Geochemistry of selenium

Since its discovery in 1817 by Jons Jakob Berzelius Se has been an element that has attracted little attention and study by geochemists. As its concentrations in most geological materials are very low it has been difficult to study its distribution at the Earth's surface. It was Se toxicity in the environment that first attracted substantial attention to Se geochemistry and at about the same time, in the 1930s, industrial uses for Se began to develop (e.g. invention of the photocopier and Se rectifiers), driving a demand for Se as a commodity (Table 1).

Table 1. Industrial uses of selenium (after Fordyce⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾)

With the improvement in analytical methodology in recent years low levels of Se can more easily be determined routinely in geological and other environmental samples. For example, using X-ray fluorescence spectrometry, particularly X-ray fluorescence energy dispersive spectrometry, the BGS now routinely produces regional maps of Se in soils and stream sediments based on the analysis of thousands of samples (e.g. Wales geochemical atlas⁽¹⁷⁾ and Fig. 1).

Fig. 1. Distribution of selenium in drainage sediments from Wales (British Geological Survey⁽¹⁷⁾). This map is based on 19 000 stream-sediment samples collected at a density of approximately one sample per 1·5 km² as part of the British Geological Survey's Geochemical Baseline Survey of the Environment project.

Chemical properties and production

Se (atomic number 34, atomic mass 78·96 and period group 16) has chemical and physical properties that are intermediate between those of metals and non-metals. In its chemical behaviour it resembles S and exists in similar oxidation states (see Table 2). The oxidation state of Se is critical in determining its availability in the food chain. For example, in neutral-to-alkaline soils Se⁶⁺ (selenate) is the dominant species. This form of Se is generally more soluble and mobile in soils and is readily available for plant uptake. Selenite (Se⁴⁺) has lower solubility and a greater affinity for adsorption on soil particle surfaces than Se⁶⁺ and hence selenites are less bioavailable in agricultural crops⁽Reference Mikkelsen, Page, Bingham and Jacob¹⁸⁾.

Table 2. Chemical forms of selenium in the environment (after Fordyce⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾)

There are no mines in the world that specifically extract Se; instead it is a by-product of the production of other metals (e.g. refining of Pb and Cu) or recovered from the sludge accumulated in H₂SO₄ plants.

Distribution and cycling

As rocks and minerals at the Earth's surface are the primary source of Se in the terrestrial environment, knowing its natural abundance in these sources is important to understanding Se distribution. Average crustal abundances of Se are generally very low (0·05–0·09 mg/kg)⁽Reference Taylor and McLennan¹⁹⁾. Intrusive igneous rocks, e.g. granites, rarely exceed these concentrations. For extrusive (volcanic) rocks Se concentrations are more complex. Ash and gases associated with volcanic activity can contain substantial quantities of Se⁽Reference Fergusson²⁰⁾ but this process leaves behind volcanic rocks such as basalts and rhyolites that have relatively very low concentrations of Se⁽Reference Flemming and Davis^21–Reference Neal and Alloway²⁴⁾. Generally, sedimentary rocks are richer in Se than are igneous rocks but limestones and sandstones will rarely contain >0·1 mg/kg⁽Reference Neal and Alloway²⁴⁾. As it is estimated that igneous rocks, sandstones and limestones account for 96% of the rock in the upper 16 km of the Earth's crust⁽Reference Clarke and Washington²⁵⁾, generally low concentrations of Se are therefore to be expected over much of the Earth's surface.

However, there are exceptions to the generally low levels of Se in some types of sedimentary rocks and deposits. Very high concentrations (>300 mg Se/kg) are reported in some phosphatic rocks, reflecting similarities in the PO₄^3– and SeO₄^2– anions⁽Reference Flemming and Davis^21–Reference Neal and Alloway²⁴⁾. Coal and other organic-rich deposits tend to have higher levels of Se, with typical ranges from 1 mg/kg to 20 mg/kg, although values of >600 mg/kg are reported for some black shales⁽Reference Jacobs²²⁾. A summary of the Se content in some common rock types is given in Table 3.

