Uptake of Metals from Soil into Vegetables

Abstract

The consumption of locally-produced vegetables by humans may be an important exposure pathway for soil contaminants in many urban settings and for agricultural land use. Hence, prediction of metal and metalloid uptake by vegetables from contaminated soils is an important part of the Human Health Risk Assessment procedure. The behaviour of metals (cadmium, chromium, cobalt, copper, mercury, molybdenum, nickel, lead and zinc) and metalloids (arsenic, boron and selenium) in contaminated soils depends to a large extent on the intrinsic charge, valence and speciation of the contaminant ion, and soil properties such as pH, redox status and contents of clay and/or organic matter. However, chemistry and behaviour of the contaminant in soil alone cannot predict soil-to-plant transfer. Root uptake, root selectivity, ion interactions, rhizosphere processes, leaf uptake from the atmosphere, and plant partitioning are important processes that ultimately govern the accumulation of metals and metalloids in edible vegetable tissues. Mechanistic models to accurately describe all these processes have not yet been developed, let alone validated under field conditions. Hence, to estimate risks by vegetable consumption, empirical models have been used to correlate concentrations of metals and metalloids in contaminated soils, soil physico-chemical characteristics, and concentrations of elements in vegetable tissues. These models should only be used within the bounds of their calibration, and often need to be re-calibrated or validated using local soil and environmental conditions on a regional or site-specific basis.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

The transfer of metal (and metalloid) contaminants from soil, through plants, to humans, may be an important exposure pathway in some urban and residential environments. This exposure pathway may also be important for agricultural land-use scenarios to manage exposure of the general population to metals and metalloids. This chapter summarises our current understanding of metal/metalloid behaviour in soils, uptake and transport of these elements by plants, and the current models and concepts used to predict exposure of humans to metals and metalloids through a food-chain pathway.

2 Metal and Metalloid Chemistry in Soil

In describing the chemistry of metals and metalloids in soil, distinctions need to be drawn between the behaviour of cationic and anionic elements given the importance of surface charge to the fate and behaviour of elements in soils (Sposito 1981, 1989). Most topsoils have a net negative surface charge, with exceptions being soils rich in iron, aluminium or manganese oxides and depleted in organic matter and phosphorus (P) or sulphur (S). These latter soils, which may have a significant net positive charge, are found in tropical regions, particularly in subsoils (Wong and Wittwer 2009). Soils with net negative charge retain cationic metals more strongly, soils with net positive charge will retain anionic metals more strongly. Hence we will discuss the behaviour of cationic and anionic metals separately. A further key property of metals which controls environmental fate is oxidation state, as this may affect charge e.g. chromium (Cr) – Cr³⁺ present as this trivalent cation in most soils, and Cr⁶⁺, present as an oxyanions as CrO₄ ²⁻ or HCrO₄ ⁻.

2.1 Cationic Metals

Elements in this group include cadmium (Cd), cobalt (Co), chromium (Cr), copper (Cu), mercury (Hg), nickel (Ni), lead (Pb) and zinc (Zn). These elements may have different valence states, but are generally present in cationic form in soil, either as free metals in solution (Mⁿ⁺), metals complexed to inorganic or organic ligands (ML) or as solid phases (e.g. precipitated as minerals, or sorbed to negatively charged surfaces) (McLaughlin 2002 – Table 8.1). The distribution coefficient (K _d ) is a measure of the relative distribution of elements between the solid phase and pore water (water held in pore space between soil particles), and is an important property to consider in assessing the potential availability of metals to plants:

$$K_d = M_{{\textrm{solid}}}/M_{{\textrm{pore\,water}}}$$

((8.1))

where M _solid = concentration in the solid phase (mg·kg_dw ^–1) and M _{pore water} is the concentration in the pore water (mg·L^–1).

Table 8.1 Physical and chemical properties of the cationic metals

K _d values for cationic metals vary widely across soils, and are also dependent on metal properties such as valence state, e.g. Co³⁺ has much higher K _d values in soils than Co²⁺. Metal partitioning in soils can be due to both sorption/desorption reactions, as well as precipitation/dissolution reactions (Fig. 8.1). This obviously depends on the degree of and type of soil contamination. Soils highly contaminated by soluble metal sources are more likely to have distinguishable metal precipitates, as are those soils contaminated by sparingly soluble materials. Sorption/desorption processes are more likely to control metal partitioning in soils contaminated by lower concentrations of soluble contaminants, or in soils contaminated by highly soluble solid materials. The impact of both reactions combined is represented by the K _d .

Fig. 8.1

Relationship between cationic metals in discrete mineral phases (MX), free cationic metal ions in pore water (Mⁿ⁺), complexed metals in pore water (ML), metals sorbed to soil charged surfaces, and metals occluded within minerals or organic matter in soil. Reaction represents sorption/desorption reactions and is described by a sorption coefficient, reaction represents precipitation/dissolution reactions and is described by a solubility product, reaction represents a solution complexation reaction and is described by an association constant, and reaction is generally termed an ‘ageing’ reaction whereby sorbed metals become less available with time (McLaughlin 2001b)

There have been numerous studies of metal partitioning in soils and how K _d values are affected by soil mineralogy, particle size, pH, salinity, redox status, metal loading, etc. and the reader is referred to several review articles that summarise these findings (Buchter et al. 1989; Degryse et al. 2009; McBride 1989; Sauvé et al. 2000). The ranges of K _d values found for various metals are shown in Table 8.2.

Table 8.2 Range of K _d values for cationic metals and anionic metals/metalloids in soils

The most important factor found to influence the dissolution of cationic metal precipitates, and the release of cationic metals by negatively charged soil surfaces, is soil pH. Hence higher concentrations of cationic metals are found in the pore waters of acidic soils. Moreover, raising soil pH by liming will significantly reduce cationic metal concentrations in soil pore waters. Soil pH is therefore a master variable in controlling the distribution of cationic metals between the solid phase and the pore water, and hence their availability to plants. Soil particle size and mineralogy also influence K _d values, with sandy soils or soils having low concentrations of high surface area minerals having low K _d values. Sandy acidic soils are therefore most at risk of having high concentrations of cationic metals in soil pore water, and hence a risk of plant uptake of these contaminants. Due to the inverse relationship between metal loading to soil and metal K _d (Hendrickson and Corey 1981), highly contaminated soils are likely to have lower metal retention than soils contaminated by lower metal loads.

While K _d values are normally determined experimentally in the laboratory, for many metals and metalloids approximate K _d values can be estimated from empirical relationships derived using total metal concentrations in soil, and the main factors affecting partitioning, usually pH, and soil clay/oxide or organic matter content (Anderson and Christensen 1988; Buchter et al. 1989; Sauvé et al. 2000). These models do not require detailed input data (as do more mechanistic models of metal partitioning). Therefore, they can be a useful first screening tool in first tier Risk Assessments to assess potential metal availability at a contaminated site, using very simple soil analytical data – total metal concentrations and pH and/or clay/organic matter content.

Cationic metals may also react with soluble natural or anthropogenic ligands (e.g. chloride, dissolved organic matter or synthetic chelates such as ethylenediaminetetraacetate – EDTA) to form solution complexes in soil pore waters (Fig. 8.1). This may reduce or even reverse the charge on the cationic metal and hence markedly reduce the K _d value, increasing metal availability, but not necessarily increasing metal uptake as this is governed by the free metal ion concentration in the pore water and other factors (see Section 8.4.1).

2.2 Anionic Metals/Metalloids

Several of the metals and metalloids may be found in anionic form in soils, due to their ability to combine with oxygen to form oxyanions (Table 8.3). These ions may not be held as strongly by soil due to the net negative charge in most soils, and hence they may have relatively low K _d values compared to the cationic metals. Hence these elements are much more readily available for plant uptake from soil compared to cationic metals with high K _d values (e.g. lead). Exceptions to this rule are those anions that can form strong bonds through ligand-exchange with soil minerals, e.g. phosphate, fluoride, but there are few metals or metalloids with this property. Arsenate is one of the few metalloids that may associate with soil minerals through ligand exchange reactions, and hence have stronger binding than would be suggested by consideration of charge alone.

