Introduction

Bangladeshi’s have amongst the highest consumption rates of rice globally, circa. 450 g/d for an adult (Meharg and Zhao 2012). Although rice is high in calories it is generally low in macro- and micro- trace elements, and in vitamins, leading to the population of Bangladesh being widely impacted by malnourishment (ICN2, 2014). Globally, Bangladesh has amongst the highest micro-nutrient deficiencies of any country (Bromage et al. 2016; ICN2, 2014).

Dietary dependence on rice can be considered as an exemplar of hidden hunger, i.e., where a population receives sufficient calories but the diet is deficient in essential elements and vitamins (FAO 2014; Kimmons et al. 2005; Murgia et al. 2011). With respect to essential elements, rice contributes to phosphorus, calcium, copper, iron, potassium, and zinc, amongst others (Narukawa et al. 2014). For vitamins it is B1, B3, B5, B6 (Lui et al. 2017; Kam et al.; 2012; Nur Haqim et al. 2013; Rezaei et al. 2022) and E (Aguilar-Garcia et al 2007; Goufo and Trindade 2017) where rice is thought to be an important dietary contributor. The vitamin E analogues, the oryzanols, as their name suggests, are elevated in rice (Aguilar-Garcia et al 2007; Liu et al. 2019; Xu et al. 2001). Rice consumption in Bangladesh, besides hidden hunger, is also of concern as rice is elevated in the toxicants and carcinogens arsenic and cadmium compared to other dietary staples at concentrations where ill-health effects should be observed in high consuming populations (Meharg and Zhao 2012; Narukawa et al. 2014; Shi et al. 2020).

It is becoming clear that both nutrients and toxicants, particularly for elements, vary genetically (Mangel et al. 2022; Norton et al. 2012) and geographically (Gashu et al. 2021; Meharg et al. 2022; Perera et al. 2022; Shi et al. 2020). This is true for arsenic and cadmium, in particular, and especially so in a Bangladeshi context as arsenic is elevated in rice in the lowland Holocene regions, while cadmium is elevated in the raised Pleistocene Tracts (Gupta et al. 2023; Williams et al 2009). There is evidence that where soil arsenic is high in Bangladesh, that arsenic impacts negatively trace element nutrition for nickel, selenium and zinc (Williams et al 2009). Regional variation in vitamins in rice is much less studied, in Bangladesh and elsewhere (Rezaei et al. 2022). This is due in part to the fact that vitamins, particularly for B and E, exist in complex suites of analogues, and need careful development of HPLC-MS methodologies for adequate qualification and quantification (Nur Haim et al. 2013; Rezaei et al. 2022; Shobana et al. 2011; Xu et al 2001).

Elemental nutrients can be fortified in rice through fertilization and breeding (Murgia et al. 2011), vitamins through breeding (Goyer 2017; Strobbe and Van Der Straeten 2018), arsenic and cadmium reduced through field management (Mlangeni et al. 2020), and cadmium, potentially, through breeding (Lu et al. 2019). Thus, interventions to address hidden hunger and hidden toxicity can be addressed, as long as it is known where issues persist. Here we present a study where the western districts, contiguously from the far north to the far south, of Bangladesh (Fig. 1) where surveyed for market purchased white rice, the primary route and form of consumption in Bangladesh. Where appropriate, nutrient levels in rice where related to their Dietary Reference Intakes (DRI) (National Academies 2006). Also, it established which nutrients were interlinked to geography through Principle Components Analysis (PCA).

Fig. 1
figure 1

District level map of sum of arsenic species, cadmium and zinc for western Bangladesh along with change in altitude on a south to north transect

