Nine varieties of Calabash chalk, selected on the basis of their obvious differences in appearance (such as colour, lamination and shape), were purchased from markets located in Jos (Plateau State, Nigeria) and Zaria (Kaduna State). Seventeen market vendors and 526 women who were in hospital or were attending antenatal clinics, were questioned about the origins and use of Calabash chalk in Nigeria. It is the intention to publish the findings of this survey at a later date, but we briefly report some pertinent information here. Pregnant women from the Igbo tribe – a large ethnic group of eastern Nigeria – were recorded as the main users, and comments were made on the effectiveness of the ingested soils/rocks in limiting vomiting during pregnancy and reducing over-salivation: there is no suggestion that the Calabash chalk is being consumed to aid mineral nutrient supplementation. The daily amount consumed varies, but generally is c. 5–10 g. However, differences are apparent between pregnant and non-pregnant women: the former tend to ingest more (up to 20 g/day) with consumption occurring either before or after mealtimes to prevent vomiting, whereas the latter consume soil generally when in a fasted state. For both groups of women, consumption occurs by gnawing chunks of the geophagical material.

The Undongo sample was obtained from a small supermarket in the Ukambani hills, Kenya. This material is purchased in labelled polythene bags (Figure 1) that contain blocky units of ‘roasted’ soil some 50 g in total. The labelling highlights the richness of Fe in the soil and its value to pregnant women and their foetus, although no details about how the material should be consumed (e.g., in a fasted or fed state; how much should be ingested/day) is provided.

Samples were oven-dried at 40°C for 48 hrs prior to disaggregation using a porcelain mortar and pestle and sieving through a 2000 µm aperture nylon mesh. Subsamples of the <2000 µm fine earth fraction were then: (I) further sieved through a 250 µm aperture nylon mesh and, (II) ground and sieved through a 150 µm aperture nylon mesh. The <2000 µm, <250 µm and <150 µm subsamples were retained for subsequent analysis.

The <150 µm subsamples were subjected to a nitric-perchloric acid digestion following the procedure described by Thompson and Wood [35]. These authors note the effectiveness of this acid mixture in decomposing clay minerals and a number of primary minerals, though some soil constituents (if present) are not fully digested, and consequently the method is often regarded as a procedure that determines, in association with appropriate instrumentation, pseudo-total (i.e., an approximation) rather than true total concentrations. However, our use of fine (i.e., <150 µm) material in this experimental procedure facilitates sample decomposition and optimises the release of chemical elements into solution. Analysis of the sample solutions was undertaken using inductively coupled plasma mass spectrometry (ICP-MS) for six soil trace elements (Cd, cobalt [Co], chromium [Cr], Cu, nickel [Ni] and Pb), whilst another trace element, Zn, and a number of major (Ca, Fe, potassium [K], Mg) and minor (Mn, sodium [Na]) elements were determined by atomic absorption spectrometry.

For our assessment of the bioaccessible concentrations, we used the Fed ORganic Estimation human Simulation Test (FOREhST) procedure. This method was initially developed to assess the bioaccessibility of soil polycyclic aromatic hydrocarbons (PAHs, widespread organic pollutants) under simulated fed conditions. Most IVBA procedures are concerned with inorganic soil constituents, and simulate fasted conditions that – relative to the fed state – are associated with lower pH conditions, so providing the most conservative estimate of bioaccessibility [36]. However, since pregnant women who consume Calabash chalk ingest the material either just before or after mealtimes, we used the FOREhST method to quantify soil bioaccessibility in a fed-state where the geophagist is consuming food. To investigate the impact of ingested foodstuffs on bioaccessibility per se, we also subjected the soil samples to the FOREhST procedure digestion pH and transit conditions but did not add the foodstuff component of the experimental method. To summarise, soil chemical element bioaccessibility was: (I) determined using the FOREhST method, a fed-state experimental procedure that is appropriate to use since the main consumers in this investigation are pregnant women who ingest the geophagical materials just before or after mealtimes, and (II) assessed using the FOREhST method where the food component of the in vitro procedure is omitted, but where other experimental variables (transit times, solution pH, enzyme concentrations) are the same as (I), so enabling the effect of food alone on bioaccessibility to be evaluated.

