Barley (Hordeum vulgare L.), genetically modified with the Al3+ resistance gene of wheat (TaALMT1), was compared with a non-transformed sibling line when grown on an acidic and highly phosphate-fixing ferrosol supplied with a range of phosphorus concentrations. In short-term pot trials (26 days), transgenic barley expressing TaALMT1 (GP-ALMT1) was more efficient than a non-transformed sibling line (GP) at taking up phosphorus on acid soil, but the genotypes did not differ when the soil was limed. Differences in phosphorus uptake efficiency on acid soil could be attributed not only to the differential effects of aluminium toxicity on root growth between the genotypes, but also to differences in phosphorus uptake per unit root length. Although GP-ALMT1 out-performed GP on acid soil, it was still not as efficient at taking up phosphorus as plants grown on limed soil. GP-ALMT1 plants grown in acid soil possessed substantially smaller rhizosheaths than those grown in limed soil, suggesting that root hairs were shorter. This is a probable reason for the lower phosphorus uptake efficiency. When grown to maturity in large pots, GP-ALMT1 plants produced more than twice the grain as GP plants grown on acid soil and 80% of the grain produced by limed controls. Expression of TaALMT1 in barley was not associated with a penalty in either total shoot or grain production in the absence of Al3+, with both genotypes showing equivalent yields in limed soil. These findings demonstrate that an important crop species can be genetically engineered to successfully increase grain production on an acid soil.
Acid soils are prevalent in many regions of the world and can limit plant production by the accumulation of toxic concentrations of trivalent aluminium (Al3+) in the soil solution (von Uexküll and Mutert, 1995). The soluble Al3+ inhibits root growth, which subsequently affects a plant's ability to take up water and nutrients. In some crop species, there is genetic variation for Al3+ resistance (exclusion of aluminium) or tolerance (ability to tolerate internal aluminium), which has been exploited by breeders to develop cultivars that maintain productivity on acid soils (Garvin and Carver, 2003). The potential for developing improved cultivars for acid soils has recently been augmented by the isolation of genes responsible for the natural Al3+ resistance of some species (Sasaki etal., 2004; Furukawa etal., 2007; Magalhaes etal., 2007; Wang etal., 2007), as well as the over-expression of genes that enhance a plant's tolerance to Al3+ (Ezaki etal., 2000; Basu etal., 2001; Tesfaye etal., 2001; Ryan etal., 2007). Other genes with roles in aluminium tolerance identified from mutant screens of Arabidopsis (Arabidopsis thaliana) include ALS1, ALS3 and AtATR (Larsen etal., 2005, 2007; Rounds and Larsen, 2008), and these also have the potential to increase the ability of plants to grow on acid soils when over-expressed or mutated.
Two families of genes responsible for the natural variation in Al3+ resistance of several species encode membrane proteins that are activated by Al3+, resulting in the efflux of organic anions from roots (Delhaize etal., 2007). The organic anions secreted by roots chelate and detoxify Al3+, and this type of mechanism appears to be widespread amongst plant species (Matsumoto, 2000; Ma etal., 2001; Ryan etal., 2001; Kochian etal., 2004). The membrane proteins identified to be responsible for Al3+ resistance belong to the aluminium-activated malate transporter (ALMT) and multidrug and toxic compound extrusion (MATE) families. Although members of these families of proteins confer the efflux of organic anions, the families possess little or no sequence similarity to one another. Expression in planta of Al3+ resistance genes encoding ALMT (Delhaize etal., 2004) or MATE (Furukawa etal., 2007; Magalhaes etal., 2007) proteins confers an Al3+-activated efflux of malate and citrate, respectively. Where polymorphisms within a species have been identified between alleles encoding these proteins, they can be used as molecular markers in marker-assisted selection for