Table 8.3 Physical and chemical properties of the anionic metals/metalloids

In contrast to cationic metals, the oxyanions have higher K _d values in acidic soils compared to alkaline soils, and different redox species of the same metal/metalloid may have widely differing K _d values (Bartlett and James 1988; Bowell 1994; Goldberg 1997; Goldberg and Forster 1998; Nakamaru et al. 2005). Arsenic is a good example, where As⁵⁺, usually present in soils as the arsenate ion H_xAsO₄ ^−(3−x) generally has higher K _d values in soil than As³⁺, present as H₃AsO₃ (Bowell 1994).

Of the metals normally found at contaminated sites, chromium perhaps is unique in that a valency change causes charge reversal, and this markedly affects not only chromium partitioning (Bartlett and James 1988), but also toxicity (McGrath 1982), with the Cr⁶⁺ species being more available and toxic compared to Cr³⁺.

2.3 Effects of Soil Redox

As outlined above, changes in soil oxidation/reduction potential have the capability to alter the valence state of metal/metalloids present in soil. However, soil redox has a more important effect on metal chemistry in that the major components of soil which are active in metal/metalloid retention, Fe and Mn oxyhydroxides, are also redox sensitive principally through the Fe²⁺/Fe³⁺ and Mn²⁺/Mn⁴⁺ couples. Both Fe²⁺ and Mn²⁺ species are much more soluble than their oxidised species, so that reduction of soil, e.g. due to waterlogging, will cause reductive dissolution of Fe and Mn oxides, and often destruction of the surfaces active in the retention of many metals and metalloid ions (Plekhanova 2007). Thus, water logging of soil will lead to the release of elements strongly retained by these surfaces, e.g. arsenic concentrations are much higher in pore waters of waterlogged soils than aerobic soils (Bowell 1994; Marin et al. 1992).

3 Plant Acquisition of Metals and Metalloids from Soil

3.1 Root Uptake Pathway

In this section, we will elaborate on the processes by which metals and metalloids are taken up from soil by plant roots and how these elements are subsequently distributed with the plant tissues.

3.1.1 Speciation and Ion Uptake Rate

Plant roots absorb metals/metalloids via the root bathing pore water. The metals/ metalloids commonly enter the plant as ions (e.g. Cd²⁺, H₂AsO₄ ⁻) via ion channels or carriers that have the capacity to concentrate the elements from solution. Non-essential elements enter the plants using the uptake systems of nutrients that resemble the contaminants in terms of charge and ionic radius. Passive uptake of the elements through water uptake rarely explains the observed uptake of several nutrients (Barber 1995) and the same is true for contaminants such as cadmium and lead. The uptake rate generally increases with increasing concentration in pore water. Short-term ion uptake studies with roots demonstrate that uptake of contaminants follows a concentration-dependent pattern that is similar to enzyme kinetics. In a similar, but not identical, fashion it is observed that plant tissue concentrations rise as soil concentrations rise. Such patterns are important in Risk Assessment where the concept is to identify tolerable soil concentrations at which plant concentrations are below target values. Figure 8.2a summarises the general patterns for non-essential and essential contaminants. The tissue concentrations of non-essential elements such as arsenic, cadmium, mercury, and lead rise almost proportionally to their concentrations in soil at low concentrations. This pattern is the basis of using the BioConcentration Factor (BCF) concept in Risk Assessment, i.e. a constant plant tissue:soil concentration relationship. As soil concentrations rise, however, there are saturating processes and tissue concentration levels off, resulting in BCF values that are lower than that at low concentrations. The saturating processes are root uptake or translocation processes or potential feedback mechanisms under toxic conditions. For essential elements, such as the micronutrients boron (B), copper (Cu), manganese (Mn), molybdenum (Mo), and zinc (Zn), the pattern is distinctly different. Tissue concentrations are maintained within narrow limits at widely different external concentrations through homeostatic control mechanisms on ion uptake and translocation. At elevated supply, however, the control mechanisms break down and tissue concentrations readily increase with soil concentrations and this increase is generally associated with the onset of phytotoxic conditions. The homeostatic mechanisms generally act stronger on shoot than on root concentrations.

Fig. 8.2

(a) Generalized concentration dependent uptake for metals:homeostatic processes maintain internal concentrations of essential metals at adequate supply, whereas these processes break down at phytotoxic concentrations. Uptake of non-essential metals increases proportionally to soil concentrations up to a level where saturation occurs. (b) the concentration dependent uptake of non-essential metals is masked by differences in soil metal bioavailability in soils with different soil properties. Data in (a): zinc (micronutrient) uptake by corn shoot from Zn²⁺ salt spiked soil (data of Maclean 1974) and cadmium (non-essential) uptake in soybean shoot from Cd²⁺ spiked soil (Haghiri 1973); in (b): cadmium uptake in wheat seedlings (Smolders et al. 1999)

The soil-plant concentration relationships depicted in Fig. 8.2 are only observed in experiments where soils are enriched with soluble forms of the metal or metalloid. These patterns are termed the ‘metal or metalloid salt linear response relationships’ (Brown et al. 1998). In the environment, no such clear pattern is found, because the different bioavailabilities of the metal or metalloid in soil obscure the underlying patterns. Figure 8.2b illustrates that the concentration dependency can even be completely invisible when merging data from various soils. Variable bioavailability is related to differences in metal or metalloid speciation, interionic effects on ion uptake from pore water and indirect effects of soil properties on translocation within the plant.

The general paradigm in metal and metalloid uptake in soil is that roots absorb elements through pore water and the concentration of the dissolved elements affects the uptake rate. Dissolved metals and metalloids can be present in different forms (‘species’) and it is known that root uptake rate changes with the type of species present. Numerous experiments have shown that uptake of cationic metals from pore water generally decreases as the metal is complexed by chelating agents. Figure 8.3a illustrates this for the uptake of cadmium in a plant grown in solution culture (hydroponics). The addition of the cadmium chelator reduces the uptake of cadmium at constant cadmium concentrations, illustrating that free ion is the preferred species. This observation can be explained by the fact that ions are absorbed through their specific uptake systems while complexes are too large to pass the root cell membranes. This model is the basis of the free ion activity model (FIAM) that relates root uptake rate to the free metal ion activity in solution (Parker et al. 1995). For that reason, experimental methods have been developed to measure the free metal ion activity in soil. The FIAM has been contested, however, and it is now clear that the free metal ion is generally the preferred species, but that the complexed (and adsorbed) species contribute to availability as well, depending on timescales considered. For example, it is observed that metal uptake increases in soil when the concentration of a complexed metal increases at constant free metal activity (e.g. Smolders and McLaughlin 1996a, b). Such is also illustrated in Fig. 8.3b that presents the same data as in Fig. 8.3a but now plotted in a free metal ion activity basis. Here, it is shown that the complexed metals have an ‘apparent availability’ in nutrient solution. Experimental evidence has now clarified that this apparent availability is related to the ability of a metal complex to buffer the free metal ion uptake at the root surface (Degryse et al. 2006a). In soils and even in stirred nutrient solution, there is a zone adjacent to roots where the rapid intake of the metal or metalloid ion is not readily replenished by the flow of water to the roots. As a result, the free ion activity decreases near the root surface (Fig. 8.4). With increasing concentrations of soluble complexes and at constant free ion activity, such depletions are less pronounced and bioavailability of the metal is enhanced. Effectively, this means that the apparent availability of the metal complex is not by direct uptake of the complex, but a consequence of the lack of a fully mixed system in the root environment. This concept now also explains why root exudates enhance both solubility as well as bioavailability of the metals copper and zinc, despite the fact that the exudates do not change the free metal ion activity in soil (Degryse et al. 2008). There is also evidence that intact metal complexes can be absorbed (Collins et al. 2001, Tandy et al. 2006a). For example, synthetic chelators added to soil increase shoot lead concentrations (Tandy et al. 2006a). The suspected mechanism here is passive (non-selective) uptake of metal-complexes through breaks in the endodermis of roots, and transfer to above ground parts in the xylem (McLaughlin et al. 1997; Tandy et al. 2006b). In that scheme, uptake increases linearly with pore water concentration and has no saturable component.