Full size image

Materials and Methods

Sample Collection

During spring 2020 market white rice was collected in 36 out of the 64 districts that comprise Bangladesh, all districts were situated in the west of the country (Fig. 1). All samples were parboiled, the post-harvesting procedure favoured by the populace (Rahman et al. 2019). A total of 556 samples were collected, averaging 15 samples per district, with cultivars recorded, and an average of ~ 11 cultivars per district. The cultivars collected, with N in parenthesis following, where: Balam (3), Basmoti (27), BR10 (2), BR11 (11), BR14 (1), BR16 (18), BRRI dhan28 (63), BRRI dhan29 (45), BRRI dhan34 (13), BRRI dhan39 (3), BRRI dhan49 (13), BRRI dhan52 (2), BRRI dhan63 (2) Cheni (18), Dega (13), Guti Sworna (28), Kajol (39), Kalizera (12), Katari (31), Lakkhi (4), Lal Sworna (6), Mala (2), Miniket (49), Mota (4), Najirsail (21),Pajam (11), Parja (11), Ronjit (5), Saila (2), Sakkha (15), Shesumoti (2), Sompa Katari (4), Sumon Sworna (12), Suval Lata ((7), Sworna (26), Sworna5 (2), Tapi Boro (1), Tej Gold (1), Thai Balm (2), Vaha manik (1), Zira Mala (1), Zira Miniket (5), Zira Sail (2) and Ziraminiket (3). The more northern districts tended to be predominantly on Pleistocene substrate, while the southern are mainly Holocene. However, as river valleys and floodplains are Holocene, the rice growing regions can be a mix of Holocene and Pleistocene. As the rice used in the experiments was market purchased, the location where it is grown was not known, beyond being of local origin, and cannot be assigned Holocene or Pleistocene with respect to the sediment that formed the paddy fields were any given batch of rice originated from. Samples were dried at 70 °C and then milled.

Inorganic Characterization

Elemental concentrations and arsenic speciation were determined as per the detailed methodologies laid out in Carey et al. (2020). Arsenic speciation and elemental totals analysis are reported briefly, to avoid self-plagiarism by reporting standardised protocols from previously published methods. For total elemental analysis, powdered rice (~ 0.1 g) was accurately weighed into 50 ml polypropylene centrifuge tubes and 2 ml of concentrated nitric acid followed by 2 ml of analytical grade hydrogen peroxide, and left to sit overnight to pre-digest. Aristar nitric acid was added, and the sample microwave digested, in batches of 40, using a CEM MARS 6 instrument, with the used digestion program given in Carey et al. (2020). CRM NIST 1568b, 3 replicates, were included in every batch of 40, as well as 3 replicates of analytical blank. On cooling samples were diluted to 30 ml with double deionized and distilled water. Elements were quantified by ICP-MS (Thermo ICAP-Q) against a set calibration series. For arsenic speciation, sample preparation was the same except that 10 ml of 0.1% nitric acid. Aristar nitric acid was used throughout as the extractant, and the microwave programme given by Carey et al. (2020). QA/QC procedures were also the same as for total elemental analysis, including the use of CRM NIST 1568b which is also certified for arsenic speciation in addition of a range of elemental totals. 1 ml of digestate was treated with 0.01 ml of hydrogen peroxide, to oxidize any arsenite to arsenate, and the samples run against calibrated standards using ion chromatography (Thermo IC5000) fitted with a Thermo AS7 and an AG7 guard column. Mobile phases and run-procedures are given in Carey et al. (2020), and interfaced with the ICP-MS as the detector.

Vitamin Characterization

Extraction and analysis were adapted from the literature. (Goh 2010; Goldschmidt and Wolf 2010; Fanali et al. 2017; Zhang et al. 2018; Pokkanta et al. 2019; Greer et al. 2021). To extract B-vitamins rice circa. 37.5 mg of dried and powdered rice was weighed into 1.5 ml Eppendorf tubes and 1.5 ml of 99:1 water:methanol mixture in 10 mM ammonium formate, 22 mM formic acid (starting mobile phase), was added and the sample then sonicated (CamLab camSonix C1275) for 10 min. The samples were then centrifuged (Hettich Rotina380R) at 15,000 rpm for 10 min, syringe filtered (Millipore Millex LG PTFE 0.2 µm) and then diluted twofold and decanted into analysis vials (Waters 9 mm 1.8 ml amber, pre-slit PTFE septa). The extraction procedure for E and K vitamins was identical to that of the B-vitamins except that a mixture of 1:99 water:methanol in 22 mM formic acid (starting mobile phase), was used for the solvent. Pure standards for B, E and K classes of vitamins were purchased from Greyhound Chromatography and Allied Chemicals, Merck and ChemService, and stock standards solutions of 1000 mg/L were prepared. Working standard sets were made up in the respective starting mobile phases. Triple quadrupole mass spectrometry (Waters ACQUITY UPLC® System Xevo® TQ MS) was used to characterize parent ions and parent/daughter relationships at optimized capillary voltage, source temperature, cone and collision cell voltages. All compounds except the 4 oryzanol analogues produced adequate signal response in positive mode, whereas the 4 oryzanol analogues performed better in negative mode.

