In an alumina refinery, bauxite ore is processed by the Bayer Process whereby Al-containing minerals are dissolved in sodium hydroxide at high temperatures and pressures. For each tonne of alumina produced, one to two tonnes of alkaline, saline, insoluble solids (bauxite processing residues) are also produced. Global use for this residue is estimated to be only 2% - 3% and the bulk of the material is stockpiled in areas close to the refinery [1]. As part of the environmental approval conditions for the newly-built alumina refinery at Al Taweelah in the UAE, Emirates Global Aluminium (EGA) is required to find practical applications for all the residue produced. For this reason, a novel manufactured soil product (Turba) has been developed in association with EGA [2]. Turba is made by mixing acidified, leached bauxite residue with 10% w/w compost and then drying and crushing the product to form aggregates [2][3]. Laboratory and greenhouse studies have demonstrated that Turba is an excellent plant growth medium and superior to that of dune sand (the main plant growth medium in the UAE) [4][5]. As a result, a multi-million dollar pilot plant, taking residue directly from the waste stream of the refinery, has recently been be completed to produce Turba for its large-scale field evaluation in the UAE [6].

Preliminary observations have suggested that over time, even under grass, aggregate size of Turba decreases and the decrease is worsened by disturbances such as tillage (R. J. Haynes personal communication). This is in contrast to natural soils where grass root growth through the medium typically increases aggregate size and stability [7]. Initially, aggregation in bauxite residue is the result of the pozzolanic cementing actions of Ca aluminates and silicates [8] and this mechanism is used to produce Turba aggregates when the acidified residue is dried. In natural soils, microaggregates form through linkage of clay particles and humic material and then these are assembled into larger units (macroaggregates) by the enmeshing action of roots, root hairs and mycorrhizal fungal hyphae and temporary polysaccharide binding agents exuded from plant roots and bacteria and fungi in the rhizosphere [9]. Macro-aggregation is a dynamic process and cycles of formation and breakdown occur continuously [10]. It is unknown whether such a plant root-effect occurs in Turba or whether pozzolanic solidification is the main/only aggregation process. Aggregation is a vital aspect of the condition of a manufactured soil because it dictates pore space and pore structure and these properties control how the soil stores and transports water, air and nutrients thereby impacting plant and microbial growth and activity [9]. Macropores (>29 μm dia.) are important since they allow for rapid water drainage and aeration as well as root growth through compacted layers.

A likely initial use for Turba is for culture of grass turf for golf courses, playing fields, landscaped lawns, etc. Golf courses in the UAE usually use Bermuda grass which is over-sown with perennial ryegrass during the cooler winter season (when peak use occurs). In this study, perennial ryegrass was grown for 6 months in Turba and in a regularly cropped, physically degraded natural soil in order to compare plant growth effects on aggregation in the two media. Unplanted (fallow) treatments were included for comparison. Two sizes of aggregates were used, 2 - 4 mm and <2 mm in order to gauge if plant root growth would induce aggregation of small aggregates (<2 mm) into larger aggregate sizes and/or stabilize larger (2 - 4 mm) aggregates. Parameters measured included total porosity, pore size distribution and plant available water in <2 and 2 - 4 mm aggregates, ryegrass shoot and root dry matter yields, aggregate size distribution after ryegrass growth and aggregate stability of 2 - 4 mm aggregates using wet or dry sieving techniques. Microbial properties including soluble C, Microbial biomass C., basal respiration were also measured since these relate to microbial activity in the ryegrass rhizosphere that would potentially promote soil aggregation.

Chemical properties of Turba in relation to its suitability as a growth medium have been discussed in detail elsewhere [2][4][5][19]. They are considered suitable for growth of ryegrass and results have shown greater dry matter yields than in dune sand [4][5]. The exchangeable Na concentrations rapidly decline due to leaching and plant uptake and ESP decreases to below 20% within a few months [5]. The soil was low in available K (critical level 2 - 5 mmol_c·kg⁻¹) [20] but fertilizer K along with other macro and micronutrients was applied regularly to also both soil and Turba. For the above reasons, it is unlikely that soil chemical properties/nutrient availability were a limiting factor to ryegrass growth in either Turba or soil.

