Sample collection
A total of 181 samples of nine crop species were randomly collected from farmers from the North Central Province of Sri Lanka during two growing cycles of the year 2016. The basic information such as sampling location, common name, variety etc. was collected at same time. The samples were prepared as per general procedures followed by consumers. The inedible parts were first removed, the edible portion was then washed thrice in distilled water, dried at room temperature, packed in polythene bags, sealed and transported to the laboratory for further analysis. While people in the area use groundwater or tap water to prepare the cereals in the kitchen, in the study we used distilled water to avoid any contamination from water. The type and number of samples were as follow; mustard (10), maize (45), finger millet (50), sesame (26), cowpea/black-eyed pea (17), Urad dal/ split bean (11), foxtail millet/Thanahal (7), long bean/Mea (5) and green gram (10).
Sample preparation and analysis
Samples were ground using a domestic food blender and sieved. Then 0.5 g of each sample was mineralized in Teflon digestion vessels in a closed microwave accelerated system (CEM, MARS 6, USA) using nitric acid (Sigma, USA) as a reagent. Analysis of As and Cd was carried out by atomic absorption spectrophotometry, AAS (Varian 240 FS, Victoria, Australia) assisted by graphite tube atomizer (GTA 120). The calibration curve range (μg/L), wavelength (nm), slit width (nm) and lamp current (mA) were as follow; for Cd, 0.5–3.0, 228.8, 0.5, 4 and for As, 5.0–50.0, 193.7, 0.5, 10 respectively.
In order to check the purity of chemicals and glassware, blank samples were run with each batch. Analytical quality control was achieved by using the Certified Reference Materials (CRM) IRMM 804 (rice flour) from the Institute for Reference Materials and Measurement (IRMM), Belgium. All reagent blanks, samples, and CRM were run in duplicate. Validation of the analytical method was performed using limit of detection, LOD (3 x standard deviation, SD) and limit of quantification, LOQ (10 SD), correlation coefficient of calibration curve (0.999 or better), relative standard deviation, RSD between duplicates (10% or below) and recovery of CRM (80–120%).
The data were analysed using Microsoft Excel (MS office, 2010) and SPSS (Version 17) software. The mean, median, minimum-maximum value and standard deviation were calculated for As and Cd in each cereal/legume type. The mean comparison ANOVA test was applied to find the significant difference between species and metals (p-value < 0.05). All samples were considered for the statistical analysis and when the sample reading was below the LOQ values, half of LOD values were considered as a sample reading.
Health risk assessment
The health risk assessment based on the legumes and cereals were conducted according to the Integrated Risk Information System (IRIS) of the US Environmental Protection Agency (US EPA) guidance (IRIS, 2018).
Provisional Tolerable Weekly Intake (PTWI) of As and Cd through consumption of cereals and legumes was calculated using the following equation (Popovic et al., 2018).
$$ \mathrm{PTWI}=\frac{C\times WC}{BW} $$
(1)
C = concentration of metals in legumes and cereals (mg/kg).
WC = average per capita cereal and legume consumption per week (kg/week).
BW = average body weight (kg).
The exposure rate of Cd and As through the legumes and cereals was calculated using the following equation (Hensawang and Chanpiwat, 2017)
$$ \mathrm{Exposure}\ \mathrm{rate}\ \left(\mathrm{ER}\right)=\frac{C\times IR\times ED}{BW\times AT} $$
(2)
C = concentration of metals in legumes and cereals (mg/kg).
IR = amount of legumes and cereal consumed (kg/year).
ED = period of time in contact with the metal (year).
BW = average body weight (kg).
AT = average time or period of exposure (day).
Non-carcinogenic health risk was calculated based on the Hazard Quotient (HQ) value using the equation is below (Hensawang and Chanpiwat, 2017).
$$ \mathrm{HQ}=\frac{ER}{RfD} $$
(3)
RfD = reference dose for Cd and As.
The carcinogenic risk was also calculated based on the following equation (Hensawang and Chanpiwat, 2017).
$$ \mathrm{AELCR}=\frac{Exposure\ rate\times SF}{DL\times 365} $$
(4)
AELCR = annual excess lifetime cancer risk.
SF = cancer slope factor (mg/kg, day).
DL = average human longevity.
Due to non-availability of the per capita consumption data, we assume that our population consume World Health Organization (WHO) recommended amounts for pulses, nuts, and seeds groups, viz., 30 g/day at least as a part of the 400 g of fruit and vegetables (Nishida et al., 2004). The mean body weight of Sri Lankan male is 61.4 kg and the female is 54.6 kg (MHNIM, 2015). The average Sri Lankan life expectancy is 75 years (CBSL, 2017). The RfD values for As (inorganic) and Cd are 3 × 10− 4 and 1 × 10− 3 mg/kg.day respectively and SF for As is 1.5 mg/kg.day (Hensawang and Chanpiwat, 2017).