Table 3. Selenium (mg/kg) in some common rock types (modified from Fordyce⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾)

As a chalcophile element (i.e. an element that prefers a sulfide host rather than a silicate one) Se is often found in sulfide mineral deposits in which it substitutes for S in the mineral crystal lattice (e.g. pyrite, chalcopyrite and sphalerite). S and Se have similar ionic radii (Se^2– 0·184 nm and S^2– 0·198 nm) and similar electronegativities (Se 2·4 and S 2·5)⁽Reference Pauling²⁶⁾. There are Se minerals such as crookesite (Cu, Tl, Ag₂Se), clausthalite (PbSe) and berzelianite (Cu₂Se), but these minerals are rare compared with sulfides. Elemental Se⁰ is occasionally reported⁽Reference Flemming and Davis^21, Reference Neal and Alloway^24, Reference Tokunaga, Pickering and Brown²⁷⁾.

The geographical distribution of Se in an area with diverse geology is likely to be quite variable, reflecting the occurrence of different rock types or sulfide mineral deposits. Higher levels of Se would be expected in environments in which the underlying bedrock consists of the high-Se sedimentary rocks described earlier or because of the occurrence of sulfide mineralisation. This feature is well illustrated by the Wales regional geochemistry map for Se in stream sediments (Fig. 1). This map has been produced by the BGS's Geochemical Baseline Survey of the Environment regional geochemical mapping programme⁽Reference Johnson, Breward and Ander²⁸⁾ and is based on the X-ray fluorescence spectrometric analysis of approximately 19 000 stream-sediment samples collected at an average density of one sample per 1·5 km². This map of Se shows regional high levels associated with sulfide mineralisation (Parys Mountain, Snowdon and the Harlech Dome) and with rock types that are relatively high in Se (Pembrokeshire and South Wales coalfields and Mercia Mudstone). These levels contrast with the lower concentrations associated with the sandstones and siltstones of mid-Wales. Similar regional geochemical maps for Se in soils and stream sediments are available for central and eastern England⁽Reference Ander, Johnson and Fordyce²⁹⁾ and Northern Ireland.

An overview of the cycling of Se⁽Reference Haygarth, Frankenberger and Benson¹³⁾ looks at the various compartments in the global cycle to describe the annual Se flux (Table 4). This flux pattern demonstrates two important facts, i.e. that the marine system constitutes the main natural pathway for the cycling of Se but it is anthropogenic activity that is a major source of Se migration. Natural release of Se is estimated to be 4500 t/year compared with the estimated anthropogenic release of 76 000–88 000 t/year, which represents a biospheric enrichment of 17 and indicates the important influence of man in the cycling of Se⁽Reference Allan and Nriagu³⁰⁾. Common anthropogenic sources of Se in the environment are shown in Table 5. Important sources include the combustion of coal and petroleum fuels, metal extraction processes and the use of phosphate fertilisers and sewage sludge in agriculture.

Table 4. Global selenium fluxes (after Fordyce⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾ based on data from Haygarth⁽Reference Haygarth, Frankenberger and Benson¹³⁾)

Table 5. Common anthropogenic sources of selenium in the environment (after Fordyce⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾)

A combination of natural and anthropogenic controls can be seen in the soil Se from northern Europe (Fig. 2). Although this map from the geochemical atlas of the agricultural soils of Northern Europe⁽Reference Reimann, Siewers and Tarvainen³¹⁾ is based on low-density sampling (one site per 2500 km²), it still delineates both natural and anthropogenic features that control the distribution of Se. The higher values in Russia can be related to black shales. On the west coast of Norway high levels of Se are caused by a combination of organic-rich soils and input of Se via sea spray. In Finland high concentrations are associated with agricultural soils (fertiliser supplemented with selenate⁽Reference Aro, Alfthan and Varo³²⁾) and in the north very-organic-rich soils (peatland) turned into agricultural fields. The most intriguing feature of this map is that for the part of northern Germany that was sampled there is a clear boundary between low and moderate Se in the top soils. This pattern follows the old East–West Germany border and probably reflects the use of Se-rich fertilisers in the west.

