3. Bangladesh

3.1 Arsenic in irrigation water, soil and crops

Irrigation water

Bangladesh is mainly known for periods of flooding and not so much for drought. However, a lack of water during the dry season and spells of drought at the beginning and end of the rainy season are a threat to agricultural production in Bangladesh. It is also feared that more areas will become drought prone as a result of climate change associated with global warming. During the last three decades, many hectares (ha) of land have been brought under Boro rice cultivation in the dry season by using STWs for irrigation. This is one of the main reasons that the country is self-sufficient in rice production. Boro rice receives the most irrigation water of all crops, with an estimated amount of 1000 mm/cycle. The total area under irrigation is 4 million ha and 75 percent is covered by groundwater resources: 2.4 million ha via 924 000 STWs and 0.6 million ha via 23 000 deep tubewells (DTWs). In the dry season, 3.5 million ha is used for Boro rice, 0.23 million ha for wheat and 0.27 million ha for other crops. Classifying the divisions according to the area under irrigation gives the following ranking: Rajshahi (39 percent) > Dhaka (27 percent) > Chittagong (13 percent) and Khulna (12 percent) > Sylhet (7 percent) and Barisal (2 percent). The area under Boro rice production follows the same pattern. Wheat and other crops follow a somewhat different pattern with Rajshahi being the most important area followed by Dhaka and Khulna (BADC, 2004).

With regard to drinking-water, the As-affected areas are mainly located in the south and southwest, i.e. Khulna, Dhaka and north Chittagong. With an estimated 20 percent of the drinking-water STWs having As concentrations above the Bangladesh drinking-water standard of 0.050 mg/l, it can be expected that a substantial percentage of irrigation STWs also have high As levels. The exact percentage is unknown because the spatial distribution of irrigation STWs is not similar to that of drinking-water STWs. In groundwater, only AsIII and AsV have been found and levels are within the same order of magnitude.

Data from Jessore showed that 87 percent (74 out of 85 tested wells) of irrigation DTWs contained more than 0.050 mg/l (JICA/AAN, 2004). The average As concentration in those DTWs was 0.21 mg/l. This value is very high because DTWs generally have As concentrations below 0.050 mg/l. One of the probable reasons for this is that those irrigation DTWs are tapping water from shallower depths than the drinking-water DTWs in the same area (~100 versus 200 m). Of the irrigation STWs that were at the same depth as the drinking-water STW, 24 percent (59 out of 246 tested wells) contained more than 0.050 mg/l. The average concentration of those wells was 0.07 mg/l.

The terminology STW and DTW is used both in the drinking-water and agricultural sector but it is important to realize that tubewells for irrigation do not necessarily tap water from the same depth as tubewells for drinking-water. The distinction between DTW and STW is first of all based on the capacity of the pump and, although related, not on the depth from which the water is withdrawn. STW pumps use suction-mode (centrifugal) pumps that have a maximum lift of about 7-8 m. The pipes used are less than 10 cm in diameter and irrigate about 4 ha. DTW pumps can, in principle, tap groundwater from any depth, use force-mode pumps, have pipes up to 25 cm in diameter and irrigate up to about 25 ha.

Farid et al. (2005) studied seasonal and temporal variation in As concentrations in a single STW. During the monitored Boro season, a small seasonal effect was observed: Starting in January, the concentrations slowly increased reaching the highest concentrations in early March and then declining again to reach the original level in June. The difference between the highest and lowest concentration was only 5 percent. No diurnal variation was found.

Based on available data on drinking-water STWs, it has been estimated that 900 000-1 360 000 kg As per year is brought onto the arable land via groundwater extraction for irrigation (Ali, 2003). The deposition of As on the arable land is high, especially in southwest and south Bangladesh. The northwestern part of the country, which has relatively low As concentrations in the shallow aquifer but has a very high intensity of using irrigation STWs, is also extracting a considerable amount of As from the aquifer. Other sources like P-fertilizer and manure are likely to be minor sources of As, but this needs confirmation.

According to Meharg and Rahman (2003), 150-200 (up to 900) mm water is used for land preparation before planting, and crop growth requires 500-3 000 mm. Conservatively, they assumed 1 000 mm groundwater/year (1 000 l/m²/year). If the irrigation water would contain 0.1 mg/l and the As would retain in the first 10 cm of soil (assuming soil density of 1 kg/l), the water input would cause a yearly increase of 1 mg As per kg soil. These figures depend strongly on the permeability of the soil. A clayey paddy field may only need water every three days, whereas a sandy field needs water every day. Irrigation of a clayey paddy field is usually stopped a few weeks up to a month before harvest, whereas water input on sandy soil is continued until a few days before harvest. Wheat, maize and vegetables are produced on a smaller scale and require much less water.

