the overall risk assessment for the populace in the vicinity of mining areas. In addition, a stepwise multivariate linear regression (MVLR) model to evaluate the concentration of metals in the leafy vegetables from metals in soil, irrigation water, and atmospheric dust was also imperative. The species-specific MVLR models need to be versatile such that they can be applicable in diverse mining areas. The MVLR model can establish the sources of the metals in the leafy vegetables also. In view of these facts, the present study was carried out to examine the metal concentrations in the locally growing leafy vegetables in iron and copper mining areas along with soil, irrigating water, and atmospheric dust.

The objectives of the study envisaged to (1) investigate the metal contamination in the intricate system of soils, irrigating water, atmospheric dust, and leafy vegetables in two metal mining environment; (2) determine the contribution of the different matrices, i.e., soil, water, and dust on the concentration of metals in the leafy vegetables using stepwise multivariate linear regression (MVLR); (3) to predict the concentration of metals in the leafy vegetables using MVLR; and (4) quantify the health risk to the local populace considering the multi pathways of metal exposure, i.e., ingestion, dermal contact, or inhalation of dust and soil; ingestion and dermal contact of water; and consumption of leafy vegetables.

Materials and method

Study area

The study was taken up in the iron mining areas of Singhbhum craton and copper mining areas of Singhbhum Shear Zone known for their high-grade ores of iron and copper. Both the study areas are situated in the Southern Jharkhand of Eastern India (Fig. 1) and form the India’s most mineralized and industrialized zones.

Fig. 1
figure 1

Map of the study area with sampling locations (modified from Giri et al. 2020)

Full size image

The Singhbhum craton consists of Precambrian rocks that are circumscribed by the arcuate Singhbhum Shear Zone in the north and Sukinda thrust in the south. The craton is a part of volcano-sedimentary Archean Iron ore group (IOG) that harbors one of the largest iron and manganese deposits of India (Upadhyay et al. 2010). A number of phases of Mesoarchean granitic entities, i.e., Singhbhum granite and Bonai granite are found within the craton. The Singhbhum granite is the most widespread unit and includes synformal keels of banded iron formations (BIF)–bearing greenstone belts (Saha 1994). The minerals that are found in the area consist of hematite, magnetite, goethite, chromites, romanechite, cryptomelane, pyrolusite, gibbsite, kaolinite, and quartz.

The Singhbhum shear zone forms an arcuate vastly deformed linear zone in the Singhbhum crustal province and is known to be one of the most potential sulfide-bearing stretch of India. The copper mineralization along Singhbhum copper belt is located along the Dhanjori group of rocks at south of shear zone and Singhbhum group of rocks at north of shear zone. The main rock types of the region are soda granite, Dhanjori lavas, and epidiorite-schist. The sulfide minerals of the zone consists predominantly of chalcopyrite, followed by pyrite and pyrrhotite, accompanied by magnetite and minor pentlandite, millerite, violarite, sphalerite, and molybdenite (Dunn and Dey 1942; Mukherjee 1968).

Collection of samples and processing for metal analysis

Locally grown leafy vegetables (31 samples) that are commonly consumed by the local population were collected comprising of 4 varieties (Red amaranthus, Spinach, Drumstick leaves, and Mustard leaves) from six villages each from both the study areas. Supplementary Table 1 displays the local, English, and scientific names and the collection sites for each. For each leafy vegetable, multiple picks were collected and then composited to one sample. The leaves were washed many times with the tap water to remove the dust adhered on to the leaf surfaces. Final wash was done with the ultrapure water, and then, the samples were air-dried followed by drying in an oven at 80 °C to a constant weight. 0.5 g of the grinded and homogenized leaves were subjected to acid digestion in a microwave digestion system (model: Anton Paar Multiwave 3000) using concentrated HNO3 and 30% H2O2.

