Printer Friendly

Modelling the effects of soil properties on the concentration of Cd extracted by 10mM Ca[Cl.sub.2] from soils of the Sydney Basin.


Exposure of the human population to cadmium (Cd) is a health concern (Staessen et al. 1999; Eklund and Oskarsson

1999; Hartwig et al. 2002) and plant products contribute almost half the Cd ingested by Australians (Anon. 1996). Some of the leafy vegetables grown on the peri-urban fringe of the Sydney Basin contain appreciable concentrations of Cd and the concentration of Cd in the soils on vegetable farms is typically ~20 times that in their unfarmed counterparts (Jinadasa et al. 1997). The accumulation of Cd in intensively used agricultural soils is commonplace (Williams and David 1976; Moorcroft et al. 1982; Holmgren et al. 1993; Zanders et al. 1999) and Cd is also a contaminant of industrial sites; consequently there is a need for general models to predict Cd transport and bioavailability in soils (van der Zee and van Riemsdijk 1987; Boekhold and van der Zee 1990; Palm 1994; del Castilho and Chardon 1995; Peijnenburg et al. 2000; Sauve et al. 2000b).

Two lines of evidence link the concentrations of certain Cd fractions in soils and Cd bioavailability. Firstly, sequential chemical extraction defines some 5 Cd fractions in acidic soils, typically: (i) soluble plus exchangeable; (ii) specifically sorbed; (iii) strongly sorbed to either organic matter (OM) or easily reducible oxides of Mn and Fe; (iv) very strongly sorbed to OM or other oxidisable species; and (v) mineral (Calvet et al. 1990; Jeng and Singh 1993). Extraction procedures may discriminate less than perfectly between the nominated Cd pools (Jeng and Singh 1993; Ma and Uren 1998; Benitez and Dubois 1999; Ho and Evans 2000); however, there is a sharp contrast in bioavailability between the extremes. For example, fractions I and 2 are likely sources of the Cd extracted by dilute solutions of Ca[Cl.sub.2], and the concentration of Cd in these extracts is one of the more reliable predictors of Cd uptake by plants (Sauerbeck and Styperek 1985; Whitten and Ritchie 1991; He and Singh 1993; Rayment 1994; Krishnamurti et al. 1995; McLaughlin et al. 1997a). In contrast, fraction 5 is unrelated to Cd uptake by either plants (Sloan et al. 1997) or microbes (Rogers 1996). Cd uptake by plants has also been related to Cd sorption parameters (Boekhold and van der Zee 1990; Palm 1994; del Castilho and Chardon 1995). Consequently, soil properties that affect the concentration of Cd in solution may also affect its bioavailability (McLaughlin et al. 1997b).

A lack of essential thermodynamic data has impeded mechanistic modelling of the effects of soil properties on the Cd species and their concentrations in solutions in contact with soil (Hesterberg et al. 1993; Benedetti et al. 1996; Xia 1997; Kinniburgh et al. 1999). Such modelling assumes equilibrium and the consequences of this assumption in the soil/water system are untested. Empirical regression modelling, informed by the results of mechanistic studies, offers an alternative approach that may lead to predictive models.

Many mechanistic studies of the sorption of metal ions by (hydrous)oxides use goethite (FeOOH) as a model substrate. Goethite is relatively stable and its reactions with Cd are particularly relevant here because goethite is the most abundant (hydrous)oxide of Fe(III) in aerobic mineral soils (Schwertmann 1964). Hydroxyl ion sorbed at the goethite/water interface gives it a negative charge that increases with pH, i.e. the charge is pH-dependent. The negative charge attracts metal ions from the surrounding solution towards the interface (Barrow 1993). For such surfaces, sorption of the ions of a metal increases sharply above a characteristic pH (edge), which for Cd on goethite occurs near pH 5.6 (Johnson 1990; Backes et al. 1995). However, patchiness in the density of surface charge, surface defects, and lattice vacancies in goethite provide sites that vary widely in Cd affinity and reaction rate (Barrow and Shaw 1979; Johnson 1990; Rudzinski et al. 1993; Ainsworth et al. 1994; Backes et al. 1995; Axe and Anderson 1997; Trivedi and Axe 2000). The effects of the slower reactions are readily observed between cycles of sorption and desorption (Barrow 1993); consequently, in soils the phytoavailability of Cd might decrease with increased time of contact, but other factors may intervene (Jensen and Mosbaek 1990; McLaughlin et al. 1996).

Surface charge also affects sorption of Cd by soils (Naidu et al. 1994) whose main charged constituents are clays, OM, and the (hydrous)oxides of Fe, Mn, and Al. This complexity provides many possibilities for Cd binding (Amacher et al. 1986; Boekhold et al. 1993; Mann and Ritchie 1994; Sloan et al. 1997; Hamon et al. 1998; McLaren et al. 1998; Smoulders et al. 1999). As is the case for goethite, the surface charge on other oxides, OM, and some clay minerals is pH-dependent, so decreasing the pH of soils increases the solubility of Cd (Barrow 1987; Anderson and Christensen 1988; Boekhold et al. 1993; Naidu et al. 1994; Temminghoff et al. 1995). In modelling these effects, the soil properties that might usefully be included are the concentration of sorbed Cd, or the operational estimate of sorbed Cd, i.e. labile Cd (C[d.sub.1] pH, surface charge, and the constituents that contribute to variable charge.

In the aqueous phase, the mechanistically appropriate measures for reactions of Cd are the activities of its species, which are affected by complexation and ionic strength. In simple systems, such as suspensions of goethite in dilute NaCl, activities of the Cd species are readily estimated (Boekhold et al. 1993; Lores and Pennock 1998); however, these calculations are not applicable to solutions that have been in contact with soils, which contain organic macromolecules like those that complex Cd in aquatic systems (Benedetti et al. 1996; Nimmo and Fones 1997; Xia 1997; Lores and Pennock 1998; Capodaglio et al. 1998; Kinniburgh et al. 1999). in addition, measurements of Cd speciation in aquatic systems have been largely inconsistent with those in soil extracts, and the latter appear method-dependent (Holm et al. 1995a, 1995b; Andrewes et al. 1996; McBride et al. 1997; Ge et al. 2000; Sauve et al. 2000b; Temminghoff et al. 2000). Nevertheless, there has been considerable interest in the effects of dissolved organic matter (DOM) on the concentration of Cd in solution and on Cd transport through soils (del Castilho et al. 1993a; Hesterberg et al. 1993; Romkens and Salomons 1998; Elzinga et al. 1999; Sauve et al. 2000b; Weng et al. 2002). The composition of the aqueous phase also affects Cd binding through other mechanisms, e.g. multivalent ions modify surface charge (Pyman et al. 1979; Kuo and McNeal 1984; Undabeytia et al. 1998; Bolan et al. 1999; Hamon et al. 2002) and divalent cations compete with Cd for sorption (Hendrickson and Corey 1981 ; Christensen 1987; Boekhold et al. 1993; Temminghoff et al. 1995; Wang et al. 1997; Echeverria et al. 1998; Undabeytia et al. 1998; Wilkins et al. 1998). Therefore, the solution properties that may be useful in Cd solubility models include pH, Cl, multivalent ions, and DOM.

Interest in the effects of soil properties on the behaviour of Cd developed concurrently with much of the cited mechanistic research. For example, Anderson and Christensen (1988) examined the sorption of a single dose of Cd using 38 soils from Denmark and 3 pH values for each soil. They noted that the [log.sub.10] transformation of the ratio of the concentration of sorbed Cd to that in solution ([log.sub.10] [sup.Cd] [K.sub.d] was strongly associated with some individual soil properties, and similar associations have been reported in other studies (Gerritse and van Driel 1984; Buchter et al. 1989; Jopony and Young 1994; Lee et al. 1996; Gray et al. 1999). Anderson and Christensen (1988) then used multiple linear regression to describe the partial effects on [log.sub10] [sup.Cd] [K.sub.d] of pH and the [log.sub.10] transformations of [Mn.sub.ox], organic C, and [Al.sub.dt] (n = 117, [R.sup.2] = 0.93). (The subscripts dt and ox denote the fractions extracted by dithionite and oxalate.) In other studies, the dependent variables have included the coefficients of the Freundlich equation and the concentration of Cd that is either sorbed or in solution (Rattan and Sehgal 1989; Basta et al. 1993; McBride et al. 1997; Romkens and Salomons 1998; Springob and Bottcher 1998b; Elzinga et al. 1999; Gray et al. 1999; Schug et al. 1999; Celardin 1999; Sauve et al. 2000a). Enriched stable or unstable Cd isotopes, and facilities to measure isotope dilution, are needed to estimate [Cd.sub.1]; however, hot concentrated HN[O.sub.3] and HCI, and their mixtures, efficiently extract soil Cd and the extracted Cd is readily measured by a variety of methods. Consequently, the total concentration of Cd ([Cd.sub.t]) has been the most common measure of Cd load in the models. Soil pH is almost invariably used as the explanatory variable and the other properties used, in decreasing frequency, are: OM > surface charge > (hydrous)oxides of Fe, Mn, and Al. The results for the (hydrous)oxides were highly inconsistent (McBride et al. 1997; Gray et al. 1999; Schug et al. 1999; Sauve et al. 2000a) and in this paper we explore why.

The sorption of Cd added to soil has been studied more often than desorption of 'native' Cd (Gray et al. 1999). However, soils and some of their constituents slowly fix added Cd (Amacher et al. 1986; Ainsworth et al. 1994; Backes et al. 1995; Franchi and Davis 1997; McLaren et al. 1998; Smoulders et al. 1999). The conditions during sorption and desorption experiments also tend to differ systematically, e.g. sorption experiments generally use briefer periods of contact between Cd and the soil, greater Cd loads, and more dilute suspensions of soil (Hendrickson and Corey 1981; Sumner 1994; Springob and Bottcher 1998a; Wilkins et al. 1998; Chang and Wang 2002). Consequently, we concur with Sauve et al. (2000a) that the effects of soil properties on Cd behaviour may differ between sorption and desorption studies. We address this question in the Discussion.

Cadmium from the labile pool can be desorbed using batch or serial procedures that inevitably involve some large dilutions, and this may affect the composition of the aqueous phase and Cd desorption (Undabeytia et al. 1998). The dilution effects can be minimised by maintaining the unique composition of the aqueous phase in contact with each soil; however, this becomes onerous when processing large numbers of soils. A common, if mechanistically simplistic, compromise is to use a background electrolyte at a constant, narrow mass to volume ratio, e.g. 1 g soil/2-10mL (Jopony and Young 1994; Celardin 1999; Chang and Wang 2002), and make only 2 Cd measurements on each soil, i.e. [Cd.sub.t] and the concentration of Cd in solution ([Cd.sub.s]). Then [Cd.sub.t], [Cd.sub.s] or their ratio is used as the dependent variable in multiple linear regression models whose independent variables are drawn from among the soil properties listed earlier (Celardin 1999; Elzinga et al. 1999; Sauve et al. 2000a). The variables in these models, other than pH, are usually [log.sub.10]-transformed.

We determined the effects of a mechanistically informed sequence of soil properties on the concentration of Cd extracted by 10mM [CaCl.sub.2] from a suite of 41 acidic soils from the Sydney Basin, by fitting multiple regression models of increasing complexity to the data. These models are compared with published models in the Discussion. Our samples were drawn from 5 soil taxa and were matched taxonomically between farmed and unfarmed sites, so the influences of farming and soil taxonomy were also assessed in the models.

