MURDOCH RESEARCH REPOSITORY

This is the author’s final version of the work, as accepted for publication following peer review but without the publisher’s layout or pagination.

The definitive version is available at http://www.publish.csiro.au/paper/SR08201.htm

Thiyagarajan, C., Phillips, I.R., Dell, B. and Bell, R.W. (2009) Micronutrient fractionation and plant availability in bauxite-processing residue sand. Australian Journal of Soil Research,

47 (5). pp. 518-528.

http://researchrepository.murdoch.edu.au/3328/

Copyright: © 2009 CSIRO.

It is posted here for your personal use. No further distribution is permitted.

Micronutrient fractionation and plant availability in bauxite-

processing residue sand

Chitdeshwari Thiyagarajan A B, I. R. Phillips C, B. Dell D, Richard W. Bell B E

A Department of Soil Science and Agricultural Chemistry, Tamil Nadu Agricultural

University, Coimbatore 641 003, Tamil Nadu, India.

B School of Environmental Science, Murdoch University, Murdoch, WA 6150, Australia.

C Alcoa World Alumina Australia, PO Box 172, Pinjarra, WA 6208, Australia.

D School of Biological Science and Biotechnology, Murdoch University, Murdoch, WA

6150, Australia.

Abstract

Bauxite-processing residue must be disposed of in specifically designed facilities for long-

term management. Consideration of alkalinity, salinity, sodium content, and poor nutritional

status is essential for successful rehabilitation of residue disposal areas (RDA). The aim of

this study was to examine the availability and distribution of the micronutrients, B, Cu, Fe,

Mn, and Zn, in (i) fresh bauxite-processing residue sand (particle size >150 μm) with and

without gypsum amendment, and (ii) aged residue sand from a 4-year-old rehabilitated RDA

that had received past gypsum and fertiliser addition. Samples of fresh residue sand from

India and Australia exhibited high alkalinity, high salinity, and sodicity. Gypsum addition

significantly lowered pH, soluble Na, and alkalinity. Aged residue sand had low levels of all

micronutrients, with low extractability for Zn and Mn followed by B, Cu, and Fe.

Fractionation showed that 30–78% of Zn and Mn and 40–60% of B existed in non-available

(residual) forms. The next most dominant fractions were the Fe and Mn oxide-bound and

carbonate-bound fractions. Plant-available fractions (i.e. exchangeable and organically

bound) contributed <1% of the total concentration. Total concentration was found to be a

reliable indicator for Zn, Cu, and B extractability but not for DTPA-extractable forms of Fe

and Mn. Leaf analysis of vegetation grown on aged residue sand indicated deficiencies of Mn

and B. Results demonstrated that bauxite-processing residue sand contained very low levels

of B, Mn, and Zn and these concentrations may be limiting to plant growth. Distribution of

micronutrients among chemical pools was significantly influenced by pH, organic carbon,

exchangeable Na, and alkalinity of residue. Nutrient management strategies that account for

the characteristics of residue sand need to be developed for residue rehabilitation.

Importantly, strategies to limit the conversion of nutrients to non-available forms are required

to minimise micronutrient disorders.

Keywords: alkalinity, bauxite residue sand, fractionation, micronutrient availability, spiking.

Introduction

Globally, >70 Mt of bauxite-processing residue is produced annually, and this production is

projected to increase over the coming decade with rising demand for aluminium (Anon.

2007). In Western Australia, Alcoa World Alumina Australia (Alcoa) separates residue into 2

main fractions: sand (>150 μm) and mud (<150 μm). Residue sand is used for constructing

residue disposal areas (RDAs) for long-term storage of residue mud, and the outer sand

embankments are progressively rehabilitated for managing dust emissions and improving

aesthetic value of operating and closed RDAs (Li 1998; Courtney and Timpson 2005).

Residue sand represents the primary growth medium for rehabilitating Alcoa’s RDAs.

However, successful establishment of vegetation in residue sand remains a challenge because

of its extreme chemical, physical, and microbiological properties (Meecham and Bell

1977; Gherardi and Rengel 2001, 2003a; Bell et al. 2003; Snars et al. 2003; Courtney and

Timpson 2005). Macro- and micronutrient deficiencies have been reported for plants growing

in residue sand, even after fertiliser addition (Bell et al. 1997; Jasper et al. 2000; Eastham and

Morald 2006). This has been attributed to the very low organic carbon and a dominance of

hydrous Fe and/or Al oxides, and the alkalinity of the residue, which strongly influences

nutrient reactions with the residue components (Ge et al. 2000; Soumaré et al. 2003).

Obtaining a better understanding of the mechanisms controlling the distribution of nutrients

in response to residue properties will help to develop sustainable fertiliser strategies in

achieving successful revegetation.

To date, much of the research on bauxite-processing residues has focused on the role of

residue mud in pollution control and micronutrient availability, with little published

information on the properties of residue sand. In Western Australia, residue sand represents

the primary growth medium for rehabilitating Alcoa’s RDAs. Consequently, more

information on nutrient availability in this material is required. The objectives of this study

were to: (1) quantify micronutrient forms using well-established fractionation schemes, (2)

use this information to identify potential micronutrient limitations to plant growth, (3) use

plant and soil analysis to verify that micronutrient limitations are manifested as nutrient

disorders in the field, and (4) identify the most important residue characteristics controlling

micronutrient availability in residue sand.