Fig. 8.3

Speciation affects shoot cadmium concentration in Thlaspi caerulescens grown in nutrient solution. Left: the addition of the Cd²⁺ chelator nitrilotriacetic acid (NTA) decreases cadmium uptake when compared at equal soluble cadmium concentration. These data illustrate that the free ion is preferred compared to complexed cadmium. Right: same uptake data as left but plotted to the predicted free ion activity in solution, illustrating that the chelator addition increases the availability of the free metal ion, i.e. the Free Ion Activity uptake Model (FIAM) is not fully valid. Note that the curves merge at large free ion activity suggesting that the deviation of FIAM at lower activities are not due to partial uptake of Cd-NTA but due to the mechanism of buffering Cd²⁺ in the unstirred layer adjacent to roots (see text for argumentation; unpublished data from F. Degryse)

Fig. 8.4

Generalised scheme of dissolved metal or metalloid concentration gradients in the rhizosphere as a result of the ion uptake by roots, mass flow of water and diffusion to the roots. Depleted conditions occur for metals that are taken up with high absorption power such as cadmium. Rhizosphere pore water cannot be sampled equivocally, therefore it becomes more difficult to interpret metal uptake from bulk soil solution data when concentration gradients are more pronounced. The range of estimated gradients is based on an assumed averaged water use of 200 l/g dry matter and on solution culture data (e.g. Checkai et al. 1987; Degryse et al. 2006b; Weng et al. 2003) or field based plant-soil solution concentration ratios (Chen et al. 2009). The estimates for lead are most uncertain due to the lack of solution culture data at background lead exposure and since field data are confounded by atmospheric contribution of lead to aerial plant parts

Recent large-scale surveys of plant metal concentrations and speciation of metals in the associated soils demonstrated that pore water concentrations or free ion activities of the metals did not explain the crop concentrations (Chen et al. 2009; Hough et al. 2005). This observation does not invalidate the general concept that pore water concentrations and free ions are the directly available forms of these elements. Two processes may be invoked here that complicate the analysis in practice, i.e. rhizosphere processes that alter the pore water composition compared to the solution that can be sampled, and ion competition effects which affect the uptake rate of the free ion.

3.1.2 Rhizosphere Processes

The rhizosphere is the soil environment influenced by the roots and the rhizosphere conditions effectively control the supply of contaminants to plant roots. Unfortunately, the rhizosphere cannot be sampled unequivocally and this is a known drawback in soil bioavailability research. The physicochemical processes that alter metal speciation in the rhizosphere are the gradients in pH, soluble organic matter, depletions or accumulation of ions and redox gradients. We refer the interested reader to reviews dedicated to this topic (McLaughlin et al. 1998) and only explore the causes of concentration gradients, since this has practical consequences as shown below. Concentration gradients in the rhizosphere are the result of the balance between the ion uptake rate and resupply by the uptake of water. A simple calculation allows estimating if the contaminant is either accumulated or depleted in the rhizosphere: a plant typically transpires about 200 L of water per kg dry weight produced (Barber 1995). The product of the transpiration and the concentration of the contaminant in the pore water is termed the mass flow of elements to the plant root. If mass flow matches uptake perfectly, then the ratio of the contaminant concentration in the plant (mg/kg_dw) to that in pore water (mg/L) should be 200. Larger ratios mean that mass flow is not sufficient to match the rate of element uptake by the root, and concentrations of that element will have been depleted in the rhizosphere. The concentration depletion is followed by a diffusion flux towards the roots and this flux can be several fold larger than mass flow. Conversely, if concentration ratios are lower than 200, then mass flow (induced by transpiration) has exceeded root uptake and, consequently, the contaminant may have accumulated around roots. Field-based data for several plants and soils show that cadmium and lead concentration ratios exceed this threshold by 1–2 orders of magnitude, while arsenic concentrations ratios are typically 1 order of magnitude below that (Chen et al. 2009). This means that cadmium and lead are, on average, depleted in the rhizosphere, while arsenic generally accumulates around the roots. The differences are related to differences in the so-called root absorption power (uptake rate per unit concentration), for example a low value for arsenic uptake per unit time from the pore water due to the strong competition with phosphate ions. Similar data for other metals and metalloids allows ranking as given in Fig. 8.4. The practical consequence of these gradients is that the causal relationship between pore water concentrations and tissue concentrations, observed in stirred solution, may not be detectable in soils anymore because we fail to measure the pore water concentrations in the rhizosphere. The rhizosphere conditions can be estimated by modelling the diffusion and mass flow (Barber 1995). If the element is depleted in the rhizosphere, then it is replenished by either solid or liquid complexed forms. Practically, this means that a fraction of the complexed ions is also part of the directly available forms in soil, provided that the dissociation rate is sufficiently rapid. For that reason, bioavailability is always a complex function of both the activity in the pore water and a fraction of the labile bound forms. It is also logical that assessments of diffusive fluxes (i.e. fluxes under conditions of zero-sink) correlates well with the uptake, provided that the metal or metalloid is indeed depleted in the rhizosphere (Nolan et al. 2005).

3.1.3 Ion Competition Effects for Metal and Metalloid Uptake

Ion uptake is furthermore affected by interionic effects, i.e. the uptake rate of the ion decreases or increases as the concentration of an ion competing with the same uptake site is increased or decreased, respectively. One of the most striking examples of this is the effect of pH on uptake of cationic metals. For example, different solution culture studies have shown that metal uptake increases when H⁺ activity is lowered (pH increases). For example, Weng et al. (2003) showed that nickel uptake in oats increased by about a factor of 3 per unit pH increase between pH 4–6, at constant Ni²⁺ activity. Concentrations of cadmium in soybean shoots was shown to increase by a factor of 1.9 between pH 5–7 at constant Cd²⁺ activity (Smolders and Helmke unpublished), while shoot cadmium increased markedly larger, with factors of 4–13 in unbuffered nutrient solution between pH 5–7 for ryegrass, lettuce, cockfoot and watercress (Hatch et al. 1988). Numerous other interaction effects in plants have been identified for metal and metalloid uptake. Without attempting to be complete, we cite those that are relevant for contaminated site Risk Assessment, i.e. Ca²⁺:Cd²⁺ (Tyler and McBride 1982), Zn²⁺:Cd²⁺ (McKenna et al. 1993), H⁺:Cu²⁺ (Chen and Allen 2001), H₂PO₄ ^–:H₂AsO₄ ^– (Khattak et al. 1991) and SO₄ ^2–:SeO₄ ²⁻ (Hopper and Parker 1999).

The ion interaction effects are required to interpret effects of soil properties on metal availability to plants. For example, increasing pH invariably decreases the free metal ion activity in soil (see previous section), which is in contrast to the above mentioned effects on the uptake of free metal ions from pore water into plants. It is tempting to predict the net effect of pH on metal bioavailability by properly ‘adding up’ the interactions on both sides. Such calculations are the basis of the Biotic Ligand Model (BLM – see next section) for predicting toxicity of metals and this model is an extension of the FIAM by taking ion interactions into account. The elegant study of Weng et al. (2003) is an example of this. That study showed that the nickel concentrations in oats grown at different pH were reasonably well described by combining solution culture data with data on free metal ion activities in pore water. The difference between predictions and observation were largest (factor 2) at lower nickel supply in soil, which may be due to lack of accounting for other ion interaction effects or lack of modelling rhizosphere conditions. A similar attempt was made to predict liming effects on cadmium uptake (Smolders and Helmke, unpublished). However, that study showed that the ion interactions studied separately did not add up, but predictions overestimated effects of liming (pH increase) and only correctly predicted the trend. The general trend emerging from large surveys are suggesting that the H⁺ interaction for Cd²⁺ uptake is important. Field data on cadmium uptake by numerous plants (Table 8.4), for example, show that the net effect of increasing soil pH on reducing cadmium bioavailability is, on average, only a factor of 1.6 per unit pH increase. This impact is distinctly smaller than that effect on reducing solubility, i.e. a factor of 3.6 per unit on average (Degryse et al. 2009). Hough et al. (2005) similarly concluded from a large set of pot-trial data with ryegrass that H⁺ decreased the availability of pore water Cd²⁺ and Zn²⁺ to an extent that the net effect of soil pH on decreasing crop cadmium is smaller than a factor of 1.5 per unit pH increase.