Fig. 2. Selenium in (a) top (0–25 cm) and (b) bottom (50–75 cm) soils from northern Europe. (After Reimann et al.⁽Reference Reimann, Siewers and Tarvainen³¹⁾.)

Selenium in soil

Soil is of fundamental importance in the food chain and is of great importance in determining the Se status of crops and livestock. The world mean value for soils is 0·4 (general range 0·01–2) mg/kg⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾. With the exception of instances of anthropogenic modification, in most cases there is a very strong correlation between the concentration of Se in the geological parent material and the soils derived from them. While the geological source of Se is a primary control on soil Se concentration, there are a number of biological and physico-chemical properties that control the Se bioavailability (i.e. the mobility and the uptake by plants and animals) and hence its concentration in foods. These factors include the prevailing pH and redox conditions, the form or speciation of Se, soil texture and mineralogy, organic matter content and the presence of competitive ions⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾. Many of these variables and their relationship are illustrated in Fig. 3.

Fig. 3. Schematic diagram showing the main controls on the chemical speciation and bioavailability of selenium in soils. – – – ▸, Increasing mobility. (After Fordyce⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾.)

A practical example of the importance of soil Se bioavailability is illustrated by considering two areas of high-Se soils from the USA. In the northern Midwest there are Se toxicity problems in plants and livestock from soils containing 1–10 mg Se/kg because ⩽60% of the element is readily bioavailable in the semi-arid alkaline environment. In contrast, soils in Hawaii containing as much as 20 mg/kg are not associated with similar toxicity problems because the Se is held in Fe and Al complexes in the humid lateritic soils⁽Reference Oldfield³³⁾.

Health impacts of selenium deficiency and toxicity

The health impacts of Se deficiency and toxicity on both human subjects and livestock are well documented elsewhere (for example, see Rayman⁽Reference Rayman⁹⁾, Fordyce⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾, Reilly⁽Reference Reilly³⁾ and Scientific Advisory Committee on Nutrition⁽³⁴⁾). On a global scale overt Se toxicity in human subjects is far less widespread than deficiency⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾. Consequently, the present paper is more focused on deficiency than toxicity.

The recognition of the detrimental effects of too much Se on livestock in the Midwest states of the USA was inevitably followed by studies of Se toxicity in human subjects⁽Reference Reilly³⁾. While initial studies showed elevated urinary Se concentrations in the population but no definite links to clinical symptoms of selenosis, a higher incidence of gastrointestinal problems, poor dental health, diseased nails and skin discolouration, now recognised as symptoms of Se poisoning in human subjects, has been reported⁽Reference Smith and Westfall³⁵⁾. Endemic selenosis in China became highlighted in the scientific literature during the 1980s⁽Reference Yang, Wang and Zhou^36–Reference Tan³⁸⁾, although outbreaks of endemic human selenosis occurred in the Enshi District, Hubei Province during the 1960s⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾. This condition was found to be associated with high-Se crops grown on soils derived from coal containing >300 mg Se/kg. The main symptoms of Se poisoning in Enshi were observe to be hair and nail loss but disorders of the nervous system, skin, poor dental health, garlic breath and paralysis were also reported and between 1961 and 1964 morbidity rates reached 50% in the worst-affected villages⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾. A more modern Se toxicity problem can be related to mineral supplement intake. In the USA in 1984 twelve cases of selenosis were reported when tablets containing 27–31 mg Se (approximately 182 times more than concentrations stated on the label) were taken. Total individual doses estimated to be consumed by the individuals ranged from 27 mg to 2387 mg⁽³⁹⁾. The low observed adverse effect level for Se is 1540 (se 653) μg/d⁽Reference Whanger, Vendeland and Park⁴⁰⁾.