Duxbury and Zavala (2005) estimated that ten years of irrigating paddy fields with As-contaminated water would add 5-10 mg/kg soil to 41 percent of the 456 study sites included in their study. Based on existing national data for As in STWs used for drinking-water and the distribution of Boro rice production, Ross et al. (2005) estimated that 76 percent of the Boro rice is grown in areas where STWs usually contain less than 0.050 mg/l, 17 percent in areas with 0.050-0.100 mg/l, and 7 percent in areas with more than 0.100 mg/l. In a case study in West Bengal (India), data on As in irrigation water and the paddy soil profile indicated a yearly As input of 1.1 mg/kg to the top soil (Norra et al., 2005).

The bulk production of Boro rice, which is mainly distributed to Dhaka, seems to take place in areas with low As in the shallow aquifer. Although less rice production takes place in the areas with high As in the shallow aquifer, the rice that is being produced is likely to be used for personal/local consumption (A.A. Meharg, personal communication, 2005).

Soil and crops

Meharg and Rahman (2003) carried out a preliminary survey of As in rice and soil from Bangladesh. A total of 71 soil samples was collected throughout the country. The highest measured soil concentration was 46 mg/kg, whereas less than 10 mg/kg was found in areas with low As in irrigation water. The western part of Bangladesh seems to have the highest soil concentrations (>30 mg/kg), followed by the central belt, which is in agreement with groundwater concentrations. At various locations with high As levels in groundwater, low concentrations were found in the soil. However, they did find a correlation between soil concentrations and irrigation water concentrations when the age of the water-well is taken into account. A positive correlation between As concentrations in rice and soil was also found.

Following up on the Meharg and Rahman (2003) study, Williams et al. (2006) did an extensive sampling of rice throughout Bangladesh, collecting 330 samples of Aman rice and Boro rice. Importantly, a positive correlation was found between As in the groundwater and As in the rice. This correlation was stronger for Boro rice than for Aman rice. Highest As concentrations in rice were all from districts in the southwest, namely Faridpur > Satkhira > Chuadanga > Meherpur. For detailed results on reported As concentration in rice refer to paragraph 3.2.

In agreement with Meharg and Rahman (2003), data from a preliminary nationwide survey of As in soil, crops and irrigation water indicate that the soils in the west and southwest part of Bangladesh contain the highest As concentrations (Miah et al., 2005). In these parts, irrigated soils had higher levels of As compared to adjacent non-irrigated soils. In the irrigated soils, the first 0-15 cm had the highest levels of As. In other parts of the country, irrigated and non-irrigated soils did not differ in As concentrations. The differences in soil concentrations were, however, not reflected by As levels in the rice plants.

Islam et al. (2005) studied As levels in water, soil and crops at 456 locations in five upazilas. The average As concentration in the soil was 12.3 (ranging from 0.3 to 49 mg/kg) and the thanas were classified according to soil concentrations: Faridpur > Tala > Brahmanbaria > Paba > Senbag. Of all soil samples, 53 percent contained less than 10 mg/kg, 26 percent contained between 10.1 and 20 mg/kg, and 18 percent contained more than 20 percent. Concentrations both between and within thanas were highly variable. The same was observed at the command area and paddy field level. In some cases this correlated with the distance to the tubewell used, in other cases the variation seemed to be random or related to micro-elevation. They also found a high seasonal variation in As soil concentrations. At the end of the Boro (dry) season the soil concentration had increased sharply when irrigated with As-rich water. Most of it was again removed after the Aman season, i.e. after flooding. There are various explanations for this phenomenon: 1) As desorbs to the standing water and is then removed laterally; 2) the top layer may be eroded and run off during heavy rainfall; 3) volatilization of As during prolonged periods of flooding; and 4) leaching of standing water desorbing and transporting As from the topsoil to deeper layers. These different processes have not been quantified yet. The general opinion is that leaching is an unlikely process because of the slow percolation rate, 2-4 cm/day (Brammer, 2005; Islam et al., submitted; Islam et al., 2005). This explains why soil concentrations in the first 15 cm are generally highest compared to the rest of the soil column.

