The lyophilized vegetables and oven-dried market foods were ground into powder of 0.25 mm with a laboratory grinder (IKA® MF 10 grinder, IKA, 79219 Staufen, Germany) and stored at ambient temperature before being digested. The main digestion method used in this study to extract As from the vegetables or dish samples in the laboratory at 2iE main campus was adapted from a protocol developed by Bhatti et al. (2013). Briefly, 0.5 g of powder sample was introduced into clean and dry quartz tubes and 5 ml of HNO₃ (Fluka TraceSELECT, purity > 69.0%) was added. Samples were predigested at ambient temperature overnight under a fume hood. The subsequent day, samples were placed in a heating water bath at 80°C for 2–3 h. After this time, no fumes were visible and the volume had reduced to 1–2 ml. The samples were subsequently cooled down to ambient temperature, diluted with 10 ml of distilled water directly added to the quartz tubes, filtered through a Whatman filter paper No. 42, and further diluted with distilled water to a final volume of 25 ml.

For comparison, a second digestion method was carried out in the laboratories of Eawag for both the dried vegetable powder samples (as quality control) and potted soil samples.

In this method, each digestion tube contained 50 mg of sample, 6 ml of concentrated nitric acid (65% Suprapur, Merck AG, Switzerland), and 1 ml of hydrogen peroxide (30%, Merck AG, Switzerland). The tubes were placed in a pressurized ultraCLAVE microwave oven (MLS GmbH, Germany) and subjected to 1,000 Watts of energy for 15 min at 240°C. After cooling, the samples were removed from the microwave and diluted in 50 ml of ultrapure water (NanoPure, Barnstead, ThermoFisher Scientific Inc.).

The two digestion methods were compared for the digestion of spinach leaves and okra pods (N = 46 samples in total, including replicates, see Supplementary Figure 1 and Supplementary Table 2). For the important concentration range from 0.5 to 20 μg/g, the two digestion methods agreed sufficiently well for the determination of concentrations of As in the food samples with either of the two digestion methods. Standard reference materials (rice flour, NIST 1568b, and tomato leaves, NIST 1573a) obtained from the National Institute of Standard and Technologies (Gaithersburg, USA) were used to calibrate the digestion methods of As in our food samples. Good agreement with the mean certified values was observed for NIST 1568b (118% mean recovery) and for NIST 1573a (125% mean recovery) (Table 1).

The estimated daily intake (EDI) of As was calculated using Equation 2:

where EDI is the estimated daily intake of As (μg/kg body mass/day), As_{ediblepart, FM} is the concentration of As in fresh edible part of the vegetable (μg/g, FM), CR_vegetable is the daily vegetable consumption rate (g FM/person/day), and BM_average is the average body mass of 60 kg (Walpole et al., 2012). The concentration of As in the fresh edible part of the vegetable was calculated as a product of the concentration of As in the dry edible part of the vegetable (μg/g, DM) and the mean water content of edible part of each vegetable species according to Equation 3:

Calculation of the daily intake requires the quantification of the average daily food consumption rate. However, estimates of daily intake vary among countries and data were unavailable at the time of publication for Burkina Faso. The weighted average portion estimates of okra and tomato (respectively, 10 and 70 g FM/person/day) were taken from the study of Lehmann et al. (2017). For the other vegetables, the daily consumption rate was assumed to be 10 g/person/day for amaranth and spinach, 30 g/person/day for lettuce, and 100 g/person/day for carrot and green bean. Those assumptions (see Supplementary Table 7) were compared with the Global Nutrition and Policy Consortium (Friedman and Friedman, 2019) that indicates consumption of total vegetables of 120.5 g/person/day for Burkina Faso in 2010. This is in the same range of what is reported by Ganry and Le Guilloux (2007) for Ivory Coast and Benin, two neighboring countries, indicating an average daily vegetable consumption of 40 and 44 kg/person/year, which represents 110 and 121 g/person/day, respectively. For bean/niébé, gari, tô, rice, and yam, the daily consumption rate was assumed to be 150 g of cooked food/person/day (FAO, accessed on 13 April 2021).

The cadmium (Cd) monthly intake was calculated using Equation (4)

where EMI_Cd is the estimated monthly intake of Cd (μg/kg body mass/month), Cd_{food, FM} is the concentration of Cd in cooked food (μg/g, FM), CR_food is the daily food consumption rate (g FM/person/day), EF_r is the exposure frequency (365 days/year), BM_average is the average body mass of 60 kg, and month is the number of months (12 months/year).