The chromatographic column for both water- and fat-soluble group separations was a Waters Acquity UPLC® HSS T3 1.8 µm, 2.1 × 100 mm. The B-vitamins required a gradient program where the starting mobile phase was a 99:1 water(A):methanol(B) mixture in 10 mM ammonium formate, 22 mM formic acid, at a constant flow rate of 0.4 ml/min. Starting at 99:1 A:B for 1 min, the mixture was changed (all changes were linear) to 60:40 A:B within 4 min, then changed to 30:70 A:B within 2 min, then to 10:90 A:B in 6 s and held for 54 s, then back to 99:1 A:B in 6 s and held for a final 54 s, giving a total time of 9 min. The last component came off around 6.1 min. The fat-soluble E and K vitamins required an isocratic program of 1:99 water: methanol in 22 mM formic acid, at a flow rate of 0.3 ml/min for a total time of 8 min. The last component came off around 6 min. The column temperature was kept at 40 °C for both groups.

For both methods samples were extracted and run in batches of 30, with additional solvent and quality control (QC) sample blanks in duplicate, and solvent and QC sample spikes at 2 concentrations, in triplicate. The QC sample was an in-house brown basmati rice. For B-vitamins a standard calibration set of 7 standards was made in the range 0 to 0.25 mg/L. All sample extracts were below the top standard. Calibrations curves for each component were optimized to the measured component concentrations in post analysis processing. QC samples and solvent spikes, which contained all ten components, were dosed, in triplicate, at 0.017, and 0.033 mg/L.

For E and K vitamins a standard calibration set of 10 standards was made in the range 0 to 5 mg/L. Calibrations curves for each component were optimized to the measured component concentrations in post analysis processing. QC samples spikes, which contained all ten components, were dosed, in triplicate, at 0.167, and 0.333 mg/L.

Statistical Procedures

Statistical analysis and graphical plotting were conducted in Prism (v. 9) and geographical plotting using Datawrapper (https://www.datawrapper.de). Data were first averaged by cultivar by region, and then averaged by cultivar average by region to give a district average. The descriptive statistics for each compound quantified are given in Table 1. As this study is primarily interested in geographical trends in nutrients and contaminants, and their interrelatedness, multiple variable analysis, Spearman’s Rank and Principle Components Analysis (PCA) were used to investigate the data, as well as pairwise regressions were appropriate.

Table 1 Certified reference material (CRM) recoveries for element as and arsenic species, and limits of detection (LoD) for all parameter reported with the exception of XRF data (Ca, Cl, Fe, K, Mg, S, Zn)
Full size table

Results

Table 1 lists the LoDs and recoveries, where appropriate of each analyte, while Table 2 lists averages, medians and ranges of quantifiable analytes. Reported recoveries for elements and arsenic species ranged from 60.7% (zinc) to 106% (rubidium) (Table 1). Twelve elements/arsenic species gave good recoveries with low LoDs including: inorganic arsenic, cadmium, zinc, copper and phosphorus. For the vitamins B1 (thiamine), B3 (nicotinic acid, nicotinamide), B5 (panthothenic acid), B6 (pyridoxamine, pyroxidine) and vitamin E (α-tocopherol), and its analogues), were all above LoD. Vitamins B2, B6 pyrodoxal, B9 and K were included in the chromatographic run but where not detected. A range of vitamin E analogues specific to rice, the oryzanols, were detected above LoD (Tables 1 and 2).

Table 2 Mean and standard error for grain concentrations of elements, arsenic species and vitamins plus
Full size table

Figure 1 shows the distribution of arsenic sum of species (inorganic arsenic plus DMA), cadmium and zinc across the regions investigated, where north–south trends in the data are apparent from this plotting. One-way-analysis of variance showed (Table 2) that the district averages varied highly significantly between each other (P < 0.05) for most analytes, with only potassium (P = 0.115) and inorganic arsenic (P = 0.074) being non-significant. Figure 2 also shows the variation in elements, arsenic species and vitamins between districts, and their association with latitude, for elements that had a significant positive, zinc (P = 0.006) and cadmium (P = 0.002) and negative, copper (P = 0.012), DMA (P = 0.0005), inorganic arsenic (P = 0.012) and sum of oryzanol (P = 0.030) with latitude. All other parameters had a nonsignificant, P > 0.05, regression with latitude. With respect to longitude only iron (P = 0.007), potassium (P = 0.0002) and B6 pyridoxine (P = 0.037) were significant with longitude, all having negative relationships west to east (Fig. 3).