Al3+ resistance (Raman etal., 2008). Alternatively, the genes themselves can be used across species to enhance Al3+ resistance through gene technology. To date, most of the genes found to enhance Al3+ resistance or tolerance have only been assessed in model species, such as Arabidopsis, with few examples of their use in important crop species. Transgenic barley (Hordeum vulgare L.) expressing the wheat (Triticum aestivum L.) Al3+ resistance gene, TaALMT1 (originally named ALMT1), shows an Al3+-activated efflux of malate that is absent from wild-type barley, and a level of enhanced Al3+ resistance similar to that found in Al3+-resistant wheat cultivars (Delhaize etal., 2004). Although an Al3+ resistance gene has been identified in barley (Furukawa etal., 2007; Wang etal., 2007), the variation within this species is comparatively low, and it is among the most Al3+ sensitive of the small-grain crops (Zhao etal., 2003). Expression of the SbMATE Al3+ resistance gene isolated from sorghum (Sorghum bicolor) in a sensitive wheat cultivar similarly conferred Al3+-activated citrate efflux and increased Al3+ resistance (Magalhaes etal., 2007). Over-expression of a manganese superoxide dismutase in Brassica napus conferred enhanced Al3+ tolerance, which was attributed to an improved ability to deal with the oxidative stress caused by Al3+, and is a mechanism independent of the efflux of organic anions (Basu etal., 2001). The increase in Al3+ resistance in these studies was based on a measure of root elongation in hydroponic culture (Basu etal., 2001; Delhaize etal., 2004; Magalhaes etal., 2007) or in soil (Delhaize etal., 2004) over several days, and did not assess the benefits that the expression of these genes might confer to shoot biomass and grain production of plants grown on acid soils.
In this paper, we investigate the interaction between Al3+ resistance and phosphorus nutrition in wild-type and transgenic barley expressing TaALMT1 when grown in a highly phosphorus-fixing and acidic soil. When supplied with a range of phosphorus concentrations, an interaction between Al3+ resistance and phosphorus nutrition was evident. In a longer term experiment, the transgenic barley showed significantly greater root growth, shoot biomass and grain yield compared with a non-transgenic control when grown to maturity on the same acid soil.
Short-term root growth
All three transgenic barley lines expressing TaALMT1 showed similar root growth when grown for 5 days on acid soil (Figure 1). On the same soil, the root growth of the three sibling lines that lacked the transgene, as well as the wild-type parental cultivar, was inhibited by about 30% relative to the transgenics. All genotypes showed similar root lengths when grown on limed soil. As the three transgenic lines showed a similar level of aluminium resistance when assessed in hydroponics and on acid soils (Delhaize etal., 2004; this study), one pair of sibling lines was used for more detailed analysis, subsequently referred to as GP-ALMT1 (line 5T) and GP (line 5S).
In the absence of added phosphorus, both transgenic and non-transgenic seedlings became severely phosphorus deficient and produced little shoot yield, regardless of whether or not the soil had been limed (Figure 2a). When phosphorus was applied, both genotypes responded equally in limed soil, whereas, on acid soil, GP-ALMT1 showed a much greater growth response than GP to phosphorus application. This was particularly evident at the lower rates of phosphorus application on acid soil, where the GP shoots remained severely phosphorus deficient and showed yields that were not significantly different (P < 0.05) from those of plants grown without phosphorus. The growth of GP responded at the two higher phosphorus treatments on acid soil and attained a shoot biomass that was equivalent to that of GP-ALMT1 only at the highest phosphorus treatment. In contrast, GP-ALMT1 responded significantly to phosphorus application on acid soil with a shoot biomass at the two lowest phosphorus applications of about one-half of that of the plants grown on limed soil (Figure 2a).