Islam et al. (2000) reported total As concentrations of 5-33 mg/kg with an average of 17 mg/kg for some soil samples from Nawabganj, Rajarampur, Jessore, Jhenidah and Comilla. However, the study has some limitations: No information was provided about the use of the sampling locations (e.g. fallow land, paddy field, other crops, residential area, etc.); the chemical analyses were not described in detail; and the use of CRMs was not mentioned.

Islam et al. (submitted) collected 100 samples of irrigation water, soil (composite sample of five randomly taken samples at 0-15 cm depth) at 100 STW command areas in Chapai Nawabganj (Sadar upazila) during the Boro season of 2003. The irrigation water contained 0.025-0.352 mg/l (mean 0.075 mg/l).

Soil concentrations were 5.8-17.7 mg/kg (mean 11.2 mg/kg), straw contained 1.48-17.6 mg/kg (mean 5.88 mg/kg) and rice grain contained 0.241-1.298 (mean 0.759). Poor correlations were found between grain and soil and between grain and water only. A good correlation was found between grain and straw. The total soil As was correlated with As in irrigation water indicating As accumulation in soil because of As rich irrigation water input. In terms of quality, the study had some limitations: Rice cultivars were not identified; pretreatment of straw and grains (e.g. rinsing with As free water to remove dust) was not mentioned; and CRMs were not included.

Rapid adsorption of As from irrigation water to soil may explain the spatial patterns found in irrigation canals and some paddy fields (Farid et al., 2005). An alternative hypothesis is that Fe²⁺ present in irrigation water is rapidly oxidized to Fe³⁺ when the water is exposed to air, resulting in the precipitation of AsV (Islam et al., submitted).

Five soil profiles of 15 m depth were collected at Deuli village (near Samta village, southwest Bangladesh) (Yamazaki et al., 2003). Although not specifically mentioned in the paper, it seems that samples were collected from fallow land, not used for agriculture. Results showed that the soil concentrations were dependent on the type of sediment. Sandy sediments contained 3-7 mg/kg (median: 5 mg/kg), clayey sediments contained 4-18 mg/kg (median: 9 mg/kg), whereas peaty and peaty clay sediments contained 20-111 mg/kg. The first 6 m contained sandy and clayey layers whereas the peaty layers were found at a depth of 7-10 m. Below 10 m, a sandy layer was dominant. The general pattern was that the As concentrations were relatively constant except for the 7-10 m layer, which contained a much higher total As.

Twenty-five locations in five thanas (Chapai Nawabganj Sadar, Kushtia Sadar, Bera, Ishurdi and Saishabari) of four districts were sampled (Alam and Sattar, 2000). Soil concentrations ranged from below detection limit to 56.7 mg/kg. Ten out of 25 locations contained As concentrations of more than 20 mg/kg. The As concentrations in the adjacent water-wells ranged from below detection limit to 0.071 mg/l, i.e. low to moderate. A positive correlation was found between As concentrations in the soil and water. In terms of quality, the paper had a number of limitations: Land use of the sampling locations was not described, which is an important feature because irrigation varies according to the type of crop; there was no mention of whether the sampled wells were for drinking-water or irrigation water; and in the methodology section quality assurance/quality control (QA/QC) was not described and CRMs were not used.

Das et al. (2004) collected soil samples (n = 18) in three upazilas: Kachua, Hajiganj (both in Chandpur district) and Sharishabari (in Jamalpur district). Composite soil samples (15-45 cm depth) were probably taken from arable land, but specific land use was not mentioned. A CRM for soil was not included. Soil concentrations ranged from 7.3 to 27.3 mg/kg with an average of 15.7 ± 6.6 mg/kg. Apositive correlation was found between the As in STWs and soil.

Except for some experimental work with rice exposed to artificially contaminated soils (see Section 2.2), hardly any work has been done on the potential risk of As in irrigation water to crop production. Duxbury et al. (2003) studied As concentrations in rice in relation to yield and panicle sterility in Bangladesh. They did not find any correlation and concluded that there was no indication of toxic effects under the current field conditions. Regarding the study design, the authors acknowledged that a potential important factor, the effect of different rice varieties, was not studied. This may have hidden a possible correlation between yield and As concentrations in grains. The authors stated that because of the continuous input of As-rich irrigation water, toxic effects cannot be excluded in the future.