Materials and Methods

Residue sand sources

Residue sand was collected from aluminium refineries in Australia (Alcoa World Alumina

Australia’s Pinjarra Refinery) and India (MALCO Ltd, Mettur). Three samples were

collected from Alcoa: (1) untreated sand direct from the Bayer Process (FRS), (2) sand from

a 1-year-old rehabilitated site (AGARS1), and (3) sand from a 4-year-old rehabilitated site

(AGARS2). Representative portions of the 2 FRS materials were also mixed with gypsum at

a rate of 1% (w/w basis) (GARS). Comparing the results for fresh (i.e. no gypsum, FRS) and

gypsum-amended (GARS) residue sand should identify the role of cations (Ca v. Na) and

alkalinity on nutrient availability.

Due to the very low (<0.5 mg/kg) micronutrient concentrations in the residue sands, samples

were spiked at rates equivalent to those levels added in the field (per ha mixture of 15 kg Zn,

15 kg Mn, and 10 kg Cu as the sulfate forms of fertilisers and 1.5 kg B as borax incorporated

to a depth of 0.25 m). Spiking was undertaken by placing 1 kg of air-dry (<2 mm) residue

sand into a polythene bag, applying the micronutrients in solution form, and mixing

thoroughly. The bags were moistened to field capacity using triple-deionised water (TDI) and

incubated aerobically at room temperature for a week. The moisture content was monitored

daily by weighing and TDI was added as required. After 7 days, the samples were air-dried

and analysed for micronutrients and basic physical and chemical properties as outlined

below.

Soil analysis

Particle-size analysis of the residue sand sources was determined using the pipette method

(Gee and Bauder 1987). Water-soluble carbonate, bicarbonate, and total alkalinity of residue

sand sources were estimated in 1 : 5 residue sand/deionised water extracts by titrating against

standard acid (0.005 m HCl) and alkali (0.50 m NaOH) to a pH equivalence point of 8.30 and

4.50, respectively, using mixed indicators (Rayment and Higginson 1992). All micronutrient-

spiked and non-spiked residue sand samples were extracted with 0.005 m DTPA–TEA–

calcium chloride extractant (pH adjusted to 7.3, Lindsay and Norvell 1978) and analysed for

Zn, Fe, Mn, Cu, and Ni (data not shown). Hot 0.01 m CaCl2 extractable B was determined in

1 : 2 residue to extractant suspension after refluxing for 30 min (Bingham 1982). Total

micronutrients were extracted by aqua regia digestion (1 : 3 HNO3 to HCl) (Melaku et al.

2005). All filtrates were analysed for micronutrient concentrations using ICP-AES.

Sequential fractionation of micronutrients

Sequential chemical fractionation of micronutrients in residue sands was carried out as

described in Benitez and Dubois (1999) for micronutrient metals. Residue sand (1 g) was

weighed into a 50-mL polypropylene centrifuge tube and each fraction was determined

operationally (Table 1). The supernatant from each step was evaporated to near dryness and

diluted with 10 mL of 1 m nitric acid. For B fractionation, the sequential extraction procedure

outlined by Datta et al. (2002) was used (Table 2).

All the samples were extracted in triplicate and analysed for their micronutrient

concentrations using ICP-AES. Simple correlation analysis was performed to examine the

relationship between micronutrient fractions, their availability, and residue properties.

Plant and soil sampling from revegetated areas

The relationship between micronutrient availability in residue sand and plant nutrient status

was investigated by sampling leaves and sand from the 2003 rehabilitated area at Alcoa’s

Pinjarra refinery. Rehabilitation established in 2003 had received 225 t gypsum/ha

incorporated to a depth of ~1.5 m and inorganic fertiliser at 2.57 t/ha incorporated to a depth

of ~0.25 m. The inorganic fertiliser was principally di-ammonium phosphate (DAP) at

1500 kg/ha (supplying 300 kg P and 265 kg N/ha) plus granulated K2SO4 at 300 kg K/ha;

CuSO4(10 kg Cu/ha); granulated ZnSO4 at 16 kg Zn/ha; MgSO4 at 30 kg Mg/ha; granulated

MnSO4 at 15 kg Mn/ha; NaMoO4 at 0.25 kg Mo/ha; and granulated borax at 1.5 kg B/ha.

Residue rehabilitation used many of the native species found in coastal south-west Western

Australia. Of the >60 species present, 4 were chosen to reflect differences in growth form and

physiology, and a range of leaf symptoms indicative of nutrient deficiencies. The plants

selected were Hardenbergia comptoniana (Hc), a vigorous creeping ground cover

legume; Acacia cyclops (Ac), a legume shrub; Grevillea crithmifolia (Gc), a proteaceous

shrub; and Eucalyptus gomphocephala (Eg), a tree.

Plants were sampled from both the upper and lower half of the embankment. Recently

matured leaf blades from 15–25 plants were sampled for each species (Reuter and Robinson

1997). If a species displayed potential nutrient-deficient symptoms, separate leaf samples

were taken from plants with and without symptoms for comparison and to provide a guide to

critical nutrient levels because deficiency criteria for coastal species of Western Australia are

generally not available. Plant samples were initially washed then oven-dried at 70°C and

ground before N, P, K, S, Na, Ca, Mg, Cl, Cu, Zn, Mn, Fe, and B analysis. Residue sand

samples were taken from 0–0.10 m depth from under plant canopies and in gaps between

canopies. Soil samples from each sampling location (n = 10) were bulked and air-dried before

analysis. Soil extractions were undertaken using standard methods for: total N (Bremner and

Mulveny 1982), NO3-N, NH4-N (Rayment and Higginson 1992); Colwell extractable P

(Rayment and Higginson 1992) and K; KCl-extractable S (Rayment and Higginson 1992);

organic carbon (Rayment and Higginson 1992); electrical conductivity (EC, 1 : 5); pH in

CaCl2 and water (1 : 5); DTPA-extractable Cu, Zn, Fe, and Mn (Lindsay and Norvell 1978);

phosphorus retention index (PRI, Rayment and Higginson 1992); exchangeable Al, Ca, Mg,

Na, and K (Tucker 1985); hot-CaCl2 extractable B (Bingham 1982). Simple correlation and

regression studies were performed to relate the influence of residue properties on

micronutrient transformation and availability.