The FOREhST procedure is a IVBA test, carried out at 37°C to simulate human body temperature, and utilising end-over-end rotation that replicates the churning of soil and fluid in the gut. The stages involved in the methodology represent the saliva/gastric and intestinal (i.e., duodenal and bile) phases of the human GI system. Briefly (for a detailed description, see Cave et al. [36]), 0.3 g of each soil/shale sample was weighed into an individual extraction bottle to which was placed 4.5 ml of simulated saliva solution (pH = 6.8±0.5). After 5 min, 9 ml of simulated gastric solution (pH = 1.4±0.5) was added to produce a mixed saliva/gastric solution phase that had a final pH of 1.6±0.2. Following 2 hr of rotation in a water bath calibrated to 37°C, sample solutions were retrieved from each extraction bottle, and retained for chemical element quantification. This procedure was then repeated on another set of weighed soil/shale samples but, following the 2 hr of rotation with the saliva/gastric solution, 9 ml of simulated duodenal (pH = 8.1±0.2) solution and 4.5 ml of bile fluid (pH = 8.2±0.2) was then added to each extraction bottle, producing a final solution that had a pH of 6.0±0.5. The extraction bottles were rotated in the water bath for a further 2 hrs after which the sample solutions were retrieved for chemical analysis. This version of the methodology leads to leachate sample solutions being collected both at the end of what we term the saliva/gastric and intestinal phases of extraction. Chemical element quantification of these solutions was undertaken using ICP-MS instrumentation. Because of analytical considerations (e.g., a variety of Na-containing reagents are used in the FOREhST procedure, with deleterious implications for the determination of bioaccessible Na), we restrict our focus to the bioaccessible concentrations of eight elements (Co, Cr, Cu, Fe, Mn, Ni, Pb and Zn): these bioaccessible concentrations associated with the geophagical materials are reported in units of mg/kg.

Since the bioaccessibility of inorganic soil constituents is dependent upon, amongst other variables, particle size, the FOREhST procedure was applied to sieved materials of <2000 µm and <250 µm. We chose these particle sizes firstly because the geophagists are ingesting the bulk material (which explains our use of the <2000 µm fraction), and secondly because IVBA methods undertaken for human health risk assessments typically use <250 µm particles. Soil chemical element bioaccessibility also depends on whether foodstuffs have been ingested and, as previously mentioned, we subject the geophagical materials to (I) a scenario of fed-state (F-S) conditions where a freeze-dried oatmeal and rice porridge infant food supplemented with sunflower oil is used as the foodstuff in the experimental procedure as described by Cave et al [36], and (II) simulated fed GI pH and transit conditions but where no food has been consumed by the geophagist (i.e., no rice porridge/sunflower oil is used in the laboratory method: a fed-state, no food [F-SNF] scenario that enables the effect of food alone on bioaccessibility to be evaluated).

To summarise the IVBA methodology, eight solutions were obtained for ICP-MS chemical element quantification from each geophagical sample subjected to the FOREhST procedure. These solutions represent the saliva/gastric and intestinal phases of extraction on <2000 µm and <250 µm particle sizes, and simulated F-S and F-SNF conditions.

To provide some background information about the geophagical materials, a number of procedures were undertaken in the laboratory. A 1∶2.5 w/v distilled water suspension was used for pH determination undertaken on the <2000 µm geophagical samples, whilst the method used to assess the cation exchange capacity (CEC) was based on that of Bascomb [37]. Organic carbon (OC) content was quantified from <150 µm material following a modified version of the procedure described by Walkley and Black [38] whereby the OC is oxidised with an acid dichromate solution. Soil colour was measured by comparison with a colour chart [39].

Trow [40] details the appropriate AQC procedures that were undertaken to measure the robustness of our analytical data. For example, two CRMs were included in the analysis to assess the accuracy of the determined pseudo-total concentrations. Replicates of these CRMs and two of the geophagical samples allowed the calculation of precision, whilst detection limits (calculated as 3 x standard deviation of mg/l blank values multiplied by appropriate dilution factor) were determined from the results derived from the analysis of blank samples.

A CRM was included when undertaking the FOREhST procedure and one geophagical sample was randomly chosen for replication. Blank samples were also included and, where appropriate, the bioaccessible concentrations of the in vitro solutions associated with the geophagical materials were ‘blank deducted’.

To aid the evaluation of the importance of the geophagical materials in supplying chemical elements to humans we have calculated MAP values for Co, Cr, Fe, Mn and Ni. Such calculations are based on the following assumptions:

Table 1 provides a summary of the AQC results associated with the determination of the pseudo-total concentrations. Some problems in the quality of the data are evident – e.g., raised concentrations of Ca in the blank samples, the poor accuracy of Na determined from the two CRMs – but the results following the analysis of the geophagical materials are discussed in light of the insight that the AQC procedures have provided.