Composite soil samples were prepared for each site. About 1 kg soil sample was collected from the plough layer (0–15 cm) from various points from where the vegetables were collected. These soil samples were mixed together, air-dried at room temperature (25–30 °C), and sieved through 2 mm screen to separate out the pebbles. The fraction that passed through the 2 mm screen was further quarter coned to reduce the size to 100 g which was oven-dried at 110 °C for 24 h and crushed and the whole quantity was passed through a 200 mesh screen. 0.25 g of the sieved soil samples was digested in a microwave digestion system using a digestion solution of HNO3 (4.5 ml), HF (2 ml), HCl (1 ml), and H2O2 (0.5 ml) using the EPA 3052 method (USEPA 1996).

Atmospheric dust samples were also collected from the selected sites using the gravity technique (BIS 2006). Dustfall sampling devices were assembled with high-density plastic buckets mounted on an iron tripod stand 1.2 m above the sampling floor to avoid the collection of dust picked up by wind eddies. It was exposed to the atmosphere for a month with a bird-shield. After the collection, the samples were oven-dried at 110 °C for 24 h and crushed, and the whole quantity was passed through a 200 mesh screen. The digestion procedure was the same as the soil samples.

From each site, the water used for irrigating the vegetable crops was collected, filtered, and preserved at pH < 2 by adding suprapure HNO3 for heavy metal analyses (Radojevic and Bashkin 1999).

Heavy metal analyses

As, Cd, Co, Cr, Cu, Fe, Mn, Ni, Pb, and Zn were determined in the processed vegetable, soil, dust, and water samples using inductively coupled plasma mass spectrometer (ICP-MS, Model: Perkin Elmer Elan DRC-e). The stability of the calibration curve was checked and confirmed after every 15th sample using calibration blank and a standard not used in the calibration.

Quality assurance

Spinach leaves standard reference material (NIST SRM 1570a) was used for validation of the analytical procedure. The analytical results for water were validated using standard reference material of natural water (NIST-1640a) while for soil and dust, estuarine sediment (NIST-1646a) was used. The % recovery of the metals for NIST-1570a, NIST-1640a, and NIST-1646a varied from 95.6 to 105.9, 97.1 to 104.0, and 90.9 to 108.8, respectively (Supp. Table 2).

Non-carcinogenic risk assessment

There are multiple pathways of metal exposure to the local population. The pathways which are considered in the study are as follows:

  1. i.

    Ingestion pathway for the leafy vegetables

  2. ii.

    Ingestion and dermal pathways for water

  3. iii.

    Ingestion, dermal, and inhalation pathways for soil

  4. iv.

    Ingestion, dermal, and inhalation pathways for dust

Non-carcinogenic risk posed by each metal present in different matrices (vegetable, water, soil, and dust) was typified by a hazard quotient (HQ) which is the ratio of average daily dose (ADD) to the reference doses for ingestion, dermal, or inhalation pathways (RfDing or RfDderm or RfCi) (Eq. 1)

$$ \mathrm{HQ}=\mathrm{ADD}/{\mathrm{RfD}}_{\mathrm{i}\mathrm{ng}}\ \mathrm{or}\ {\mathrm{RfD}}_{\mathrm{derm}}\ \mathrm{or}\ {\mathrm{RfC}}_{\mathrm{i}} $$
(1)

Expressions for calculating HQ for different pathways have been listed in Table 1. Reference doses for the ingestion pathway (RfDing in mg/kg/day), dermal pathway (RfDderm in mg/kg/day), and inhalation pathway (RfC in mg/m3) are obtained from USEPA (2011). These represent the safe doses with reference to a particular metal that a human body can bear without any probable adverse health effects. The input parameters for calculating HQ are provided in Supp. Table 3. All the input values originated from USEPA (USEPA 2002, 2004, 2011, 2017) except for IRveg, IRwater, and BW which were taken in Indian context from NSSO 2014; Dang et al. (1994); and NFI 2010, respectively.