In agreement with Takahashi et al. (2004), As concentrations in soil water from flooded paddy fields increased with the duration of flooding reaching levels above 3 mg/l (Loeppert et al., 2005). During both Boro and Aman seasons, the soil water concentrations thus largely exceeded the irrigation water concentrations. This is important to realize, particularly for Aman rice, which is rainfed and contains As below the detection limit. It has been hypothesized that As uptake and toxicity to rice is much better correlated with As in the soil water than with As in irrigation water or total As in the soil.

In the 2006 Boro season, a small but detailed pilot study was conducted in which phytotoxicity to rice was studied at the field level in a paddy field in Faridpur contaminated by twenty years of irrigation. Preliminary data indicate a clear negative correlation between As in the soil water and plant growth (G.M. Panaullah, personal communication, 2006). These results emphasize the need to further investigate the possible risks of As in irrigation water to crop production.

Two studies from West Bengal have provided strong evidence for accumulation of As in topsoil because of irrigation with contaminated groundwater (Norra et al., 2005; Roychowdhury et al., 2002a). Roychowdhury et al. (2002a) reported for Domkal block that fallow lands contained 5.31 mg/kg, whereas adjacent lands irrigated with 0.082 mg/l and 0.17 mg/l contained 11.5 and 28.0 mg/l, respectively. The calculated input of As was approximately 1.6-16.8 kg/ha/yr. The age of the tubewells was not given and an estimation of the total amount of As deposited on these lands is therefore not possible. The suggestion that the samples from the fallow land and irrigated land did not originate from the same parent material, which would (partially) explain the observed differences, cannot be excluded. However, the good correlations between concentrations in STWs and soils clearly showed the effect of the contaminated irrigation water on As concentrations in the soil. On a smaller scale, measurements of As in paddy fields with increasing distance from the tubewells also gave a good correlation, again showing that soil concentrations increased because of the application of contaminated water. The same was reported by Hossain (2005) for a number of locations in Faridpur, Bangladesh. This study also found a relationship between micro-elevation and As soil concentration.

Norra et al. (2005) collected water, soil and plant samples from three fields in Kaliachack I block: one paddy field and one adjacent wheat field both irrigated with water containing 0.5-0.8 mg/l, and one reference paddy field not contaminated with As. Soil profiles were collected down to a depth of 110 cm. The upper topsoil in the contaminated paddy field contained 38 mg/kg, the less intensively irrigated wheat field grown with wheat contained 18 mg/kg, whereas a reference paddy field contained 7 mg/kg. The soil profiles of the contaminated paddy field and wheat field clearly showed a decreasing As level with increasing depth. Although, to a lesser extent, it is important to note that As did not only build up in the paddy field but also in the wheat field. This has not been reported before. According to the author, continuation of irrigation with As-rich water could result in alarmingly high soil concentrations within a few decades.

Conclusions: irrigation water, soil and crops

There are indications that soil concentrations are increasing over time because of irrigation with As-contaminated water. Data are, however, insufficient in terms of quantity and quality. It is thus still unclear under what specific conditions and over what period of time As is accumulating in the soil. Major problems with many data available in Bangladesh are the quality of the chemical analysis and the description of the methodology in general. The international symposium on the behaviour of arsenic in aquifers, soils and plants, held in January 2005 in Dhaka, recommended long-term monitoring of As in water, soil, and crops under the various conditions present in Bangladesh, along with detailed studies on the behaviour of As in paddy fields.

The risk of As-contaminated irrigation water to crop production has received little attention until now. To evaluate current and future soil concentrations, representative toxicity data for crops are needed, both for flooded and non-flooded soil conditions. Thus, field studies to test if As is one of the factors limiting growth in the field should be emphasized. Further, it should become clear what soil parameters correlate with uptake and toxicity and, based on that information, a toxicity database for different rice cultivars and other crops could be developed to set standards for As in flooded and non-flooded soils.

3.2 Human exposure

Food safety standards

When properly applied, food safety standards are a useful tool to evaluate levels of contaminants in foods. This section will therefore briefly discuss available As food safety standards before presenting published As concentrations in foods from Bangladesh.