The root growth of GP was severely reduced in acid soil with no or low phosphorus applications compared with that of GP-ALMT1, which showed similar root lengths to the genotypes grown on limed soil for all phosphorus treatments (Figure 2b). As observed for shoot biomass, the effect of soil acidity on the root length of GP was ameliorated by high phosphorus treatments, and maximum root length occurred at 500 mg P/kg and above. At the two highest concentrations of applied phosphorus, the levels of aluminium in the soil were reduced to 36% and 28% of the total cations, respectively (Table 1). The amount of phosphorus accumulated by the shoots of both genotypes grown on limed soil was similar, and increased linearly with phosphorus application (Figure 2c). On acid soil, GP-ALMT1 accumulated more shoot phosphorus than did GP at intermediate phosphorus applications (125, 250 and 500 mg/kg). When the phosphorus uptake into shoots was expressed per unit root length, the two genotypes grown on limed soil showed similar uptake (Figure 3a). By contrast, on acid soil, GP-ALMT1 showed greater phosphorus uptake (P < 0.05) per unit root length than did GP for both the 250 and 500 mg/kg phosphorus treatments. GP grown on acid soil with phosphorus applied had a greater root to shoot ratio than did GP-ALMT1 or the two genotypes grown on limed soil for all treatments, except at the highest phosphorus rate (Figure 3b). The root to shoot ratio of GP-ALMT1 on acid soil differed from the plants grown on limed soil only at the lowest phosphorus treatments (0 and 125 mg/kg). Shoot phosphorus concentrations increased as the phosphorus supply increased, with GP-ALMT1 maintaining a higher concentration than GP on acid soil across all treatments in which phosphorus was applied, except at the highest rate (Figure 3c).
Table 1. Effect of phosphorus applications on exchangeable aluminium, plant-available phosphorus and pH of limed and acid soils. Values represent single analyses taken from the bulk soil for each phosphorus treatment
Phosphorus (P) applied (mg/kg)
Aluminium (% of total cations)
Plant-available phosphorus (mg/kg soil)
Despite large differences in shoot yield and root growth on acid soil, the rhizosheaths of both genotypes were similar and responded to phosphorus application, with an increase of about twofold in the size of the rhizosheath between no phosphorus and the highest rate of phosphorus application (Figure 4). When grown on limed soil, GP-ALMT1 and GP also showed similar rhizosheaths to each other across all phosphorus rates, but, at the three lowest phosphorus treatments, the rhizosheaths were about three-fold larger than those of plants grown on acid soil.
Long-term growth experiment
Phosphorus was applied in the top 1 kg of soil at 1000 mg/kg to enable the plants to achieve maximum yield without being restricted by phosphorus nutrition, and also to simulate the ‘liming effect’ caused by phosphorus or lime applications in the surface soil when applied in the field. In this system, both genotypes showed similar grain and shoot biomass when grown on limed soil (Figure 5a,b), and showed no evidence of any nutrient deficiency (Table 2). By contrast, GP-ALMT1 on acid soil showed significantly higher grain production (P < 0.05), which was over double that of GP and about 80% of that of the limed GP-ALMT1 control (Figure 5a). Similar differences between the genotypes were apparent for total shoot weights and tiller numbers (Figure 5b,c). The largest effect of soil acidity on the distribution of GP roots in the soil profile occurred in the top 10 cm (Figure 6). This occurred despite the application of 1000 mg P/kg, which would be expected to ameliorate the effect of Al3+ toxicity on root growth, as observed in the short-term phosphorus trial (Figure 6). Root weights did not differ for any treatment or genotype combination in the 10–50-cm sections. However, below 60 cm, roots of GP grown in acid soil were almost absent, with the small amount of root material identified having grown down the side of the pot or along the crack formed where the two pieces of the pot were joined together. By contrast, the other treatments, particularly those with GP-ALMT1, grew roots down to the bottom of the soil profile, although they represented only a small proportion of the total root weight.
Table 2. Elemental concentrations in leaves of barley genotypes grown on acid or limed soil. Youngest fully expanded leaves were collected at booting and the data show the mean and the standard error in parentheses
Elemental analysis of leaf blades collected at booting and grain collected at final harvest showed that calcium concentrations were elevated in both tissues for limed treatments and manganese concentrations were elevated in both tissues for plants grown on acid soil (Table 2). The elemental concentrations of leaf blades did not identify a nutrient deficiency that might have explained the differences in yields between GP and GP-ALMT1 grown on acid soil (Reuter etal., 1997). Grain elemental concentrations were largely similar when genotypes were compared within treatments (Table 3).