As reported by Antoine et al. (2017), there are strict regulations of intakes of many trace elements by various international, national, and regional bodies. However, in 2011, the Joint FAO/ WHO Expert Committee on Food Additives (JECFA) withdrew the provisional tolerable weekly intake (PTWI) for inorganic As because it was no longer appropriate (WHO, 2011). The risks associated with consumption of foodstuff with significant levels of As were characterized following the definitions of the United States Environmental Protection Agency (US EPA). The non-carcinogenic and carcinogenic risks were characterized by the HQ and CR. A HQ of 1 or lower means adverse non-cancer effects are unlikely. The potential for adverse non-carcinogenic health effects increases for HQ > 1. The CR is “estimated as the incremental probability of an individual developing cancer over a lifetime” (Uddh-Söderberg et al., 2015). HQ and CR were calculated according to Equations 5 and 6, respectively, as defined by US EPA (IRIS, 2019),

where EDI is the estimated daily intake (mg/kg BM/day, see Equation 2), EFr is the exposure frequency (365 days/year), ED is the exposure duration (70 years), AT is the period over which exposure is averaged in days for non-carcinogens (365 days/year * number of exposure years), and RfD is the oral reference dose (0.0003 mg/kg/day). According to USEPA, RfD is “an estimate of a daily oral exposure for an acute duration to the human population (including sensitive subgroups) that is likely to be without an appreciable risk of deleterious effects during a lifetime” (US EPA glossary, accessed on 09/06/2020). OSF is the cancer oral slope factor for inorganic As (1.5 mg/kg/day) (IRIS, 2019).

In spinach, concentrations in the leaves rose sharply from <1.5 to 15.2 μg/g with increasing concentrations of As from 0 to 1,000 μg/L in the irrigation water. The mean concentrations of As in the leaves were 2.5–6.6 times higher than in the stems (Figure 1). In contrast to the expectations, the control plants irrigated with As-free water had measurable concentrations of As in their leaves (1.0 μg/g ± 0.3). Growth curves show no difference in plant heights after 52 days within the four As treatments (Supplementary Figure 5).

In lettuce, concentrations in the leaves rose from <0.5 to 6.3 μg/g with increasing As in the irrigation water. As accumulated more in the roots than in the leaves (around three times more on average). The TF (concentration of As in the shoot parts relative to the concentration in the roots) was lower than 1 (Figure 1). With increasing concentration of As in the irrigation water, leaves were drier, i.e., water content decreased.

In amaranth, concentrations in the leaves rose from <1 to 6.1 μg/g with increasing As in the irrigation water (Figure 1). Plants showed a higher concentration in their roots than in leaves (1.6 times higher on average) and stems (3 times higher). TF was lower than 1. Similarly to spinach, green bean, and tomato, concentration of As was higher in the leaf than in the stem. With increasing concentration of As in irrigation water, the water content in roots and stem decreased while it remained stable in leaves (Supplementary Figure 4).

In green bean, concentrations in the fruits rose from <1 to 3.5 μg/g with increasing As in the irrigation water. As mainly translocated to leaves, stems, and then fruits (Figure 1). For plants irrigated with As-spiked water, the concentration in fruits was on average 9.4 times lower than that in leaves and 5.4 times lower than that in the stems. The TF could not be calculated (due to a breakdown of the freeze dryer, roots could not be dried and therefore not analyzed).

In okra, concentrations in the fruits rose from <0.1 to 1.0 μg/g with increasing As in the irrigation water. Due to the same laboratory problems mentioned earlier, analysis of As was only possible in the fruits of okra. As content in fruits increased with increasing concentration of As in the irrigation water (Figure 1). The evolution of the height of the stem (measured 52 days after germination) showed that treatment with 100 and 500 μg/L seemed beneficial for plant growth, with greater heights than plants irrigated at 1,000 and 0 μg/L (Supplementary Figure 5). However, one-way ANOVA tests did not reveal that the differences in stem height were statistically significant at the p <0.05 level.

In tomato, concentrations in the fruits rose from 0.1 to 0.4 μg/g with increasing As in the irrigation water. With the exception of treatment at 100 μg/L, As mainly translocated to roots, leaves, stems, and then fruits (Figure 1). For plants irrigated with As-spiked water, the concentration of As in fruits (0.2 μg/g ± 0.1) was on average 90 times lower than in roots (22.5 μg/g ± 23.5) and 54 times lower than in leaves (13.4 μg/g ± 12.0) (Supplementary Table 5). Among all plants, the tomato had the lowest accumulation of As in its edible part with an average TF of 0.6. Tomato growth curves (Supplementary Figure 5) show that plants irrigated with As treatment at 1,000 and 500 μg/L had greater heights and greater leaf diameters compared to plants irrigated at 100 and 0 μg/L.