Table 1 Calculation of hazard quotient (HQ) (USEPA 2004, 2011)
Full size table

ADD and RfD of different metals are different in different pathways; thus, corresponding HQ(s) should also be different. Thus, total hazard index (HI) considering multiple metals and multiple pathways of exposure comprising “n” number of pathways and “m” number of metals may be expressed as Eq. 2,

$$ \mathrm{HI}={\sum}_{j=1}^n{\sum}_{k=1}^m{HQ}_j^k $$
(2)

Thus, if we consider the water matrix, comprising of four heavy metals, namely, Fe, Mn, Zn, and Cu, the comprehensive hazard index arising from ingestion and dermal absorption pathways might be expressed as,

$$ \mathrm{HI}=\left[{{\mathrm{HQ}}^{\mathrm{Fe}}}_{\mathrm{ing}}+{{\mathrm{HQ}}^{\mathrm{Mn}}}_{\mathrm{ing}}+{{\mathrm{HQ}}^{\mathrm{Zn}}}_{\mathrm{ing}}+{{\mathrm{HQ}}^{\mathrm{Cu}}}_{\mathrm{ing}}\right]+\left[{{\mathrm{HQ}}^{\mathrm{Fe}}}_{\mathrm{derm}}+{{\mathrm{HQ}}^{\mathrm{Mn}}}_{\mathrm{derm}}+{{\mathrm{HQ}}^{\mathrm{Zn}}}_{\mathrm{derm}}+{{\mathrm{HQ}}^{\mathrm{Cu}}}_{\mathrm{derm}}\right] $$

Likewise, the HI for all the considered matrices and pathways may be calculated.

HQjk > 1 implies that the health risk associated with a particular metal in a particular matrix through a particular pathway is more than the accepted level.

Similarly, HI > 1 also implies that total risk exceeds the acceptable limit even though each of its constituent components (HQjk) may be < 1.

Proposed metal transport model

Stepwise linear multiple regression analyses were used to recognize the influential environmental parameters, i.e., cultivated soils, irrigation waters, and atmospheric dust to ascertain models for predicting metal concentrations in the leafy vegetables (Xu et al. 2015; Yang et al. 2014). Stepwise MVLR checks the best combination of the independent variables to predict the dependent variable and discards the insignificant variables. F test was used in regression analysis to compare the disparity between the actual data and the modeled data derived from the fitted equation. The Durbin Watson values are also studied to observe the autocorrelation in the residuals.

The simplest possible model for describing transport of metal from soil, water, and dust to the leafy vegetables was envisaged. Since the sources of metals in vegetables are primarily soil along with irrigation water and atmospheric dust, their inter relationship may be functionally expressed as,

$$ {V}_i=f\left({X}_i,{Y}_i,{Z}_i\right) $$
(3)

where Vi, Xi, Yi, and Zi are the ith heavy metal concentration in the vegetable, water, soil, and dust, respectively.

A multivariate linear regression (MVLR) model was chosen as the function f(Xi, Yi, Zi) which may be expressed as

$$ {V}_i={a}_0+{a}_1.{X}_i+{a}_2.{Y}_i+{a}_3.{Z}_i $$
(4)

where a0 is the constant and a1, a2, and a3 are the coefficients for water, soil, and dust, respectively.

Model validation

Multiple regression modeling is a statistical tool used extensively for the prediction analysis and examines the effect of more than one independent variable over a response (DeForest et al. 2018). However, it is imperative to make certain that the prediction model works appropriately with real data and predicted data. For the performance evaluation, three assessment tools were adopted:

  1. i.

    A regression plot (scatter plot) of actual values against the modeled values allows a visual assessment of the model performance. More the scatters tend to concentrate close to the equiline (denoted as 1:1 line), better is the agreement between the actual and predicted values, and thus, better is the model.

  2. ii.

    For the validation process, model efficiency (ME) is widely used to minimize error in the selection and feasibility of a model (Eid et al. 2018). It is also known as Nash and Sutcliffe coefficient of efficiency. The ME of regression models was calculated using Eq. 5:

$$ ME=1-\frac{\sum_{i=1}^N{\left({O}_i-{P}_i\right)}^2}{\sum_{i-1}^N{\left({O}_i-\overline{O}\right)}^2} $$
(5)

where Oi and Pi are the observed values and predicted values, respectively, while Ō is the mean of the observed values. The sample size is “N.”

  1. iii.