Table 3. Elemental concentrations in grain of barley genotypes grown on acid or limed soil. Grain was collected from matured plants and the data show the mean and the standard error in parentheses
Transgenic barley expressing TaALMT1 showed improved phosphorus uptake efficiency compared with wild-type barley when grown on acid and highly phosphorus-fixing soil (Figures 2 and 3). This can largely be attributed to GP-ALMT1 maintaining root growth on acid soil, whereas the root growth of GP was severely constrained, particularly at low phosphorus treatments. Phosphate is generally poorly mobile in soil and a plant's ability to acquire phosphorus relies, to a large extent, on an effective root system able to explore the soil. In addition to TaALMT1 conferring a larger root system in acid soil, the uptake of phosphorus per unit root length was also greater for GP-ALMT1 than for GP at the two intermediate phosphorus treatments (Figure 3a; 250 and 500 mg P/kg). The difference in phosphorus uptake did not appear to be caused by differences in root to soil contact, as indicated by equivalent rhizosheath development by the two lines (Figure 4). The greater uptake by GP-ALMT1 roots might be attributable in part to the secretion of malate, which either solubilized soil phosphorus directly, making it more readily available for uptake, or indirectly, as a consequence of root function being protected by malate, such that phosphorus uptake in GP-ALMT1 was less affected by Al3+ toxicity. Despite GP-ALMT1 having larger roots on acid soil, the root to shoot ratio of GP was significantly greater than that of GP-ALMT1 for all treatments in which phosphorus was applied, except at the highest rate. A high root to shoot ratio is a symptom of plants suffering from phosphorus deprivation, and this further suggests that poor phosphorus nutrition limited the growth of GP plants. Shoot phosphorus concentrations also indicated that phosphorus nutrition limited the growth of GP on acid soil with concentrations (Figure 3c) that would be considered deficient (125 and 250 mg P/kg treatments) or critical (500 mg P/kg treatment) in whole shoots of young barley seedlings (Reuter etal., 1997). In addition, the difference between genotypes in their ability to acquire phosphorus and attain maximum biomass was abolished at the highest phosphorus rate (Figure 2). The ameliorative effect of phosphorus application can be attributed to the formation of Al–P complexes which detoxify Al3+ (Pellet etal., 1996). The increase in the root length of GP with increasing phosphorus application in acid soil is consistent with this hypothesis. Furthermore, soil analysis showed that the amount of exchangeable aluminium expressed as a percentage of total cations decreased in the soil solution at the two highest phosphorus rates (Table 1).
Although GP-ALMT1 showed greater phosphorus uptake efficiency than GP when grown on acid soil, it was not as effective at low phosphorus applications as the limed controls (Figures 2 and 3a). This difference could not be attributed to the inhibition of root growth, as the root lengths of GP-ALMT1 grown on acid soil were similar to those of plants grown on limed soil at all phosphorus treatments (Figure 2). However, roots grown in acid and limed soils differed markedly in their ability to form rhizosheaths. Rhizosheaths consist of soil that adheres firmly to roots as a result of the presence of root hairs and mucilage that forms a ‘glue’ to bind the soil particles (McCully, 1999). In some species, the rhizosheaths extend to the length of the root hairs and are therefore an indirect measure of root hair length. Indeed, Haling (2007) showed, in a range of barley genotypes that included GP and GP-ALMT1, that a strong correlation exists between root hair length and rhizosheath size in the same soil as used in the current study. Root hair length is an important plant attribute for the effective uptake of phosphate from soil (Gahoonia etal., 1997), and the smaller rhizosheath for GP-ALMT1 on acid soil is consistent with a lower phosphorus uptake efficiency caused by shorter root hairs compared with plants grown on limed soil. Interestingly, although the expression of TaALMT1 protected roots from Al3+ toxicity, root hairs were apparently not protected to the same extent, which may indicate that insufficient malate was effluxed from mature regions of the root to protect root hairs from Al3+, or that the root hairs were sensitive to another factor in the soil, such as soluble manganese or the acidity itself.
The GP-ALMT1 line produced substantially more grain than the GP control on acid soil, whereas both genotypes produced similar amounts of grain when grown on limed soil (Figure 5). Similarly, the genotypes did not differ in a range of attributes when grown on limed soil in the phosphorus trial. These findings indicate that, although the transgene confers an advantage on acid soil, this is not associated with a yield penalty to the plant when grown in the absence of Al3+ stress. Malate efflux from roots only occurs in the presence of Al3+, which avoids the loss of carbon from roots when it is not required (Delhaize etal., 1993). Furthermore, despite the use of a promoter that is expressed constitutively throughout the plant, malate efflux in transgenics is largely confined to the root apex (Delhaize etal., 2004), which is the site primarily affected by Al3+ toxicity (Ryan etal., 1993).