That after 6 months grass growth, aggregate sizes separated by sieving were less than the original sieved diameters for Turba and greater for soil (Table 2) reflects the contrasting mechanisms of aggregation in the two materials. In natural soils, it is a dynamic process whereby macro-aggregation continually occurs under the influence of physical binding (plant roots and associated mycorrhizal hyphae) and temporary cementing agents (e.g. polysaccharides exuded from roots and rhizosphere microorganisms) [9]. Macroaggregates are continually being broken down while new ones are being formed by the above processes. Indeed, observations revealed that plant roots were ramifying through recovered soil aggregates (Figure 2) and aggregate stability was greater in planted than fallow treatments (Table 3)

By contrast, the major aggregating mechanism for Turba is a permanent, one-off solidification which occurs the first time the material is dried. It shrinks and solidifies through the action of pozzolanic agents (Ca silicate and Ca aluminate gels) which cement the material together [8] to form a solid mass. This is a similar reaction to that which occurs when cement dries [21]. This material is then crushed and sieved to form aggregates. Over time, external forces acting on the material cause cleavage particularly at the edges of aggregates so their size tends to decrease. After 6 months ryegrass growth 100% of aggregates added as <2 mm dia. were still present as <2 mm dia. so no aggregation induced by plant root growth and microbial activity in the rhizosphere was evident.

Even though, over the 6 month experimental period, aggregate size increased for soil and decreased for Turba, both wet and dry sieving demonstrated that 2 - 4 mm dia. Turba aggregates were markedly more stable than the corresponding soil aggregates (Table 3). Thus, pozzolanic cementation permanently binds Turba aggregates together more strongly and in a different way to that of temporary soil binding agents such as roots, root hairs, fungal hyphae and short-term organic cementing agents (e.g. polysaccharides) in natural soils.

For natural soils, wet sieving typically yields a lower MWD than dry sieving because hydraulic forces (e.g. slaking, dispersion and breakdown from air pressure during wetting) are more destructive than physical abrasion and mechanical forces occurring during dry sieving [22][23]. Stable soils can typically have a MWD_wet/MWD_dry of >0.7 while very unstable soils have a ratio of <0.4 [23][24]. Thus, the unstable soil used here had a ratio of 0.33 under fallow and 0.44 under grass. By contrast, the Turba aggregates were extremely water stable and were more affected by mechanical action of dry sieving so the ratio was greater than 1.0. The cementing action of Ca aluminates and silicates results in aggregates that are extremely stable in water but when dry they can be brittle and are decreased in size by mechanical action (albeit still more stable to dry sieving than the soil aggregates).

Because of the very low microporosity and solid nature of Turba aggregates they are not easily penetrated by growing roots and few or no roots were observed passing through Turba aggregates (Figure 2). In general, a lack of macroporosity is known to restrict root growth [25]. Roots tend to proliferate in the macroporosity between Turba aggregates and grow down water films on the outside of aggregates. Root growth therefore has little effect on aggregation and aggregate stability was therefore unaffected by the presence of grass. Indeed, the larger macroporosity in 2 - 4 than < 2 mm dia. Turba aggregates (Table 1) reflects a larger macroporosity occuring between the larger aggregates and that macroporosity occurs mainly between rather than within aggregates. Macroporosity in Turba can be managed by manipulating aggregate size and this inter-aggregate porosity.

In contrast to results presented here, increases in aggregation and aggregate stability with time (along with decreases in ESP) have been recorded on aging/revegetating bauxite residue deposits by a number of workers [26][27]. This is attributable to Na⁺-induced dispersion which predominates in recently deposited residues (the ESP is typically 70% - 90%) (26). Over time, Na is released from Na-containing minerals and leached out while Ca is released from Ca-containing minerals (e.g., calcite) thus resulting in a progressive decrease in ESP and replacement of the monovalent Na⁺ with divalent Ca²⁺ as the dominant exchangeable cation [28]. In the present study, the process of OBxR manufacture involves acidification and leaching of excess salts (including Na) so that the ESP of the original Turba was only 42% and exchangeable Ca²⁺ was already present at higher concentrations than exchangeable Na⁺.