    To quantify the prediction error, the root mean square error (RMSE) is used which is calculated as Eq. 6

$$ \sqrt{\frac{\sum_{i=1}^N{\left({O}_i-{P}_i\right)}^2}{N}} $$
(6)

where the variables Oi , Pi, and N are same as in Equation 5.

The criteria for the goodness of fit evaluation for ME and RMSE are given in Supplementary Table 4.

Results and discussion

Heavy metals in leafy vegetables, irrigation water, soil, and atmospheric dust samples

The concentrations of the 10 metals (As, Cd, Co, Cr, Cu, Fe, Mn, Ni, Pb, and Zn) in the different environmental matrices for both the mining areas are depicted in Tables 2 and 3. The samples were collected from six locations each from the iron mining areas of West Singhbhum (Table 2) and copper mining areas of East Singhbhum (Table 3). It is a common observation that the concentrations of the metals in the environmental matrices do not follow a normal distribution which can be attributed to the extreme values that are often encountered. Since the central tendency is better represented by the geometric mean for a skewed data, it has been used in the study to signify the average concentrations of the metals. Compared to the Indian standards for food (Awasthi 2000), none of the metal concentrations in the leafy vegetables from both the areas exceeded the standards except for Cu in some samples of Red amaranthus collected from copper mining areas.

Table 2 Concentration of metals in leafy vegetables, water, soil, and dust in the iron mining areas of West Singhbhum (unit: water—μg/L; soil, dust, food—mg/kg)
Full size table
Table 3 Concentration of metals in leafy vegetables, water, soil, and dust in the copper mining areas of East Singhbhum (unit: water—μg/L; soil, dust, food—mg/kg)
Full size table

Water, soil, and free-fall atmospheric dust samples were collected from the same locations at the same time as that of the collection of the leafy vegetables. The metal concentrations of the soil and dust samples were compared to the metals in the average shale values (Turekian and Wedepohl 1961) to reveal that almost all the metals surpassed the values which can be attributed to the metal bearing deposits of both the study areas in conjunction with the mining and industrial activities. The irrigation water that was used for the vegetable crops in the region was mostly groundwater, which was also used for drinking purposes. However, in one location of copper mining area (Kendadih), the mine water from the copper mine was used for the irrigation. Mine water is the water emanating from the mines and is often loaded with metals in the case of metal mines since it is being dewatered from the working face of the mines. The metal concentrations did not exceed the irrigation water standards (FAO 1985) for any of the locations but for the mine water from Kendadih which exceeded the standards for Cu, Mn, and Ni. There is a marked difference in the metal concentrations between the two study areas as for most of the matrices; higher Cu, Ni, and Co concentrations were observed in the copper mining areas while Fe and Mn were higher in the iron mining areas. These groups of metals are associated with the particular geological formations of both areas along with the respective metal mining activities.

Human health risk assessment

The health risk was assessed for 9 different pathways for the local population. The pathways comprised of ingestion of vegetables, water, soil, and dust; dermal contact with water, soil, and dust; and inhalation pathway for soil and dust. The hazard quotients for the metals for considering all the pathways of exposure did not exceed unity for the iron ore mining area except for Co and Cr at one location and Fe at two locations through ingestion of vegetables (Fig. 2a). In the case of the copper mining areas, the HQ for the metals As, Co, Cu, and Mn surpassed the reference value of one for some of the pathways at some of the locations. All of these incidences were related to ingestion of leafy vegetables and water which implied considerable risk to the local population due to the consumption of metal-contaminated vegetables and water (Fig. 2b).