In carrot, concentrations in the roots rose from <1 to 4.6 μg/g with increasing As in the irrigation water. For plants irrigated with As-spiked water, the concentration of As in leaves (7.5 ± 5.5 μg/g) was on average 2.6 times higher than in roots (2.9 ± 1.8 μg/g) (Figure 1) and TF was >1.

The As content in the dried edible parts of each plant increased approximately linearly with the concentration of As in irrigation water (Figure 2; see also Supplementary Figure 2 for concentrations of As based on the FM). When examining all levels of As-spiked irrigation water used, the greatest average uptake of As based on DM was observed in spinach (8.1 ± 5.6 μg/g), lettuce (3.6 ± 2.5 μg/g), and amaranth (3.0 ± 2.5 μg/g), whereas the lowest concentrations were observed in green bean (1.5 ± 1.5 μg/g), okra (0.7 ± 0.4 μg/g), and tomato (0.3 ± 0.2 μg/g) (Supplementary Table 5). On the whole, the edible parts of leafy vegetables irrigated with As-spiked water had a higher average As content (4.9 ± 4.5 μg/g) than root vegetable (2.9 ± 2.0 μg/g) and fruit/pod vegetables (0.8 ± 1.1 μg/g). At the mid-point level of spiked irrigation water (500 μg/L), the concentrations of As in the edible parts (from highest to lowest) were as follows: spinach (6.6 ± 0.5 μg/g); lettuce (3.9 ± 0.1 μg/g); carrot (3.5 ± <0.1 μg/g); amaranth (2.2 ± <0.1 μg/g); okra (0.9 ± <0.1 μg/g); green bean (0.8 ± <0.1 μg/g); and tomato (0.2 ± <0.1 μg/g) (Supplementary Table 4).

Amaranth, green bean, lettuce, spinach, and tomato verified the accumulation trend orders of organs (root > leaf > stem > fruit) described in previous studies (Arslan et al., 2016). Carrot also verified the trend order leaf > root as described by Mayorga et al. (2013).

The As content in all edible parts of the plants increased as the concentration of As increased in irrigation water. Similar observations are well-documented in the literature studies (Table 5). The As content in edible parts of okra and tomato was in the range of what was previously reported in the literature studies. The As content in green bean was also in the range of previous studies. However, in one study carried out in India (Chandra et al., 2017), using contaminated soil at 20 mg/kg, the As content was 2.6 times higher than in the plants irrigated with 1,000 μg/L in our study. For carrot roots, amaranth leaves, and spinach leaves, the As contents in our study were 1, 8, 2.2, and 10.9 times higher than the highest values reported in the literature studies (Table 5). For lettuce leaves, the As content was higher than what was previously reported in the literature studies. However, the As content in the leaves of lettuce irrigated with 1,000 μg/L in this study was 0.7 times lower than one study carried out in Argentina (Yanez et al., 2019) using soil at 8 mg As/kg and As-contaminated water at 1,440 μg/L.

There are several possible reasons explaining the higher levels of As observed in carrot and leafy vegetables: First, leafy vegetables have a greater capability to accumulate high levels of trace metals and minerals from soil than other vegetables (Jamali et al., 2008; Baig and Kazi, 2012). The climate of Burkina Faso, characterized by low relative humidity and high ambient temperatures, could have increased the transpiration rate and caused the higher As uptake of leafy vegetables compared to non-leafy vegetables (Tani and Barrington, 2005; Luo et al., 2011; Hao et al., 2012).

Second, the sandy loam texture of the soil used in our experiment may explain the overall higher As contents observed among the plants in this study. The high soil density sometimes resulted in a flooded pot (after irrigation, water remained above the soil surface), which could have created anaerobic conditions. According to Bhatti et al. (2013), excessive use of irrigation water will develop an anaerobic condition in soil, which will result in increased solubility and release of As from As-binding minerals.

Third, experimental features of this study could explain the differences observed in the uptake of As as compared to previously reported values. These features include the relatively low volume of pots, the use of As-contaminated water directly for germination, and the irrigation water dispersion on the leaves.

Finally, it is possible that the rinsing of vegetables (without hydrochloric acid) was not sufficiently thorough to remove all the As present on the leaves. However, this simple rinsing with water reflects a more realistic scenario after the plants are harvested. Further details on the reasons explaining higher As content in carrot and leafy vegetables can be found in the Supplementary Material.