Relative to the pseudo-total contents, the bioaccessible concentrations determined from the leachate solutions following the in vitro FOREhST procedure are more difficult to interpret from the perspective of AQC. Many concentrations are close to or below detection limits – thus, for example, having implications for quantifying the precision of our analysis determined from replicate samples – whilst CRMs that can be used to evaluate accuracy are very limited for this type of analysis. The British Geological Survey (BGS) reference soil no: 102 has a bioaccessible guidance value – determined using the Unified BARGE Method (UBM) IVBA procedure [31] – of 13±6 mg Pb/kg for the stomach phase (note this method simulates unfed conditions). This CRM was used in our analysis, but the two blank sample solutions associated with the F-SNF saliva/gastric phase – the closest simulated conditions that we employed compared to the UBM method – of our experimental procedure were found to have notably higher Pb concentrations than the BGS reference material solutions (indeed the solutions of all the geophagical samples associated with this F-SNF phase had a Pb content that was less than those of the blank solutions). We conclude that the CRM soil, and all the geophagical materials, are adsorbing Pb and consequently lowering the simulated saliva/gastric solution concentrations of the F-SNF scenario (a reaction that, as we highlight later, is also apparent for some of the other determined elements, most notably Cu and Zn). The UBM procedure is run at a stomach pH of 1.2, lower than the FOREhST method we employed where the pH following the saliva/gastric phase of the F-SNF scenario is typically 1.6±0.2. The latter pH range was observed for the solutions of the geophagical samples following the saliva/gastric F-SNF extraction in our work, but notably the pH of the BGS CRM soil solutions following this phase was 4.5–4.8. The increased pH will be a significant factor in accounting for the solution concentrations of Pb observed in our analysis of this CRM, with the implication being that we cannot assess the accuracy of the Pb concentrations determined in our work because of the significant pH differences compared to the UBM procedure from which the bioaccessible guideline value of 13±6 mg Pb/kg is obtained.

Eight samples were replicated when undertaking the IVBA procedure in order to determine the precision of this analysis. However, because a number of solution concentrations were below the limits of detection, the reproducibility of analysis can only be quantified from a more restricted number of duplicated samples: Table 2 provides a summary of the precision estimates determined in our study.

Most of the Calabash chalk samples are clay-rich soil materials that have been dried and/or baked into blocky or spherical units, though some are laminated shales (i.e., argillaceous sedimentary rocks; Figure 2). Table 3 records some details relating to these materials. All of the Calabash chalk samples have a very low OC content ranging from <0.1–0.4% (median  = 0.1%) suggesting that the geophagical soils have been excavated from sub-surface horizons rather than (more organic enriched) topsoils. These inorganic geophagical materials have a CEC rating that we interpret as mainly varying from very low (i.e., <6 cmol_c/kg) to medium (i.e., within the range 12–25 cmol_c/kg), and all samples of Calabash chalk are acidic in reaction (minimum-maximum pH = 3.4–6.4; median = 5.1). In contrast, Undongo has a high CEC (32 cmol_c/kg) and an alkaline reaction (pH = 7.7), though like the Calabash chalk it is associated with a very low OC content (0.1%).

A summary of the pseudo-total concentrations determined from the samples of Calabash chalk is provided in Table 4. The concentrations of Pb approximate or slightly exceed typical ‘average’ values that are associated with shales and present-day soils [43]–[44], and the contents are similar to Calabash chalk analysis that has been previously reported by Dean et al. [16]. However, a notable feature of the majority of the results in Table 4 is that while they are mostly within the normal range of total soil/shale concentrations, for most of the elements – Ca, Cd, Co, Cr, Cu, Fe, K, Mg, Mn, Na, Ni and Zn – the contents in many samples are below what may be regarded as an average value (in coming to this conclusion, it needs to be appreciated that normal/typical total concentrations of soil chemical elements are mostly reported from topsoils – that the Calabash chalk soils appear not to be – which are likely to be enriched because of natural processes such as adsorption by surface organic matter or low-level anthropogenic pollution). The pseudo-total concentrations of some pieces of Calabash chalk are an exception to this general observation with, for example, two samples – both shales – containing the highest amounts of Ca (1769 mg/kg and 2473 mg/kg, respectively), Co (both 17 mg/kg), Fe (37070 mg/kg and 46670 mg/kg) and Mg (12936 mg/kg and 21176 mg/kg) recorded in our study. However, no concentration of any chemical element in any sample of Calabash chalk can be regarded as excessively high. In contrast, the Undongo sample contains elevated pseudo-total concentrations of Cr and Ni (152 mg/kg and 126 mg/kg, respectively) compared to many soil materials, and is distinct from the Calabash chalk in having a low Pb and K content (8 mg/kg and 210 mg/kg, respectively).