Materials and methods

Soils and measurements

The samples collected to study Cd accumulation in the cultivated layer of soil (0-15 cm) on vegetable farms in the Sydney Basin (Jinadasa et al. 1997) were drawn from the 5 soil taxa most used for vegetable production and comprise taxonomically matched composites from 29 farmed and 12 unfarmed sites. The duration of farming for our sites was 5 to ~ 100 years and contemporary management practices include the use of lime and gypsum, heavy rates of fertilisers and poultry manure, intensive tillage, irrigation, and multiple cropping (Jinadasa et al. 1997). Samples were air-dried at 40 C and crushed to pass a 2-mm sieve. Texture was estimated and total C ([C.sub.t] was measured by combustion. Exchangeable Ca, Mg, K, Na, and A1 ([Ca.sub.ex, [Mg.sub.ex], etc.) were leached using 10 mM [BaCl.sub.2] (Milham and Vimpany 1987) and measured using inductively coupled plasma-optical emission spectroscopy (ICP-OES). The formal charges on these cations were summed to give the effective cation exchange capacity (ECEC), which is an estimate of surface charge (Gillman 1979). Ammonium oxalate was used in the dark to extract P ([P.sub.ox]) and the (hydrous)oxides of Fe ([Fe.sub.ox]) and Al ([Al.sub.ox]) (Schwertmann 1964), and their concentrations were measured using ICP-OES.

A solution of hydroquinone in 1 M ammonium acetate was used to reduce the (hydrous)oxides of Mn (Sherman et al. 1942). The extracted Mn ([Mn.sub.hq]) was measured by atomic absorption spectrophotometry (AAS). Soil (10g) was suspended in water (50mL) and shaken end-over-end at 30 r.p.m. for l h at 25[degrees]C. The supernatant was analysed for electrical conductivity (EC) and for Cl using Ag ions generated electrically. Soil (10g) was suspended in 10mM [CaCl.sub.2] (50mL) and shaken as for EC. The pH was measured on ~30mL of the suspension ([]). The remainder was centrifuged then the supernatant was analysed for DOM using a combustion analyser, and for Cd as described below ([]). 'Total' elements were extracted from a finely ground subsample of each soil (0.2-1.0g) using reverse aqua regia (5 mL), and heated in a sealed, polyfluorocarbon vessel, using microwave energy (Long and Martin 1989). The digests were diluted and centrifuged, and the supernatants were analysed for P ([P.sub.t]) and Zn ([Zn.sub.t]) using ICP-OES, and for Cd ([Cd.sub.t]) as described below.

The procedure for [Cd.sub.t], had been validated and was in routine use prior to this study. This was not the case for the procedure used to measure [], which we validated using a subset of 4 of our soils. First, shaking times in the range 0.5-6 h were shown to have little effect on the measurement. This finding is consistent with the results of other studies on acidic soils (Amacher et al. 1986; Franchi and Davis 1997; Wang et al. 1997; Harper et al. 1998: Gray et al. 1999), although the rate of Cd release from alkaline soils may be slower (O'Connor et al. 1984; Barrow et al. 1989; McLaren et al. 1998; Ramachandran and D'Souza 1999). Secondly, filtering the supernatants of centrifuged extracts through 10 kDa polyethersulfone, cut-off filters (Millipore, USA) did not affect the Cd concentrations. On this basis, we shook the suspensions for 1 h, centrifuged, and analysed the supernatant for []. Colloids may nonetheless have been present (del Castilho et al. 1993b; Jopony and Young 1994; Kretzschmar et al. 1999) and any Cd associated with them would have contributed to the measurements. [Cd.sub.t] and [] were measured using furnace-AAS (FAAS) with Zeeman background correction and a matrix modifier consisting of Mg(N[O.sub.3])2 and N[H.sub.4] [H.sub.2] P[O.sub.4]. Results were confirmed using ICP-mass spectrometry and between-method discrepancies >10% were resolved mainly by optimising the conditions in the graphite furnace to minimise Cd losses through volatilisation prior to the measurement step. Acid digests and [CaCl.sub.2] extracts were analysed within 48 h of preparation to minimise the risks of contamination and losses (Sekaly et al. 1999). Reported [Cd.sub.t] and [] values are means of measurements on at least 2 independent extracts (Table 1).

Statistical methods

The [log.sub.10] scale was used for all soil properties other than []. [Log.sub.10] [] was used as the dependent variable in the base multiple regression model in which [Cd.sub.t] and [] were the independent variables. Other soil properties were added sequentially as independent variables and the order of entry, as informed by the chemical literature, was [C.sub.t], ECEC, [Fe.sub.ox], and [Mn.sub.hq]. Additional soil properties including [Al.sub.ox], [P.sub.ox], [P.sub.t], DOM, [Ca.sub.ex], [Mg.sub.ex] EC, and Cl were then tested. A pair of models was fitted at the entry of each property. The first consisted of only the linear effects of the soil properties. The second was an extended model that also included the quadratic effects of the soil properties, the effects of farming and soil taxonomy, and the interactions of farming and soil taxonomy with the linear effects. Each model was reduced by removing all terms that were non-significant (P > 0.05); however, model hierarchy was preserved for the extended models, e.g. if a quadratic term was significant, the corresponding linear term was necessarily included (see model 4). All models are presented with [R.sup.2] and residual standard deviation (r.s.d.) values and the regression coefficients are followed by standard errors in parentheses.

Multiple linear regression modelling adjusts each regression coefficient for the effects of correlations with the other independent variables to produce a partial regression coefficient. The partial coefficient estimates the slope of the linear relation between the dependent variable and the uncorrelated part of the independent variable; consequently, the value of a partial coefficient usually differs from that of the simple linear regression relation between 2 variables. When an independent variable has very high correlations with other independent variables, the estimate of the partial regression coefficient may become unreliable. To detect any such adverse effects we examined the variance inflation factors of the partial coefficients [Montgomery and Peck 1982) and also used added variable plots (Cook and Weisberg 1982) to assess the partial regressions and the patterns of residuals for each soil property.


All of the measured properties varied widely, e.g. the textural range (Table 1) corresponded to approximate clay contents of 5-40% (Northcote 1979). Farming generally increased [Cd.sub.t], [], [], and [P.sub.t], and decreased [C.sub.t] (Table 1). The concentration ranges of properties for which data are not presented for individual soils were (with means in parentheses): [Zn.sub.t] 5-196 (45)mg/kg, [Al.sub.ox] 500-2300 (1000)mg/kg, Pox <10-3800 (1300)mg/kg, DOM 20-120 (49)mgC/L, [Ca.sub.ex] 0.24-28 (7.8) [cmol.sub.(c)]/kg, [Mg.sub.ex] 0.18-8.3 (2.54) [cmol.sub.(c)]/kg, EC <0.01-0.63 (0.23) mS/cm, and Cl 12-463 (107) mg/kg.

The models

We modelled the effects of soil properties on [log.sub.10] [] (mg/L). The explanatory variables in the base model were [log.sub.10] [Cd.sub.t] and [log.sub.10] [] ([R.sup.2] = 0.885, r.s.d. = 0.245):


[log.sub.10] [] = 1.39 ([+ or -] 0.30) + 0.98 ([+ or -] 0.06) [log.sub.10] [Cd.sub.t] - 0.66 ([+ or -]0.06) []

Farming and soil taxonomy are factors in our data, and when added to model 1, both had strong effects on [log.sub.10] [] (P < 0.001), but there were no interactions with either [log.sub.10] [Cd.sub.t] or [] (P > 0.). Inclusion of the farming and soil taxonomy effects improved the values of [R.sup.2] (0.951) and r.s.d. (0.171).

For our acidic soils, [C.sub.t] is a measure of organic C (Baldock and Skjemstad 1999), a property often used as an independent (explanatory) variable in empirical models of Cd solubility. Therefore, in model 2 we included [log.sub.10] [C.sub.t] as an independent variable ([R.sup.2] = 0.914, r.s.d. = 0.214):


[log.sub.10] [Cd.sub.Ca] = 2.02 ([+ or -] 0.32) + 0.94 ([+ or -]0.05) [log.sub.10] [Cd.sub.t] - 0.64 ([+ or -]0.05) [] - 0.61 ([+ or -]0.17) [log.sub.10] [C.sub.t]

When added to model 2, the farming and soil taxonomy factors behaved as in model 1, and inclusion of their effects again improved the values of [R.sup.2] (0.963) and r.s.d. (0.151).

We extended model 2 by including [log.sub.10] ECEC, and in the extended model the partial regression coefficient for [log.sub.10] ECEC (-0.69 [+ or -] 0.15) was significant (P<0.001), while that for [log.sub.10] [C.sub.t] ([+ or -] 0.21) was not (P>0.5). Therefore, [log.sub.10] ECEC replaced [log.sub.l0] [C.sub.t] in model 3 ([R.sup.2] = 0.946, r.s.d. = 0.170):


[log.sub.l0] [] 1.36([+ or -]0.21) + 1.10([+ or -] 0.05) [log.sub.10] [Cd.sub.t] - 0.53 ([+ or -]0.04) [] - 0.64 ([+ or -] 0.10) [log.sub.10] ECEC

When added to model 3, farming interacted with [log.sub.10] [Cd.sub.t] (P<0.01) and appeared to interact with [log.sub.10] ECEC (P < 0.05), while soil taxonomy had no effect (P > 0.1). However, the farming interaction with log 10 ECEC was not as strong as the quadratic effect of [log.sub.10] ECEC (P<0.01). Inclusion of the latter term gave model 4 ([R.sup.2] = 0.974, r.s.d. = 0.122):


[log.sub.l0] [] = 0.77 ([+ or -] 0.20) [1.34 ([+ or -] 0.05) [log.sub.10] [sup.f] [Cd.sub.t] or [1.01 ([+ or -] 0.04) [log. sub.10] [sup.uf] [Cd.sub.t] - 0.43 ([+ or -] 0.04) [] - 0.05 ([+ or -] 0.21) [log.sub.10] ECEC - 0.47([+ or -] 0.13) [[log.sub.10] ECE[C.sup.]2

where [sup.f] [Cd.sub.t] and [sup.uf] [Cd.sub.t] are values of [Cd.sub.t] for farmed and unfarmed soils. When model 3 was extended by the addition of [log.sub.10] [Fe.sub.ox], this new term was significant (P < 0.05), but only for [] values >5.6 ([sup.pH>5.6][Fe.sub.ox]). Only 12 farmed soils and 1 unfarmed soil had [pH.sub.Ca] values >5.6 (Table 1). To enable a direct comparison of the coefficients in the extended model (model 5) with those in the earlier models, all 41 soils were retained and the model includes a slope coefficient for [log.sub.10] [sup.pH>5.6][Fe.sub.ox] and a corresponding constant term, and a separate constant term for the 28 soils with pH [less than or equal to] 5.6. Model 5 is ([R.sup.2] = 0.952, r.s.d. = 0.163):

(5) [log.sub.10] [Cd.sub.Ca] = [1.33 ([+ or -] 0.20)for [pH.sub.Ca] [less than or equal to] 5.6] or [2.48 ([+ or -] 0.56) for [pH.sub.Ca] > 5.6] + 1.09 ([+ or -] 0.04)[log.sub.10] [Cd.sub.t] - 0.54 ([+ or -] 0.04)[pH.sub.Ca] - 0.56 ([+ or -] 0.10)[log.sub.10] ECEC - 0.33 ([+ or -] 0.15)[log.sub.10][sup.pH>5.6][Fe.sub.ox]

The effect of [log.sub.10] [sup.pH>5.6][Fe.sub.ox] was also significant when added to model 4 (P < 0.01), but the farming interaction with [log.sub.10] ECEC, or alternatively the quadratic effect of [log.sub.10] ECEC, was not significant (P > 0.05). Model 6 is ([R.sup.2] = 0.974, r.s.d. = 0.121):

(6) [log.sub.10][Cd.sub.Ca] = [1.18([+ or -] 0.15) for [pH.sub.Ca] [less than or equal to] 5.6] or [2.69 ([+ or -] 0.42) for [pH.sub.Ca] > 5.6] + [1.33 ([+ or -] 0.05)[log.sub.10] [sup.f][Cd.sub.t]] or [1.01 ([+ or -] 0.04)[log.sub.10] [sup.uf][Cd.sub.t]] - 0.49 ([+ or -] 0.03)[pH.sub.Ca] - 0.69([+ or -] 0.08)[log.sub.10] ECEC - 0.43([+ or -] 0.12)[log.sub.10] [sup.pH>56][Fe.sub.ox]

When the effect of [log.sub.10] [Mn.sub.hq] was added to model 6, the coefficient (-0.062 [+ or -] 0.035) was significant at P = 0.10, but not at P = 0.05. In model 6, little variation remained to test the effects of the additional properties [Al.sub.ox], [P.sub.ox], [P.sub.t], [Z.sub.nt], DOM, [Ca.sub.ex], [Mg.sub.ex], EC, and CI and of differences in the effect of [log.sub.10] [Fe.sub.ox] as pH increased above 5.6; each was non-significant (P > 0.10). Finally, the farming effect on the coefficient for [log.sub.10] [Cd.sub.t] did not change appreciably when we adjusted [Cd.sub.t] for the Cd desorbed into 10 mM Ca[Cl.sub.2], i.e. 5 [Cd.sub.Ca] (mg/kg).