The health impacts of Se deficiency have been discussed in more detail elsewhere⁽Reference Rayman^9, Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley^15, Reference Rayman⁴¹⁾ and are summarised in Table 6 with conditions being considered under the following headings: mortality; immune function; antiviral effects and HIV; fertility and reproduction; cancer; thyroid effects; CHD.

Table 6. Summary of human health conditions that have been related to selenium deficiency

GPx, glutathione peroxidise; NHANES III, Third National Health and Nutrition Examination Survey; RCT, randomised controlled trials.

Se has a lower reference nutrient intake value of 40 μg/d for males and females, and a reference nutrient intake of 60 μg/d for females and 75 μg/d for males⁽⁴²⁾. Optimal intakes must consider the species of Se ingested, the medical condition being addressed, the overall nutritional adequacy of the group or population, the extent to which genomic differences between individuals or populations may be relevant, other risk or lifestyle factors that may be present and, most importantly, the background Se intake in that population (for discussion, see Rayman⁽Reference Rayman⁹⁾).

In the body Se is used as part of a group of molecules known as selenoproteins that contain the amino acid selenocysteine. These selenoproteins have a variety of functions including: glutathione peroxidise, antioxidant defence; thioredoxin reductase, cell redox control; selenoprotein SEPS1, anti-inflammatory; selenoprotein P, Se transport in the plasma⁽Reference Papp, Lu and Holmgren⁴³⁾.

Keshan disease (KD) is a medical condition in human subjects that is associated with Se deficiency and the subject of a case study highlighted later in the present paper. It is an endemic cardiomyopathy (heart disease) that derives its name from substantial occurrences in Keshan County, north-east China, with annual prevalence exceeding forty cases per 100 000 of the population and 1400–3000 deaths per year⁽Reference Tan³⁸⁾. Affected populations are characterised by very low Se status as indicated by hair Se contents. Like many of the conditions associated with Se deficiency there are believed to be other contributory factors to KD. More recent studies (for example, see Li et al.⁽Reference Li, Peng and Yang⁴⁴⁾) have demonstrated a high prevalence of the Coxsackie B virus in patients with KD, which is now thought to be a cofactor in the disease.

Selenium status in human subjects is linked to the environment

Se status in human subjects can be assessed by determining Se levels in blood (whole blood, serum or plasma), hair, nails and urine⁽Reference Reilly³⁾. The geochemistry of Se described earlier provides a guide to the relative amounts of Se that may be expected in a particular geological setting and studies of Se in the soil, drainage sediments, water and crops give an indication of environmental Se status.

It has been noted that there is a marked variation in Se intake and status (as measured in blood, plasma or serum, toenails or hair) from one part of the world to another and even between different parts of the same country⁽Reference Rayman^41, Reference Rayman⁴⁵⁾. Fig. 4 shows serum or plasma Se levels from a compilation of European studies and puts these data in the context of the optimal levels required for glutathione peroxidise and selenoprotein P activity. These data for Europe, especially when put into the context of US serum Se levels of 125–137 μg/l (as measured in recent US National Health and Nutrition Examination Surveys⁽Reference Niskar, Paschal and Kieszak^46, Reference Bleys, Navas-Acien and Laclaustra⁴⁷⁾), demonstrate that European populations are generally of low Se status. Hair analysis has been used in several studies of Se status in China⁽Reference Chen, Yang and Chen⁴⁸⁾ and is also the method that was used by the BGS's Se case studies carried out in China in the 1990s⁽Reference Johnson, Ge and Green^16, Reference Fordyce, Zhang and Green^49, Reference Appleton, Zhang and Green⁵⁰⁾. Whilst the methodology for hair sampling is not generally standardised and the hair can be contaminated from external sources (e.g. Se-containing shampoos), it is a technique that can be easily carried out by non-medical personnel.

Fig. 4. Concentrations of selenium in human serum and plasma in Europe. GPx, glutathione peroxidise. (After Rayman⁽Reference Rayman⁴¹⁾.)

The importance of understanding Se speciation in relation to its distribution and behaviour in soils has been discussed earlier. An understanding of food-chain Se and health equally needs knowledge of more than only total Se concentrations. Health effects, both beneficial and toxic, thought to be associated with specific Se species are described elsewhere⁽Reference Rayman⁵¹⁾.