Table 8.4 The net statistical effect of soil pH on cadmium concentrations in vegetables based on a survey of crops and associated soils in contaminated and non-contaminated soils in Belgium and the Netherlands with pH roughly between 5–7. The effects given are all statistically significant at P ≤ 0.05

While the BLM concept improves our understanding of metal toxicity over the FIAM, it is still incomplete and not ready for practical use. First of all, the BLM is currently applied to pore water data while it is conceptually most correct to do that for the solution in the rhizosphere. The ionic composition in the rhizosphere differs dramatically from that in the bulk soil, in terms of concentrations of metals, accumulation of dissolved organic matter, accumulation of Ca²⁺ ions and pH gradients (McLaughlin et al. 1998). Secondly, nutrient solution culture has also its limitations as a model system for separately studying ion interactions: the ion activity ratios in nutrient solution are not necessarily identical to those at the root surface where the free metal ion activity is a complex function of uptake rate, diffusion and metal-ligand dissociation rate (see above). Thirdly, interactions in soil are usually complex with several factors involved. The effects of pH are interpreted and modelled as H⁺:M²⁺ competition at the biological membrane, while the physiology behind this relationship is unclear. Modifying soil pH alters the availability of many elements that also can compete with the metal uptake process and it becomes tedious (and difficult) to take all such interactions into account. Finally, BLM modelling is not yet of practical use for Risk Assessment, both because it lacks precision (Antunes et al. 2006) and it requires numerous parameters (see below).

3.1.4 Translocation of Metals and Metalloids in the Plant

The long distance transport of solutes in plants takes place in the vascular system of xylem and phloem. Translocation is an important process in determining trace metal concentrations in plant tissues. For instance, Florijn and Van Beusichem (1993a) found that internal distribution rather than root uptake explained the genotypic differences in cadmium accumulation in shoots of maize inbreds. Figure 8.5 shows three typical patterns of metal partitioning between shoots and roots. In Fig. 8.5a, the root and shoot concentrations increase proportionally and are of similar magnitude (‘non-shoot excluder’; Florijn and Van Beusichem 1993a). In contrast, ‘excluders’ species retain most of the metals in the root, though the ability to retain the element in the root may strongly diminish beyond a threshold concentration (Fig. 8.5b). Figure 8.5c illustrates the partitioning of essential elements such as zinc. In the lower concentration range, the shoot concentrations are strongly regulated, and do not fall beneath a critical concentration. At low supply, the growth rate rather than the plant concentration decreases with decreasing metal supply.

Fig. 8.5

Typical relationships between metal concentrations in the exposure solution and concentrations in shoot and root tissue for (a) a non-shoot cadmium excluder and (b) a shoot cadmium excluder (graphs based on results from Florijn and Van Beusichem, 1993b), and (c) for uptake of the essential element Zn. Panel (c) also shows the yield response (based on results of Degryse et al. 2006b)

Since the translocation to the shoot depends strongly on the genotype and external conditions – such as the availability of the metal – the translocation factor (usually defined as the ratio of shoot to root concentration) is not constant for a given element and plant species (Fig. 8.6). As a result, BCF values for vegetables expressed between shoot and soil reflect the combined impact of bioavailability of the contaminant in soil and the effect of soil conditions on the internal translocation. Table 8.5 gives a non-exhaustive overview of metal distribution in plants. This table illustrates the dependence of translocation on the metal, plant species and variety, external concentration of the element and solution composition (pH, other elements). Below we briefly discuss these major trends.

Fig. 8.6

Typical relationships between concentration in shoots and in roots for a non-shoot cadmium excluder, a shoot cadmium excluder, and for the essential elements zinc and copper (full lines) The dotted lines give the relations for the given ratios of root to shoot concentration (R/S)

Table 8.5 Concentrations of metal(loid)s in plant tissues for given concentrations of the element and other relevant conditions of the exposure solution. The ratio of the root to shoot concentration (R/S) is calculated based on concentration in the total shoot or in the leaves (when analyzed separately)

Metal or metalloid type, concentration and speciation: The results of Guo and Marschner (1995) clearly illustrate that translocation is element-specific. Nickel showed strong translocation to shoots in bean, curly kale, and to lesser extent, rice, and was excluded from the shoot of maize, while the opposite was observed for cadmium. The available data indicate that translocation of mercury to shoots is generally limited. The results of Göthberg et al. (2004) suggest that translocation of lead is relatively large at background concentrations. However, the much larger concentration in the leaves than in the stem suggests that lead in the leaves was mainly derived from aerial deposition. Most likely, translocation of lead from roots to shoots is limited, since the endodermis acts as a barrier (Sobotik et al. 1998), unless under specific conditions, e.g. after addition of synthetic chelators (EDTA) at high concentrations (Luo et al. 2005; Tandy et al. 2006b). Translocation of redox-sensitive metals or metalloids may depend on the redox state of the element in the pore water. Hopper and Parker (1999) found that translocation of selenium was much larger when the plants were exposed to selenate (Se(VI), SeO₄ ²⁻) compared with selenite (Se(IV), HSeO₃ ⁻) (Table 8.5). Wang et al. (2002) showed that the uptake rate of arsenic by Pteris vittata was smaller, but the translocation efficiency larger, for arsenite (As(III), H₃AsO₃) than for arsenate (As(V), HAsO₄ ²⁻/H₂AsO₄ ⁻).

Plant species and variety: Table 8.5 illustrates the large variation of cadmium translocation to shoots for plants that were grown under similar conditions (e.g. Guo and Marschner 1995; Hatch et al. 1988). Gramineae (maize, barley, oat, ryegrass, cocksfoot, rice) are often shoot cadmium excluders, though this cannot be generalized. For instance, some maize inbred lines show large cadmium translocation to shoot. Also for the Leguminosae (pea, beans, etc.), the translocation of cadmium to the aerial parts is usually small, whereas the Solanaceae (potato, tomato), Asteraceae (lettuce), Brassicaceae (cabbage, watercress) and Amaranthaceas (spinach) show in general relatively large translocation to the shoot (Kuboi et al. 1986).

Pore water composition: As discussed above, the availability of an element will affect its translocation. For instance, the translocation of essential metals (copper, zinc) is usually larger at small external concentration than at large external concentration (Fig. 8.6). The concentrations of other elements may also affect the translocation. For instance, Florijn and Van Beusichem (1993b) found that uptake of cadmium increased with increasing pH, presumably due to diminishing competition with H⁺ ions for binding on the root surface. The translocation of cadmium to the shoot increased with increasing pH in the excluder plant species, whereas translocation was unaffected in the non-excluders (Table 8.5). Hatch et al. (1988) also found reduction of cadmium uptake with increasing pH, but found no consistent effect of pH on the translocation factor of cadmium in excluder (grasses) or non-excluder species. Increasing phosphate concentrations decreased the uptake of As(V), indicating competition for uptake between phosphate and arsenate, but increased the arsenic translocation to the shoot (Geng et al. 2005; Quaghebeur and Rengel 2003).

It is clear that many factors may affect the distribution of metals and metalloids within plants. Plants have a complex network of homeostatic mechanisms, including transport, chelation and sequestration, in order to maintain the concentrations of essential elements within the physiological limits and to minimize the detrimental effects of non-essential elements. The exact mechanisms are only starting to be unravelled. Most molecular insights obtained so far are on the cellular level, and very little is known about mechanisms controlling metal distribution on the level of the plant (Clemens 2001). The transport of metal(loid)s from roots to shoots is probably largely through the xylem. Guo and Marschner (1995) showed that the cadmium and nickel concentrations in the shoot dry matter were positively correlated with the concentrations in the xylem sap. The soluble fraction of cadmium in the roots was much larger for maize than for the other plant species tested, which was in agreement with its higher mobility in the plant (Table 8.5). Similarly, the higher soluble fraction of nickel in the roots extracts of bean compared with maize was in accordance with the higher nickel mobility in bean plants.