Results and discussion

Characterisation of residue sand sources

Samples from both Australia and India were predominantly sand (>950 g/kg), although the

Indian residue sand had higher silt and clay content (>20 g/kg) than the Australian source

(>8.0 g/kg). The water content at field capacity varied between 110 and 161 g/kg. The Indian

sample retained more water than the Australian samples due to its higher proportion of fine

material. The fresh residue sand samples from both Australia and India were extremely

alkaline, saline, and sodic (Table 3).

Adding gypsum lowered the initial pH and EC irrespective of the residue source. In the

Indian sample, gypsum addition reduced pH from 10.3 to 9.79 and in the Australian sample

from 10.3 to 9.80. This reduction may be attributed to precipitation of hydroxides and

carbonates by Ca2+ ions (Barrow 1982; Bell et al. 1997; Courtney and Timpson 2005).

Samples from 1- and 4-year-old rehabilitation exhibited a lower pH (7.98 and 8.18), which

reflects the effects of past gypsum incorporation and rainfall leaching on removal of water-

soluble alkalinity. Gypsum addition produced a 4-fold reduction in EC relative to fresh

samples, presumably via the formation of insoluble calcium carbonates and hydroxides. The

EC1 : 5 varied between 0.10 and 1.70 dS/m and ageing reduced the salt status to favourable

levels for plant growth. Values of EC1 : 5 were consistent with those reported in other residue

characterisation studies (Meecham and Bell 1977; Bell et al. 1997; Courtney and Timpson

2004, 2005).

Organic carbon content was very low in both the Australian and Indian samples (<4.0 g/kg),

but the AGARS1 samples had higher values than fresh sand. This may be attributed to

additions of poultry manure (from past management programs), wood mulch used for dust

suppression, and recycled plant litter.

Samples of fresh residue sand from India and Australia exhibited very high total water-

soluble alkalinity (61.8 and 56.6 g CaCO3/kg). These concentrations are 6-fold higher than

the optimal range reported for plant growth (<10 g CaCO3/kg, Bell 1981). Carbonate

represented >85% of total water-soluble alkalinity of the FRS, while bicarbonate was the

dominant source of alkalinity in GRS. Gypsum dissolution and subsequent precipitation of

CO3 as CaCO3 is considered the most likely cause for the decline in water-soluble carbonate

(Barrow 1982;Courtney et al. 2003). Aged residue sand exhibited water-soluble alkalinity

levels 10–15 times lower than fresh samples, with bicarbonate being the dominant inorganic

carbon ion. Water-soluble carbonate alkalinity was nil in aged residue sand sources, which

indicated the presence of sufficient Ca ions supplied by gypsum to precipitate CO3, in

addition to removal by drainage water (Courtney and Timpson 2004; Xenidis and Harokopou

2005).

Plant-available micronutrients

With the exception of Fe, micronutrient concentrations were well below levels regarded as

inducing plant deficiencies (Munshower 1994; Gherardi and Rengel 2001). This finding

supports the practice of micronutrient addition for rehabilitation of bauxite-processing residue

areas. However, no significant increase in concentration of extractable micronutrients was

observed, with additions equivalent to current field application rates of B, Cu, Mn, and Zn

(Table 4). Gypsum amendment reduced B and Cu extractability by more than 40 and 14%,

respectively, while it increased the extractability of Mn, Fe, and Zn in the Australian residue

sand sample. The increased extractability of Fe, Mn, and Zn following gypsum addition may

be associated with the removal of these cations from weakly held forms associated with the

carbonate-bound fraction.

Negligible organic matter, coupled with abundant carbonate and hydrous Fe and Al oxides

and hydroxides, suggests that inorganic reactions may be largely responsible for poor

availability of micronutrients (Fuller and Richardson 1986; Gherardi and Rengel 2003c).

Extractable levels of all micronutrients were lower in 4- than 1-year-old sand and the

decrease in extractability was 20% for Fe, 50% for Cu, 92% for Zn, 89% for Mn, and 58%

for B. This suggests that declining nutritional status over time may be an issue with respect to

optimal plant growth.

Manganese deficiency has been cited as an important limiting factor for vegetative plant

growth on residue sand by many workers (Gherardi and Rengel

2001, 2003a, 2003b; Courtney and Timpson 2004); however, the current study indicates that

in addition to Mn, deficiencies of Zn and B may also need to be addressed in the

rehabilitation of RDAs. The Indian sample had higher extractable Mn, Cu, and B, while the

Australian sample was richer in Zn and Fe, suggesting that bauxite ore mineralogy has a

significant bearing on the levels of micronutrients in residues. Correlation analysis of

extractable micronutrients with residue properties showed that with the exception of Fe, all

micronutrients were negatively correlated with increasing pH. Poor correlation was observed

between total Mn and Fe and DTPA-extractable forms of these elements. Despite this, total

concentration was found to be a reliable indicator for Zn, Cu and B extractability. Alkalinity

of the residue sand was negatively correlated with Zn and Cu but was not related to levels of

other micronutrients. More than 45% of variation in Zn and Cu extractability was explained

by total alkalinity and exchangeable Na. Despite the low levels of organic matter in the

residue, there was a positive and significant correlation of organic carbon with Zn, Mn, and B

availability.