Since the soil had been under arable cultivation for 25 years it had a low organic matter content (6.14 g C kg⁻¹) while Turba had a higher content of 15.4 g C kg⁻¹. This level is in the range similar to sandy soils under grassland, prairie and shrubs [29]. As a result, Turba had a greater soluble C, microbial biomass C and basal respiration than the soil. As expected, planted treatments of Turba and soil had higher soluble C contents than fallow treatments (Table 4). This reflects rhizodeposition of organic material through normal growth and senescence of root segments and root hairs as well as exudation of a wide range of organic substrates by roots and the associated microbial community present in the rhizosphere [30]. In pots, grass root growth is restricted to the volume of the pot and over time the bulk of the soil volume effectively becomes rhizosphere soil and is affected by the dense, ramified grass root system [7]. Changes in total organic C content were not detected in soil and Turba due to the large background level present. For soil, the increased soluble C resulted in greater size (microbial biomass C) and activity (basal respiration) of the microbial community and in Turba there was an increase in microbial biomass. While the size and activity of the microbial community was significantly greater in Turba than soil, there was no indication of aggregation occurring in Turba (the <2 mm fraction remained as <2 mm dia. after 6 months fallow or under ryegrass) suggesting that microbially-induced macro-aggregation is not a significant mechanism for Turba.

The main factor influencing plant yields is likely to have been the available water holding capacity of the growth media [4][5]. Although the pots were watered to saturation every 2 - 3 days, in the period between irrigation events the plants in the treatments holding less available water are likely to have suffered from incipient water stress thus limiting their growth. Greatest yields were in Turba < 2 and this treatment held by far the greatest plant available water (Table 3). In addition, yields were greater in the <2 than 2 - 4 mm treatments in both Turba and soil and in both cases the <2 mm treatment held more available water than the 2 - 4 mm one.

The lack of continual aggregation in Turba means its management will involve conservation of aggregates. External forces acting on Turba aggregates can cause their disruption (particularly at the edges) and a reduction in their diameter over time. In the field, such forces might include tillage, machinery and foot traffic, and raindrop impact. Minimizing disturbance by growing permanent crops (turf grass, forages plantation crops) and using minimum/zero tillage will be important considerations. In the present greenhouse study aggregates used were <4 mm dia. but when research is extended to the field it will be appropriate to use much larger aggregates (e.g. > 25 or >50 mm dia.). The appropriate size will need to be assessed in the field and the fact that aggregates tend to decrease in diameter over time will need to be considered. This will also be important from the viewpoint of maintaining microporosity, and thus aeration, since Turba aggregates have very low macro-porosity and most of the macropores in the Turba medium are between aggregates. Factors that increase the strength of binding within Turba aggregates will also need to be optimized during their manufacture. This will involve maximizing the concentration of pozzolanic materials present prior to drying. Since both Ca aluminates and silicates are highly acid soluble increasing acid additions during Turba manufacture have been found to weaken aggregate binding so there is a balance between lowering pH to promote plant growth while not producing brittle aggregates.

During revegetation of bauxite processing waste areas, the material is left to dry and it solidifies, shrinks causing cracks and macropores to form to depth [31]. Over time the solidified material close to the surface breaks into smaller peds and the surface layer can be ripped and tilled to form a tilth suitable for plant establishment. In the vast majority of cases permanent vegetation is planted (grass, herbs, bushes). It is very likely that plant roots grow predominantly between aggregates (as occurs with Turba) rather than through them as occurs in a natural soil. As has been demonstrated on a number of field sites this is not a problem and revegetation can be highly successful [31][32]. However, the material is probably not suitable for long-term production of arable crops because tillage will break peds into progressively smaller aggregates until aeration and downward water movement are inhibited.

Due to the influence of external forces acting on them, Turba aggregates tend to decrease in size over time. Root growth and microbial activity had no measurable effect on aggregate size or stability in Turba because roots grew between rather than through the solid, compact, cemented Turba aggregates. Important factors in Turba manufacture will include maximizing the concentration of pozzolanic binding agents present and optimizing aggregate size while management practices need to minimize aggregate disruption.

We thank Dr. S. Rosenberg and Dr. M. Graefe of Emirates Global Aluminium for their support and comments throughout the study and Professor Bernard Schroeder of the University of Southern Queensland for supplying the soil used in the experiment. The study was funded by Emirates Global Aluminium through a research contract with The University of Queensland.