Phytochelatins (PCs) are metal-complexing peptides that play an important role in metal tolerance. The possible roles of PCs in heavy metal detoxification and homeostasis have been reviewed by Cobbett and Goldsbrough (2002). Phytochelatin synthesis is induced upon exposure to a variety of metals and metalloids (Grill et al. 1989). Overexpression of phytochelatin synthase was found to increase tolerance to cadmium, mercury and arsenate (Vatamaniuk et al. 1999). In some plants, sulphides also seem to play a role in the detoxification of cadmium by PCs. It was shown that phytochelatin-cadmium complexes of tomato grown at high cadmium concentration (100 μM) contained cadmium-S-peptide aggregates of ca. 2 nm diameter that consisted of a CdS crystallite core coated with PCs (Reese et al. 1992). Phytochelatins may be sequestered in the vacuole (Salt and Rauser 1995; Vogeli-Lange and Wagner 1990), but they may also be transported in the xylem (Gong et al. 2003). Translocation of cadmium in the xylem has been found to be independent of PC production (Florijn et al. 1993; Salt et al. 1995). Limited translocation of cadmium from shoot to root seems therefore not to be related to the presence of PCs, but is most likely due to sequestration of metals in the vacuoles of root cells, either as complexes with PC, as free ion or in another form. The transport from leaves to other plant tissues (e.g. grains, tubers) can occur in phloem only. For instance, cadmium in potato tubers and peanut kernels is not directly taken up from the soil, but is first transported in the xylem to the shoot, and then back down through the phloem (Popelka et al. 1996; Reid et al. 2003). Page and Feller (2005) showed that nickel and zinc were redistributed from older to younger leaves in wheat plants, indicating high phloem mobility, whereas Mn remained in the old leaves. Also split-root experiments and foliar application of ⁶⁵Zn demonstrated significant phloem transport of zinc from leaves to other plant parts (Haslett et al. 2001).

Differences in translocation of metals are an important factor determining the concentrations in plant tissues. The selection of plant species or cultivars with relatively small or large (in case of essential elements) concentrations in the harvested products can be used to manage metal concentrations in food crops. For instance, McLaughlin et al. (1994a) showed that potato cultivars grown commercially in Australia exhibited significant, nearly two-fold, differences in tuber cadmium concentration. Plant-breeding can also be an important tool to reduce the concentrations of potentially harmful elements, such as cadmium. For instance, low-cadmium durum wheat cultivars and sunflower hybrids have been developed for this purpose (Grant et al. 2008).

3.2 Foliar Uptake of Metals

In air, non-gaseous metals and metalloid are associated with small particulates (<10 μm) that can be deposited on plants. These aerosols may adsorb to plants or eventually also be absorbed and the metals and metalloids translocated. The mechanisms of the absorption processes are still unclear. Washing or even peeling vegetables in preparation for cooking does not fully remove the airborne metals (Dalenberg and Van Driel 1992). As a result, airborne metals can be a significant source of metals in the food chain. As shown below, this fraction can be even more important than that derived from soil.

Metals may also enter plants and the food chain through gaseous exposure through soil emissions (e.g. mercury). Little is known of the potential for uptake of metals by plants through gaseous routes of exposure (as opposed to aerosol or particulate exposure). Mercury (Hg) can be present in air as gaseous forms (Hg°) or as particulates (Hg-P). Plants can absorb mercury from soil via root uptake, from air as of Hg° via stomatal uptake and from air via adherence of Hg-P similar to the mechanisms described above. Plants can also be a source of mercury, releasing mercury when grown in low air mercury and high soil mercury. As a result, there is a so-called compensation point, the air concentration where no net flux of mercury vapour occurs and this point increases as soil mercury concentration increases (e.g. Ericksen and Gustin 2004).

There are very few data on uptake of airborne metals or metalloids and the methodologies to assess these are limited. Harrison and Chirgawi (1989) estimated the atmospheric contribution from the differences in metal (cadmium, chromium, nickel, lead and zinc) concentrations between plants grown in cabinets with either filtered or unfiltered air. An alternative method uses soils enriched with stable or radioactive isotopes and the isotope dilution in the crop as the basis for estimating airborne metal contribution (Mosbaek et al. 1989; Tjell et al. 1979). Dalenberg and Van Driel (1990, 1992) combined both methods to reduce the uncertainty related to the isotope ratio of bioavailable metals in soil. Finally, surveys of soil, plant and atmospheric concentrations combined with statistical tools allows a statistical, indirect, estimate of the fractions metal derived from atmosphere and soil (Voutsa et al. 1996). A compilation of data for cadmium and lead shows that most (generally >80%) of the lead in above-ground plant tissues of crops derives from the atmosphere (even in washed plant tissue), whereas variable results are obtained for cadmium (Table 8.6). The larger percentages for lead than for cadmium are related to the lower availability of soil lead relative to cadmium (largest sorption for lead). The airborne percentages of nickel and chromium are intermediate between cadmium and lead whereas the percentages for zinc are similar to those of cadmium (Harrison and Chirgawi 1989). The statistical method applied in an industrialized area in Greece confirmed that concentrations of cadmium, chromium and lead in washed vegetables (cabbage, lettuce and endive) were mainly related to air contamination rather than soil contamination, and the same was found for arsenic (Voutsa et al. 1996). The consequence of the large atmospheric contribution is that crop metal concentrations can become unrelated to soil metal contamination, even when collating data with contrasting soil metal contaminations (Fig. 8.7). The relative contributions from soil and air obviously depend on the concentrations in air and soil, the bioavailability of soil metal or metalloid and the food preparation method (washing, boiling, etc.), i.e. generalizations cannot be made. A further consequence of foliar uptake is that soil remediation may not be completely successful for reducing metal concentrations in garden vegetables if atmospheric uptake is significant. Douay et al. (2008) examined the concentrations of metals in seven types of vegetables from remediated (complete soil replacement) and non-remediated soils in the vicinity of a metal smelter. Complete soil replacement reduced concentrations of cadmium in all vegetables (by 50–90%), but concentrations of lead were only reduced slightly, or not at all (Douay et al. 2008).

Table 8.6 The fraction of airborne cadmium or lead in different crops calculated in experimental studies that used ‘ambient air’ (period 1975–1990) and ‘background soils’ Experiments were based on isotope labelling studies or comparison between crops grown in filtered air versus that in ambient air. All studies were performed < 1990, current air concentrations are lower and result in lower atmospheric contribution (see theoretical analysis in Table 8.7)

Fig. 8.7

Data of Dutch and Belgian surveys of metal concentration in field grown maize (whole shoots). Concentrations in maize generally correlate better with soil metal concentrations for cadmium (left) than for lead (right); lead concentrations in arial parts are often more reflecting air concentrations rather than soil concentrations, i.e. atmospheric contribution is larger than soil contribution in the case of cadmium. Data collated by Jansson et al. (2007) and Römkens and Rietra (unpublished)

The transfer of metals and metalloids from air to plant can be aggregated in the so-called air accumulation factors (AAFs, with units m³/g), calculated as the ratio of concentrations in the plant to that in air (μg/g divided by μg/m³). Other expressions relate the plant concentrations to metal deposition via bulk precipitation (g/ha/year). Hovmand et al. (1983) used the isotope dilution method and bulk precipitation data to show that up to 70% of deposited cadmium could be incorporated in field-grown carrots, but the authors acknowledged that airborne submicron-sized particles containing cadmium, not sampled via the bulk precipitation, could also be filtered out by plant surfaces. Such interpretation suggests that AAF values may be a more suitable basis for modeling the transfer of metals and metalloids from air to plant. The AAF values observed by Harrison and Chirgawi (1989) ranged from 2 to 40 m³/g for different plants and metals, typically about 10 m³/g. Field-based estimates, using isotopically labeled soils, have yielded values that sometimes exceed 100 m³/g for plant leaves (values for cadmium based on Hovmand et al. 1983). The AAF values are obviously not metal-specific values, but depend on plant type, aerosol properties, climatic factors, et cetera.