Total micronutrients

Iron was the most abundant micronutrient element followed by Mn > Cu > Zn > B. With the

exception of B, the Indian samples contained much higher total micronutrient concentrations

(Fe, Mn, Zn, Cu) than the Australian samples. Gypsum addition increased total Zn while it

reduced Cu content. The increase in Zn may be directly related to gypsum addition since the

gypsum used in this study contained Zn (see Table 3). Except for B, all micronutrients were

higher in 1- than 4-year-old residue sand, which may be due to the recent fertiliser input and

less-complete plant uptake.

Sequential fractions of micronutrients

Zinc

Zinc fractionation followed the order: residual > Fe and Mn oxide bound > organic matter

bound = carbonate bound > exchangeable fraction. The contribution of the residual fraction

to total Zn was >75% in the Indian source sand and <40% in Australian source (Table 5).

The predominance of Zn in the residual fraction has been reported by other workers for soils

(Yasrebi et al. 1994; Ma and Uren 1995). Ma and Uren (1995) found that although Zn was

primarily concentrated in the residual fraction, it was also present at significant

concentrations in the Fe–Mn oxide bound fractions. Higher concentrations of Zn in Fe and

Mn oxide fraction were associated with aged residue sand followed by fresh and gypsum-

amended Indian samples. Fe and Mn oxide bound Zn accounts for >60 and 20% of total Zn in

fresh and gypsum-amended Australian residue sand, respectively.

Gypsum addition produced an inconsistent change of Zn levels in the Fe oxide fraction, but

did increase Zn in the Mn oxide bound fraction. Preferential sorption of Zn by Fe and Mn

oxides under alkaline pH has been reported by others for soils (Shuman 1985; Luo and

Christie 1998; Silveira et al. 2006). A positive and significant correlation was found between

Fe and Mn oxide bound Zn with pH (r = 0.793**, 0.816**), organic carbon (r = 0.696**,

0.655**), and total alkalinity (r = 0.471**, 0.509**). Levels of Zn bound to organic matter

ranged from 1.67 to 6.67 mg/kg. These values are regarded as being very low, and contribute

to <1% of the total Zn in the Indian sample and 5–6% in the Australian sample. However, the

extractability of organically bound Zn declined between 1- and 4-year-old sand sources,

possibly through re-distribution into less-available fractions. Zn bound to carbonates ranged

from 2.30 to 34.3 mg/kg, with highest concentrations in the 1-year-old residue sand. These

results indicate that Zn has a moderate affinity for carbonate surfaces, particularly

CaCO3and/or it has precipitated as Zn carbonate (Prasad et al. 2006).

A strong positive correlation was found between carbonate bound and Fe–Mn oxide bound

fractions of Zn (r = 0.914** and 0.943**, Table 6). For non-spiked samples, concentrations

of Zn in the exchangeable fraction were negligible (<0.02 mg/kg). These findings are

consistent with those reported by Luo and Christie (1998). Low extractability of Zn from

carbonate and exchangeable fractions might be due to higher Fe and Mn oxide bound Zn,

which acts as major active sites for Zn sorption under alkaline pH (Hseu 2005; Silveira et al.

2006). A positive and significant correlation was observed with organic carbon (r = 0.620**),

DTPA Zn (r = 0.857**), carbonate (r = 0.862**), and Fe–Mn oxide bound Zn fractions

(r = 0.778** and 0.811**). The pH and total alkalinity of residue sources was correlated with

all Zn fractions except the residual fraction. Total Zn and organic carbon content of residue

sand significantly and positively correlated with Zn fractions.

Manganese

The extractability of Mn followed the order: residual fraction > Fe oxide bound > Mn oxide

bound > carbonate bound > organic matter bound > exchangeable (Table 7). On average, 40–

66% of total Mn was associated with the residual fraction and 12–45% in Fe–Mn oxide

bound fractions. Strong retention of Mn by the oxide fraction has previously been found for

both soil and bauxite residue sand, and has also been related to the low affinity of Mn for

organic compounds (Goldberg and Smith 1984; Tong et al. 1995; Gherardi and Rengel

2001). Samples from India exhibited higher Mn concentrations in the residual and Fe–Mn

oxide bound fractions than the Australian sample. Oxide-bound fractions showed significant

positive correlations among themselves (r ≥ 93%) and also were highly correlated with

exchangeable Na (r = 0.873**, 0.896**) and total Mn status (r = 0.962**, 0.975**).

Exchangeable and organically bound Mn contributed <0.5% of the total Mn, while carbonate-

bound fraction accounted for 2–7% of total Mn. A relatively high proportion of Mn was

associated with carbonate-bound forms, which might be due to higher CaCO3 content of

residue sand sources, or through precipitation as MnCO3. Strong association of Mn with the

carbonate fraction was reported for soils by Tong et al. (1995) and for bauxite residue

by Gherardi and Rengel (2001). There was a strong relationship between carbonate bound

Mn and DTPA Mn (r = 0.883**, Table 6) and exchangeable Mn (r = 0.895**), suggesting

that carbonate bound Mn might be an important source of plant-available Mn.

Adding gypsum significantly increased the concentration of exchangeable and residual Mn,

while it markedly lowered organic matter and oxide-bound fractions. Displacement of Mn by

Ca from exchange sites and further precipitation as MnCO3 under alkaline pH could be the

possible reasons for enhanced extraction of the carbonate-bound fraction and not the other 2

plant-available fractions (McBride 1994; Gherardi and Rengel 2001,2003a; Zinati et al.

2001). One-year-old residue sand samples had higher extractability of all fractions except the

Mn oxide bound fraction, which may be due to enhanced rhizosphere activities and Mn

reduction in the presence of root exudates (Uren 1981; Gherardi and Rengel 2001). Further

ageing reduced extractability of all Mn fractions but increased residual Mn, which indicated

the possible conversion of all plant-available fractions into non-available forms with time.