Properties of the soils

The wide range of values of all the measured properties was a consequence of intrinsic differences between soil taxa and of the effects of farming practices (Table 1, and Jinadasa et al. 1997). Some unfarmed soils had values of [Cd.sub.t] approaching the limit of detection for the method (~0.01 mg/kg). At the other extreme was soil 17A, in which [Cd.sub.t] was unusually high for what appears to be a pristine soil (1.93 mg/kg); however, this soil had a considerable shale influence and its Cd concentration fell in the range for other shale soils (Alloway 1990). The values of [Cd.sub.t] for the farmed soils were within the range 0.1-11 mg Cd/kg reported for other agricultural soils (Sanchez-Camazano et al. 1994); nevertheless, some values of [Cd.sub.t] and [Cd.sub.Ca] (Table 1) considerably exceeded those for surface soils from other Australian farms (Rayment 1994; McLaughlin et al. 1997a, 1997b) and for fertilised pasture soils from New Zealand (Gray et al. 1999). The values of [Fe.sub.ox], [Mn.sub.hq], and [Al.sub.ox] were within the ranges reported for other Australian soils (Osborne et al. 1988; Vimpany et al. 1997). In contrast, the concentrations of Ct were lower in the farmed soils than in their unfarmed counterparts (Table 1), which supports earlier evidence that intensive vegetable farming in the Sydney Basin may exploit C reserves (Sarooshi et al. 2002).

Associations between pairs of soil properties may have occurred for intrinsic reasons (Norrish and Rosser 1983; Basta et al. 1993) and as a result of our sampling strategy. The family of associations with ECEC is an example that is pertinent to the previous modelling. [Log.sub.10] ECEC was positively associated with [log.sub.10][C.sub.t], [log.sub.10][Fe.sub.ox], and [pH.sub.Ca] (Table 2), and with the approximate clay content (r = 0.81, data not presented) estimated from soil texture after Northcote (1979). We attribute these 4 associations to the surface charge on soil OM, the hydrous oxides of Fe and kaolinite--probably the most abundant clay mineral in the soils (Michael Maguire and Ian Vimpany, unpublished data)--and to the pH-dependence of that charge. An association also occurred between [log.sub.10] ECEC and [log.sub.10][P.sub.t] (Table 2). This association, and the influence of farming on the association, may have been due to an underlying relation of Fe and P that is typical of Australian surface soils (Norrish and Rosser 1983) and the large amounts of P applied to the farmed soils (Jinadasa et al. 1997). Finally, [log.sub.10] ECEC and [log.sub.10] [Ca.sub.ex] were associated (r = 0.96), because [Ca.sub.ex] averaged ~66% of ECEC (data not presented).

The background electrolyte

Composition and soil/solution ratio

The ions [Na.sup.+], [K.sup.+], N[O.sub.3.sup.-], and Cl[O.sub.4.sup.-] have little effect on the (de)sorption of Cd by either competition or complexation, and consequently have been the preferred constituents of background electrolytes for studies of Cd interactions with a variety of substrates (Garcia-Miragaya and Page 1976; Hendrickson and Corey 1981 ; Zachara et al. 1993; Ainsworth et al. 1994; McBride et al. 1997; McLaughlin et al. 1997b). In contrast, the ions [Ca.sup.2+] and [Cl.sup.-] affect the (de)sorption of Cd (Gerritse and van Driel 1984; Boekhold et al. 1993), are present at appreciable concentrations in many soils, and have increasingly been used as components of background electrolytes (Jopony and Young 1994; Backes et al. 1995; Temminghoff et al. 1995; Wang et al. 1997; Gray et al. 1999).

We used 10mM Ca[Cl.sub.2] as the background electrolyte because: its ionic strength is similar to that of soil solutions (Houba et al. 1996); it limits the effects of dilution on suspension pH (Sumner 1994); and the concentration of Cd it extracts is related to that in the soil solution (Degryse et al. 2003) and to Cd availability to plants (Sauerbeck and Styperek 1985; Whitten and Ritchie 1991; He and Singh 1993). However, the choice of a soil to solution ratio of 10g/50 mL was more pragmatic--these conditions are widely used in Australia for the measurement of pH (Rayment and Higginson 1992) and the concentration of Cd in the extract was sufficient for [Cd.sub.Ca] to be measured directly using FAAS.

Effects on the modelling

The large natural variations in the concentrations of Cl and Ca (not presented) might have had significant effects in model 6 (McLaughlin et al. 1997b; Elzinga et al. 1999) had 10mM Ca[Cl.sub.2] not been used as the background electrolyte and [pH.sub.Ca], [Cd.sub.t] and ECEC explained less of the variation in [Cd.sub.Ca].

The concentration of Ca in the background electrolyte (10 mM) would also have limited the concentration of DOM in the soil suspensions (Romkens and Dolfing 1998) and presumably, the formation of complexes between DOM and Cd (Holm et al. 1995a, 1995b; Xia 1997). This might partly explain the lack of significance of [log.sub.10] DOM as an independent variable when added to model 6 (P > 0.1); however, as noted in the preceding paragraph, the residual variation in model 6 was too small to provide an adequate test of additional properties. The presence of Ca in our background electrolyte may not have been the only reason we observed no effect of DOM, since DOM also had no effect in a study in which no background electrolyte was added (Romkens and Salomons 1998). in contrast, the effects of DOM are evident for soils in which organic matter dominates metal binding (del Castilho et al. 1993a, 1993b; Impellitteri et al. 2002).

The dependent variable

There is a long tradition of modelling partition coefficients such as [log.sub.10] ([Cd.sub.t]/[Cd.sub.Ca]) to describe the results of sorption experiments, although for soils, the more operationally convenient, if mechanistically simplistic, expression [log.sub.10] ([Cd.sub.t]/[Cd.sub.Ca]) is more often used (Gerritse and van Driel 1984; Buchter et al. 1989; Houba et al. 1996; Sauve et al. 2000a). To provide a link with those studies we use the latter expression in model 7 ([R.sup.2] = 0.919, r.s.d. = 0.169):

(7) [log.sub.10]([Cd.sub.t]/[Cd.sub.Ca]) = [-1.15([+ or -] 0.18) for pH [less than or equal to] 5.6] or [-2.38([+ or -] 0.58) for pH > 5.6] + 0.53([+ or -] 0.04)[pH.sub.Ca] + 0.48([+ or -] 0.10) [log.sub.10] ECEC + 0.35([+ or -] 0.16) [log.sub.10] [sup.pH>5.6][Fe.sub.ox]

where the units of [log.sub.10] ([Cd.sub.t]/[Cd.sub.Ca]) are L/kg and the form of the model is as described for model 5.

Use of the ratio [Cd.sub.t]/[Cd.sub.Ca] in model 7 does not allow for a difference in the effect of [Cd.sub.t] between farmed and unfarmed soils; therefore, there are some biases in the estimates of the regression coefficients for other soil properties. The use of models such as model 7 assumes that the numerator and denominator of the ratio in the dependent variable, in this case [Cd.sub.t] and [Cd.sub.Ca], vary in strict proportion. The values of the partial regression coefficients of [log.sub.10] [Cd.sub.t] in model 6 have values of 1.01 [+ or -] 0.04 for the unfarmed soils and 1.33 [+ or -] 0.05 for the farmed soils, i.e. the assumption was applicable to the unfarmed but not the farmed soils. The validity of the assumption was also questionable for a metastudy of the effects of soil properties on Cd behaviour (Sauve et al. 2000a). Therefore, [log.sub.10]([Cd.sub.t]/[Cd.sub.Ca]) may generally be a less suitable dependent variable than either [log.sub.10] [Cd.sub.t] or [log.sub.10] [Cd.sub.Ca].

The farming effect

We found only one other modelling study of Cd behaviour which declared that the samples had been drawn from both farmed and unfarmed sites, and it is not apparent whether the regression analysis tested for farming effects (Romkens and Salomons 1998). A possible effect of farming on the coefficient of [log.sub.10] ECEC in model 4 was resolved in model 6 by the inclusion of [log.sub.10][sup.pH>5.6][Fe.sub.ox]. Nevertheless, the effect of farming on the coefficient of [log.sub.10] [Cd.sub.t] in model 6 remained highly significant (P < 0.01), even after the model had been extended to allow for Cd desorption, i.e. by using [log.sub.10] ([Cd.sub.t] - 5 [Cd.sub.Ca]) in place of [log.sub.10] [Cd.sub.t].

The values of the coefficients of [log.sub.10] [Cd.sub.t] in model 6 were 1.33 4 [+ or -] 0.05 (farmed soils) and 1.01 [+ or -] 0.04 (unfarmed soils), i.e. after the effects of [pH.sub.Ca], [log.sub.10] ECEC, and [log.sub.10] [sup.pH>5.6][Fe.sub.ox] had been adjusted for, [log.sub.10] [Cd.sub.Ca] changed more rapidly with changes in [log.sub.10][Cd.sub.t] for farmed than unfarmed soils. Farming may have affected the Cd-complexing properties of OM (McArthur et al. 2001). However, the effect is also consistent with the shorter contact time and larger Cd load for the farmed soils, as described in the Introduction and with evidence of specific binding at the oxide/water interface (Rudzinski et al. 1993) and to DOM (Xia 1997). A similar effect of Cd load is also evident in a compilation of data from studies of Cd sorption (fig. 1 of Elzinga et al. 1999). Therefore, Cd added in the laboratory and during farming may behave differently from 'native' Cd and perhaps from Cd present at low concentrations.