The above-mentioned data are at least 15 years old and there are little recent data measured under current atmospheric air quality, which typically demonstrates lower air metal concentrations in recent years due to environmental regulations (e.g. the ban on leaded gasoline). A simple sensitivity analysis allows modeling of the current atmospheric contribution, using the experimentally determined AAF and BCF values, the latter observed in ‘clean air’. As shown in Table 8.7, we predict that still 50% of lead present in lettuce leaves in rural areas is derived from the atmosphere. A quantification of the pathway by which metals or metalloids enter the plant, i.e. by air or soil, has obvious consequences for Risk Assessment. For example, it is generally known that managing food chain contamination by metals could be achieved by managing air contamination for lead and, conversely, soil contamination for cadmium. However, the analysis in Table 8.7 suggest that human exposure to cadmium around smelters in the past, with cadmium concentrations in air often exceeding 100 ng/m³, could have predominantly originated from atmospheric contributions to plants.

Table 8.7 A sensitivity analysis of predicted atmospheric contribution to plant metal concentrations, using current air quality conditions in rural and contaminated areas. Assumed air accumulation factors (AAF: plant:air concentration ratio, m³/g_dw) and soil-plant bioconcentration factors (BCF, dry weight based) for lettuce are based on experimental data of Harrison and Chirgawi (1989). The 4 scenarios of soil and air concentrations are based on expert judgement, reflecting current air concentrations in rural areas in EU and background soil concentration for lowest values, and air concentrations near metal smelters and contaminated soils, for high concentrations

4 Integrating Factors Affecting Metal/Metalloid Accumulation by Vegetables

There are a number of metal/metaloid-specific, soil and plant factors that can affect the accumulation of metals/metalloids by vegetables in urban gardens (Chaney 1973). Much of the information in this area is derived from studies of the use of sewage biosolids on land, with only a few studies examining contaminant uptake by vegetables in soils contaminated by industrial, mining or urban sources of contaminants. The most important factors controlling metal/metalloid accumulation by vegetables are outlined in the sections below.

4.1 Type of Metal/Metalloid

From the preceding sections, it is evident that the behaviour of metals/metalloids in urban/industrial soils and their uptake by plants are highly metal/metaloid specific. Cationic elements are generally more firmly held by soils than anions, and plants accumulate cations more readily than anions due to the negative electrical potential across the root membrane. Combining these properties of metals/metalloids, Chaney developed the ‘soil-plant barrier concept’ (Chaney 1980) where metals/metalloids were classified into groups for food-chain Risk Assessment. The classification of metals/metalloids depends on their partitioning behaviour in soil, their propensity for root uptake, and their propensity to accumulate in edible portions of plants in relation to critical phytotoxicity concentrations versus critical food concentrations for humans (Table 8.8). Using this concept, metals/metalloids which are highly insoluble, or are retained very strongly by plant roots, are in Group 1. Examples of such metals are Cr³⁺ and also titanium, zirconium, tin, yttrium and silver. Group 2 comprises metals/metalloids that can be absorbed by roots, but are not readily translocated to edible plant parts, like arsenic, mercury and lead. Since 1980, it has become apparent that arsenic may be a food chain hazard in paddy rice production systems (due to the redox-induced mobilization of arsenic as described in Section 8.2.2), but in urban residential scenarios, arsenic can be regarded as a low hazard in terms of plant uptake. Group 3 comprises the elements boron, copper, nickel and zinc, which are easily taken up by plants, but are phytotoxic to plants before significant exposure to humans occurs. Group 4 are metals/metalloids which most likely pose a food-chain risk, like cadmium, cobalt, molybdenum and selenium, as plants can readily absorb and translocate these to edible portions, and they are not highly phytotoxic.

Table 8.8 Maximum tolerable levels of metals/metalloids in plants in relation to maximum levels tolerated by animals in forages (modified from Chaney 1983)

It should be noted that this approach classifies elements on the basis of transfer of contaminants via the soil-root-shoot/edible portion pathway. In urban contaminated soils, aerial deposition of contaminants (Section 8.4.2) provides a by-pass of the ‘soil-plant barrier’, so that elements such as arsenic, lead and other metals/metalloids may pose a risk to human or animal health where there is a continuing source of metal/metalloid deposition to soil.

4.2 Vegetable Species

Vegetables vary widely in their ability to accumulate metals, either from soil or from atmospheric deposition. Numerous studies have examined this in both glasshouse trials, where soil-root uptake dominates, to field studies and surveys where both soil and atmospheric uptake pathways are important. Uptake by different vegetables varies according to metal/metalloid (and is not always consistent across sites), and in Table 8.9 vegetables are ranked according to their potential to accumulate contaminants. Cobalt and molybdenum are not listed, despite being in the group of elements likely to pose risks through food chain accumulation, due to the lack of data for these elements. It can be seen that in general leafy vegetables tend to be ‘accumulators’, and fruiting vegetables tend to be ‘excluders’ (Preer et al. 1980). This is likely because leaves accumulate ions through transpiration and aerial deposition, and non-essential elements may therefore accumulate in these tissues if the element is transported across the root membrane or deposited on the leaf surface. Fruits are principally formed by movement of nutrients in phloem (the nutrient transport system) into the fruit across a membrane from the xylem (the water transport system) within the plant, and transfer of many elements from xylem to phloem may be controlled, or affected by many counter ions, similar to root uptake processes (Section 8.4.1). For aerial deposition, which is mainly important for lead, species morphology (e.g. hairiness of leaves, surface roughness of aerial tissues) may be important in the trapping of atmospheric particulates and aerosols (Tiller et al. 1976, 1997).

Table 8.9 Relative differences in uptake of metals/metalloids by vegetable species. Note these may include both soil and atmospheric pathways. Studies focussing on contaminated urban soils were targeted, and where available, data from field-grown crops used

4.3 Vegetable Cultivar

Most of the work on effects of vegetable cultivar on accumulation of metals or metalloids has focused on cadmium, since this is the element that is most likely to pose a food-chain hazard. There may be significant differences in cadmium uptake by different cultivars of the same vegetable species grown on a contaminated site, as was demonstrated for lettuce (Florijn et al. 1991; Thomas and Harrison 1989; Wang et al. 2007; Xue and Harrison 1991; Yuran and Harrison 1986), potato (Harris et al. 1981; McLaughlin et al. 1994b), curcurbits (Mattina et al. 2006), beans (Guo and Marschner 1996), carrots and peas (Alexander et al. 2006). Cultivar effects vary from insignificant up to a 3- to 4-fold difference in cadmium concentrations. Cultivar effects are therefore less pronounced than the influence of the type of vegetables, but selection of non-accumulating culivars may provide an important Risk Management action for minimizing exposure to humans through consumption of vegetables.

4.4 Soil Physical/Chemical Properties

In considering the effect of soil chemical and physical properties on metal/metalloid availability to plants, the type of contaminant, valence and charge, and other properties that determine fate in soil (Section 8.3) need to be considered. Anionic metalloids, for example, will respond differently in terms of crop accumulation compared to cationic metals. The principal factors governing vegetable uptake are discussed below.

4.4.1 pH

Soil pH is perhaps the master soil variable governing the availability of metals and metalloids to plants (Page and Chang 1985). The major effect that pH has on both partitioning of metals/metalloids in soil (Section 8.3), and on response of plant roots to uptake of ions (Section 8.4.1), means that this variable accounts for most of the variation in metal-metalloid concentrations in vegetable tissues. Acidic soils will tend to increase uptake and accumulation of the cationic metals cadmium, copper, mercury, nickel, lead and zinc, while uptake and accumulation of the anionic elements arsenic, molybdenum and selenium will be reduced. As noted in Section 8.4.1, the effect of pH on accumulation of cationic metals is not predicted well just by consideration of changes in partitioning, as H⁺ competition for root uptake mitigates the effects of pH on metal partitioning (Hough et al. 2005). Hence, liming of acidic soils will generally reduce uptake by vegetables of cationic metals in urban soils (Preer et al. 1980, 1995).

4.4.2 Soil Texture and Depth of Contamination

Soil texture (content of sand, silt and clay) will affect metal availability to crops as finer textured soils (clays) have a greater cation exchange capacity (CEC) and hence a greater ability to retain cationic metals (higher K _d ) compared to sandy soils. Given the same total metal concentration in soil, clay-rich soils will produce crops with lower (cationic) metal concentrations. Depth of soil is also important as shallow contamination is likely to have a lesser effect on metal concentrations in vegetables compared to deeper contamination (Tiller et al. 1997), as plants may extend their root systems into less-contaminated subsoils.