Many papers refer to the Australian maximum permissible concentration (MPC) for As in foods (1 mg/kg) to evaluate their results (Abedin et al., 2002a; Abedin et al., 2002b; Das et al., 2004; Islam et al., 2004b; Jahiruddin et al., 2004; JICA/AAN, 2004). The Australian MPC has however various limitations in the context of Bangladesh and other countries where rice is a staple food. Australia does not have a rice based diet, and thus the Australian food safety standard is probably too high for countries where rice is a staple food. Further, the MPC has been set for total As only, which does not take into account the great differences in toxicity between organic and inorganic As species. It is therefore advised not to use the Australian MPC for evaluating As concentrations in rice and other foods in the Bangladesh or Asian context. To evaluate As in foods, it is recommended to develop a guideline value for inorganic As in foods and specifically for rice taking into account dietary consumption patterns.

Table 3.1 Chinese food safety standard for inorganic As (mg/kg) in various products

In October 2005, the Ministry of Health of China adopted a new food safety standard on As in foods, GB 2762-2005 (Table 3.1). This standard, specified for a variety of food products, has been set for inorganic As, and not for total As. This is an important step as it recognizes that total As in foods is not appropriate for evaluating food safety. This standard may serve as a guideline for other countries in the region that could derive their respective national standards from it. A re-examination of the data of Abedin et al. (2002a), Abedin et al. (2002b), Das et al. (2004), Islam et al. (2004b), Jahiruddin et al. (2004), JICA/AAN (2004), which used the Australian food safety standard, shows that most of their results exceed the Chinese food safety standard.

Arsenic in foods from Bangladesh

Inorganic As

The first data on As speciation in rice from Bangladesh have been published recently (Williams et al. 2006; Williams et al., 2005). Williams et al. (2005) collected 15 samples of various rice cultivars from the wholesale market in Dhaka and analysed for total As, DMA and inorganic As (Figure 3.1). The average total As concentration was 0.13 ± 0.02 mg/kg (ranging from 0.03 to 0.30 mg/kg). The method to extract the As species trifluoroacetic acid (TFA) had an efficiency of approximately 80 ± 12 percent. This means that a relatively small portion of the As is unaccounted for and its speciation is unknown. Assuming that the speciation pattern of this portion is equal to the other portion (A. Meharg, personal communication, 2005), the average percentage of inorganic As was 80 ± 3 percent. It also implies that three out of 15 samples exceeded the Chinese food safety standard.

Compared to other countries, rice from Bangladesh (and India) had the highest percentage of inorganic As (80 percent), against 42 percent in rice from the USA. This indicates that the percentage of inorganic As in rice is not a constant factor geographically and probably depends on cultivar and growth conditions.

Source: Williams et al. (2005)

Note: All samples bought at the wholesale market in Dhaka. Cultivars are shown vertically below the x-axis whereas districts are shown horizontally below the x-axis.

Figure 3.1 As speciation in 15 rice samples of various cultivars and districts

Williams et al. (2006) analysed another 21 rice samples from Bangladesh (seven different cultivars, both Boro and Aman) for As speciation of which approximately half of the samples exceeded the Chinese food safety standard. The observed As speciation pattern was similar to Williams et al. (2005). Combining the data from Williams et al. (2005) and Williams et al. (2006) shows a strong positive correlation between total As and inorganic As in rice from Bangladesh, and indicates that 80 percent of inorganic As in rice may be representative of use within the country (Figure 3.2).

As speciation analysis on a number of vegetables (arum stolon and tuber, potato, bitter gourd, ribbed gourd, pointed gourd, teasel gourd, plantain banana and long yard bean), pulses and spices indicated that all As was present in the inorganic form (Williams et al., 2006). The extraction efficiency with TFA varied from a reasonable 79 percent to 128 percent for vegetables, 70 percent for spices, to a low 45 percent for pulses. All detected As was in the inorganic form.

Total arsenic

With only two studies published on inorganic As in rice and vegetables from Bangladesh, some data on total As in foods from Bangladesh and its neighbour West Bengal, India are presented here as well.

Williams et al. (2006) collected a large number of samples (rice: 330, vegetables: 94, pulses and spices: 50) throughout the country. For rice, the results clearly showed that the highest levels of As were found in the southwestern part of the country, and there was a positive relationship between As levels in rice and As levels in groundwater (Figure 3.3). With a factor of 1.3, Boro rice contained significantly more As than Aman rice. This could be caused by a difference in rice cultivars grown during the Boro and Aman seasons. An alternative explanation is that Aman rice is mainly rain fed, while Boro rice is irrigated with groundwater containing As. Concentrations in Boro rice were in the range of 0.04 to 0.91 mg/kg, whereas Aman rice contained <0.04 to 0.92 mg/kg. Assuming that 80 percent of the total As was in the inorganic form, a substantial number of samples exceeded the Chinese food safety standard.