To alleviate As pollution in local vegetables, it is suggested to avoid planting leafy vegetables in areas with As concentrations in irrigation water. It has been shown that rinsing the leafy vegetables does not seem to be an effective measure to remove contaminants like As (HAS, 2020). For root vegetable, such as carrot, one measure is to peel the skin as it accumulates a higher As concentration than edible root (Bhatti et al., 2013). Therefore, fruit vegetables and pod vegetables could be considered a safer option in As-contaminated areas, particularly tomato or okra.

Cooking As-contaminated vegetables with As-free water reduces the As content in the cooked vegetables and, therefore, decreases the HQ and CR associated with their consumption. As concentrations in edible parts decreased after cooking amaranth, okra, and spinach irrigated with 500 and 1,000 μg/L. Similar results were found by Devesa et al. (2008), reporting that vegetables cooked with As-free water had lower inorganic As contents than those detected in the product prior to cooking. Surprisingly, amaranth released more As to cooking water than spinach. This might be explained by the higher surface to volume ratio of amaranth.

Cooking with As-contaminated water transfers As to the prepared foods. Adapting the cooking method could be a solution; for instance, the transfer of As, CR, and HQ are lower if yams or rice are cooked with steam rather than in As-contaminated boiling water.

We observed unexpected Cd levels in cooked foods (Supplementary Figure 8). With the highest observed Cd concentration in samples of tô (17.4 μg/g), the monthly intake of Cd associated with the consumption of 150 g of tô per day would be 132 μg/kg/month. This is 5.3 times higher than the oral provisional tolerable monthly intake (PTMI) defined by WHO/JECFA (25 μg/kg/month). The non-carcinogenic and carcinogenic risks associated with the consumption of Cd are high. It is possible that high concentrations of Cd in cooked foods could originate from the cooking materials (e.g., metal pot and cooking spoons). Cheap and low-quality materials are abundant in low-income countries such as Burkina Faso. However, the source of the contamination with Cd has to be investigated in future studies, as contamination during sample collection or sample processing (especially grinding tô a powder and digestion) in the laboratories in Burkina Faso cannot be excluded. Furthermore, Hashempour-Baltork et al. (2019) reported high levels of mercury in imported rice. This might be a concern in Burkina Faso, where large quantities of rice are imported.

To limit CR, irrigation water containing more than 100 μg/L As should be avoided for the cultivation of carrot. For the cultivation of green bean, spinach, lettuce and amaranth, irrigation water should not contain more than 500 μg/L. In areas where those As concentration thresholds in irrigation water cannot be ensured, or information regarding the concentration of As in groundwater is unavailable, it is recommended to cultivate resistant varieties such as tomato or okra and to avoid planting leafy vegetables.

To limit CR, cooking water containing more than 100 μg/L As should be avoided for boiling food. In this case, our results show that the CR is lower when the food is steamed. However, cooking water containing more than 500 μg/L As should be avoided for any cooking method. Spinach had lower CRs compared to other foods (the CR exceeded 1 in 10,000 for spinach only when it was cooked at the highest As concentration of 1,000 μg/L). This is due to the estimates of small spinach portion (consumption rate of 10 g FM/person/day compared to 150 for the other prepared foods, see assumptions in Supplementary Table 7).

According to As measurements in 1,498 rural drinking water boreholes in Burkina Faso collected by Bretzler et al. (2017), 2.3% of wells were affected by As > 50 μg/L and only 1% affected by As > 100 μg/L, which is within the permissible limit of FAO for irrigation water. Groundwater irrigation for smallholder farmers is growing in extent and importance (Villholth, 2013), and a study in Kenya underlined that drip irrigation could reduce the drought vulnerability of households (Wens et al., 2020). Shallow groundwater represents a neglected opportunity for small-scale irrigated agriculture in sub-Saharan Africa (Gowing et al., 2020). In northern regions of Burkina Faso concerned with high concentrations of As in groundwater, access to a sufficient quantity of surface water throughout the year and especially during the dry season could be problematic. Thus, these regions may be more likely to depend on groundwater than southern and eastern regions and, therefore, at risk for exposure to elevated As. Additional at-risk populations in Burkina Faso include those living next to mining sites, where high concentrations of heavy metals can be released locally into the environment without proper treatment, leading to contamination of soil and water. Above all, the risks from vegetable consumption may be higher in areas where As-contaminated water is used for both drinking and cooking.