4.4.3 Soil Organic Matter

Soil organic matter (OM) plays a similar role to clays in affecting metal concentrations in vegetables, as OM is a major contributor to the pH-dependent negative charge in soils which gives rise to soils’ ability to retain cationic metals (represented by the CEC). Addition of OM to soil in the form of compost can therefore markedly reduce cationic metal uptake by plants (Farfel et al. 2005; Jones et al. 1987; Narwal and Singh 1998; Traulsen and Schonhard 1987; Verloo and Willaert 1988). Addition of OM has also been found to reduce arsenic accumulation by vegetables from contaminated soils (Cao and Ma 2004).

Many vegetable gardens use imported OM (e.g. composts, mulches or other soil amendments) and these may be useful in minimising uptake of cationic metals by vegetable crops. Also, high OM contents in many garden soils compared to agricultural soils means that soil to plant transfer factors developed from agricultural surveys (which may not identify OM as an important factor controlling plant uptake) may overestimate metal uptake in garden scenarios.

However, it should also be noted that high concentrations of dissolved OM in a soil may lead to increased potential for leaching of cationic metals, due to complexation by dissolved OM reducing K _d values (Sauvé et al. 2000). This is particularly important for copper.

4.4.4 Salinity

Salinity may play a role in enhancing uptake of cadmium from soil by vegetables. Chloro-complexation of Cd²⁺ ions reduces the charge of the cadmium ion (CdCl⁺ or CdCl₂ ⁰) and hence increases its mobility through soil (Hahne and Kroontje 1973). Chloro-complexation also increases the diffusive flux of cadmium to root surfaces (Degryse et al. 2006a; Smolders and McLaughlin 1996a, 1996b). As a result of this increased mobility in soil and increased diffusive flux of cadmium to root uptake sites, cadmium concentrations in vegetables are increased when soil salinity increases, and this has been demonstrated in field-grown crops (Helal et al. 1998; McLaughlin et al. 1994a, 1997).

Theoretically, soil salinity should also have a large effect on uptake of mercury, but this has not been demonstrated in any uptake studies by vegetables in soil. Effects of salinity on the uptake of other metals should be minimal, and for the anionic metals/metalloids, the chloride ion may actually inhibit uptake of these ions, although this has not been demonstrated in practice.

4.4.5 Redox Potential

Redox potential, through the effects on reductive dissolution of Fe and Mn oxides (Section 8.2.2), can have a marked influence on the uptake of several metals by plants. For example, uptake of arsenic by rice and uptake of cobalt by pasture plants (Adams and Honeysett 1964) has been found to be significantly enhanced under low redox conditions, again due to reductive dissolution of Fe/Mn oxides, and in the case of cobalt, reductive dissolution of insoluble Co³⁺ to soluble Co²⁺.

There is little information in the literature on effects of low redox on metal uptake by vegetables. It should be noted that most plants are not tolerant of waterlogged soil conditions which create low redox potentials, and generally die due to lack of oxygen. Redox may therefore not be a major risk factor for affecting metal uptake by vegetables in urban gardens.

4.4.6 Nutrient Status

It is well known that soil nutrient status can influence metal uptake by plants, and the effects can operate in several ways. Low levels of addition of nitrogen and phosphorus in soil have been shown to enhance cadmium uptake by crops (Grant and Sheppard 2008; Maier et al. 2002a, b; Williams and David 1973). Higher rates of addition of N fertiliser are likely to acidify soils and lead to enhanced uptake of cationic metals (Eriksson 1990; Grant et al. 1996; Lorenz et al. 1994), while high rates of P addition may reduce uptake of cationic metals through precipitation reactions (McGowen et al. 2001). High P addition, however, may promote uptake of anionic metals/metalloids, due to displacement of these ions from sorbing surfaces in soil (Cao and Ma 2004; Creger and Peryea 1994). High additions of potassium (K) fertiliser as KCl can enhance cadmium uptake by crops due to the chloro-complexation issue described in Section 8.4.4.4 (Sparrow et al. 1994).

In many urban garden soils, fertilisers are often added in excess of plant requirements, so this factor may play an important role in metal accumulation by vegetables. In particular, it is important that N (from either manufactured fertilisers or organic sources of N) is not used in excess, as this rapidly acidifies the soil and will enhance cationic metal uptake by vegetables.

5 Models to Predict Contaminant Uptake by, or Toxicity to, Vegetables

Different models with increasing complexity can be used to predict metal concentrations in vegetables (Table 8.10). Among the easiest to apply are those that assume a constant plant metal concentration or the ones that relate the plant metal concentration to certain soil parameters: an equilibrium approach. In more process-oriented approaches, the concentration is not necessarily constant throughout the growing period. Here also the uptake of metals is a function of certain soil parameters, but also specific plant parameters that for instance regulate water uptake, such as root growth and length.

Table 8.10 Overview of model approaches to predict heavy metal contents in vegetables

An important aspect related to Risk Assessment is that only the metal concentrations in the edible products of the plant (root, stem or fruit) are considered relevant, although metals are present in all plant parts. Identification of those processes responsible for uptake of metals have been described in papers such as by Clemens (2001) and Hall (2002) and will not be dealt with here. Also models that are used to predict toxicity for plants (e.g., Thakali et al. 2006) will not be dealt with here. Although models to predict toxicity are similar to metal uptake models, they do not describe the metal concentrations or translocation to the food product.

The metal uptake models differ strongly and vary in complexity from very simple models to detailed growth models with a high requirement for input data. Also the scale from which they have been used differs from pot experiments to models used on a national scale (Table 8.10).

5.1 Model Characteristics

5.1.1 Constant Heavy Metal Content for Each Plant Species

This assumption is used in most site-specific studies when the metal concentration is assessed by actual site measurements (e.g. Sipter et al. 2008). In principle it is assumed that the metal concentrations do not differ within a year, between years and are not influenced by the applications of soil amendments e.g. lime or changes in vegetable type over time.

5.1.2 Soil-Plant Transfer Models

A simple and commonly-used approach is the bioconcentration factor (BCF) or bioaccumulation factor (BAF). The terms BCF and BAF originate from studies of contaminant concentrations in water (mg/L) and biota (mg/kg_dw). Originally BCF or BAF are expressed as the ratio between the concentration in the biota and water and are therefore given in L/kg. However several soil Risk Assessment models such as CLEA (Environmental Agency 2008) and CSOIL (Brand et al. 2007) use dimensionless BCF or BAFs, most commonly expressed as the ratio between the metal/metalloid concentration in the vegetable (mg/kg) and that in soil (mg/kg). Compilations of BCF values are given by Sauerbeck and Styperek (1985) and Bechtel Jacobs (1998). The metal concentration is either expressed on a fresh or dry weight basis. The BCF approach is rather easy to apply, but it has been shown that BCF values are not constant, but depend on the level of soil contamination (see Section 8.4.1.1). Often BCF values decrease with increasing levels of soil contamination (Wang et al. 2004).

Metal concentrations in plant tissues are often linked to those in various soil extracts by regression analysis. Linear or log transformed data for metal concentrations in soil, soil solution or soil extracts are linked to measured plant concentrations. For metals like cadmium in the soil (M _soil) and the plant (M _plant), Freundlich type functions are often used (Efroymson et al. 2001; Krauss et al. 2002).

$$M_{{\textrm{plant}}} = 10^{\textrm{a}} M_{{\textrm{soil}}} ^{\textrm{b}}\ {\textrm{or}}\ {\textrm{log}}\,[M_{{\textrm{plant}}} ] = {\textrm{a}} + {\textrm{b}}\log [M_{{\textrm{soil}}} ]$$

((8.2))

where M _plant is metal concentration in the plant, M _soil is metal concentration in the soil, and a and b are constants.