Figure 3.2 The correlation between total As and inorganic As in rice from Bangladesh

Source: Williams et al. (2006)

Note: Figures in parentheses represent the mean As concentrations in the shallow aquifer of each district given in mg/l.

Figure 3.3 The average concentrations of total As in rice (both Boro and Aman cultivars) collected from 25 districts

In Table 3.2, other data on total As concentrations in rice from Bangladesh and West Bengal have been summarized. Roychowdhury et al. (2002b) found that cooked rice had approximately twice the level of As as raw rice. This is likely to be because of parboiling and/or boiling the rice in As-contaminated water. Bae et al. (2002) reported that As concentrations in rice after boiling in As-contaminated water (0.223-0.373 mg/l) were increased from 0.178 mg/kg dw to 0.228-0.377 mg/kg dw. On the other hand, Duxbury et al. (2003) found that processing (parboiling and milling) of rice grown and processed in low and high As areas reduced As concentrations by ~20 percent. Unpublished data from Meharg and co-workers showed the same pattern: processed rice bought at markets had lower As concentrations than rice sampled in the field (A. Meharg, personal communication, 2005).

Table 3.2 Total As concentrations (mg/kg dw) in rice from Bangladesh and West Bengal

^a Lowest concentration in BR11: 0.043 mg/kg dw.

^b Only inorganic As and DMA were found.

1 = (Meharg and Rahman, 2003); 2 = (Duxbury et al., 2003); 3 = (Das et al., 2004); 4 = (Roychowdhury et al., 2002b);

5 = (Mandal et al., 1998); 7 = (Watanabe et al., 2004); 8 = (Islam et al., submitted); 9 = (Williams et al., 2005).

A summary of the results of Williams et al. (2006) on total As in vegetables, roots and tubers, pulses, and spices is presented in Table 3.3. Recalculating the presented concentrations to wet weight shows that a number of samples exceed the Chinese food safety standard. More specifically, the mean total As concentration of the following items exceeds the standard: radish leaf, plantain banana, long yard bean, arum tuber, giant taro, potato, and arum stolon. The authors do not give a clear indication of any correlation between As in these products and As levels in the shallow aquifer in the area where the samples were collected.

Table 3.3 Total As concentrations in vegetables, roots and tubers, pulses, and spices

Source: Williams et al. (2006)

Three other papers were found on total As in non-rice foods from Bangladesh and West Bengal, namely Alam et al. (2003), Das et al. (2004) and Roychowdhury et al. (2002b). They all described the methodology reasonably well and certified reference materials were included. Most samples were collected from a few locations known for high As in the shallow aquifer. Concentrations in vegetables, fruits, spices, and freshwater fish ranged from less than 0.04 mg/kg dry weight (dw) to 3.99 mg/kg dw, with most samples having less than 0.5 mg/kg dw. Das et al. (2004) reported that total As concentrations in fish were below 1 mg/kg dw. However, data from Taiwan Province of China showed that fish cultivated in As-rich water may lead to high levels of inorganic As (Huang et al., 2003). Alam et al. (2003) mistakenly did not convert dry weight concentrations to wet weight before data interpretation, including the estimation of the daily As intake. This has caused a fivefold overestimation of the daily exposure to As and other metals that were included (corresponding author E. Snow, personal communication, 2004).

On a dry weight basis, a number of vegetable samples contained higher As concentrations than rice. However, this does not necessarily mean that As in vegetables poses a higher risk to human health than As in rice. From a food safety perspective, water contents and food consumption data need to be taken into account. Usually, food consumption data are on a raw weight basis, i.e. fresh vegetables (usually containing 70 to 90 percent water) and uncooked rice (containing approximately 13 percent water). Comparing As concentrations in rice and vegetables on a raw weight basis shows that As levels in rice are usually higher.

Concerning dairy products and meat, various researchers have expressed concern about possible transfer of As from water and straw to cattle (Abedin et al., 2002a; Jahiruddin et al., 2005; Panaullah et al., 2005). However, no peer-reviewed publications on this issue have been found, indicating a need to investigate this issue.