The reduced shoot biomass of GP compared with GP-ALMT1 was associated with a reduced root mass in the 0–10-cm layer, suggesting that the genotypic difference in the growth of this fraction of the root system was responsible for the differences in yield. Although the phosphorus trial showed that high rates of phosphorus application ameliorated Al3+ toxicity over 26 days of growth, a longer growing period appears to allow the effects of residual Al3+ toxicity to become apparent. Table 1 shows that, when phosphorus was applied at 1000 mg/kg, more than one-half of the exchangeable aluminium remained in the soil, where it may have been sufficiently toxic to inhibit root growth over the 156-day growth period. The elemental concentrations of leaf blades indicated that phosphorus deficiency did not limit the growth of GP on acid soil in this experiment, in contrast with the phosphorus-limited growth observed at lower phosphorus supplies in the phosphorus trial (Reuter etal., 1997). The amount of phosphorus supplied in the long-term experiment appeared to be sufficient to meet plant demand, as indicated by tissue phosphorus concentrations of both leaf and grain (Tables 2 and 3). Furthermore, the growth of GP on acid soil did not appear to be limited by a deficiency of other mineral nutrients. The greater calcium concentrations in the tissues of plants grown on limed soil reflect the increased concentrations of calcium in the soil contributed by the lime. Manganese concentrations in tissues of both genotypes were increased on acid soil, reflecting the greater availability of this element in the soil solution. Indeed, the leaf concentrations were sufficiently high such that the plants may have suffered from marginal manganese toxicity (Reuter etal., 1997), and illustrate that TaALMT1 specifically conferred resistance to Al3+ and not to other factors, such as manganese, which may be present at toxic concentrations in acid soils. In the current experiment, we surmise that manganese toxicity did not pose a severe constraint to plant growth, as GP-ALMT1 yields were only marginally smaller on acid than on limed soil (Figure 5).
Tang etal. (2001) showed that, in a pair of near-isogenic wheat lines that differed in Al3+ resistance, the ability of the sensitive line to take up water from depth in reconstructed soil columns was severely impaired when compared with the resistant line. In the current experiment, root weights in the 10–50-cm soil layers did not differ between genotypes grown on acid soil, but the roots of GP appeared thicker and had fewer laterals than those of GP-ALMT1 (data not shown). Below 50 cm, GP plants effectively had no roots and the plants would be impaired in their ability to acquire water at depth, which could further restrict the yield potential of plants when grown in the field. In the experiments described here, water was applied to the surface, and any differences in water use efficiency as a result of deep rooting would not be apparent. Tang etal. (2002) demonstrated in field experiments that aluminium resistance in wheat enhances the ability of plants to take up water from depth in acidic soils, and a similar benefit is likely to be conferred on the transgenic barley expressing TaALMT1. Clearly, the application of lime is a preferred strategy for enhancing crop yields on acid soils, but the ability to incorporate lime to depth is problematic. Even if the surface layer of a deeply acidic soil profile is ameliorated with lime or with a high-phosphate application, grain yields can still be restricted by the Al3+ toxicity to roots that occurs at depth. Several studies have concluded that a combination of liming and the use of aluminium-resistant cereal germplasm is an effective strategy for managing soil acidity when it occurs at depth (Scott etal., 2001; Tang etal., 2003).
In the current study, we assessed the performance of transgenic barley expressing TaALMT1 when grown on a highly phosphorus-fixing and acidic soil. Although such soils are not typical of Australian cropping systems, they are common in tropical regions throughout the world, including South America where cereals are grown. For instance, a recent study has shown that field-grown rice in central Brazil requires the application of 500 kg P/ha to attain maximum yield (Fageria and Santos, 2008). This is similar to a value extrapolated from Figure 2a for the attainment of the maximum growth of barley when grown on our soil. Moreover, we expect that the acid soil tolerance of transgenic barley will similarly be apparent on a range of soils, including those with lower phosphorus-fixing properties. The transgenic barley expressing TaALMT1 has a similar level of aluminium resistance to a range of aluminium-resistant wheat cultivars. Many of these wheat cultivars have been assessed in the field on a range of soil types, including acid-sandy soils with low phosphorus-buffering capacity (Scott etal., 2001; Garvin and Carver, 2003; Tang etal., 2003), and it is expected that transgenic barley would provide similar benefits to yield.