The Freundlich-type equation can be extended, using soil parameters such as soil pH and soil organic matter (S) (Adams et al. 2004; Efroymson et al. 2001; Hough et al. 2003; Plette et al. 1999).

$$\log \,[M_{{\textrm{plant}}} ] = {\textrm{a}} + {\textrm{b}}\log \,[M_{{\textrm{soil}}} ] + {\textrm{c}}\,[pH] + {\textrm{d}}\log \,[S]$$

((8.3))

Efroymson et al. (2001) have shown that incorporation of soil pH improved the model for cadmium, mercury, selenium and zinc. Incorporation of other factors such as CEC, total metal concentrations in soil and extractable concentrations in soil using Freundlich-type equations has been reported for various vegetables and metals (Wang et al. 2004). In most cases the use of soil parameters, like pH, organic matter, CEC or texture, improves the model performance compared to those based on extractable or total concentrations only. For example in Fig. 8.8 and in Table 8.11 it can be seen that the cadmium concentrations in leek in the sandy soils in the Belgian-Dutch Kempen region depend on the cadmium concentration of the soil and soil pH. Using Eq. 8.3, with soil cadmium, soil pH and soil organic matter, most of the variation can be explained. Such a model can be used, for example, to predict the threshold soil cadmium concentrations as a function of pH above which the food cadmium standards would be exceeded.

Fig. 8.8

Concentrations of cadmium in leek (Allium ampeloprasum L.) as a function of soil cadmium concentration (0.43 M HNO₃ extraction) (a), as a function of soil pH (b), predicted versus measurements (c), and predicted soil concentrations (model No. 4 in Table 8.11) as a function of pH where the limit value for cadmium in leek is exceeded (d). Results are from several studies near the Dutch zinc smelter in Budel: experimental fields, vegetable gardens, liming experiments and farms. The dotted line is the European limit value for cadmium in leek

Table 8.11 Soil to plant transfer models for cadmium in leek on the basis of Eq. (8.3) for the data in Fig. 8.8. Given for each model are the coefficients of determination (R ²), and standard errors (se). SOM = soil organic matter content (%)

An advantage of Freudlich-type soil-plant transfer relations is the simplicity and the applicability. Most equations use variables that are available from soil investigations, such as total metal content, pH, organic matter and CEC. However these soil-plant equations should not be used for soils where concentrations of metals are outside the range from which the regressions were derived. Römkens et al. (2009) studied the quality of the soil-plant transfer equations for rice and concluded that only models for cadmium and zinc gave good predictions. While predictions for cobalt and nickel where not as good, and prediction for copper and lead were not possible.

5.1.3 Fiam

The free ion activity model (FIAM) (Morel 1983) is based on the assumption that metal uptake or toxicity is related directly to the free ion activity in the pore water. While early reports suggested this was a promising model (Sauvé et al. 1996), later studies have questioned whether free ion activities alone can improve predictions of plant metal uptake, given effects of diffusional limitations adjacent to and near the plant root (Degryse et al. 2006a, b; Hudson 2005; McLaughlin 2001a) and ion competition and other factors that can markedly affect plant metal/metalloid acquisition (Section 8.4.1). Free metal activities could not explain nickel uptake by oats in glasshouse experiments (Weng et al. 2003, 2004) nor copper uptake and toxicity to barley and tomato, also in glasshouse experiments (Zhao et al. 2004). Nolan et al. (2005) examined a wide range of techniques to predict cationic metal (cadmium, copper, lead, and zinc) uptake by wheat (again in glasshouse trials), including free metal ion activities, and found consideration of elemental free ion activities did not improve the predictions. McLaughlin et al. (1997) examined the relationship between free Cd²⁺ ion activities and concentrations of cadmium in potato tubers grown in field soils, and again found that consideration of free ion activities did not improve prediction of metal accumulation in tubers. This is not to say that knowledge of elemental speciation is not important in explaining metal accumulation by plants, but the analytical effort of separating and determining free metal ion activities in pore water may not be justified in terms of improving predictions of metal uptake by plants. This is more fully explained in Nolan et al. (2003).

5.1.4 Biotic Ligand Model

Originally the BLM described metal uptake and toxicity as a function of binding to certain biotic ligands present on surfaces of aquatic organisms (Di Toro et al. 2001). The applicability of the BLM has been studied for a range of aquatic organisms and heavy metals (lead, copper, zinc, nickel) and has shown potential to be used to define water quality standards (Comber et al. 2008). Toxicity has been described for soils using a BLM approach for copper and nickel toxicity to barley (Thakali et al. 2006), or nickel toxicity to oat (Weng et al. 2003; 2004). However zinc uptake by algae could not be predicted by surface-bound zinc or solution zinc chemistry, probably due to internal regulation of zinc (Hassler and Wilkinson 2003). Also the competition between different cations in a BLM could not improve the prediction of cadmium and zinc uptake by Lolium perenne (Hough et al. 2005), or describe the competition between copper and lead and zinc on algae (Hassler et al. 2004).

A problem of the FIAM and the BLM is that if these models are used to predict effects of various cations on metal uptake by roots (Cheng and Allen 2001), one still needs to describe the translocation of the metal from root to shoot or fruit with a translocation coefficient, as was done by Cheng and Allen (2001) for copper. However, translocation of metals from root to shoot is not constant, especially not for essential metals such as copper and zinc (Kalis et al. 2006).

5.1.5 Physiological Models

Uptake of any contaminants from soil is related to the water uptake and, for some contaminants also the concentration or free ion activity in the pore water (Ingwersen and Streck 2006; Peijnenburg et al. 2000). In models for neutral organic xenobiotics it is often assumed that uptake is a passive process related to water transpiration (see Chapter 9 by Trapp and Legind, this book). Ultimately, the metal content of plant tissues is a function of plant growth, water transpiration, metal concentration or speciation in the pore water, selectivity/ion competition effects at the root surface and diffusional limitations to uptake. An advantage of such models is that they can explain differences in metal uptake between years based on differences in weather. In practice this still seems to be rather complex, because other parameters (e.g. root depth) also vary between years (Ingwersen and Streck 2006). The physiological model can be extended to include competitive effects of other cations, such as the effect of H⁺ and Zn²⁺ on Cd²⁺ uptake by plants, similar to BLMs (Ingwersen and Streck 2006).

5.1.6 Barber-Cushman Mechanistic Model

A commonly-used model to predict the uptake of nutrients by plants is the Barber-Cushman model (Barber 1995). This model has also been used to describe the uptake of heavy metals, including uptake of zinc by rice (Adhikari and Rattan 2000), cadmium by maize and Thlaspi (Sterckeman et al. 2004) and zinc by Thlaspi (Whiting et al. 2003). The Barber-Cushman model includes plant and soil parameters, such as root geometry, growth and kinetic parameters for the uptake of ions, parameters which can only be determined in detailed, small scale experiments. In general, application of this kind of model helps to identify which factors determine metal uptake by plants from a research perspective, but is less useful for generic predictive modelling.

5.2 Application of Models

In most Risk Assessment studies, soil-plant transfer models have been used. Also the physiological model of Ingwersen and Streck (2006) has been applied in a small region. Human exposure due to vegetable consumption has been described in Elert et al. (Chapter 11 of this book). Soil-plant transfer models have been used in:

• large-scale Risk Assessment on the basis of national soil data (Brus and Jansen 2004; Brus et al. 2002; De Vries et al. 2008);

• derivation of critical soil concentrations (Brus et al. 2005; De Vries et al. 2007) on the basis of limit values for crops and fodder;

• local and regional Risk Assessment on the basis of standard soil analysis in CLEA or CSOIL (Römkens et al. 2005); and

• local and regional Risk Assessment on the basis of soil analysis and soil maps (Hough et al. 2004).

Soil-to-plant transfer models for metals can be applied in Risk Assessment studies on a regional and local scale, either as such or in combination with additional sampling of soil and vegetables. However, parameters of current models very much depend on soil and plant data on which they are based which renders most models suitable for local applications only. The applicability of generic soil-to-plant transfer models to predict heavy metal contents in local circumstances should be checked. Ideally the models are verified with measurements on soil and plants. A practical use of the models in Risk Assessment of vegetable gardens is, for example, to calibrate the models for the crops which have the highest contribution to metal intake by humans (e.g. potato, leafy vegetables) and to use generic models for all other crops. In the next chapters it will be shown that the use of the model also depends on the questions asked.