Food consumption in Bangladesh

Only one study reported food consumption on a gram/capita/day basis in Bangladesh (Hels et al., 2003). That study reported data from two villages, Falshatia (Manikganj) and Jorbaria (Mymensingh), covering 1981/1982 and 1995/1996. Twenty-four hour food weighing data were collected from October to November and from January to March. Data were collected at household level from which consumption per capita values were derived. Corrections were made for the number of meals consumed outside the house. All members of a household were treated equally. In Table 3.4, food consumption data are presented for the period 1995/1996. For most data, the standard error was less than 20 percent. Seasonal variation in food consumption was observed, particularly for rice and vegetables. There were also differences between the villages.

Table 3.4 Estimated food consumption for two villages in Manikganj and Mymensingh (g/capita/day) in 1995/1996

¹ n = number of surveyed households; average household size = 5; average consumption unit is 4.5. Source: Hels etal. (2003)

Drinking-water consumption in Bangladesh

Watanabe et al. (2004) studied water intake by adult men and women in two As-affected areas and reported a total water intake of 4.6 and 4.2 l/day, respectively. Two methods were used to assess drinking-water intake, namely 24-hour self report and interviews with frequent visits. The range of water intake was 1 to 6 l/day. There was no difference between direct intake of drinking-water between men and women (both ~3 l/day). Water intake via food preparation determined by field experiments added another 1.6 l/day (men) and 1 l/day (women) per day.

Mandal et al. (1998) estimated the drinking-water intake of a small number of people in an As-affected village and the average intake by adults was 4 l/day. The highest intake values for some individuals were 7 and 8 l/day. However, reliability of the data is unknown because the methodology was not described. In contrast with Watanabe et al. (2004), a great difference in water consumption between the sexes was reported. The general impression is that adults consume 3 l/day and an additional 1 l/day from foods.

Dietary exposure to arsenic

Williams et al. (2006) concluded that rice is the predominant source of inorganic As from foods. This was based on a daily consumption of 500 g rice, 130 g vegetables, 12 g pulses and 5 g spices (all weights based on unprepared products) and data on inorganic As and total As in a range of food items from Bangladesh. Most of the Boro rice samples collected contributed at least 50 percent to the provisional maximum tolerable daily intake (PMTDI) for inorganic As (0.126 mg/day for a 60 kg person). That leaves only 0.66 mg/day or less to other sources of exposure including drinking-water. Assuming a realistic level of inorganic As of 0.2 mg/kg in rice, a drinking-water concentration of 0.050 mg/l (Bangladesh drinking-water standard) and a water consumption of 3 l/day, the total daily intake of inorganic As would be 0.25 mg/day, exceeding the PMTDI by a factor of two. Rice would contribute 40 percent of total daily intake of As.

Food items other than rice only make a minor contribution. Even for a worst case scenario (consumption of 130 g/day of a vegetable with the highest As level on a wet weight basis, which in this study is potato with 0.23 mg/kg ww), the contribution is only 0.03 mg/day (25 percent to the PMTDI). In a number of cases, arum stolon has been receiving particular attention because of the reportedly high levels of As. The data show that high levels of As in arum stolon (in this study 1.93 mg/kg dw, i.e. 0.193 mg/kg wet weight (ww)) would only contribute 0.025 mg/day. This emphasizes the need to consider As concentrations in food items from the perspective of the overall dietary intake of inorganic As. Also the important nutritional value of vegetables like arum should be taken into account before conclusions are drawn on the risks of As in such food items.

Roychowdhury et al. (2002b) estimated the daily intake of total As via water and food for two locations in West Bengal. The intake via foods was approximately 180 and 97 |ig/day for adults and children (10 years old) respectively. Adults and children were exposed to approximately 400 and 200 ug/day via drinking-water. Drinking-water counted for ~70 percent of the exposure whereas rice contributed ~30 percent. The authors poorly described the method of collecting data on water and food consumption, and compared to Hels et al. (2003), they used substantially higher values for food consumption.

Watanabe et al. (2004) estimated that the daily intake of total As by adults was approximately 600 ug/day (male: 674 ug/day, female: 515 ug/day) with 70 percent via drinking-water and 10 percent via rice. The food consumption data seemed to be a rough estimation only. Neither Roychowdhury et al. (2002b) nor Watanabe et al. (2004) included any variation in concentrations, consumption and seasonal effects in their exposure assessment.