Baseline geochemical maps (e.g. Fig. 1) give an immediate impression of whether there is high or low Se in the environment, although it is only in recent years that such baseline maps have been produced. Statistical analysis of such large data sets when they become available will enable good comparisons between the environmental Se status of different countries or regions of the world. Fig. 5 demonstrates that where a good dependence on the local environment can be shown, other environmental indicators such as plant Se concentrations and Se-deficiency diseases in livestock (e.g. white muscle disease) can be used to indicate environmental Se status⁽Reference Muth and Allaway⁵²⁾.

Fig. 5. Indicators of environmental selenium status for the USA. (), Selenium-rich environments; (•), plant selenium >50 mg/kg; (○), white muscle disease incidence. (From Fordyce⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾, adapted from Muth & Allaway⁽Reference Muth and Allaway⁵²⁾.)

Case study: Zhangjiakou District of China

In the late 1990s the BGS carried out a project on ‘prediction and remediation of human Se imbalances’ in China supported by the UK Department for International Development. The objective of the project was to develop a methodology for delineating where Se deficiency or toxicity posed a health risk. The environmental geochemical controls on the distribution of Se-responsive diseases were also evaluated. The case study carried out in the KD belt in the Zhangjiakou District⁽Reference Johnson, Ge and Green¹⁶⁾ serves as a good example of how the environmental and human Se status can be determined and compared.

On a regional scale the area investigated is part of the KD belt shown as being a Se-deficient ecological landscape⁽Reference Tan³⁸⁾ (Fig. 6). However, within this belt some villages were found to be more susceptible to KD than others. The case study in Zhangjiakou District was designed to look at villages on a local scale in order to determine which factors could account for high incidences of KD in some villages whilst other villages in the same county had moderate or no incidence of the disease. The study was constrained by budget to a 10 d period of fieldwork and was timed to coincide with the harvest so that the wheat and the soil on which it was grown could be sampled simultaneously.

Fig. 6. Map showing the location of the Zhangjiakou case study area. (), Keshan disease belt. (Adapted from Tan⁽Reference Tan³⁸⁾ with permission.).

The investigation has been described in more detail⁽Reference Johnson, Ge and Green¹⁶⁾ and the following is a summary account. The project was a collaboration with the Chinese Institute of Rock and Mineral Analysis and depended on the excellent medical records that documented the health of every villager in the target communes. The sampling strategy targeted fifteen villages within the Se-deficiency belt: five with no incidence of KD; five with moderate incidence (>0–3%); five with high incidence (>3%). From each village five samples of soil and grain and one sample of the village drinking water were collected to represent the Se status of the local environment and five human hair samples to represent the human Se status.

Much of the area was quite mountainous, making access difficult. Villages (populations of 200–400) were quite isolated and connected by dirt tracks that were impassable during bad weather. This isolation of villages and the near subsistence level of existence made the region ideal for investigating how local environmental factors were influencing the rates of incidence of KD.

Se was determined using hydride-generation atomic fluorescence spectroscopy and a summary of the Se results for the four sample media is shown in Fig. 7. It is clear that both the human Se status (as indicated by hair) and the environmental Se status (as indicated by water, grain and soil) are Se deficient. The grain values are less than half those reported for UK wheat and soil values are one-quarter to half the UK average. The thresholds for environmental deficiency shown in Fig. 7 are based on data from the The Atlas of Endemic Diseases and their Environments in the People's Republic of China ⁽Reference Tan³⁸⁾ and are discussed in more detail elsewhere⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾. For the hair, drinking water and grain results the trend is as would be expected, with the villages showing a high incidence of KD having the least Se. However, soils show the opposite trend; whilst levels are still very low, the highest concentrations of soil Se are found in fields surrounding villages with the highest incidences of KD.