Soil properties as independent variables


[Log.sub.10] [Cd.sub.t] is included in most models that describe the effects of soil properties on Cd behaviour (Celardin 1999; Elzinga et al. 1999; Sauv6 et al. 2000a; Degryse et al. 2003) and in all our models i-6, [log.sub.10] [Cd.sub.t] had a strong positive effect. It is mechanistically reasonable to suggest that [Cd.sub.t] should be adjusted for Cd desorption. Since the units of [Cd.sub.Ca] are mg/L and of [Cd.sub.t] are mg/kg, and the ratio of soil to background electrolyte is 1 g/5 mL, the corrected value is ([Cd.sub.t] - 5 [Cd.sub.Ca]) mg/kg. However, logl0 ([Cd.sub.t] - 5 [Cd.sub.Ca]) was no more effective an independent variable than [log.sub.10][Cd.sub.t], perhaps because [Cd.sub.Ca] typically constituted so small a proportion of [Cd.sub.t] (Table 1). The other adjustment that might be made on mechanistic grounds would be to use [Cd.sub.1] in place of [Cd.sub.t] on the assumption that [Cd.sub.Ca] is drawn from the labile pool. [Log.sub.10] [Cd.sub.1] may be a superior independent variable to [log.sub.10] [Cd.sub.t] if, as it seems, [Cd.sub.1] may constitute a highly variable proportion of [Cd.sub.t] (Gerritse and van Driel 1984; Nakahone and Young 1993; Springob and Bottcher 1998a; Wilkins et al. 1998; Smoulders et al. 1999; Young et al. 2000; Degryse et al. 2003). This was so in the only study that compared the efficacy of [Cd.sub.1] and [Cd.sub.t] as explanatory variables (Degryse et al. 2003).


The simple correlation between the [log.sub.10] [Cd.sub.Ca] and [pH.sub.Ca] was weak (-0. 31, Table 2); however, the correlation between these properties was strong for soils that had a narrower range of properties than ours (Whitten and Ritchie 1991;El-Falaky et al. 1991;Rayment 1994). Therefore it is not surprising that, after adjusting for the effects of [log.sub.10] [Cd.sub.t], [log.sub.10] ECEC, and [log.sub.10] [sup.pH>5.6][Fe.sub.ox] in our data (model 6), the partial correlation for [pH.sub.Ca] was also strong (-0.93).

The pH is usually included among the independent variables in multiple regression models of the effect of soil properties on Cd behaviour, and the value of its coefficient may be affected by the other properties chosen as independent variables, as occurred in models 1-6. Furthermore, the choice of dependent variable may, of necessity, change the sign of the coefficient, as it did between models 6 and 7. Standard errors of the coefficients are often not given, making it unclear how reliably the coefficients may have been estimated; even so, recent estimates of the coefficient for pH appear to be converging towards a consensus value of ~0.50 (Elzinga et al. 1999; Sauve et al. 2000a; Degryse et al. 2003). This is very similar to our final estimate of 0.49 [+ or -] 0.03 (model 6) and to the mean of 0.56 for the 11 cited multiple regression studies that include the effects of pH on Cd behaviour (de Haan et al. 1987; Anderson and Christensen 1988; del Castilho et al. 1993a; Springob and Bottcher 1998b; Jopony and Young 1994; Celardin 1999; Romkens and Salomons 1998; Elzinga et al. 1999; Gray et al. 1999; Sauve et al. 2000a; Degryse et al. 2003). In these studies, there was no discernable effect on the value of the pH coefficient whether the range of pH values was natural, or artificially extended by adjustment in either the field or laboratory. This may be no coincidence, since for sorption by goethite, the species being sorbed appears to determine the proton stoichiometry, and for Cd under a variety of conditions the proton coefficients were ~0.50 (Spark et al. 1995).

[C.sub.t] and ECEC

The effect of [log.sub.10][C.sub.t] was significant (P < 0.01) in model 2, and the sign of the partial regression coefficient was negative. Both of these results are consistent with other regression models (Gerritse and van Driel 1984; Anderson and Christensen 1988; Basta et al. 1993; McBride et al. 1997; Gray et al. 1999; Sauv6 et al. 2000a; Degryse et al. 2003), with the proposition that much of the Cd in soils may be bound to OM (Calvet et al. 1990; Jeng and Singh 1993), and with spectroscopic evidence that -COOH and -SH groups on OM bind Cu and Zn, and presumably Cd (Xia 1997). Nevertheless, the value of the coefficient of [log.sub.10][C.sub.t] is inconsistent between studies, e.g. in model 2 and in the model of Sauve et al. (2000a) the values are -0.61 ([+ or -] 0.17) and -0.81 ([+ or -] 0.05). Contrast this difference with that in the following paragraph, where the explanatory variable is [log.sub.10] ECEC instead of [log.sub.10][C.sub.t].

A measure of surface charge, rather than OM, is a key variable in some mechanistic and empirical models of Cd (de)sorption (Kuo and McNeal 1984; Barrow 1993; Romkens and Salomons 1998; Celardin 1999; Elzinga et al. 1999; Gray et al. 1999; Springob and Bottcher 1998b; Degryse et al. 2003). Possible alternative measures of surface charge in soils are CEC and ECEC. We chose ECEC because it is estimated at a pH similar to that which pertains during Cd desorption in 10mM Ca[Cl.sub.2], rather than CEC, which is estimated at a fixed, elevated pH (Rayment and Higginson 1992). In models 3-6, the effect of [log.sub.10] ECEC on [log.sub.10][Cd.sub.Ca] was highly significant (P < 0.01), and the sign of the partial regression coefficient was negative. In addition there was remarkable consistency between the values of the coefficient of [log.sub.10] ECEC in model 8 when the model was applied to data compiled from adsorption studies (Elzinga et al. 1999) and to our data, i.e. 0.63 ([+ or -] 0.03) and 0.67 ([+ or -] 0.09). The closeness of this comparison contrasts with that in the previous paragraph for the property [C.sub.t].

We compared the efficacy of OM and ECEC as independent variables, and in model 3, [log.sub.10] ECEC displaced [log.sub.10][C.sub.t]. Holm et al. (2003) obtained a similar result for 49 soils from Denmark at pH 5.3; however, the result was reversed at pH 6.7. The result of comparisons between OM and ECEC also differed in other studies (Anderson and Christensen 1988; Basta et al. 1993; Romkens and Salomons 1998; Gray et al. 1999; Degryse et al. 2003). Soil OM and surface charge are often related, as occurred in our data (Table 2); nevertheless, surface charge may sometimes displace Ct from the models, because the -COOH groups on soil OM contribute to surface charge and may complex a considerable proportion of [Cd.sub.t] (Calvet et al. 1990; Jeng and Singh 1993; Xia 1997), but they do not directly influence estimates of soil OM concentration.

Oxides of Fe, Mn, and Al

Oxides such as goethite (Fe), hausmannite (Mn), and alumina (Al) have large ratios of surface area to mass, and avidly sorb Cd when the pH exceeds their respective sorption edges, i.e. 5-6 (Fe), 3-4 (Mn), and 7-8 (Al) (Backes et al. 1995; Spark et al. 1995). In this respect, the (hydrous)oxides of these metals that occur naturally in soils appear to behave similarly (Zachara et al. 1992; Murphy and Zachara 1995; Tessier et al. 1996); furthermore, sequential chemical extraction of soils indicates that a considerable portion of [Cd.sub.t] may be associated with the (hydrous)oxides of Fe and Mn (Jeng and Singh 1993; Benitez and Dubois 1999). Therefore, increased concentrations of the (hydrous)oxides of Fe, Mn, and Al in soils should, if anything, be associated with increased sorption. Their apparent lack of effects on Cd (de)sorption in some modelling studies might be attributed to the concentrations being too small or varying too little between samples, or to the use of extraction procedures that result in the measurement of more abundant, less reactive forms of the oxides (McKeague et al. 1971 ; Parfitt and Childs 1988; Buchter et al. 1989; Zachara et al. 1993; Agbenin 2003; Holm et al. 2003). The extraction conditions we used are thought to selectively extract the hydrous fractions of the oxides of Fe, Mn, and Al (Sherman et al. 1942; Schwertmann 1964) and our soils contained a considerable range of concentrations of the hydrous oxides (Table 1). In addition, we are the first to apply the concept of sorption edges to Cd behaviour in soils.

Six studies reported effects of Fe, Mn, and Al (hydrous)oxides on Cd behaviour using multiple regression models (Anderson and Christensen 1988; del Castilho et al. 1993a; Bolton and Evans 1996; Gray et al. 1999; Schug et al. 1999). As noted by Tessier et al. (1996) and Sauve et al. (2000a), some of the effects were inconsistent with the known sorption properties of the (hydrous)oxides of Fe, Mn, and Al. Consequently, we used 3 criteria to exclude studies that were likely to be unreliable and applied the criteria in the order: (i) was a common value used for any key property for more than 1 sample, e.g. samples drawn from different plots within a field trial; (ii) did the concentrations of the (hydrous)oxide(s) span a sufficiently wide range (McBride et al. 1997); (iii) were the signs of all the statistically significant, simple regression coefficients consistent with known sorption properties? Only the studies of Bolton and Evans (1996) and Schug et al. (1999) survived this cull and both included effects of (hydrous)Fe oxides that were consistent with mechanistic expectations. For our soils, [log.sub.10][Fe.sub.ox] affected [log.sub.10][Cd.sub.Ca] in the expected manner, but only when [pH.sub.Ca] exceeded ~5.6 (P < 0.05). This pH agrees well with the approximate sorption edge for Cd on goethite (Johnson 1990; Ainsworth et al. 1994; Axe and Anderson 1997); however, we had too few soils with [pH.sub.Ca] values >5.6 (n = 13) to permit a more detailed modelling of the pH dependence. The correspondence of the [pH.sub.Ca] value above which [log.sub.10][Fe.sub.ox] affected [log.sub.10] [Cd.sub.Ca], and the sorption edge for goethite indicates that more widespread application of the sorption edge concept may enhance the quality of empirical models of cation behaviour in soils.

No multiple regression studies of the effects of Mn and Al (hydrous)oxides on Cd behaviour in soils survived the cull. In the case of Al the reason may be that the sorption edge occurs at too high a pH value (7-8) (Spark et al. 1995) for an effect to have been observed in most studies. In contrast, if hausmannite is an appropriate model for the (hydrous)oxides of Mn in soils, one might have expected the effects to have been observed, given that the sorption edge is pH 3-4 (Backes et al. 1995). In an extension of model 6, [log.sub.10] [Mn.sub.hq] had a mechanistically consistent, weak effect (P < 0.1), whereas [log.sub.10] [Al.sub.ox] had no effect (P > 0.1). However, as stated earlier, there was not enough residual variation in model 6 to adequately test any additional properties. Further studies should use selective methods to extract the hydrous oxide fractions from soils, and investigate the effects of substitution and humic coatings on the sorption edges (Mann 1985; Cowan et al. 1992; Zachara et al. 1992; Zachara et al. 1994; Spark et al. 1995).

Measured and fitted values of [log.sub.10] [Cd.sub.Ca]

The [R.sup.2] values were presented with the models in most of the following 10 multiple regression studies (Anderson and Christensen 1988; McBride et al. 1997; Romkens and Salomons 1998; Springob and Bottcher 1998b; Celardin 1999; Elzinga et al. 1999; Gray et al. 1999; Schug et al. 1999; Sauve et al. 2000a; Degryse et al. 2003). The [R.sup.2] value is an estimate of the proportion of the variation explained by the model. In the preceding studies, [R.sup.2] values of 0.7-0.9 were common and model 6 had an [R.sup.2] value at the upper end of this range (0.974).

For multiple regression models the r.s.d. conveys information about the agreement between fitted and measured values (Montgomery and Peck 1982) and r.s.d. values were presented with only 2 of the 10 cited models, both of which described the effects of soil properties on the concentration of desorbed Cd. The first, for 30 contaminated samples, had an r.s.d. of 0.30 [log.sub.10] units (Jopony and Young 1994). The second, for a compilation of data from ~70 studies, had an r.s.d. value of 0.62 [log.sub.10] units (Sauve et al. 2000a). These r.s.d. values were larger than that for model 6 (0.12 [log.sub.10] units), for which there was good correspondence between the measured and estimated values of [log.sub.10][Cd.sub.Ca] (Fig. 1). The measured values of [Cd.sub.Ca] all fell within [+ or -] 58% of those obtained by back-transforming the estimates of [log.sub.10] [Cd.sub.Ca] (model 6) and the median difference was [+ or -] 17%.