Fig. 2
figure 2

Boxplots of hazard quotient (HQ) of metals considering the 9 pathways in a iron mining areas of West Singhbhum, b copper mining areas of East Singhbhum

Full size image

The combined risk due to the multiple metals was evaluated by summation of the HQs to obtain the hazard index (HI) (Table 4). The HI for the different pathways for the different locations of the two study areas depicted that the most important route of exposure to the metals was the ingestion of the leafy vegetables, irrespective of locations and area. Intake of the leafy vegetables contributed 71.9% and 80.7% of the total risk for the iron and copper mining areas, respectively. In the iron mining areas, the next significant pathway was ingestion of soil (8.6% of total risk) followed by ingestion of dust and drinking of water. In the area, the groundwater is used for irrigating the vegetable fields which is also used for drinking as well. In the case of copper mining areas, the risk apportionment is a little different, where the second most relevant pathway for exposure of metals was the drinking water pathway contributing 10.6% of the total risk. This has been calculated excluding the risk from the water pathway of the Kendadih location where mine water was used for irrigation and which is not used for drinking. The other important pathways were the ingestion of dust and soil. Ingestion of soil and dust is found to be an important pathway of metal exposure for both the study areas. Ingestion exposure can occur via the intentional or inadvertent non-dietary ingestion of soil and dust on surfaces or objects that are contacted via hand-to-mouth or object-to-mouth activity (USEPA 2017).

Table 4 Location wise hazard index for different pathways for iron mining areas of West Singhbhum and copper mining areas of East Singhbhum
Full size table

Considering the HIs, it can be inferred that for both the mining areas, the ingestion pathway is the primary exposure route irrespective of the matrix. Dermal and inhalation routes form the secondary exposure mode. Considering both the mining areas, the overall risk was higher for the locations of the copper mining area compared to the iron mining areas. The overall risk as represented by HI was more than double for the copper mining area (12.1) than the iron mining area (5.13) which may be attributed to the higher concentrations of Cu and Co in environs of the copper mining areas. Cu and Co are known to have a greater risk on human health compared to Fe and Mn which are found to be at large in iron mining areas.

Regression models

Regression models were formed to estimate the metal concentrations in the leafy vegetables based on the metal concentrations in different environmental matrices. Stepwise multivariate linear regression analysis was performed on the basis of the three factors, i.e., metal concentrations in the soil, irrigation water, and atmospheric dust. The results and prediction precisions of the models are depicted in Table 5. The ANOVA results revealed that the correlation between the measured and predicted metal concentrations by model equations was statistically significant with high F values and low P values (level of significance P < 0.001). The Durbin Watson values are also between 1.5 and 2.5 suggesting the absence of autocorrelation in the residuals (Rafiee et al. 2010). The high (> 0.7) coefficient of determination values (R2) which is an indicative of the goodness of fit of the developed models is supportive for the prediction of effective metal uptake by the leafy vegetables from the different environmental components. The R2 of 0.725 was obtained for the model when metals for all the samples of leafy vegetables irrespective of the species were considered cumulatively. However, the R2 values further improved when the different species of the leafy vegetables were considered individually thus advocating for species-specific regression models for metal uptake. The R2 for all species-specific models ranged between 0.821 and 0.980, the highest being for the Spinach.

Table 5 Results and prediction precisions of the species-specific multivariate linear regression (MVLR) models for predicting metals in leafy vegetables
Full size table

The constant term or the intercept (a0) is insignificant for all the models and can be omitted implying that the concentration of metals in the leafy vegetables is totally dependent on the concentrations of metals in soil, water, and dust. The possibility of other sources of metals in the leafy vegetables is negligible. In view of the coefficients of the models, it can be suggested that each and every one of the predictor variables (metal concentrations of water, soil, and dust) did not have a significant contribution to the metal uptake of all the leafy vegetables. The concentration of metals in the mustard leaves and red amaranthus was dependant on all three matrices. However, the highest contribution is made from the metals in the irrigation water. Same is the case when the model is developed irrespective of the species. However, in the case of spinach, the metal uptake was contributed from water and soil and the role of atmospheric dust was insignificant. Foliar deposition of metals through atmospheric dust is more pronounced in rough leaf surfaces compared to smooth ones (Schreiber and Schönherr 1992). The spinach leaves are smooth compared to red amaranthus and mustard leaves, and thus, the effect of atmospheric dust was insignificant in spinach leaves. For the drumstick leaves, the metal content in the soil was the only causative factor of the metal concentration in the leaves. The three leafy vegetables other than drumstick leaves are short life–cycled plants; however, the drumstick is perennial trees with smooth leaves which are used as leafy vegetable by the local populace. Being a perennial tree, it is not needed to be irrigated as in the case of other three species. This might be the explanation for fact that the stepwise regression model for drumstick leaves only retained the metal content in the soil as a contributing factor eliminating the water and dust factors. The same models relevant for both the mining areas prove its versatility of being useful in different environmental conditions with variable concentrations of metals in different matrices like soil, water, and atmospheric dust. However, the models need to be species-specific. Thus, the models for the prediction of metals in the 4 considered leafy vegetables are depicted as Equations 7–10.