Fig. 2
figure 2

Correlation of elements, arsenic species and vitamins in rice grain that were significantly (P < 0.05) with latitude

Full size image
Fig. 3
figure 3

Correlation of elements and vitamins in rice grain that were significantly (P < 0.05) with longitude

Full size image

Multiple variable analysis is needed to further interpret the data due to the interdependence of parameters. First, Spearman’s Rank analysis of the average analyte by district data showed that inorganic arsenic and DMA were strongly positively correlated with each other, and also with copper, and with the oryzanols (Fig. 4). Arsenic species were negatively correlated with zinc, rubidium, cadmium, α-tocopherol and some B-vitamins, pyridoxamine and nicotinamide and most vitamin E analogues. Cadmium and zinc tended to have the mirror image with respect to regressions as arsenic species, i.e., being negatively correlated with inorganic arsenic and DMA. The oryzanols were all closely positively associated with each other, and this was generally true for the E- and B-vitamins, but less strongly so. The oryzanols had weak associations with other vitamins. Arsenic species and oryzanol had strong negative relationships with altitude, but low-lying regions are found in the geographic south of Bangladesh.

Fig. 4
figure 4

Spearman’s rank correlation between all measured parameters for rice grain

Full size image

PCA factor loading analysis (Fig. 5) found a clear split in the elements along PCA1, with cadmium, calcium, chlorine, zinc and rubidium having a positive PCA1 and clustering with elevation and degrees north, with the terrain in Bangladesh being more elevated in the north. The only vitamin to cluster with group was the vitamin E analogue δ-tocotrienol. A negative separation was observed for PCA for arsenic species, potassium, iron and all B-vitamins quantified, and these were diagonal to high northing and elevation, i.e., associated clustering with low elevation and low latitude. Copper also showed negative separation on PC1, slightly separated from the previously described cluster, but relatively close in space to inorganic arsenic. Vitamin E analogues, excluding δ-tocotrienol, and vitamin B compounds clustered and were mainly negative on PCA, though negative on PCA2 where the elements were positive, along with northing and elevation, though northing and elevation were positive on PCA1. Phosphorus and sulphur were at the origin on PCA1 but positive in PCA2, where copper and cadmium were at the origin for PCA2, but at the opposite ends of PCA1.

Fig. 5
figure 5

Principle components analysis between all measured parameters for rice grain

Full size image

The PCA regression analysis (Fig. 5) showed, where districts were divided into quadrats based on median longitude and latitude that the NW districts tended to separate, with only one district from each region overlapping, from the SE, with other quadrats, SW and NE being intermediate. The NW cluster, from the factor loading plot, was associated with elevated zinc and cadmium with low arsenic species and low organic vitamin content, while the SE quadrat was the opposite, low zinc and cadmium with high arsenic species and organic vitamins.

DRIs (Table 3) were calculated for elements and vitamins that had dietary reference intakes set by the US National Academy of Science (2006), assuming a per capita rice consumption rate of 0.45 kg for Bangladesh, (Meharg and Zhao 2012) for adult males as they tend to have higher RDAs than females, with the exception of iron, where females DRI is 18 mg/d as compared to males 8 mg/d. For and adult male Bangladeshi, rice contributes significantly, in decreasing order for medians, for copper (97%), phosphorus (77%), B1 (59%), zinc (37%), iron (27%), B3 (21%), B6 (20%), potassium (10%), and poorly for E (1.2) and calcium (0.7%). The ranges in contribution to DRI are important with copper being the widest 89–177%, the only nutrient to oversupply RDA from rice consumption alone. B1 also has a very wide range, 56%, ranging from 27 to 83%. Similarly, medians for arsenic (61 μg/d) and cadmium (13 μg/d) (Table 4) also had wide ranges, 46 and 19 μg/d, respectively.