Comparing the effects of soil properties during sorption and desorption

In the Introduction we postulated that soil properties may affect Cd behaviour differently during sorption and desorption because of systematic differences in the reaction conditions. We tested this possibility for the properties pH and ECEC by comparing the values of the coefficients for pH and [log.sub.10] ECEC in a model used to describe a large compilation of Cd sorption data (Elzinga et al. 1999) with their values when the model was fitted to our data. Elzinga et al. (1999) used the concentration of sorbed Cd as the dependent variable, which we approximated by ([Cd.sub.t] - 5 [Cd.sub.Ca]) (mg/kg), and in our notation the model is (n = 1125, [R.sup.2] = 0.78):

(8) [log.sub.10]([Cd.sub.t] - 5 [Cd.sub.Ca]) = -3.22([+ or -] 0.10) + 0.87 ([+ or -] 0.01) [log.sub.10] [Cd.sub.Ca] + 0.45 ([+ or -] 0.01) [pH.sub.Ca] + 0.63 ([+ or -] 0.03) [log.sub.10] ECEC - 0.47 ([+ or -] 0.02) [log.sub.10 Ca

where Ca represents the concentration of [Ca.sup.2+] in solution. The concentration of Ca varied over a wide range in the data compiled by Elzinga et al. (1999), whereas in our background electrolyte it was ~10 mM; nevertheless, model 8 fitted our data well ([R.sup.2] : 0.946, r.s.d. = 0.170) and the coefficients of [pH.sub.Ca] and [log.sub.10] ECEC were 0.49 ([+ or -] 0.04) and 0.67 ([+ or -] 0.09). These values are remarkably similar to those in model 8 (Elzinga et al. 1999). However, the value of our coefficient for [log.sbu.10] ECEC was slightly affected by farming, as it was in model 4, so we sought another desorption modelling study that included pH and [log.sub.10] ECEC among the independent variables. Degryse et al. (2003) had used such a model to describe the partitioning of Cd under desorption conditions for 74 polluted European soils ([R.sup.2] = 0.87) and estimated the coefficients for pH and [log.sub.10] ECEC as 0.45 and 0.63, again remarkably similar values to the preceding estimates. That is, the coefficients of pH and [log.sub.10] ECEC appear to differ and their values seem independent of whether the observations are made during sorption or desorption experiments.


Empirical models that describe the effects of soil properties on Cd behaviour for many soils from the Northern Hemisphere also do so for a wide range of acidic soils from the Sydney Basin. In our models, the signs of the partial regression coefficients are consistent with the expected effects of the soil properties and model 6 gives estimates of [log.sub.10] [Cd.sub.Ca] that are close to the measured values. The models presented should be regarded as descriptive rather than predictive until they have been validated using a sufficient body of independent data. The modelling raises questions of appropriate choice of soil properties as independent variables and the methods used to measure them. We note that the effects of the (hydrous)oxides of metals, such as Fe, need further investigation. Comparisons between studies would be facilitated if reports consistently included at least the primary data, standard errors, and the r.s.d. in addition to the [R.sup.2]. Finally, although the behaviour of Cd may vary with contact time and load, the effects of pH and ECEC on its behaviour appear independent of whether they are evaluated during processes of sorption or desorption. These observations and inferences may be pertinent to modelling the behaviour of other heavy metals.
Table 1. Selected properties of the surface soils (0-15 cm) from 29
farmed and 12 unfarmed sites in the Sydney Basin

Brief description of methods: p[H.sub.Ca] and [Cd.sub.Ca] are pH and
concentration of Cd in the 1 : 5 (w/v) suspension in 10 mM
Ca[Cl.sub.2]; [C.sub.t] is total C, measured as C[O.sub.2] after
combustion; ECEC (effective cation exchange capacity) is the sum of
exchangeable Na, K, Ca, Mg, and Al; [Cd.sub.t] and [P.sub.t] were
extracted by reverse aqua regia; [Fe.sub.ox] is oxalate-extractable
iron; [Mn.sub.hq] is Mn reducible by hydroquinone. Soil texture
abbreviations: KS, coarse sand; S, sand; LS, loamy sand; SL, sandy
loam; L, loam; CL, clay loam; SC, sandy clay; LC, loamy clay; MC,
medium clay

Sample Texture p[H.sub.Ca] [Cd.sub.Ca] [Cd.sub.t]
i.d. (mg/L) (mg/kg)

 Soil type 1: Strongly weathered soils on Triassic shale.
 Gently undulating topography


4 CL (with grit) 5.4 0.0012 0.36
6 CL to LC 6.6 0.0002 0.34
7 CL 6.0 0.0012 0.85
8 CL to LC 5.2 0.0002 0.14
17 Gravelly LC 5.7 0.0372 6.37
25 CL to LC 6.4 0.0014 1.88
27 SCL 4.4 0.0542 3.15
28 LC 6.8 0.0040 6.30
29 LC 4.9 0.0400 2.25


6A CL 4.3 0.0020 0.06
7A CL 4.7 0.0004 0.04
17A CL 4.6 0.0598 1.93
25A CL 4.3 0.0014 0.04
29A LC 5.0 0.0004 0.04

 Averages, soil type 1

 Farmed 5.7 0.0155 2.40
 Unfarmed 4.6 0.0128 0.42
 Farmed and unfarmed 5.3 0.0145 1.70

 Soil type 2: Mildly weathered soils on a recent river terrace


5 CL 6.4 0.0028 1.84
9 L (fat) 4.6 0.0274 1.41
10 S 4.2 0.0086 0.19
12 LS or SL 5.9 0.0088 1.95
13 CL 4.4 0.0200 1.12
14 L 4.1 0.0584 1.59
15 CL 4.9 0.0218 1.95
16 SL to LS 5.1 0.0130 1.12


5A CL 5.7 0.0004 0.12
9A L 5.2 0.0046 0.67

 Averages, soil type 2

 Farmed 5.0 0.0201 1.40
 Unfarmed 5.5 0.0025 0.40
 Farmed and unfarmed 5.1 0.0166 1.20

 Soil type 3: Moderately weathered soils on a late Pleistocene
 river terrace


1 M-K S 5.8 0.0016 0.24
2 M-K S 5.6 0.0006 0.11
3 S 4.6 0.0034 0.24
11 SL 5.9 0.0214 2.15
26 KSC 6.4 0.0008 0.39


11A SL 4.7 0.0024 0.07

 Averages, soil type 3

 Farmed 5.7 0.0056 0.63
 Unfarmed 4.7 0.0024 0.07
 Farmed and unfarmed 5.5 0.0050 0.53

 Soil type 4: Strongly weathered soils on triassic interbedded
 shales and sandstone: hilly topography


20 SL to L 4.7 0.0084 0.49
24 L with KS 6.0 0.0012 0.35


20A LS 4.3 0.0010 0.04
24A LS 4.3 0.0010 0.02

 Averages, soil type 4

 Farmed 5.3 0.0048 0.42
 Unfarmed 4.3 0.0010 0.03
 Farmed and unfarmed 4.8 0.0029 0.23

 Soil type 5: Strongly weathered soils on Triassic sandstone
 and sandy early, Pleistocene alluvium


18 S 4.5 0.0114 0.22
19 S 5.8 0.0026 0.31
21 KS to KLS 4.6 0.0160 0.56
22 KS 4.5 0.0048 0.14
23 KS +clay 5.2 0.0104 0.68


18A S 4.3 0.0016 0.02
19A LS 4.3 0.0012 0.02

 Averages, soil type 5

 Farmed 4.9 0.0090 0.38
 Unfarmed 4.3 0.0014 0.02
 Farmed and unfarmed 4.7 0.0069 0.28

 Averages, all soil types

 Farmed 5.3 0.0132 1.33
 Unfarmed 4.6 0.0063 0.26
 Farmed and unfarmed 5.1 0.0112 1.02

Sample Texture [C.sub.t] ECEC [Fe.sub.ox]
i.d. (g/kg) (cmol(+)/kg)

 Soil type 1: Strongly weathered soils on Triassic shale.
 Gently undulating topography


4 CL (with grit) 27.8 22.4 6300
6 CL to LC 23.2 33.7 6000
7 CL 28.5 20.4 5900
8 CL to LC 25.2 19.5 7300
17 Gravelly LC 18.4 22.2 3200
25 CL to LC 22.3 26.6 6700
27 SCL 19.2 13.6 5600
28 LC 30.5 38.8 5200
29 LC 17.7 15.1 4200


6A CL 29.8 10.8 5900
7A CL 21.6 14.3 5400
17A CL 21.7 5.3 2600
25A CL 26.3 6.5 3000
29A LC 18.8 8.7 3400

 Averages, soil type 1

 Farmed 23.6 23.6 5600
 Unfarmed 23.6 9.1 4060
 Farmed and unfarmed 23.6 18.4 5050

Soil type 2: Mildly weathered soils on a recent river terrace


5 CL 16.2 20.2 6800
9 L (fat) 9.6 8.8 2500
10 S 8.0 3.0 1500
12 LS or SL 12.5 13.5 2800
13 CL 17.1 13.3 5100
14 L 14.2 9.4 4500
15 CL 16.5 14.5 4400
16 SL to LS 11.4 11.0 2900


5A CL 30.7 18.1 5700
9A L 28.6 18.0 3500

 Averages, soil type 2

 Farmed 13.2 11.7 3812
 Unfarmed 29.6 18.1 4600
 Farmed and unfarmed 16.5 13.0 3970

Soil type 3: Moderately weathered soils on a late Pleistocene
 river terrace


1 M-K S 18.6 5.6 1000
2 M-K S 7.1 3.1 720
3 S 9.2 3.9 1100
11 SL 5.5 7.8 1700
26 KSC 9.3 12.1 2200


11A SL 16.2 4.0 1200

 Averages, soil type 3

 Farmed 9.9 6.5 1344
 Unfarmed 16.2 4.0 1200
 Farmed and unfarmed 11.0 6.1 1320

Soil type 4: Strongly weathered soils on triassic interbedded
 shales and sandstone: hilly topography


20 SL to L 22.3 9.0 3900
24 L with KS 18.1 10.6 2400


20A LS 39.2 7.4 1800
24A LS 31.7 6.7 2500

 Averages, soil type 4

 Farmed 20.2 9.8 3150
 Unfarmed 35.5 7.0 2150
 Farmed and unfarmed 27.8 8.4 2650

 Soil type 5: Strongly weathered soils on Triassic sandstone
 and sandy early, Pleistocene alluvium


18 S 8.5 2.1 540
19 S 12.8 5.4 660
21 KS to KLS 10.6 3.3 520
22 KS 11.2 1.9 480
23 KS +clay 14.4 5.9 410


18A S 13.4 1.0 460
19A LS 19.6 1.9 980

 Averages, soil type 5

 Farmed 11.5 3.7 522
 Unfarmed 16.5 1.5 720
 Farmed and unfarmed 12.9 3.1 579

 Averages, all soil types

 Farmed 16.1 13.0 3329
 Unfarmed 24.8 8.6 3037
 Farmed and unfarmed 18.6 11.7 3243

Sample Texture [Mn.sub.hq] [P.sub.t]
i.d. (mg/kg)

 Soil type 1: Strongly weathered soils on Triassic
 shale. Gently undulating topography


4 CL (with grit) 524 2379
6 CL to LC 283 3553
7 CL 617 2609
8 CL to LC 888 1420
17 Gravelly LC 99 1930
25 CL to LC 785 3623
27 SCL 1303 517
28 LC 166 7514
29 LC 34 829