For both the saliva/gastric and intestinal phases of the F-SNF scenario, all the in vitro leachate solutions have Pb concentrations less than the sample blanks, suggesting an adsorption of this PHE by soil/shale constituents (furthermore, for the intestinal phase, any Pb not adsorbed by the geophagical material is likely to be precipitated/complexed by the increased pH and enzyme concentration [45]–[46]). The blanks associated with the F-S scenario of our in vitro experimentation have a notably lower Pb concentration when compared with the blank solutions of the F-SNF scenario. These results are indicative of adsorption by the food component of the experimental procedure, a reaction exacerbated by the pH of the blank solutions (the blank solutions associated with the saliva/gastric part of the F-S scenario [median pH = 4.4] are notably less acidic than those of the F-SNF [pH = 1.8] phase). Binding of Pb by some of the geophagical materials is also apparent from the F-S in vitro results (i.e., the Pb content of the geophagical solutions are less than those determined from the blanks), especially by particles <250 µm in size. However, some materials are also a source of Pb since associated leachate solution concentrations are greater than those of the blanks, though these solution concentrations are close to or below the limits of detection (which are 0.25 mg/kg and 0.7 mg/kg for the gastric/stomach and intestinal phases of the F-S scenario, respectively). A maximum concentration of 0.65 mg Pb/kg (sample 1, <2000 µm soil) and 2.5 mg Pb/kg (sample 9, <2000 µm soil) was recorded from the F-S saliva/gastric and intestinal phases, respectively.

The results associated with the Cu and Zn bioaccessible concentrations are similar to those observed for Pb. The blank solution Cu contents associated with the F-SNF scenario are elevated (22 and 8 ng/ml for the saliva/gastric and intestinal phases, respectively), and many leachate solutions contain less than these concentrations and so are indicative of adsorption by the geophagical materials. Some intestinal leachate solutions contain more Cu than the F-SNF blanks, but at concentrations that can not be detected with confidence. Similar conclusions apply to the Cu concentrations of solutions associated with the F-S scenario, with only one sample (number 4) yielding more (1.1 mg/kg linked with the saliva/gastric phase) than can be robustly detected. The majority of the Zn concentrations obtained from the F-SNF scenario are below the content found in the blanks (8.1 and 10.1 ng/ml for the saliva/gastric and intestinal phases, respectively), though nearly all of the <2000 µm geophagical materials associated with the F-SNF saliva/gastric phase yield leachate solution concentrations slightly above the blank concentrations (but none are greater than the robust limit of detection which is 5.3 mg/kg). For the F-S scenario, again the majority of the leachate solution Zn concentrations are either below the blank solution contents or below the limits of detection (the latter are 1.2 mg/kg and 2.4 mg/kg for the saliva/gastric and intestinal phases of this scenario, respectively). However, one F-S leachate solution (Undongo, <250 µm soil, saliva/gastric phase) yielded a detectable concentration of 4.7 mg Zn/kg.

For the remaining chemical elements (i.e., Co, Cr, Fe, Mn and Ni) considered in the bioaccessibility study, whilst a number of leachate solutions have concentrations below those of the blank solutions, many are above so indicating the potential of ingested soils/shales to be sources of these mineral elements that humans can subsequently absorb. Tables 5 and 6 provide a summary of these bioaccessible concentrations, with a number being reported as below the limits of detection.

Any ingested geophagical material has the potential to release mineral nutrients and PHEs when they come in contact with digestive fluids. However, our investigations following the use of two scenario’s associated with the FOREhST IVBA procedure, indicate that at rates of consumption of 20 g/day, the Calabash chalk materials on the whole are not a significant source of mineral nutrients or PHEs to humans (this finding applies also to Undongo despite the Fe enrichment stated on the labelling of the product). Indeed, with many in vitro leachate solutions containing concentrations of, for example, Pb that are less than the blanks employed in our analysis, the geophagical materials are capable of adsorbing chemical elements, so preventing their absorption. Our results further suggest that foodstuffs can be a sink for Pb in the GI environment. Whilst the reported concerns about ingested Calabash chalk may be lessened with such findings we would still advocate caution about the use of such materials: their microbial sterility, for example, is questionable and needs to be investigated.