In summary, we have shown that, when TaALMT1 is expressed in barley, it improves the ability of plants to acquire phosphorus from an acid soil. Even when phosphorus was not limiting, TaALMT1 conferred a substantial yield benefit to plants grown on the same acid soil. Furthermore, TaALMT1 expression was not associated with a yield penalty in the absence of Al3+ stress. These findings demonstrate the effectiveness of genetically engineering an important crop species with TaALMT1 for greater yield potential on acid soils.
Soil and plant material
Plants were grown in an acidic red ferrosol obtained from the Robertson region of New South Wales, Australia (34°35′S, 150°36′E). The pH of a 0.01 m CaCl2 extract (one part dry soil to five parts solution) was 4.1 and the exchangeable aluminium comprised 46% of total cations (Table 1). Although the soil had a high total phosphorus content, it was low in plant-available phosphorus, as indicated by extraction with 0.5 m NaHCO3 (~10 mg P/kg) and by a low resin phosphorus (Liao etal., 2008). The high phosphorus-fixing capacity of this soil is also evident by the need to apply the equivalent of 700 kg P/ha to attain the maximum growth of wheat (Liao etal., 2008). The three transgenic barley lines expressing the TaALMT1 gene and their sibling azygous controls have been described in Delhaize etal. (2004). These lines were generated by Agrobacterium tumefaciens transformation of the barley cultivar Golden Promise with the TaALMT1 gene under the control of the ubiquitin promoter. Line 5, which harbours a single insert (GP-ALMT1), and its sibling azygous line (GP), were used for the phosphorus, rhizosheath and yield experiments.
A range of phosphorus treatments was used on unamended (acid) and limed soil to assess the response of the genotypes to phosphorus fertilizer. For limed treatments, CaCO3 was added at a rate of 4 g/kg dry soil to yield a pH of about 4.6 across all phosphorus treatments (Table 1). The application of lime reduced the amount of soluble aluminium to 6% or less of the total cations in all phosphorus treatments, which was less than 14% of that in acid soil (Table 1). Basal nutrients were added to all soils at rates described by Liao etal. (2008), except that the treatments were mixed as 13.5-kg batches of soil. Phosphorus was then applied as finely ground triple-superphosphate (20.7% phosphorus) at rates of 0, 125, 250, 500 and 1000 mg P/kg, and the soil was again mixed thoroughly. Water was then added to bring the soil to 80% field capacity and re-mixed before 1.4-kg aliquots were dispensed into cylindrical pots (diameter, 9 cm; height, 12 cm). The amount of plant-available phosphorus in the soil, as estimated by Colwell extraction (2.5 g soil mixed in 100 mL of 0.5 m NaHCO3 over 16 h), increased as the phosphorus rate was increased, and was similar in both acid and limed soils for a given phosphorus treatment (Table 1).
Treatments consisted of four replicate pots for each of the phosphorus rates, lime treatments and plant lines. Seed was treated for 10 min with a solution of sodium hypochlorite (4.2 g/L), rinsed with deionized water and then germinated on wetted filter paper in Petri dishes. A single pre-germinated seed was planted in each pot, and pots were watered daily and taken to weight every second day to maintain the soil at 80% field capacity. The pots were arranged randomly on benches within each of four blocks in a naturally lit glasshouse. After 24 days of growth, the plants were harvested to determine shoot dry weight, root dry weight, root length and phosphorus content of tissues. Roots were washed free of soil, scanned whilst immersed in water on a flatbed scanner and their lengths determined using the software package WinRHIZO Pro v. 2002c (Regent Instruments Inc., Quebec, Canada). Shoots were dried and analysed for phosphorus using the procedures described by Liao etal. (2008).