Except for carbonate-bound Mn, distribution into various fractions was not significantly

influenced by residue properties such as pH, total alkalinity, and exchangeable Ca. However,

exchangeable Na exhibited a positive relationship with all non-available pools of Mn

(r = 0.85**).

Boron

More than 40% of total B occurred in the residual pool, followed by readily soluble (16–

20%) > organically bound (13–17%) > oxide bound (6–15%) > specifically adsorbed (3.6–

6.5%) and hot water soluble fractions (2–8%,Table 8). Relative to the micronutrient metals,

readily soluble and organic matter bound fractions comprised a high proportion of total B,

while residual B was relatively low. All residue sand sources had levels of hot water soluble

boron (<0.50 mg/kg) too low for normal growth of most plants (Bell 1999) and values ranged

from <0.04 to 0.21 mg/kg.

The high proportions of readily soluble B observed with all residue sources could be due to

weak sorption on negatively charged surfaces of Fe and Al oxides under alkaline pH

(Goldberg 1997; Datta et al. 2002), as extractable B was positively related to pH

(r = 0.921**). Gypsum addition favoured extraction of the CaCl2-B fraction and aging

reduced B extractability. Total alkalinity and exchangeable Na were significantly correlated

with readily soluble B (r = 0.745**, 0.910**). Low levels of specifically adsorbed B may be

due to very low organic carbon and clay content of residue sand sources. Aged samples had

higher values of specifically adsorbed B than fresh residue, but 2003 samples had reduced

extractability of this fraction compared to 2006 samples (data not shown). No significant

correlations were observed between the specifically adsorbed B fraction and residue

properties but the specifically adsorbed B fraction was positively correlated with all other B

fractions. This indicates that the fraction might have originated from weakly bonded sites of

both organic and inorganic constituents of residue sand sources. Similar relationship between

soil properties and specifically adsorbed B was reported by Xu et al. (2001) for a range of

Chinese soils.

Oxide-bound fractions of B were higher in fresh residue sand than gypsum-amended and

aged samples, which can be explained with the positive effect of Fe2O3 in adsorbing B as

B(OH3) and B(OH4) (Su and Suarez 1995;Xu et al. 2001; Datta et al. 2002). Positive

correlations were found between oxide-bound and all other B fractions. None of the residue

sand properties was significantly correlated with the oxide-bound B fraction. Considerable

proportions of organically bound B were extracted from all residue sand sources despite the

low soil organic carbon status. The highest levels of organically bound B were associated

with 1-year-old residue sand, which may be due to poor decomposition and mineralisation of

the added organic amendments (Hou et al. 1994). Ageing reduced extractability of the

organically bound B fraction by 50% and was positively and significantly correlated with

organic carbon (r = 0.550**), exchangeable Na (r = 0.521**), and CaCl2 B (r = 0.701**). A

strong positive correlation existed between residual fraction and total B (r = 0.935**) but

surprisingly there were no significant correlations between residual B and properties of

residue sand sources. Gypsum addition slightly increased the readily soluble fraction but

decreased other fractions such as oxide-bound, organically bound, residual, and specifically

adsorbed B. Indian residue sand had lower extractability of all B fractions than Australian

residue sand. A positive response to B addition was observed for specifically adsorbed,

residual, and total B content of residue sand sources but not on other fractions.

Plant nutrient status

In Acacia cyclops, reduced leaf size, curling, crinkling, and malformation of leaves

resembling B deficiency were observed with concomitant low leaf B values. Other Acacia

cyclops plants showed symptoms resembling Mn deficiency and contained low leaf Mn

values. By contrast, Hardenbergia comptoniana and Eucalyptus gomphocephala exhibited

vigorous growth and dark green leaves with no specific symptoms resembling micronutrient

deficiencies.

Based on previous studies (Bell et al. 1997; Gherardi and Rengel 2003a, 2003b; Eastham et

al. 2006), Mn deficiencies were expected in plants grown on bauxite residue sand. However,

results from this study revealed little evidence of Mn deficiency, apart from Acacia cyclops.

Both cereal rye and triticale, which showed Mn deficiency in earlier studies (Bell et al.

1997; Eastham et al. 2006), are expected to be Mn-efficient, relative to most cereal crops.

Hence, the lack of Mn deficiency in Hardenbergia comptoniana and Grevillea

crithmifolia(data not shown) suggests a high level of Mn efficiency exists in the native

species. Grevillea crithmifolia (data not shown) in particular contained very high leaf Mn

(and Fe) concentrations. Many proteaceous species produce proteoid roots that mobilise

nutrients such as Mn and P in the root-zone by organic acid secretion (Shane and Lambers

2005). Grevillea crithmifolia appears to be able to acquire Mn and Fe efficiently, suggesting

the production of proteoid roots when grown in residue sand. Acacia cyclops appeared most

susceptible to low Mn, with possible Mn deficiency in plants showing symptoms in

phyllodes. Many eucalyptus species are sensitive to Mn deficiency on alkaline soils, and leaf

concentrations <15–20 mg/kg dry weight are considered to be marginal for growth (Dell et

al. 2001). Based on this critical range, Eucalyptus gomphocephalathroughout residue

rehabilitation areas may be expected to exhibit deficient levels of Mn (data not presented).

Earlier studies by Bell et al. (1997) did not determine B levels in the test species, triticale, or

in soils. Easthamet al. (2006) reported that B concentrations in cereal rye were adequate for

growth in the first season after fertiliser application. By contrast, Eastham and Morald

(2006) found B levels in cereal rye below critical levels. Acacia in particular had low leaf B,

and leaves with symptoms in 2003 rehabilitation commonly contained very low leaf B. In

2006 and 2005 rehabilitation, low leaf B in leaves of Grevillea crithmifolia with symptoms

were indicative of B deficiency (data not shown), but in 2003 rehabilitation areas, this species

appeared to be free of B deficiency. One case of low B (along with low Zn) in Eucalyptus

gomphocephala leaves was found in 2005 rehabilitation (data not shown). The hot

CaCl2 extractable B was below detection limits in residue in 2003 rehabilitation areas.