Fig. 7. Box and whisker plots showing the selenium content of the four sample types (human hair (a), grain (b), drinking water (c) and soil (d)) classified by incidence of Keshan disease in the local population in Zhangjiakou District, China. Plots were created using UNISTAT® (v4.5) statistical software (Unistat Ltd, London, UK). Medians are represented by horizontal bars within the box and upper and lower quartiles ±1·5×quartile range are represented by the whiskers. (–), Arithmetic mean; (·), outlying values. For a discussion of marginal and deficient thresholds, see text. (After Johnson et al.⁽Reference Johnson, Ge and Green¹⁶⁾.)

An explanation for this paradox can be seen by looking at some of the other variables determined for the soil samples. The overall total soil Se content is low and with water-extractable Se representing generally <1% of the total the determination of the very low concentrations of water-extractable Se was found to be very difficult. However, the higher percentage of water-extractable Se is present in the samples from villages with no incidences of KD and the majority of samples from villages with high incidence of KD have extractable Se concentrations below detection; thus indicating that the problem in villages with a high incidence of KD is that the little amount of Se that is available in this environment is not being released from the soils. Determination of a full range of chemical elements and other variables in the soils and statistical analysis of the data shows good correlation between total Se and the organic matter content of the soil (as estimated by loss on ignition; Fig. 8). This plot is very revealing, particularly if the soil sites are classified by their colour as an indication of soil type. Clearly, the black, organic-rich soils have the least-mobile Se content. The conclusion from the Zhangjiakou study is that in an already-Se-deficient environment soil characteristics will have a very marked localised impact on Se availability to the food chain and it is very important to take into account the bioavailability of the Se in the environment.

Fig. 8. A plot of loss on ignition (representing soil organic matter content) v. total selenium in soils from the Zhangjiakou District, China classified by soil colour. (▪), Black; (•), brown–black; (), yellow–black; (▵), yellow. (After Johnson et al.⁽Reference Johnson, Ge and Green¹⁶⁾.)

Health impact of environmental selenium deficiency can be seen where food is sourced locally

The BGS case studies in China have demonstrated that the health impacts of Se deficiency in human subjects can be related to the level of Se in the environment if the local inhabitants are dependent on their immediate surroundings for sustenance. In urbanised and developed regions of the world in which populations are less dependent on their immediate environment for foodstuffs and drinking water, establishing a link is more difficult. However, maps such as Fig. 1 are invaluable in understanding the Se concentrations in crops and livestock and can be used to predict which areas will have an imbalance in Se concentrations in agricultural products. An effective application of Se through fertilisers must be demonstrably safe to the environment and monitored appropriately. In the UK the quality and quantity of baseline geochemical data will enable such monitoring to be undertaken with confidence⁽Reference Broadley, White and Bryson⁵³⁾.

The recent decline in selenium status in the UK

Data on human Se intake (for example, see data from the UK Total Diet Study⁽³⁴⁾ and from Rayman⁽Reference Rayman⁴⁵⁾) clearly demonstrate that in the UK intake has been falling and is now demonstrably less than the range of Se intakes believed to be required for the optimal activity or concentration of some essential plasma selenoproteins (see UK data in Fig. 4). The UK Total Diet Study shows a drop in Se intake from 60 μg/d in the 1980s to approximately half that value by 2000⁽³⁴⁾ (Fig. 9), leaving it well below reference nutrient intake values (60 μg/d and 75 μg/d for adult females and males respectively⁽⁴²⁾). Whilst geological controls can account for the generally low concentrations in UK crops relative to North America, it is anthropogenic activities that are thought to have led to a decline in the Se status of the UK population and most of Europe. The following summarises the possible causes of declining Se status:

1. 1. changes in bread-making technology allowing the use of low-protein low-Se EU and UK wheat rather than high-protein high-Se wheat from North America⁽Reference Whitley⁵⁴⁾;