Short-term experiments to measure root elongation and rhizosheaths
Root elongation of seedlings was measured after 5 days of growth to assess the relative acid soil tolerance of the three transgenic barley lines expressing TaALMT. Acid or limed soil with no phosphorus added was prepared as described for the phosphorus trial, and 280 g were added per pot. Seedlings were surface sterilized as described above, and a single pre-germinated seed was planted in each pot. Pots were watered by weight to 80% field capacity every day and, after 5 days of growth, the total root lengths of seedlings were measured.
A rhizosheath is defined as the soil that remains firmly adhered to a plant root when it is removed from the surrounding soil (Watt etal., 1994). The procedures used to measure rhizosheaths were modified from those of Watt etal. (1994), and are based on measurements of root and soil weight rather than the radius of the adhering soil. The structural properties of the red ferrosol were particularly well suited for measuring rhizosheaths, as it formed few soil lumps when wet and moisture was evenly distributed throughout the soil when mixed with water. To determine the ability of seedlings to develop rhizosheaths, the same experimental set-up as described for the phosphorus trial was used, except that seedlings were harvested after 7 days of growth. The measurement of rhizosheaths in older seedlings proved to be difficult and subject to variability as a result of the large root mass formed by laterals and secondary roots. Smaller seedlings, prior to developing extensive lateral roots, were tipped more easily out of pots and removed with minimal agitation, together with any adhering soil, before being excised from shoots, weighed and their lengths measured. Rhizosheaths are expressed as gram per metre of root, and included both the fresh root and moist soil weights.
Long-term growth experiment
Pots consisted of cylindrical polyvinylchloride (PVC) tubes, 10 cm in diameter and 105 cm in length, that had been cut longitudinally into two halves. The lengths of cut PVC tubing were re-assembled into cylinders, a PVC push-on cap with an internal diameter of 10 cm was inserted on to one end and the two halves were taped together. Wetted soil with basal nutrients (minus phosphorus), and either limed or unamended, was prepared as described above, and 8.15 kg were added to each pot. Phosphorus (4.8 g triple-superphosphate; equivalent to 1 g P) was mixed in 1 kg of dry soil for both the limed and unamended treatments that had basal nutrients already applied, taken to 80% field capacity with deionized water (220 mL), mixed again and then dispensed (1.22 kg/pot) above the soil that lacked phosphorus in the pots to yield a final weight of 9.35 kg wetted soil per pot. A single pre-germinated seed was planted in each pot (five replicates for each treatment and genotype combination), and the pots were watered daily and taken to weight every second day to maintain 80% field capacity. The pots were arranged as a random block design consisting of five blocks in a naturally lit glasshouse set at night and day temperatures of 10 and 24 °C respectively. The shoots were monitored for disease or insect damage and sprayed when required. Samples of the youngest fully expanded blades were taken at booting for elemental analysis by inductively coupled plasma-atomic emission spectrometry after digestion in HNO3/H2O2 (9 : 1, v/v), and for nitrogen analysis by mass spectrometry using the method described by Dumas (1981). After 156 days of growth, and when the grain was physiologically mature, water was withheld from the pots and the plants were allowed to air dry. At harvest, whole shoots were removed, air-dried and weighed. After removal, the heads were threshed to determine the grain yield. The pots were split open and the soil was divided into sections, 10 cm long, down the length of the pot. Roots were washed from the soil, collected on a 2-mm sieve, dried and then weighed.
Experiments were set up as randomized blocks (four or five depending on the number of replicates in the experiment). Data are presented as the mean of four or five replicates and, where shown, error bars represent one standard error of the mean. Significant differences between means were established using general analysis of variance (anova), and treatment means were compared by least-significant difference (LSD; P = 0.05) (Genstat v11; Rothamsted Experimental Station, Harpenden, Hertfordshire, UK). The data in Figures 2 and 3 were analysed both as untransformed data and as data transformed to log10 to take into account differences in error variances.
Phillip Taylor was supported by a Summer Studentship jointly funded by CSIRO, The Australian Pastoral Trust and the Grains Research and Development Corporation of Australia.