Eastham et al. (2006) found that cereal rye contained deficient levels of Zn in leaves despite

the addition of 15 kg ZnSO4/ha at planting. Bell et al. (1997) found no indication of Zn

deficiency in triticale when grown on residue sand with inorganic nitrogen–phosphorus based

fertiliser added, although DTPA-extractable Zn levels were noted to be low. Soil extractable

Zn levels were extremely variable, with several very high values from 2005 and 2004 areas

(data not shown) and a very low value in 2003 areas. Hence, further research is needed on the

risks and prevalence of Zn deficiency in residue sand. However, in 4-year-old residue

rehabilitation sites, Zn levels had dropped relative to those in 1-year-old residue and may

indicate a future risk of its deficiency. Like Mn deficiency, Zn deficiency is likely to be

expressed in some species more than others due to the differential Zn acquisition

characteristics of plants from residue sand.

Despite the alkaline pH and low levels of DTPA-extractable Fe, there were no confirmed

cases that suggested Fe deficiency in the revegetated plants. Hence, the forms of Fe in residue

sand apparently allow root acquisition of adequate Fe even under alkaline pH. The

extractable Cu levels were low in most residue samples. However, leaf Cu status was not

associated with deficiency symptoms except in Grevillea crithmifolia at 2006 and 2005

rehabilitation areas, which had low leaf Cu associated with symptoms (data not shown).

However, the symptoms in these plants could not be attributed to low Cu alone.

Next to Mn, B and Zn limitations were the major micronutrients affecting plant growth in the

revegetation sites. Furthermore, the mechanisms responsible for maintaining adequate

available Fe and Cu for plants under alkaline pH conditions warrant further investigation. Use

of geochemical speciation models such as PHREEQ may provide valuable insight to key

mechanisms affecting plant available forms of micronutrients.

Conclusion

Samples of fresh residue sand from both Australia and India exhibited characteristics (highly

saline, highly alkaline, and sodic) that may be expected to cause severe limitations to the

availability of plant micronutrients. Gypsum addition markedly improved residue sand

characteristics by reducing alkalinity (hence pH), and replacing Na with Ca as the dominant

solution and exchangeable cation. Fresh residue sand had higher total alkalinity (>85–90%),

which was primarily due to the dominance of water-soluble carbonate. In contrast, ageing

reduced the pH along with a shift from carbonate to bicarbonate alkalinity. Ageing reduced

extractability of all micronutrients, especially of Zn and Mn followed by B. Plant-available

micronutrients were very low in all residue sand samples despite high total concentrations,

which was also confirmed through the low leaf Zn, Mn, Cu, and B concentrations of plant

species grown in the revegetated sites especially with A. cyclops.

More than 80–90% of Mn, Zn, and Cu existed in residual and Fe and Mn oxide bound pools.

Ageing reduced all Zn and Cu fractions while it increased non-available Fe, Mn, and B

forms. Residue properties such as pH, organic carbon, exchangeable Na, and alkalinity were

correlated with Zn availability but only few significant correlations were observed with Fe,

Mn, and B. Although Fe and Mn oxide bound fractions are not expected to be immediately

available to plants, altering the residue properties like pH, alkalinity, and exchangeable Na

and redox conditions can modify these fractions resulting in enhanced micronutrient

availability and their redistribution. Management strategies based on chemical reactions

observed in this study that control plant nutrient availability should be considered central to

developing sustainable rehabilitation programs on residue sand.

Acknowledgements

The authors wish to acknowledge the support provided for this study by the Australian

Government through the Australian Leadership Award Fellowship (ALA) and for the support

provided by Alcoa World Alumina Australia.

References

Anon. (2007) Bauxite and alumina market to reach 250 million mt by 2010. Engineering and Mining Journal 6, 208. Barrow NJ (1982) Possibility of using caustic residue from bauxite for improving chemical and physical properties of sandy soils. Australian Journal of Agricultural Research 33, 275–285. Bell LC (1981) A systematic approach to the assessment of fertiliser requirements for the rehabilitation of mine wastes. In ‘Australian Mining Industry Council (AMIC) Environmental Workshop Papers’. Canberra, 20–24 Sept. 1981. pp. 163–175 (AMIC) Bell RW (1999) Boron. In ‘Soil analysis: An interpretation manual’. (Eds K Peverill, L Sparrow, DJ Reuter) pp. 309–317. (CSIRO Publishing: Collingwood, Vic.) Bell RW , Fletcher N , Samaraweera MK , Hammond I (1997) ‘The use of inorganic fertilizer in place of organic amendments for rehabilitation of gold and bauxite ore refining residue.’ Report submitted to Alcoa Australia. (School of Environmental Science, Murdoch University: Murdoch, W. Aust.) Bell RW , Fletcher N , Samaraweera MK , Krockenberger I (2003) Can inorganic fertilisers substitute for organic amendments in the rehabili-tation of saline, sodic ore refining residues? In ‘ORBIT 2003. 4th International Conference of the ORBIT Association. Biological Processing of Organics: Advance for a Sustainable Society’. 28 April–2 May 2003, Perth, W. Aust. (Eds P Pullumannapallil, A McComb, LF Diaz, W Bidlingmaier) pp. 602–611. (ORBIT Association: Perth, W. Aust.) Benitez LN, Dubois JP (1999) Evaluation of ammonium oxalate for fractionating trace elements in soils by sequential extraction. International Journal of Environmental Analytical Chemistry 74, 289–303. Bingham FT (1982) Boron. In ‘Methods of soil analysis’. Agronomy No. 9. (Eds AL Page et al.) pp. 431–442. (American Society of Agronomy/Soil Science Society of America: Madison, WI) Bremner JM , Mulveny CS (1982) Nitrogen–total. In ‘Methods of soil analysis. Part 2. Chemical and morphological properties’. Agronomy No. 9, 2nd edn. (Ed. AL Page) (American Society of Agronomy, Soil Science Society of America: Madison, WI)