2. 2. changing dietary patterns; in northern Europe meat and poultry are the main sources of dietary intake of Se. A decline in meat eating, particularly good sources of Se such as liver and kidney, will result in a decline in dietary Se intake unless replaced by foodstuffs of equally-high Se content. The Scientific Advisory Committee on Nutrition⁽³⁴⁾ reports that the main source of Se in the UK diet are breads, cereals, fish, poultry and meat and that rich dietary sources of Se include brazil nuts (Bertholletia excelsa), fish and offal;

3. 3. reduced atmospheric deposition from fossil-fuel burning has reduced Se input into the surface environment. Coal can potentially be a very rich source of Se⁽Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley¹⁵⁾;

4. 4. increased use of S fertilisers to overcome S deficiency in arable crops has meant that there is increased competition between S and Se in crop uptake⁽Reference Adams, Lombi and Zhao⁵⁵⁾. S inputs from fertilisers and atmospheric deposition have an important influence on the Se status of wheat grain⁽Reference Fan, Zhao and Poulton⁵⁶⁾.

Fig. 9. Plot of UK wheat imports from North America (, USA; , Canada) in the context of the UK daily selenium intake (○). Daily selenium intake from the UK Total Diet Study (see Table 6 from Scientific Advisory Committee on Nutrition⁽³⁴⁾) and wheat import data (starting in 1986) from FAOSTAT (Food and Agriculture Organization⁽⁸²⁾; detailed trade matrix).

Relationship between selenium status and environmental selenium from crops

The contrast between the Se status of North American and UK wheat is another good example of how environmental Se status impacts on crops. BGS data for >22 000 topsoils from central and eastern England show a median value of 0·3 mg Se/kg (British Geological Survey, unpublished results). This value can be contrasted with that for soils from the Se-deficiency belt of China (Fig. 7) that have median values of 0·13 mg Se/kg and those of soils developed on the cretaceous shale of South Dakota, Montana, Wyoming, Nebraska, Kansas, Utah, Colorado and New Mexico in the USA that have Se contents ranging from 2 mg/kg to 10 mg/kg⁽Reference Jackson and Bear⁵⁷⁾ (see Fig. 5). The Se content of wheat grain from these areas reflects the soil concentrations, with 0·025, 0·016 and 0·37–0·46 mg/kg for wheat grain from the UK⁽Reference Adams, Lombi and Zhao⁵⁵⁾, Zhangjiakou (China)⁽Reference Johnson, Ge and Green¹⁶⁾ and USA⁽Reference Wolnik, Fricke and Capar^58, Reference Hahn, Kuennen and Caruso⁵⁹⁾ respectively. There is a >10-fold difference in Se concentrations between UK and USA wheat, so it is easy to comprehend that a decline in the use of North American wheat will result in a decline in Se in the UK diet. This contrasting content of Se in wheat grain from the UK and USA is illustrated in Fig. 10. Data for wheat grain⁽Reference Eurola, Ekholm and Ylinen⁶⁰⁾ before and after Finnish supplementation of fertilisers with sodium selenate⁽Reference Aro, Alfthan and Varo³²⁾ is also illustrated in Fig. 10 to demonstrate the success of this programme in raising the wheat grain Se content. As a consequence, the Se intake in Finland has grown from about 30 μg/d to 100 μg/d⁽Reference Paivi, Maija and Pekka⁶¹⁾.

Fig. 10. A comparison of UK (□), US (▪) and Finnish () wheat selenium concentrations. Data for Finnish wheat grain⁽Reference Eurola, Ekholm and Ylinen⁶⁰⁾ represent values obtained before and after supplementation of fertilisers with sodium selenate⁽Reference Aro, Alfthan and Varo³²⁾.

North American wheat is now imported for more than UK bread making

When the UK import of wheat from North America is shown together with the UK daily Se intake over a period of several decades (Fig. 9) it can be seen that in the 1990s the daily intake levels of Se did indeed drop with the declining imports of North American wheat. However, although wheat imports at the end of the 1990s are higher than pre-1990 levels, there is no marked increase in the UK daily Se intake, as shown by the daily Se intake measured in 2000. This outcome would suggest that current Se intake levels no longer simply reflect North American wheat imports; the imported wheat must now be being put to some other use.