Duxbury et al. (2003) analysed 150 rice samples from Bangladesh. Assuming a rice consumption of 400 g/day with 0.250 mg/kg and a water intake of 4 l/day with 0.050 mg/l As (drinking-water standard Bangladesh), the total daily intake would be 0.3 mg/day. Rice would contribute 33 percent. Fourteen percent of their rice samples contained > 0.250. Meharg and Rahman (2003) assumed a rice consumption of 420 g/day with 0.5 mg/kg and a water intake of 2 l/day (the WHO default value) with 0.1 mg/l. The calculated total daily intake was 0.41 mg/day and rice contributed 50 percent. However, taking into account the climate in Bangladesh and the high consumption of rice, 2 l/day is likely to be an underestimation (Watanabe et al., 2004).

After assessing the exposure levels, the results need to be compared to a reference value such as a tolerable daily intake (TDI) value. For As, only a provisional maximum tolerable daily intake (PMTDI) is available. This provisional value of 0.0021 mg/kg body weight/day for inorganic As was established in 1988 and is commonly used to evaluate dietary intake (WHO, 1996). After almost two decades, the PMTDI still has not been ratified.

When evaluating risks to human health associated with As in foods, other sources of exposure such as drinking-water have to be taken into account as well. The WHO guideline value is 0.010 mg/l and the Bangladesh drinking-water standard is 0.050 mg/l (Duxbury and Zavala, 2005; Williams et al., 2005). Assuming a body weight of 60 kg, the PMTDI is 0.126 mg/day. A water consumption of 3 l/day with 0.050 mg/l would already exceed the PMTDI, regardless the levels of As in foods. This suggests that the PMTDI and the Bangladesh drinking-water standard need to be evaluated so that a proper assessment of As in foods can be made.

The human health risk assessment is likely to be more complicated because of the prevalence of micronutrient deficiency in Bangladesh and many other Asian countries, particularly among women and children. For example, studies have reported on the selenium and As interaction. Arsenic is toxic by itself and it also interacts with selenium, resulting in excretion of their mutual metabolite (Gailer et al., 2000). As selenium is an essential micronutrient, this confounding excretion of selenium can aggravate further micronutrient deficiency among the most vulnerable subpopulations and can thus be a health concern.

Conclusions: human exposure

It has become clear that dietary exposure can contribute significantly to the total daily intake of inorganic As. More data on food and water consumption patterns and As levels in foods are needed to refine the human exposure assessment. It is further recommended to review the status of the PMTDI and to propose a food safety standard for inorganic As.

It is important to realize that there are clear indications that As concentrations in rice are increasing over time because of the prolonged input of As-contaminated irrigation water. This may offset the efforts in the drinking-water sector to reduce human exposure to As through water consumption. This emphasizes the need to further investigate As in the food chain and develop appropriate management options.

3.3 Agricultural management options

Even though the risks to food safety and, in particular, to crop production are not yet fully understood, it can be stated generally that the input of contaminants to the environment should be avoided or, at least, minimized, and that natural resources such as groundwater should be used in a sustainable way. From this perspective, there are various topics that can be explored to address management options (Brammer, 2005; Duxbury et al., 2003; Lauren and Duxbury, 2005; Meharg, 2004; Panaullah et al., 2005; Ross et al., 2005):

• Reduce irrigation water use in rice cultivation. This will reduce As input, reduce leaching of nutrients and minimize extraction of groundwater resources. The overall aim of optimizing groundwater use should be to improve sustainable agricultural production. According to BRRI, farmers could apply 40 percent less irrigation water without any yield losses compared to current practices. If water input could be reduced to such an extent that the soil conditions become more aerobic, the solubility of As and, therefore, the uptake of As will be minimized as well. Less irrigation water means fewer costs for the farmer, but it requires a reliable water supply.
• Promote cropping patterns that require less irrigation water in areas with high As in soil and/or irrigation water. For example, depending on soil conditions, replace Boro rice with crops like wheat and maize.
• Select/breed rice cultivars that are tolerant to As and have a limited uptake of As. This could be useful only in combination with a minimized As input in the soil to avoid As buildup in the soil. Rice cultivars show variation in their response to As exposure, but only a limited number of cultivars has been screened to date. A systematic screening of the numerous cultivars could reveal differences in As uptake, tolerance, translocation and speciation.

Phytoremediation has been suggested as a means to remove As from soil. There are two main reasons why this is an unlikely option. First, there may be no need to remove As actively from the soil. Second, phytoremediation is a very slow process and thus not a pragmatic approach for agriculture.