Courtney R, Timpson JP, Grennan E (2003) Growth of Trifolium pratense in red mud amended with process sand, gypsum and thermally dried sewage sludge. International Journal of Mining, Reclamation and Environment 17, 227–233. Courtney RG, Timpson JP (2004) Nutrient status of vegetation grown in alkaline bauxite processing residue amended with gypsum and thermally dried sewage sludge – A two year field study. Plant and Soil 266, 187–194. Courtney RG, Timpson JP (2005) Reclamation of fine fraction bauxite processing (red mud) amended with coarse fraction residue and gypsum. Water, Air, and Soil Pollution 164, 91–102. Datta SP, Rattan RK, Suribabu K, Datta SC (2002) Fractionation and calorimetric determination of boron in soils. Journal of Plant Nutrition and Soil Science 165, 179– 184. Dell B , Malajczuk N , Xu D , Grove TS (2001) ‘Nutrient disorders in plantation eucalypts.’ 2nd edn, ACIAR Monograph No. 74. (Australian Centre for International Agricultural Research: Canberra, ACT) Eastham J, Morald T (2006) Effective nutrient sources for plant growth on bauxite residue I. Comparing organic and inorganic fertilizers. Water, Air, and Soil Pollution 176, 5–19. Eastham J, Morald T, Aylmore P (2006) Effective nutrient sources for plant growth on bauxite residue: II. Evaluating the response to inorganic fertilizers. Water, Air, and Soil Pollution 171, 315–331. Fuller RD, Richardson CJ (1986) Aluminate toxicity as a factor controlling plant growth in bauxite residue. Environmental Toxicology and Chemistry 5, 905–916. Ge Y, Murray P, Hendershot WH (2000) Trace metal speciation and bioavailability in urban soils. Environmental Pollution 107, 137–144. Gee GW , Bauder JW (1987) Particle size analysis. In ‘Methods of soil analysis. Part I. Physical and mineralogical methods’. pp. 377–411. (Soil Science Society of America, American Society of Agronomy: Madison, WI) Gherardi MJ, Rengel Z (2001) Bauxite residue sand has the capacity to rapidly decrease availability of added manganese. Plant and Soil 234, 143–151. Gherardi MJ, Rengel Z (2003a) Deep banding improves residual effectiveness of manganese fertiliser for bauxite residue revegetation. Australian Journal of Soil Research 41, 1273–1282. Gherardi MJ, Rengel Z (2003b) Deep placement of manganese fertiliser improves sustainability of lucerne growing on bauxite residue: a glasshouse study. Plant and Soil 257, 85–95. Gherardi MJ, Rengel Z (2003c) Genotypes of lucerne (Medicago sativa L.) show differential tolerance to manganese deficiency and toxicity when grown in bauxite residue sand. Plant and Soil 249, 287–296. Goldberg S (1997) Reactions of boron with soils. Plant and Soil 193, 35–48. Goldberg SP, Smith KA (1984) Soil manganese: E values, distribution of manganese-54 among soil fractions, and effects of drying. Soil Science Society of America Journal 48, 559–564.

Hou JL, Evans J, Spiers GA (1994) Boron fractions in soils. Communications in Soil Science and Plant Analysis25, 1841–1853. Hseu ZY (2005) Extractability and bioavailability of zinc over time in three tropical soils incubated with biosolids. Chemosphere 63, 762–771.

Jasper D , Lockley I , Ward S , White A (2000) Current approaches and future challenges for rehabilitation of bauxite residue. In ‘Proceedings of Remade Lands 2000, International Conference on the Remediation and Management of Degraded Lands’. Fremantle, W. Aust. (Eds A Brion, RW Bell) pp. 57–62. (Promaco Conventions: Canning Bridge, W. Aust.) Li LY (1998) Properties of red mud tailings produced under varying process conditions. Journal of Environmental Engineering 124, 254–264. Lindsay WL, Norvell WA (1978) Development of a DTPA soil test for zinc, iron, manganese and copper. Soil Science Society of America Journal 42, 421–428. Luo YM, Christie P (1998) Bio-availability of copper and zinc in soils treated with alkaline stabilised sewage sludges. Journal of Environmental Quality 27, 335–342. Ma YB, Uren NC (1995) Application of new fractionation scheme for heavy metals in soils. Communications in Soil Science and Plant Analysis 26, 3291–3303. McBride MB (1994) Chemisorption and precipitation of Mn at CaCO3 surfaces. Soil Science Society of America Journal43, 693–698. Meecham JR, Bell LC (1977) Revegetation of alumina refinery wastes: I. Properties and amelioration of the materials. Australian Journal of Experimental Agriculture and Animal Husbandry 17, 679–688. Melaku S, Dams R, Moens L (2005) Determination of trace elements in agricultural soil samples by inductively coupled plasma mass spectrometry: Microwave acid digestion versus aqua regia extraction. Analytica Chimica Acta 543, 117–123. Munshower FF (1994) ‘Practical handbook of disturbed land revegetation.’ (Lewis Publishers: CRC Press, FL) Prasad MBK, Ramanathan AL, Shrivastav SK, Anshumali , Saxena R (2006) Metal fractionation studies in surfacial and core sediments in the Achankovil river basins in India. Environmental Monitoring and Assessment 121, 77–102. Rayment GE , Higginson FR (1992) ‘Australian laboratory handbook of soil and water chemical methods.’ (Inkata Press: Melbourne, Vic.) Reuter DJ , Robinson JDB (Eds) (1997) ‘Plant analysis: An interpretation manual.’ (CSIRO: Collingwood, Vic.) Shane MW, Lambers H (2005) Manganese accumulation in leaves of Hakea prostrata (Proteaceae) and the significance of cluster roots for micronutrient uptake as dependent on phosphorus supply. Physiologia Plantarum 124, 441–450. Shuman LM (1985) Fractionation methods for soil microelements. Soil Science 140, 11–22. Silveira ML, Alleoni L, O’Connor GA, Chang A (2006) Heavy metal sequential extraction methods – a modification for tropical soils. Chemosphere 64, 1929–1938. Snars KR, Gilkes RJ, Hughes JC (2003) Effect of soil amendment with bauxite Bayer process residue (red mud) on the availability of phosphorus in very sandy soils. Australian Journal of Soil Research 41, 1229–1241.