$$ \mathrm{Metal}\ \left(\mathrm{mustard}\ \mathrm{leaves}\right)=0.0284\left[\mathrm{water}\right]+0.00058\left[\mathrm{soil}\right]+0.00094\left[\mathrm{dust}\right] $$
(7)
$$ \mathrm{Metal}\ \left(\mathrm{R}.\mathrm{amaranthus}\right)=0.02199\left[\mathrm{water}\right]+0.00079\left[\mathrm{soil}\right]+0.00162\left[\mathrm{dust}\right] $$
(8)
$$ \mathrm{Metal}\ \left(\mathrm{spinach}\right)=0.058\left[\mathrm{water}\right]+0.00032\left[\mathrm{soil}\right] $$
(9)
$$ \mathrm{Metal}\ \left(\mathrm{drumstick}\ \mathrm{leaves}\right)=0.00269\left[\mathrm{soil}\right] $$
(10)

where [water] is concentration of metals in irrigating water; [soil] is concentration of metals in soil; and [dust] is concentration of metals in atmospheric dust;

Model validation

The validation of the models was performed using equiline scatter plot of actual and modeled values, model efficiency (ME) values, and root mean square error (RMSE) values. The scatter plot (Fig. 3) depicts that most of the points are near the equiline with some exceptions which may be considered outliers. However, the outliers have been retained in the models since the extreme values (outliers) are often encountered in the environmental studies. Nevertheless, the outliers in our study did not affect the models drastically. The plot provides a high agreement in the positive correlation between the observed and predicted values for all five models. The ME values ranged from 0.720 to 0.959 which depicted good to very good goodness of fit for the models (Cheng et al. 2017). The ratio of standard deviation of observed values to the RMSE ranged from 1.89 (considering all species) to 5.03 (Spinach) which suggest that the models are acceptable to very good based on the quantification of prediction error.

Fig. 3
figure 3

Equiline scatter plot of actual and modeled values of regression models

Full size image

Conclusions

From the assessment of metal concentrations in the leafy vegetables and other associated environmental components collected from the metal mining areas, we could understand the impact of the mining and industrial activities on the food chain and subsequently the potential risk to the human health due to the contaminated environs inclusive of food items. Health risks of metals were assessed for multiple pathways to the local populace including ingestion of leafy vegetables. The hazard quotient of the metals exceeded the reference value of unity for some metals, however, more frequently due to ingestion pathways for vegetables and water. The combined risk due to multiple metals as assessed by hazard index suggested that the ingestion of leafy vegetables was the major route of exposure of metals contributing 71.9% and 80.7% of total risk for the iron and copper mining areas, respectively. The average HI value considering all the metals and pathways was calculated to be 12.1 and 5.13, respectively for copper and iron mining areas suggesting considerable risk to the local residents.

The present study also aimed to develop species-specific multivariate regression models for predicting the concentrations of metals in leafy vegetables from their concentration in the soil, irrigating water, and atmospheric dust. Correlations between the measured and predicted metal values, high model efficiency values, and low root mean square values reflected the goodness of fit of the model. Species-specific models for the leafy vegetables were found to be better in the prediction of metals uptake from the associated environs. The models will be useful for the prediction of metal concentrations in the leafy vegetables, and thus, the possible human risks can be identified. The models would also help to understand the effect of environmental factors on the concentration of metals in the leafy vegetables. Also, the results suggest screening of irrigating water to ensure the safety of food in particular leafy vegetables with respect to metals.