Table 3 Percentage adult male DRIs (%) for consumption of 0.45 kg/d rice calculated for elements and vitamins that had dietary reference intakes set by the US National Academy of Science (2006)
Full size table
Table 4 Daily intake or total arsenic and cadmium (µg) per day based on a per capita consumption 0.45 kg of rice per day
Full size table

Discussion

The only vitamins detected above LoD for Bangladesh market rice were B1, B3, B5, B6 and α-tocopherol (vitamin E). B1 was reported at ~ 0.4–0.7 mg/kg in two polished Chinese rice cultivars that underwent different cooking approaches (Liu et al. 2019), while a median of 1.54 mg/kg was reported here for market purchased and uncooked rice. That same Chinese study found B2 at concentrations of ~ 0.2 mg/kg, higher than our reported LoD for B2, 0.06 mg/kg, with Bangladeshi samples below this value. These higher B1 and lower B2 may be due to cultivar differences in these vitamins between studies, or due to rice processing, as the Chinese investigation was on cooked non-parboiled rice, were the Bangladesh study reported here was for uncooked parboiled rice. Liu et al. (2017) reported that there was a linear decrease in thiamine and riboflavin on milling, with up to 90 and 60% losses, respectively, when milled to 15% biomass loss. Rezaei et al. (2022) report B1, B2 and B6 in market purchased (polished) Iranian rice. All 3 vitamins varied ~ fivefold between rice types. They did not report B5, which was found to be above LoD in the current study. Here B2 was below LoD (0.06 mg/kg) using state-of-the-art triple quadrupole detection, while with their analysis the LoD was ~ 10 mg/kg, suggesting methodological differences, as they used outmoded diode array detection. Their B1 and B6 were considerably higher than the findings reported here, by at least threefold for B1 and an order of magnitude for B6. For Philippines rice, using older (non-MS) technologies, Mangel et al. (2022) reported very similar levels of B1 and B6 in polished to those reported here, finding significant differences in concentrations between cultivars. For unpolished Thai rice (Lichanporn et al. 2020), B3 was ~ 60% of that reported here, while they report B9, which we did not find above the LoD (0.36 mg/kg), again using older UV-detection technologies. B9 in Australian parboiled rice was below the limits of quantification (0.5 mg/L) (Kam et al. 2012), confirming the low levels, < LoD, reported here. B1 was reported in Malaysian uncooked polished grain at concentrations of ~ 2 mg/kg and niacin at 0.2 mg/kg or less (Nur Haqim et al. 2013). Here we report B1 at 1.5 mg/kg B2 at 8.8 mg kg 0.31 mg/kg. Thus, B1 measurements reported here are similar to Nur Haqim (2013), while B2 concentrations are an order of magnitude higher here. This may be cultivar or regent dependent, or it may be methodological. Our method used LC–MS, while theirs had spectroscopic detection. From this literature survey of B-vitamins reported in rice, there is agreement and disagreement with our current study, and this is probably due to those literature studies being based on non-mass spectroscopy detector techniques, though more work needs to be conducted to build a more comprehensive knowledge of B-vitamins in rice. Cultivar differences appear apparent, but there is little geographic information beyond our Bangladesh study reported here.

Similarly, characterization of vitamin E analogues is poor in rice, though more attention has been paid to the oryzanol analogues as they are found at high concentrations in rice bran. The polyphenol complex that represents vitamin E analogues is ~ fourfold higher in rice bran compared to brown rice flour (Aguilar-Garcia et al. 2007). The tocopherols, tocotrienols and γ–oryzanols are all thought to have antioxidant activity, and because the oryzanols dominant in rice, up to tenfold higher in concentration, they have been thought to be more important than other vitamin E analogues for rice (Aguilar-Garcia et al. 2007; Xu et al. 2001). Shobana et al. (2011) investigated the tocopherols and tocotrienols present in parboiled Indian rice that underwent various degrees of milling. The reported sum of tocopherols and tocotrienols in non-milled, non-parboiled rice was 220 mg/kg, and slightly higher, 22 mg/kg, for non-milled parboiled rice, declining to 5 mg/kg in 8% milled non-parboiled, and 4.4% milled parboiled, rice. Here we report 2.5 mg/kg for the sum of tocopherols and tocotrienols. However, it is only the 2R-stereoisomesr of α-tocopherol that can be considered as vitamin E as other analogues are not converted to α-tocopherol in the body, and have poor affinity for the liver α-tocopherol transfer protein (National Academies 2006). Furthermore, the non-vitamin E tocotrienols, tocopherols and oryzanols of rice, according to a recent overview, appear to have limited antioxidant function (Goufo and Trindade 2017). Vitamin E was reported in Thai rice ~ tenfold than the median reported here, using older analytical technologies, UV-detection (Lichanporn et al. 2020).