6A CL 538 564
7A CL 1135 341
17A CL 154 142
25A CL 150 266
29A LC n.t. 266

 Averages, soil type 1

 Farmed 522 2708
 Unfarmed 494 316
 Farmed and unfarmed 514 1854

 Soil type 2: Mildly weathered soils on a recent
 river terrace


5 CL 194 1877
9 L (fat) 116 1270
10 S 25 684
12 LS or SL 125 1320
13 CL 172 1149
14 L 120 665
15 CL 179 1131
16 SL to LS 89 887


5A CL 204 612
9A L n.t. 799

 Averages, soil type 2

 Farmed 128 1123
 Unfarmed 204 706
 Farmed and unfarmed 136 1039

Soil type 3: Moderately weathered soils on a late
 Pleistocene river terrace


1 M-K S 584 988
2 M-K S n.t. 321
3 S 205 1298
11 SL 178 1140
26 KSC 643 1354


11A SL 874 162

 Averages, soil type 3

 Farmed 402 1020
 Unfarmed 874 162
 Farmed and unfarmed 497 877

Soil type 4: Strongly weathered soils on triassic
interbedded shales and sandstone: hilly topography


20 SL to L 11 1058
24 L with KS 14 1308


20A LS 28 184
24A LS 19 216

 Averages, soil type 4

 Farmed 12 1183
 Unfarmed 24 200
 Farmed and unfarmed 18 692

Soil type 5: Strongly weathered soils on Triassic
sandstone and sandy early, Pleistocene alluvium


18 S 12 875
19 S 7 919
21 KS to KLS 11 907
22 KS 7 453
23 KS +clay 11 454


18A S 6 17
19A LS 7 31

 Averages, soil type 5

 Farmed 10 722
 Unfarmed 6 24
 Farmed and unfarmed 9 522

 Averages, all soil types

 Farmed 269 1532
 Unfarmed 312 300
 Farmed and unfarmed 277 1172

n.t., Not tested for this property because there was insufficient

Table 2. Correlations (x 100) among [log.sub.10]-transformed values of
selected properties for all 41 soils, with separate values for the 29
farmed and 12 unfarmed soils in parentheses

Values in bold are significant at P=0.05. Raw data and a brief
description of the methods are in Table 1

 [Cd.sub.Ca] [Cd.sub.t] pH

[Cd.sub.t] 67# (59#, 76#)
pH -31# (-59#, -17) 43# (19, 45)
[C.sub.t] -36# (-27, -10) -17 (27, 14) 3 (31, 9)
ECEC -9 (-18, -18) 53# (61#, 39) 59# (57#, 60#)
[Fe.sub.ox] -10 (-9, -16) 33# (50#, 33) 26 (26, 47)
[Mn.sub.hq] -19 (-29, 2) 23 (26, 40) 31# (34#, 41)
[P.sub.t] 8 (-35, -11) 68# (39#, 40) 65# (66#, 55#)

 [C.sub.t] ECEC [Fe.sub.ox]

ECEC 50# (73#, 67#)
[Fe.sub.ox] 57# (66#, 60#) 87# (88#, 93#)
[Mn.sub.hq] 25 (36#, 9) 63# (64#, 69#) 68# (68#, 69#)
[P.sub.t] 1# (59#, 65#) 72# (76#, 97#) 47# (61#, 91#)


[P.sub.t] 39# (47#, 45)

Note: Values in bold indicated with #.


We thank: the 29 vegetable growers who allowed us to sample soil from their farms; Graeme Poile, Albert Oates, Eleanor Fairbairn-Wilson, Lisa Phillips, and Paula Goodall for assistance; Mike McLaughlin, Kay Raulach, and Debbie Rae for constructive comments on the manuscript; and Fredi Celardin and Sebastian Sauve for clarification of aspects of their published work. Neel Jinadasa and Paul Milham were supported by postgraduate scholarships from the University of Western Sydney. The research was funded jointly by NSW Agriculture and the Rural Industry Research and Development Corporation (Grant UWS 4A).


Agbenin JO (2003) The distribution and transformation of iron and manganese in soil fractions in a savannah Alfisol under continuous cultivation. Nutrient Cycling in Agroecosystems 66, 259-270. doi: 10. 1023/A:1024453315497

Ainsworth CC, Pilon JL, Gassman PL, van der Sluys WG (1994) Cobalt, cadmium and lead sorption to hydrous iron oxide: residence time effect. Soil Science Society of America Journal 58, 1616-1623.

Alloway BJ (1990) Cadmium In 'Heavy metals in soils.' (Ed. BJ Alloway) pp. 100-124. (John Wiley and Sons: New York)

Amacher MC, Kotuby-Amacher J, Selim HM, Iskandar IK (1986) Retention and release of metals by soils-evaluation of several models. Geoderma 38, 131-154. doi: 10.1016/0016-7061 (86)90011-X

Anderson PR, Christensen TH (1988) Distribution coefficients of Cd, Co, Ni and Zn in soils. Journal of Soil Science 39, 15-22.

Andrewes P, Town RM, Hedley MJ, Loganathan P (1996) Measurement of plant-available cadmium in New Zealand soils. Australian Journal of Soil Research 34, 441-452.

Anon. (1996) Review of the maximum permitted concentrations for Cd in food, Information Summary, Proposal P144. (Australia and New Zealand Food Authority: Canberra)

Axe L, Anderson PR (1997) Experimental and theoretical diffusivities of Cd and Sr in hydrous ferric oxide. Journal of Colloid and Interface Science 185, 436-448. doi: 10.1006/JCIS.1996.4635

Backes CA, McLaren RG, Rate AW, Swift RS (1995) Kinetics of cadmium and cobalt desorption from iron and manganese oxides. Soil Science Society of America Journal 59, 778-785.

Baldock JA, Skjemstad JO (1999) Soil organic carbon/organic matter. In 'Soil analysis an interpretation manual'. (Eds KI Peverill, LA Sparrow, DJ Reuter) pp. 159-170. (CSIRO Publishing: Melbourne, Vic.)

Barrow NJ (1987) 'Reactions of variable charge soils.' (Martinus Nijhoff Publishers: Dordrecht, The Netherlands)

Barrow NJ (1993) Effects of surface heterogeneity on ion adsorption by metal oxides and by soils. Langmuir 9, 2606-2611.

Barrow NJ, Gerth J, Bruemmes GW (1989) Reaction kinetics of the adsorption and desorption of nickel, zinc and cadmium by goethite. II Modelling the extent and rate of reaction. Journal of Soil Science 40, 437-450.

Barrow NJ, Shaw TC (1979) Effects of soil : solution ratio and vigour of shaking on the rate of phosphate adsorption by soil. Journal of Soil Science 30, 67-76.

Basta NT, Pantone DJ, Tabatabai MA (1993) Path analysis of heavy metal adsorption by soil. Agronomy Journal 85, 1054-1057.

Benedetti MF, Van Riemsdijk WH, Koopal LK, Kinniburgh DG, Goody DC, Milne CJ (1996) Metal ion binding by natural organic matter: from the model to the field. Geochimica et Cosmochimica Acta 60, 2503-2513. doi: 10.1016/0016-7037(96)00113-5

Benitez LN, Dubois J-P (1999) Evaluation of the selectivity of sequential extraction procedures applied to the speciation of cadmium in soils. International Journal of Environmental Analytical Chemistry 74, 289-303.

Boekhold AE, Temminghoff EJM, van der Zee SEATM (1993) Influence of electrolyte composition and pH on Cd sorption by an acid sandy soil. Journal of Soil Science 44, 85-96.

Boekhold AE, van der Zee SEATM (1990) Plant uptake of cadmium as affected by variation in sorption parameters. In 'Plant nutrition--physiology and applications'. Developments in Plant and Soil Sciences 41. (Ed. ML van Beusichem) pp. 313-316. (Kluwer Academic Publishers: Boston, MA)

Bolan NS, Naidu R, Khan MAR, Tillman RW, Syers JK (1999) The effects of anion sorption on sorption and leaching of cadmium. Australian Journal of Soil Research 37, 445-460.

Bolton KA, Evans LJ (1996) Cadmium adsorption capacity of selected Ontario soils. Canadian Journal of Soil Science 76, 183-189.

Buchter B, Davidoff B, Amacher MC, Hinz C, Iskandar IK, Selim HM (1989) Correlation of Freundlich Kd and n retention parameters with soils and elements. Soil Science 148, 370-379.

Calvet R, Bourgeois S, Msaky JJ (1990) Some experiments on extraction of heavy metals present in soil. International Journal of Environmental Analytical Chemistry 39, 31-45.

Capodaglio G, Turetta C, Toscano G, Gambaro A, Scarponi G, Cescon P (1998) Cadmium, lead and copper complexation in Antarctic coastal seawater--evolution during the Austral summer. International Journal of Environmental Analytical Chemistry 7, 195-226.

del Castilho P, Chardon WJ (1995) Uptake of soil cadmium by three field crops and its prediction by a pH--dependent Freundlich sorption model. Plant and Soil 171, 263-266.

del Castilho P, Chardon WJ, Salomons W (1993a) Influence of cattle-manure slurry application on the solubility of cadmium, copper and zinc in a manured, acidic, loamy-sand soil. Journal of Environmental Quality 22, 689-697.

del Castilho P, Dalenberg JW, Brunt K, Bruins AP (1993b) Dissolved organic matter, cadmium, copper and zinc in pig slurry-and soil solution-size exclusion chromatography fractions. International Journal of Analytical Chemistry 50, 91-107.

Celardin F (1999) Semi-empirical correlations between so-called mobilizable and mobile concentrations of heavy metals in soils. Communications in Soil Science and Plant Analysis 30, 843-854.

Chang TW, Wang MK (2002) Assessment of sorbent ratio effect on adsorption using dimensional analysis and batch experiments. Chemosphere 48, 419-426. doi: 10.1016/S0045-6535(02)00053-X

Christensen TH (1987) Cadmium sorption at low concentrations: V. Evidence of competition by other heavy metals. Water, Air, and Soil Pollution 34, 293-303.

Cook RD, Weisberg S (1982) 'Residuals and influence in regression.' (Chapman and Hall: London)

Cowan CE, Zachara JM, Smith SC, Resch CT (1992) Individual sorbent contributions to cadmium sorption on Ultisols of mixed mineralogy. Soil Science Society of America Journal 56, 1084-1094.

Degryse F, Broos K, Smolders E, Merckx R (2003) Soil solution of Cd and Zn can be predicted with a Ca[Cl.sub.2] soil extract. European Journal of Soil Science 54, 149-157. doi: 10.1046/J.13652389.2003.00503.X

Echeverria JC, Morera MT, Mazkiaran C, Garrido JJ (1998) Competitive sorption of heavy metals by soils. Isotherms and fractional factorial experiments. Environmental Pollution 101, 275-284.

Eklund G, Oskarsson A (1999) Exposure of cadmium from infant formulas and weaning foods. Food Additives and Contaminants 16, 509-519. doi: 10.1080/026520399283650

El-Falaky AA, Aboulroos SA, Lindsay WL (1991) Measurement of cadmium activities in slightly acidic to alkaline soils. Soil Science Society of America Journal 55, 974-979.

Elzinga EJ, van Grinsven JJM, Swartjes FA (1999) General purpose Freundlich isotherms for cadmium, copper and zinc in soils. European Journal of Soil Science 50, 139-149. doi: 10.1046/J.13652389.1999.00220.X

Franchi A, Davis AP (1997) Desorption of Cd (II) from artificially contaminated sediments. Water, Air, and Soil Pollution 100, 181-196, doi: 10.1023/A:1018392022843

Garcia-Mimgaya J, Page AL (1976) Influence of ionic strength and inorganic complex formation on the sorption of trace amounts of Cd by montmorillonite. Soil Science Society of America Journal 40, 658-663.