Soumaré M, Tack FMG, Verloo MG (2003) Distribution and availability of iron, manganese, zinc and copper in four tropical agricultural soils. Communications in Soil Science and Plant Analysis 34, 1023–1038. Su C, Suarez DL (1995) Coordination of adsorbed boron: AFTIR spectroscopic study. Environmental Science & Technology 29, 302–311. Tong Y, Rengel Z, Graham RD (1995) Effects of temperature on extractable manganese and distribution of manganese among soil fractions. Communications in Soil Science and Plant Analysis 26, 1963–1977. Tucker BM (1985) Laboratory procedures for soluble salts and exchangeable cations in soils. CSIRO Australia, Division of Soils, Technical Paper No. 47. Uren NC (1981) Chemical reduction of insoluble higher oxide of manganese by plant roots. Journal of Plant Nutrition4, 65–71. Xenidis A, Harokopou AD (2005) Modifying alumina red mud to support a revegetation cover. JOM 57, 42–46. Xu JM, Wang K, Bell RW, Yang YA, Huang LB (2001) Soil boron fractions and their relationship to soil properties. Soil Science Society of America Journal 65, 133–138. Yasrebi J, Karimian N, Maftoun M, Abtahi A, Sameni AM (1994) Distribution of zinc forms in highly calcareous soils as influenced by soil physical and chemical properties and application of zinc sulphate. Communications in Soil Science and Plant Analysis 25, 2133–2145. Zinati G, Li Y, Bryan HH (2001) Accumulation and fractionation of copper, iron, manganese and zinc in calcareous soils amended with composts. Journal of Environmental Science & Health, Part B 36, 229–243.

Table 1. Summary of extraction procedures used for micronutrient (Zn, Fe, Mn, and

Cu) fractionation after Benitez and Dubois (1999)

Table 2. Summary of extraction procedures used for B fractionation after Datta et al.

(2002)

Table 3. Selected chemical properties of fresh and aged residue sand sources

FRS, Fresh residue sand; GFRS, gypsum-amended residue sand; AGARS1, aged gypsum-

amended residue sand, 2006 (1 year old); AGARS2, aged gypsum-amended residue sand,

2003 (4 years old). Values in parentheses are standard errors of the mean of 3 replications.

The composition of gypsum applied (by aqua regia) was (mg/kg): 640 P, 1530 K, 104 000

Ca, 26 700 Na, 10.6 Zn, 317 Fe, 0.14 Mn, 1.38 Cu, 33.1 B

Table 4. DTPA-extractable copper (Cu), iron (Fe), manganese (Mn), and zinc (Zn) and

CaCl2-extractable boron (B) status (mg/kg) in residue sand sources with and without

addition of respective micronutrients

FRS, Fresh residue sand; GFRS, gypsum-amended residue sand; AGARS1, aged gypsum-

amended residue sand, 2006 (1 year old); AGARS2, aged gypsum-amended residue sand,

2003 (4 years old). Values in parentheses are standard errors of the mean of 3 replications

Table 5. Distribution of Zn into various fractions (mg/kg) in residue sand sources with

(+Zn) and without (–Zn) addition of Zn

FRS, Fresh residue sand; GFRS, gypsum-amended residue sand; AGARS1, aged gypsum-

amended residue sand, 2006 (1 year old); AGARS2, aged gypsum-amended residue sand,

2003 (4 years old). Values in parentheses are standard errors of the mean of 3 replications

Table 6. Relationship between residue properties and Zn and Mn fractions

*P < 0.05; ** P < 0.01; n = 36 (6 residue sand sources, 2 rates of spiking, 3 replications)

Table 7. Distribution of Mn into various fractions (mg/kg) in residue sand sources with

(+Mn) and without (–Mn) addition of Mn

FRS, Fresh residue sand; GFRS, gypsum-amended residue sand; AGARS1, aged gypsum-

amended residue sand, 2006 (1 year old); AGARS2, aged gypsum-amended residue sand,

2003 (4 years old). Values in parentheses are standard errors of the mean of 3 replications

Table 8. Distribution of B into various fractions (mg/kg) in residue sand sources with

(+B) and without added B (–B)

FRS, Fresh residue sand; GFRS, gypsum-amended residue sand; AGARS1, aged gypsum-

amended residue sand, 2006 (1 year old); AGARS2, aged gypsum-amended residue sand,

2003 (4 years old). Values in parentheses are standard error of the mean of 3 replications.

n.d., Not detected