Peng et al. (2023) looked at nutrient element availability in paddy fields across Eastern China, and while there were no strong trends for iron, copper and manganese, zinc did appear to have higher availability in the north, but grain concentrations were not measured. Joy et al. (2017) reported a survey of a transect of wholegrain and polished rice along a N-S transect for Malawi for calcium, iron, selenium, zinc and arsenic species. They did not report any geospatial trends even though their sampling strategy had a good N-S distribution, but the lack of spatial tends may be due to the fact that their samples numbers were low; 33 and 21 for wholegrain and polished rice, respectively.

Abdu et al. (2022) and Gashu et al. (2021) surveyed a N-S transect of Ethiopian districts for nutritional content (calcium, iron, selenium and zinc) of barley, maize, sorghum, teff and wheat. They found that while all four of these elements varied significantly with respect to daily consumption, there was no clear geographical pattern, though a composite consumption for the grains considered was used. For zinc in Malawian maize (Botoman et al 2022), and other grains including maize (Gashu et al. 2021), concentrations increased by ~ 100% in the south as compared to the north. Soil factors did not explain well the spatial variation in grain zinc, though it was thought pH did contribute a small proportion of spatial effects. There was no correlation with soil organic carbon, but there did seem to be a positive association with temperature. There is a positive relationship between pH and grain maize zinc from a range of studies (Botoman et al. 2022). For rice, however across a transect from Vietnam to Spain, soil pH was strongly negatively correlated with grain zinc, and cadmium (Perera et al. 2022). The Pleistocene soils of Bangladesh are more leached and therefore of lower pH than the Holocene (Lu et al. 2009), and thus this negative correlation.

Chen et al. (2018) found a geographical trend in arsenic speciation in Chinese rice with higher percentages of DMA and lower inorganic arsenic compared to more southern provinces, but no change in sum of arsenic species. Here for Bangladesh, both DMA and inorganic arsenic decreased towards the north significantly, and for sum of species and total arsenic concentrations (graphs not shown), but there was no significant change in speciation.

Chen et al. (2018) also report that cadmium decreased significantly at northern Chinese latitudes, where the reverse was true here. Unlike Chen et al.’s (2018) study, there was a clear edaphic gradation in the current study with Holocene sediments dominating in the south and Pleistocene in the north (Lu et al. 2009). The more oxidized and weathered Pleistocene soils release more cadmium and less arsenic in paddy systems (Meharg and Zhao 2012), while the reverses is true for the less oxidized and less weathered Holocene soils, and this explains the different opposite trends in cadmium and arsenic observed for Bangladesh. Cadmium did not show any strong geographical trends, rather having country-specific elevation (Meharg et al. 2022; Shi et al. 2020). Otero et al. (2020) found that inorganic arsenic concentrations did not differ between South America and Spain. Al-Rmalli et al. (2012) found a mean of 0.037 mg/kg for raw rice in an eastern Bangladesh survey (Dhaka and Sylhet regions), not dissimilar to the mean of 0.032 for parboiled rice reported here.

No geographical trends were observed for arsenic species from paddy rice surveyed from southern Spain through to southern France (Signes-Pastor et al. 2016). Mu et al. (2019) surveyed total arsenic and cadmium in rice grain across the eastern growing band of China. They did not show any trend in total arsenic but found that cadmium was higher in grain in the south, which they attributed to higher soil cadmium availability rather than to total soil cadmium concentration. Characterization of a wide range of grain nutrient elements and arsenic and cadmium, from East Africa, South America, Asia and Europe found strong regional differences in grain elemental profile, but no strong north–south relationships with the exception of grain inorganic arsenic (Meharg et al. 2022), as already shown by Carey et al. (2020).

PCA analysis found that there may be potential synergistic and antagonistic interactions were observed between vitamins and elements, but there is no literature on this that we could find, unlike previous identified interactions within trace elements for Bangladeshi grown rice (Williams et al. 2009). Also, climatic and soil gradients would need to be considered in this context (Williams et al. 2009). How grain vitamin and elemental chemistry interrelate with soils and climatic gradients was beyond the scope of the present study and is identified as a key area for future investigation.

To conclude, this current study found strong geographical gradients in essential elements, vitamins and toxic elements across a north–south and east–west gradient for the western half of Bangladesh. Thus, any strategies to counteract malnutrition or toxic element exposure of those on a rice subsistence diet within this populace (Bromage et al. 2016; FAO 2014; ICN2, 2014; Kimmons et al. 2005; Murgia et al. 2011) must consider the regional differences identified here for Bangladesh market rice.