Ge Y, Murray P, Hendershot WH (2000) Trace metal speciation and bioavailability in urban soils. Environmental Pollution 107, 137-144. doi: 10.1016/S0269-7491(99)00119-0

Gerritse RG, van Driel W 0984) The relationship between adsorption of trace metals, organic matter and pH in temperate soils. Journal of Environmental Quality 13, 197-204.

Gillman GP (1979) A proposed method for the measurement of exchange properties of highly weathered soils. Australian Journal of Soil Research 17, 129-139.

Gray CW, McLaren RG, Roberts AHC, Condron CM (1999) Solubility, sorption and desorption of native and added cadmium in relation to properties of soils in New Zealand. European Journal of Soil Science 50, 127-137. doi: 10.1046/J.1365-2389.1999. 00221.X

de Haan FAM, van der Zee SEATM, van Riemsdijk 0987) The role of soil chemistry and soil physics in protecting soil quality: variability of sorption and transport of cadmium as an example. Netherlands Journal of Agricultural Science 35, 347-359.

Hamon RE, McLaughlin MJ, Cozens G (2002) Mechanisms of metal availability in in situ remediation treatments. Environmental Science and Technology 36, 3991-3996. doi: 10.1021/ES025558G

Hamon RE, McLaughlin MJ, Naidu R, Correll R (1998) Long-term changes in cadmium bioavailability in soil. Environmental Science and Technology 32, 3699-3703. doi: 10.1021/ES980198B

Harper ME Davison W, Zhang H, Tych W (1998) Kinetics of metal exchange between solids and solutions in sediments and soils interpreted from DGT measured fluxes. Geochimica et Cosmochimica Acta 62, 2757-2770. doi: 10.1016/S00167037(98)00186-0

Hartwig A, Asmuss M, Ehleben I, Herzer U, Kostelac D, Pelzer A, Schwerdtle T, Burkle A (2002) Interference by toxic metal ions with DNA repair processes and cell cycle control: molecular mechanisms. Environmental Health Perspectives 110, 797-799.

He QB, Singh BR (1993) Plant availability of cadmium in soils. I. Extractable cadmium in newly and long-term cultivated soils. Acta Agriculturae Scandinavica, Section B. Soil and Plant Science 43, 134-141.

Hendrickson LL, Corey RB (1981) Effect of equilibrium metal concentrations on apparent selectivity coefficients of soil complexes. Soil Science 131, 163-171.

Hesterberg D, Bril J, del Castilho P 0993) Thermodynamic modelling of zinc, cadmium and copper solubilities in a manured, acidic loamy-sand topsoil. Journal of Environmental Quality 22, 681-688.

Ho MD, Evans GJ (2000) Sequential extraction of metal contaminated soils with radiochemical assessment of readsorption effects. Envirionmental Science and Technology 34, 1030-1035. doi: 10.1021/ES981251Z

Holm PE, Andersen S, Christensen TH (1995a) Speciation of dissolved cadmium: interpretation of dialysis, ion exchange and computer (GEOCHEM) methods. Water Research 29, 803-809. doi: 10.1016/0043-1354(94)00205-L

Holm PE, Christensen TH, Tjell JC, McGrath SP (1995h) Speciation of cadmium and zinc with application to soil solutions. Journal of Environmental Quality 24, 183-190.

Holm PE, Rootzen H, Borggaard OK, Moberg JP, Christensen TH (2003) Correlation of cadmium distribution coefficients to soil characteristics. Journal of Environmental Quality 32, 138-145.

Holmgren GGS, Meyer MW, Chaney RL, Daniels RB (1993) Cadmium, lead, zinc, copper and nickel in agricultural soils of the USA. Journal of Environmental Quality 22, 335-348.

Houba VJG, Lexmond ThM, Novozamski I, van der Lee JJ (1996) State of the art and future developments in soil analysis for bioavailability assessment. The Science of the Total Environment 178, 21-28. doi: 10.1016/0048-9697(95)04793-X

Impellitteri CA, Lu Y, Saxe JK, Allen HE, Peijnenburg WGJM (2002) Correlation of the partitioning of dissolved organic matter fractions with the desorption of Cd, Cu, Ni, Pb and Zn from 18 Dutch soils. Environment International 28, 401-410. doi: 10.1016/S0160-4120(02)00065-X

Jeng AS, Singh BR (1993) Partitioning and distribution of cadmium and zinc in selected cultivated soils in Norway. Soil Science 156, 240-250.

Jensen H, Mosbaek H (1990) Relative availability of 200 years old cadmium from soil to lettuce. Chemosphere 20, 693-702. doi: 10.1016/0045-6535(90)90160-U

Jinadasa KBPN, Milham PJ, Hawkins CA, Cornish PS, Williams PA, Kaldor CJ, Conroy JP (1997) Survey of cadmium levels in vegetables and soils of Greater Sydney, Australia. Journal of Environmental Quality 26, 924-933.

Johnson BB (1990) Effect of pH, temperature and concentration on the adsorption of cadmium on goethite. Environmental Science and Technology 24, 112-118.

Jopony M, Young SD (1994) The solid solution-equilibria of lead and cadmium in polluted soils. European Journal of Soil Science 45, 59-70.

Kinniburgh DG, Riemsdijk WH, Koopal LK, Borkovec M, Benedetti MF, Aventa MJ (1999) Ion binding to organic matter: competition, heterogeneity, stoichiometry and thermodynamic consistency. Colloids and Surfaces A: Physicochemical and Engineering Aspects 151, 147-166. doi: 10.1016/S0927-7757 (98)00637-2

Krishnamurti GSR, Huang PM, van Rees KCJ, Kozak LM, Rostad HPW (1995) A new soil test method for the determination of plant-available cadmium in soils. Communications in Soil Science and Plant Analysis 26, 2857-2867.

Kretzschmar R, Borkovec M, Grolimund D, Elimelech M (1999) Mobile subsurface colloids and their role in contaminant transport. Advances in Agronomy 66, 121-193.

Kuo S, McNeal BL (1984) Effects of pH and phosphate on cadmium sorption by a hydrous ferric oxide. Soil Science Society of America Journal 48, 1040-1044.

Lee S-Z, Allen HE, Huang CP, Sparks DL, Sanders PF, Peijnenburg WJGM (1996) Predicting soil-water partition coefficients for cadmium. Environmental Science and Technology 30, 3418-3424. doi: 10.1021/ES9507933

Long SE, Martin TD (1989) 'Determination of trace elements in water and wastes by ICP-MS, version 4.0.' (Office of Research and Development, USEPA: Cincinnati, OH)

Lores EM, Pennock JR (1998) The effect of salinity on binding of Cd, Cr, Cu and Zn to dissolved organic matter. Chemosphere 37, 861-874. doi: 10.1016/S0045-6535(98)00090-3

Ma YB, Uren NC (1998) Transformation of heavy metals added to soil--application of a new sequential extraction procedure. Geoderma 84, 157-168. doi: 10.1016/S0016-7061(97)00126-2

Mann SS (1985) The influence of aluminium on iron oxides: XII. High-resolution transmission electron microscopic (HTRM) study of aluminous goethites. Clay Minerals 20, 255-262.

Mann SS, Ritchie GSP (1994) Changes in the forms of cadmium with time in some Western Australian soils. Australian Journal of Soil Research 32, 241-250.

McArthur DFE, Huang PM, Kozak LM (2001) Impact of long-term cultivation on the status of organic matter and cadmium in soil. Canadian Journal of Soil Science 81, 349-355.

McBride M, Sauve S, Hendershot W (1997) Solubility control of Cu, Zn, Cd and Pb in contaminated soils. European Journal of Soil Science 48, 337-346.

McKeague JE, Brydon JE, Miles NM (1971) Differentiation of forms of extractable iron and aluminium in soils. Soil Science Society of America Proceedings 35, 33-38.

McLaren RG, Backes CA, Rate AW, Swift RS (1998) Cadmium and cobalt desorption kinetics from soil clays: effect of sorption period. Soil Science Society of America Journal 62, 332-337.

McLaughlin MJ, Tiller KG, Naidu R, Stevens DP (1996) Review: the behaviour and environmental impact of contaminants in fertilisers. Australian Journal of Soil Research 34, 1-54.

McLaughlin MJ, Maier NA, Rayment GE, Sparrow LA, Berg G, McKay A, Milham PJ, Merry RH, Smart MK (1997a) Cadmium in Australian potato tubers and soils. Journal of Environmental Quality 26, 1644-1649.

McLaughlin MJ, Tiller KG, Smart MK (1997b) Speciation of cadmium in soil solutions of saline/sodic soils and relationship with cadmium concentration in potato tubers (Solanum tuberosum L.). Australian Journal of Soil Research 35, 183-198. doi: 10.1071/S96032

Milham PJ, Vimpany IA (1987) Chemical methods. In 'BCRI soil testing methods and interpretation'. (Ed. TS Abbott) pp. 19-60. (NSW Agriculture: Orange, Australia)

Montgomery DC, Peck EA (1982) 'Introduction to linear regression analysis.' pp. 291-301. (John Wiley and Sons: New York)

Moorcroft S, Watt J, Thornton 1, Wells SJ, Stehlow CW, Barltrop D (1982) Composition of dusts and soils in an apparently uncontaminated rural village in southwest England--implications to human health. In 'Trace substances in environmental health XVI'. (Ed. DD Hemphill) pp. 155-163. (University of Missouri: Columbia, MO)

Murphy EM, Zachara JM (1995) The role of humic substances on the distribution of organic and inorganic contaminants in groundwater. Geoderma 67, 103-124. doi: 10.1016/0016-7061(94) 00055-F

Naidu R, Bolan NS, Kookana RS, Tiller KG (1994) Ionic-strength and pH effects on the sorption of cadmium and the surface charge of soils. European Journal of Soil Science 45, 419-429.

Nakahone LN, Young SD (1993) The significance of (radio-) labile cadmium pools in soils. Environmental Pollution 82, 73-77.

Norrish K, Rosser H (1983) Mineral phosphate. In 'Soils: an Australian viewpoint'. (CSIRO Division of Soils) pp. 335-364. (CSIRO: Melbourne, Vic.)

Northcote KH (1979) 'A factual key for the recognition of Australian soils'. (Rellim Technical Publications: Glenside, S. Aust.)

Nimmo M, Fortes GR (1997) The potential pool of Co, Ni, Cu, Pb and Cd organic complexing ligands in coastal and urban rainwaters. Atmospheric Environment 31, 693-702. doi: 10.1016/S13522310(96)00243-9

O'Connor GA, O'Connor C, Cline GR (1984) Sorption of cadmium by calcareous soils: influence of soil composition. Soil Science Society of America Journal 48, 1244-1247.

Osborne GJ, Poile GJ, Hochman Z, Fisher J (1988) Manganese associations and va ations. In 'International Symposium on Manganese in Soils and Plants'. (Eds MJ Webb, RD Graham, RJ Hannam) pp. 11-13. (Manganese Symposium 1988 Inc.: Adelaide, S. Aust)

Palm V (1994) A model for sorption, flux and plant uptake of cadmium in a soil profile: model structure and sensitivity analysis. Water, Air, and Soil Pollution 77, 169-190.

Parfitt RL, Childs CW (1988) Estimation of forms of Fe and AI: a review and analysis of contrasting soils by dissolution and Moessbauer methods. Australian Journal of Soil Research 26, 121-144.

Peijnenburg W, Baerselman R, de Grool A, Jager T, Leenders D, Posthuma L, Van Veen R (2000) Quantification of metal bioavailability for lettuce in field soils. Archives of Environmental Contamination and Toxicology 39, 420-430. doi: 10.1007/ S002440010123

Pyman MAF, Bowden JW, Posner AM (1979) The movement of titration curves in the presence of specific adsorption. Australian Journal of Soil Research 17, 191-195.

Ramachandran V, D'Souza TJ (1999) Adsorption of cadmium by Indian soils. Water, Air, and Soil Pollution 111, 225-234. doi: 10.1023/A: 1005038325836

Rattan RK, Sehgal D (1989) Investigations on cadmium adsorption on soils differing in physical and chemical characteristics. Journal of the Indian Society of Soil Science 37, 284-289.

Rayment GE (1994) Cadmium in Queensland vegetables and associated soils. MAgricSc dissertation, University of Queensland, Australia.

Rayment GE, Higginson FR (1992) 'Handbook of soil and water chemical methods.' (Inkata Press: Melbourne, Vic.)

Rogers S (1996) Bioavailability of heavy metals in biosolids: implications for biosolid regulations. In 'Land application of wastes in Australia and New Zealand: research and practice'. Proceedings of the 14th Land Treatment Collective Meeting Canberra. (Eds PJ Polglase, WM Tunningly, RJ Hannam) pp. 42-49. (CSIRO Publishing: Melbourne, Vic.)

Romkens PFAM, Dolfing J (1998) Effect of Ca on the solubility and molecular size distribution of DOM and Cu binding in soil solution samples. Environmental Science and Technology 32, 363-369. doi: 10.1021/ES970437F

Romkens PFAM, Salomons W (1998) Cd, Cu and Zn solubility in arable and forest soils: consequences of land use changes for metal mobility and risk assessment. Soil Science 163, 859-871. doi: 10.1097/00010694-199811000-00003

Rudzinski W, Charmas R, Partyka S, Bottero JY (1993) On the nature of the energetic surface heterogeneity in ion adsorption at a water/oxide interface: theoretical studies of some special features of ion adsorption at low ion concentrations. Langmuir 9, 2641-2651.

Sarooshi RA, Cresswell GC, Tesoriero L, Milham PJ, Barchia I, Harris AM (2002) Effect of biosolids compost on two NSW coastal soils used to grow vegetables. Australian Journal of Soil Research 40, 761-774. doi: 10.1071/SR00107

Sanchez-Camazano M, Sanchez-Martin MJ, Lorenzo LE (1994) The content and distribution of cadmium in soils as influenced by soil properties. The Science of the Total Environment 156, 183-190.

Sauerbeck DR, Styperek P (1985) Evaluation of chemical methods for assessing the Cd and Zn availability from different soils and sources. In 'Chemical methods for assessing bioavailable metals in sludges and soils'. (Eds R Leschber, RD Davis, P L'Hermite) pp. 49-66. (Elsevier Applied Science Publishers: London)

Sauve S, Hendershot W, Allen HE (2000a) Soil-solution partitioning of metals in contaminated soils: dependence on pH, total metal burden and organic matter. Environmental Science and Technology 34, 1125-1131. doi: 10.1021/ES9907764

Sauve S, Norvell WA, McBride M, Hendershot W (2000b) Speciation and complexation of cadmium in extracted soil solutions. Environmental Science and Technology 34, 291-296. doi: 10.1021/ES990202Z

Schug B, HossT, During R-A, Gaith S (1999) Regionalization of sorption capacities for arsenic and cadmium. Plant and Soil 213, 181-187. doi: 10.1023/A:1004563731376

Schwertmann U (1964) Differenzierung des eisenoxyde des bodens durch extraction mit ammoniumoxalate luosung. Zeitschrift Jar Pflanzenernahrung Dungung und Bodenkunde 105, 194-202.

Sekaly ALR, Chakrabarti CL, Back MH, Gregoire DC, Lu JY, Schroeder WH (1999) Stability of dissolved metals in aqueous samples: Rideau River surface water, rain and snow. Analytica Chimica Acta 402, 223-231. doi: 10.1016/S0003-2670(99) 00529-2

Sherman GD, McHargue JS, Hodgkiss WS (1942) Determination of active manganese in soil. Soil Science 54, 253-257.

Sloan JJ, Dowdy RH, Dolan MS, Linden DR (1997) Long-term effects of biosolids applications on heavy metal bioavailability in agricultural soils. Journal of Environmental Quality 26, 966-974.

Smoulders E, Brans K, Foldi A, Merckx R (1999) Cadmium fixation in soils measured by isotopic dilution. Soil Science Society of America Journal 63, 78-85.

Spark KM, Johnson BB, Wells JD (1995) Characterising heavy-metal adsorption on oxides and oxyhydroxides. European Journal of Soil Science 46, 621-631.

Springob G, Bottcher J (1998a) Parameterization and regionalization of Cd sorption characteristics of sandy soils. I Freundlich type parameters. Zeitschrift fur Pflanzenerniihrung Dungung und Bodenkunde 161, 681-687.

Springob G, Bottcher J (1998b) Parameterization and regionalization of Cd sorption characteristics of sandy soils. I1 Regionalization: Freundlich k estimates by pedotransfer functions. Zeitschrift fur Pflanzenernahrung Dungung und Bodenkunde 161, 689-696.

Staessen JA, Roels HA, Emelianov D, Kuznetsova T, Thijs L, Vangronsveld J, Fagard R (1999) Environmental exposure to cadmium, forearm bone density and risk of fractures: prospective population study. Lancet 353, 1140--1144. doi: 10.1016/S01406736(98)09356-8

Sumner ME (1994) Measurement of soil pH: problems and solutions. Communications in Soil Science and Plant Analysis 25, 859-879.

Temminghoff EJM, Plette ACC, Van Eck R, Van Riemsdijk WH (2000) Determination of the chemical speciation of trace metals in aqueous systems by the Wageningen Dorman Membrane technique. Analytica Chimica Acta 417, 149-157. doi: 10.1016/S00032670(00) 00935-1

Temminghoff EJM, Van der Zee SEATM, de Haan FAM (1995) Speciation and calcium competition effects on cadmium sorption by sandy soil at various pHs. European Journal of Soil Science 46, 649-655.

Tessier A, Fortin D, Belzile N, De Vitre RR, Leppard GG (1996) Metal sorption to diagenic iron and manganese oxyhydroxides and associated organic matter: narrowing the gap between field and laboratory measurements. Geochimica et Cosmochimica Acta 60, 387-404. doi: 10.1016/0016-7037(95)00413-0

Trivedi P, Axe L (2000) Modelling Cd and Zn sorption on hydrous metal oxides. Environmental Science and Technology 34, 2215-2223. doi: 10.1021/ES991110C

Undabeytia T, Nir S, Rytwo G, Morillo E, Maqueda C (1998) Modelling adsorption-desorption processes of Cd on montmorillonite. Clays and Clay Minerals 46, 423-428.

Vimpany IV, Nicholls PJ, Milham PJ, Bradley J (1997) Reactive Fe controls the relative amounts of P extracted from acidic soils by NaHC[O.sub.3] and acidic fluoride. I. Soils without recent P additions. Australian Journal of Soil Research 35, 355-364. doi: 10.1071/S96058

Wang W-Z, Brusseau ML, Artiola JF (1997) The use of calcium to facilitate desorption and removal of cadmium and nickel in subsurface soils. Journal of Contaminant Hydrology 25, 325-336. doi: 10.1016/S0169-7722(96)00046-0

Weng L, Temminghoff EJM, Lofts S, Tipping E, Van Riemsdijk WH (2002) Complexation with dissolved organic matter and solubility control of heavy metals in a sandy soil. Environmental Science and Technology 36, 4804-4810. doi: 10.1021/ ES0200084

Whitten MG, Ritchie GSP (1991) Calcium chloride extractable cadmium as an estimate of cadmium uptake by subterranean clover. Australian Journal of Soil Research 29, 215-221.

Wilkins BJ, Brummel N, Loch JPG (1998) Influence of pH and zinc concentration on cadmium sorption in acid, sandy soils. Water, Air, and Soil Pollution 101, 349-362. doi: 10.1023/A: 1004909322206

Williams CH, David DJ (1976) The accumulation in soil of cadmium residues from phosphate fertilizers and their effect on the cadmium content of plants. Soil Science 121, 86-93.

Xia K (1997) Species-specific exchange reactions controlling solubility of trace elements in soil. PhD thesis, University of Wisconsin, Madison, WI.

Young SD, Tye A, Carstensen A, Resende L, Crout N (2000) Methods for determining labile cadmium and zinc in soil. European Journal of Soil Science 51, 129-136. doi: 10. 1046/J.1365-2389.2000. 00286.X

Zachara JM, Smith SC, McKinley JP, Resch CT (1993) Cadmium sorption on specimen and soil smectites in sodium and calcium electrolytes. Soil Science Society of America Journal 57, 1491-1501.

Zachara JM, Smith SC, Resch CT (1994) Influence of humic substances on [Co.sup.2+] sorption by a subsurface mineral separate and its minerologic components. Geochimica et Cosmochimica Acta 58, 553-566. doi: 10.1016/0016-7037(94)90488-X

Zachara JM, Smith SC, Resch CT, Cowan CE (1992) Cadmium sorption to soil separates containing layer silicates and iron and aluminium oxides. Soil Science Society of America Journal 56, 1074-1084.

Zanders JM, Hedley MJ, Palmer AS, Tillman RW, Lee J (1999) The source and distribution of cadmium in soils on a regularly fertilised hill-country farm. Australian Journal of Soil Research 37, 667-678.

van der Zee SEATM, van Riemsdijk WH (1987) Transport of reactive solutes in spatially variable soil systems. Water Resources Research 23, 2059-2069.

Manuscript received 2 February 2004, accepted 6 July 2004

P. J. Milham (A,B,J), K. B. P. N. Jinadasa (B), D. Collins (C), P. J. Nicholls (C), C. A. Hawkins (B), R. G. Wenzel (D), C. J. Kaldor (E), A. A. Senn (F), C. S. Humphris (G), J. Fabien (H), M. K. Conyers (I), K. Y Chan (I), P. Holford (B), and J. P. Conroy (B)

(A) NSW Agriculture, Agricultural Institute, Orange, NSW 2800, Australia.

(B) Centre for Horticulture and Plant Sciences, University of Western Sydney, LB 1797, Penrith South DC, NSW 1797, Australia.

(C) NSW Agriculture, Elizabeth Macarthur Agricultural Institute, PMB 8 Camden, NSW 2570, Australia.

(D) pacific Laboratory Medicine Services, Royal North Shore Hospital, St Leonards, NSW 2065, Australia.

(E) 77 Eddy Rd, Chatswood, NSW 2067, Australia.

(F) NSW Agriculture, LB11 Windsor, NSW 2756, Australia.

(G) 10 Thornbill Way, West Pennant Hills, NSW 2125, Australia.

(H) 25/8 Watergum Way, Greenacre, NSW 2190, Australia.

(I) NSW Agriculture, Agricultural Institute, LB Wagga Wagga, NSW 2650, Australia.

(J) Corresponding author. Email:
COPYRIGHT 2004 CSIRO Publishing
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2004 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Milham, P.J.; Jinadasa, K.B.P.N.; Collins, D.; Nicholls, P.J.; Hawkins, C.A.; Wenzel, R.G.; Kaldor,
Publication:Australian Journal of Soil Research
Date:Dec 1, 2004
Previous Article:Participatory soil quality assessment: the case of smallholder farmers in Ethiopian highlands.
Next Article:Complexation and release isotherm of arsenic in arsenic-humic/fulvic equilibrium study.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters