Histopathological assessment of infection by the crown rot pathogen Fusarium pseudograminearum in wheat seedling tissues was performed using fluorescence microscopy. The coleoptiles and leaf sheaths of four host cultivars of differing susceptibility were examined. Leaf sheaths were most frequently penetrated via stomata, indicated by initial lesions forming at the guard cells. Internally, cell wall penetration was facilitated by penetration structures which appeared as hyphal swellings or septate foot-shaped appressoria. Colonization of leaf sheaths resulted in the re-emergence of hyphae from stomata on both surfaces of the sheath. These hyphae are hypothesized to have two major roles; first as exploratory hyphae for colonization of new tissues, and secondly as sites of profuse conidial production. The formation of conidia on the leaf sheath surface was only recorded on the most susceptible bread wheat genotype. No other major differences in host–pathogen interactions were observed among these cultivars. Almost all cell types in the leaf sheath tissues were extensively colonized, except for the vascular bundles and silica cells. This investigation provides the first comprehensive assessment of F. pseudograminearum infection structures and growth patterns during the infection of wheat seedlings.
Partial resistance to crown rot has been identified in some bread wheat genotypes (Wildermuth & McNamara, 1994); however, the mechanisms involved in this resistance have not been identified. In Australia, resistance to crown rot in commercial bread wheat cultivars is intermediate at best, with only a few cultivars reaching this level (DEEDI, 2011). This partial resistance is limited in its ability to restrict crown rot disease and can be overwhelmed under high disease pressure.
Histopathological examination of plant tissues during infection provides essential information relevant to understanding pathogenesis and mechanisms of resistance. To date, several studies have reported fungal growth patterns for F. culmorum and F. graminearum during crown rot (Stephens et al., 2008; Beccari et al., 2011) and head blight infections (Pritsch et al., 2000; Jansen et al., 2005; Rittenour & Harris, 2010). However, the strategies and growth patterns employed by F. pseudograminearum for host penetration and colonization during symptom development have not been reported and there has been no direct comparison of cultivars differing in susceptibility.
This study has observed the general growth patterns of F. pseudograminearum in coleoptiles and seedling leaf sheaths during crown rot disease using a recently developed differential fluorescence staining technique (Knight & Sutherland, 2011). The progress of infection in the partially resistant bread wheat cultivar 2-49 and the susceptible bread wheat cv. Puseas were selected for significant microscopic examination in an attempt to compare the patterns of fungal growth in the host tissues and observe any differences in pathogenesis between these genotypes. These growth patterns were also compared with those observed in two commercial cultivars, EGA Wylie and EGA Gregory.
Materials and methods
Fungal strain and inoculum preparation
All experiments were performed using a highly aggressive F. pseudograminearum isolate A03#24, collected from Tara, Queensland. A pure culture was grown and maintained on spezieller nährstoffarmer agar (Nirenberg, 1976). Subcultures were grown on starch nitrate agar (Dodman & Reinke, 1982) in the dark at 25°C for 14 days. Conidia were collected by flooding plates with 5 mL of a 6% Tween20 solution before filtering through several layers of cheesecloth with high purity water (milli-Q system, Millipore), to a volume of 10 mL. The conidial suspension was quantified using a counting chamber (Weber and Sons) and diluted to a concentration of 106 conidia mL−1. The final suspension was stored for a maximum of 2 months at −70°C before inoculation.
Plant growth, inoculation and disease assessment
Four bread wheat genotypes demonstrating a range of resistances to crown rot were assessed: 2-49 (intermediate resistance), EGA Wylie (moderately susceptible), EGA Gregory (susceptible) and Puseas (highly susceptible). Seeds were surface sterilized in a 2% sodium hypochlorite solution, then rinsed. Each seed was transferred to a 5 × 5 × 10 cm pot and planted at a depth of 2·5 cm in Premium Hi-Retention growing medium (Power Blend). All seedlings were grown in an environmentally controlled glasshouse at 24°C day/15°C night for 14 days.
Seedlings were inoculated 14 days after planting using a method modified from Mitter et al. (2006). Seedlings were laid horizontally, each coleoptile was gently rubbed with a sterile toothpick at the inoculation site and then 6 μL of the 106 conidia mL−1 inoculum was placed onto the tip of the coleoptile. Control plants were inoculated with sterile water. After inoculation, seedlings were placed in a growth chamber in the dark for 48 h with two cycles of 25°C for 14 h followed by 15°C for 10 h at 100% relative humidity. After 48 h, seedlings were placed upright in the growth chamber and subsequently maintained under the same temperature regime with 14 h daylight at a photon flux density of 150 μmol m−2 s−1. Plants were positioned in a randomized block design.
Harvest of seedling material of all four genotypes was performed at weekly intervals from 0 to 35 days after inoculation (dai). Separately harvested tissues included coleoptiles and leaf sheaths. At least 10 plants of each genotype were examined at 14 dai. A minimum of 5 plants was collected at 2, 5, 7, 21, 28 and 35 dai. Leaf sheaths of each plant were individually examined microscopically, including all diseased leaf sheaths, up to and including the first non-infected leaf sheath.
Tissue preparation, staining and microscopy
Tissues that displayed the initial formation of lesions were selected for microscopic examination using bright field and fluorescence microscopy. After clearing and fixation (Hückelhoven & Kogel, 1998; Schäfer et al., 2004), the surfaces of leaf sheath tissues were observed. Transverse sections were produced by hand-sectioning using a microtome blade (Feather). Briefly, tissues were placed in a clearance solution [0·15% trichloroacetic acid (w/v) in ethanol: chloroform (4:1; v/v)] for 48 h, with the solution being changed once during this time. The samples were then washed 2 × 15 min with 50% ethanol, 2 × 15 min with 50 mm NaOH and 3 × 10 min with milli-Q water, subsequently followed by 30 min of incubation in 0·1 m Tris-HCl (pH 8·5). Samples were either immediately stained or stored in 50% (v/v) glycerol.
Tissues were stained as described by Knight & Sutherland (2011). Briefly, tissue samples were stained for 5 min in safranin O solution [0·2 g safranin O (Michrome, Edward Gurr Ltd), in 100 mL of 10% (v/v) ethanol] and washed three times in water. The samples were then stained for 10 min in solophenyl flavine 7GFE [Ciba Specialty Chemicals; 0·1% (w/v) with 0·1 m Tris-HCl (pH 8·5)] and finally washed 4 × 10 min with water. Samples were mounted in 50% (v/v) glycerol.
Tissues were observed using a Nikon Eclipse E600 fluorescence microscope. Unstained tissues were observed using bright-field microscopy. Stained tissues were viewed under fluorescence filter UV-2A (excitation filter, 330–380 nm; dichroic mirror, 400 nm; barrier filter, 420 nm). Images were captured using a MicroPublishing 5·0 RTV digital camera (QImaging, Canada) and analysis software (Soft Imaging System).
Comparisons between F. pseudograminearum hyphal widths in individual genotypes were performed by analysing values in a one-way anova and the multiple comparisons Tukey HSD test (α = 0·05) in spss v. 17.0. Homogeneity of variance was monitored using the Levene statistic.
Initial penetration and growth in the coleoptile
Inoculum was placed on the coleoptile and this tissue was subsequently examined as the first potential site of hyphal growth and penetration. Germination of the conidial inoculum was not observed until 48 h after inoculation, possibly because at earlier times the fixing and clearing procedures may have dislodged the conidia. At 48 h after inoculation thin hyphae, originating from conidia, were growing on the surface of the coleoptile tissue. Surface hyphae frequently grew appressed to the host surface. Direct penetration of the epidermal cells was facilitated by slightly thickened hyphal tips, underneath which small, brown, halo-like lesions appeared. Penetration events into the coleoptile were infrequently observed; however, in these few cases, penetration occurred either directly through epidermal cell walls or over the union of vertical cell walls. Associations with stomata were infrequent and trichomes are not present on coleoptiles. Following penetration of the coleoptile cuticle, hyphae were observed to grow within cells of the epidermis, cortical parenchyma and less frequently in cells of the vascular bundles.
Penetration of leaf sheaths
For the first 3 weeks after germination the coleoptile remained tightly wrapped around the outside of leaf sheath 1 (leaf sheath cell types/structure are described in Figure 1a,b). Hyphal growth from the coleoptile across to the abaxial surface of leaf sheath 1 followed two routes, either by growth from the coleoptile edge onto and then across the leaf sheath surface or by penetration of hyphae directly from the inner surface of the coleoptile tissue into the abaxial leaf sheath epidermis. This was first observed between 2 and 5 dai. Surface growth of hyphae on the abaxial surface of a leaf sheath was most typically covered by subtending tissues. Initial surface growth of hyphae was a frequent feature on leaf sheaths, where hyphae spread across the leaf sheath surface between trichomes. Hyphal growth around trichomes occurred from 48 h after inoculation (Fig. 1a,c) in all host genotypes. This frequently involved multiple hyphae growing to and around an individual trichome. All wheat genotypes displayed similar colonization patterns.
Hyphal penetration of trichome bases and stomatal apertures was frequently observed (Fig. 1d). Penetration was facilitated by appressoria-like structures, typically occurring as simple swellings at the end of hyphal tips but also as hook-like extensions around trichomes or into stomata (Fig. 1e,f). It appeared that the penetration of trichome bases involved hyphae forming over the union of vertical cell walls at the trichome base. On occasions, hyphae growing between tightly wrapped coleoptile and leaf sheath tissues grew in a coralliform mass on the inner leaf sheath abaxial surface, with hyphae penetrating through the aperture of adjacent stomata (Fig. 1g).
Lesion initiation was observed to coincide with sites of fungal penetration at stomata (most frequently), at the base of trichomes, at wounds and at sites of direct fungal penetration of the epidermis (least frequently). Tissues displaying initial lesions were collected from 5 to 26 dai and ranged from leaf sheath 2 to leaf sheath 5 of Puseas, 2-49, EGA Wylie and EGA Gregory. Under bright-field microscopy, initial lesion formation was frequently observed at the guard cells, which became light to dark brown (Fig. 2a). Uninfected stomata did not display guard cell discoloration (Fig. 2b). Penetrated trichome cells appeared randomly across the epidermal surface and displayed a similar discoloration (Fig. 2d). In contrast, discoloration of guard cells frequently occurred along rows of stomata, with independent, individual lesions at each stoma (Fig. 2e). Lesion formation and cell discoloration progressed into adjacent epidermal and internal parenchyma cells, resulting in the characteristic visible symptoms of crown rot infection (Fig. 2c). These cells showed decreased fluorescence and became distorted during colonization. Within guard and epidermal cells the brown discoloration appeared to be associated with cytoplasmic and cell wall granulation.
In developing leaf sheath lesions hyphae were observed penetrating stoma, epidermal cells and trichomes, all of which displayed discoloration (Fig. 2f). Epidermal cells adjacent to penetrated stomata initially appeared to become densely colonized without further penetration of the hyphae (Fig. 2g). Hyphae within such cells frequently appeared to grow towards a corner of the cell en masse (Fig. 2h). This concentration of hyphae within individual epidermal cells was observed in all four wheat genotypes. In tissues displaying widespread browning, the hyphal infection edge, predominantly occurring in epidermal cells, coincided with the margin of the discoloured zone.
Hyphal growth within leaf sheaths
Penetration of host cell walls was facilitated by swellings of the hyphal tips contacting the cell wall. Penetration structures ranged from swellings of hyphal tips (Fig. 3a) to distinct septate foot-shaped appressoria (Fig. 3b). The formation of a penetration peg was frequently observed under the foot-shaped septate appressoria during cell wall penetration, forming an obvious hyphal constriction as it passed through the cell wall, returning to normal diameter once wall penetration was successful. Hyphal bending within cells was observed along with indentations in the cell wall underneath hyphal tips attempting a penetration event (Fig. 3a).
The epidermal cells of both leaf sheath surfaces showed the highest degree of hyphal proliferation and were the cells where the lateral hyphal infection edge was located. Differences in hyphal growth patterns were observed between the abaxial and adaxial epidermal cell layers and between different cell types.
On the abaxial leaf sheath surface hyphae rapidly penetrated through stomata into the epidermal cells. After colonization of epidermal and cortical parenchyma cells, multiple hyphae re-emerged from stomata on the abaxial surface (Fig. 3c,d). These hyphae spread across the surface of the leaf sheath during favourable environmental conditions. In Puseas, masses of conidia occasionally formed on coleoptile tissues by 10 dai, with hyphae erupting through the epidermal cells. Conidia were also observed to form on hyphae emerging from abaxial stomata at the base of heavily infected leaf sheath 1 tissues of Puseas by 14 dai (Fig. 3e,f). While leaf sheaths or coleoptiles of 2-49, EGA Wylie and EGA Gregory were also heavily infected with hyphae, the formation of conidial masses was not observed. During heavy infection hyphae were occasionally observed in the bases of broken trichomes or growing inside whole trichomes. Colonization of coleoptiles and leaf sheaths was frequently more extensive along one of the vertical margins. Following infection of the epidermis, hyphae were seen to penetrate into the parenchymatous tissues. Hyphae and penetration attempts were not observed in the vascular tissues of the leaf sheaths, in contrast to those of coleoptiles. Failed penetration attempts were observed at silica cells, in which hyphae were never observed (Fig. 3h).
Hyphae readily extended laterally within the tissue through the epidermal cells located between the parallel vascular bundles. However, the vascular bundle and associated stereome tissue appeared to restrict lateral hyphal growth (Fig. 3g). On the abaxial surface the vascular bundles are covered by a single layer of epidermal cells connected by a stereome directly to the bundle sheath cells (Percival, 1921; Fig. 1b). This single layer of epidermal cells included unspecialized epidermal cells which were regularly interspaced with sinuous silica cells. These silica cells were typically most common close to the vascular bundles. When growing through the epidermis adjacent to the bundle sheath, the hyphae were observed to grow only in the unspecialized epidermal cells between the silica cells, forming a network across this layer which grew around but not through the silica cells (Fig. 3i). No difference in silica cell morphology or frequency was observed between the wheat genotypes.
In contrast to the abaxial epidermis, the adaxial epidermis of the leaf sheaths contained no trichome or silica cells and stomata were infrequent. This cell layer supported dense growth of F. pseudograminearum mycelium, both internally and on the surface. It was difficult to discern if hyphae penetrated through from the abaxial to the adaxial surface or grew around the leaf sheath edge from the abaxial surface. It is most likely a combination of both, with rates of hyphal surface growth being dependent on the environment. The leaf sheath adaxial surface was typically in loose contact with the abaxial surface of the younger leaf blade and sheath furled within it. This humid and sheltered location was the site for the formation of mycelial mats. These mats were observed most frequently on the highly susceptible cultivar Puseas. Observations of 30 leaf sheaths at 14 dai indicated hyphal mats on six 2-49, six Gregory and 20 Puseas leaf sheaths. As with the abaxial surface, cells adjacent to the vascular bundles influenced hyphal growth. The adaxial surface adjacent to a vascular bundle included epidermal cells and typically two layers of cortical parenchyma cells, which were directly adjacent to the bundle sheath. Hyphae grew laterally through epidermal cells until encountering cells adjacent to a vascular bundle. At this point hyphae either grew across the leaf sheath surface after emerging from the epidermal cells, or, more frequently, produced large quantities of very thick hyphae, with obvious swellings, which facilitated crossing of the non-vascular cell walls within the tissue. In some cases a foot-shaped appressorial structure was obvious (Fig. 3b); however, in other cases penetration structures were swollen and oblong in shape (Fig. 3j). These thickened hyphae were frequently tightly packed, filling the cell cavity, and produced multiple cell wall crossing events. It could not be determined using the described methods if these hyphae had penetrated the cell membrane, were growing around the cell membrane or whether penetrated cells were alive or dead.
Comparison of growth in different genotypes
Measurements of F. pseudograminearum hyphal widths (n =26) across leaf sheaths 1–4 at 14 dai indicated that hyphae growing in cultivars 2-49 (mean width 4·4 μm, range 1·4–12·2 μm), EGA Wylie (mean width 5·1 μm, range 2·7–11 μm), EGA Gregory (mean width 5 μm, range 2·3–11·6 μm) and Puseas (mean width 5·2 μm, range 2·1–20·5 μm) were not significantly different in width. The maximum hyphal width observed growing in Puseas was greater than those observed in the other genotypes; however, hyphae with this increased width were rare. During infection of the leaf sheath, hyphae of F. pseudograminearum could be divided into three size classes. These were thin hyphae (1–4 μm width) observed on the surface of coleoptile or leaf sheath tissues, thicker hyphae (4–8 μm width) growing within cells, typically adjacent to vascular bundles, and very thick hyphae (8–20 μm width) occurring infrequently but only in cells adjacent to a vascular bundle on the adaxial surface. These very thick and generally short, bulbous hyphae appeared to facilitate cell wall crossing in the same manner as appressoria but without the consistent foot shape observed in cells of the abaxial surface.
The trial comparing 2-49, EGA Gregory and Puseas at 14 dai indicated more rapid hyphal growth across leaf sheaths 1 to 3 on Puseas tissues compared to tissues of the other genotypes. Specifically, more hyphae were visible on Puseas leaf sheath 1 compared to 2-49 and EGA Gregory. On leaf sheath 2 of 2-49, hyphae were observed only on the abaxial surface and penetrating into stomatal apertures. In contrast, on leaf sheath 2 of EGA Gregory and Puseas, hyphae were present on both surfaces with significant quantities of hyphae penetrating through their internal tissues. Tissues of Puseas typically contained areas of more extensive hyphal growth than EGA Gregory at 14 dai. At this time no hyphae were present on leaf sheath 3 of 2-49, while leaf sheath 3 of EGA Gregory bore small quantities of hyphae on the abaxial surface. Leaf sheath 3 of Puseas varied in infection intensity between individual plants, with small amounts of surface hyphae on some leaf sheath 3 samples while on others mycelial mats were forming accompanied by significant internal growth in the adaxial epidermal cells. In all attempts to infect host plants, all inoculated plants became diseased. Hyphae were never observed on the tissues of control seedlings.
This report is the first to conduct a thorough histopathological examination of the growth of F. pseudograminearum in wheat seedlings during crown rot pathogenesis. It has documented initial hyphal interactions with the host surface, followed by penetration and internal hyphal growth patterns, culminating in faster spread of the pathogen in more susceptible cultivars. A comparison of the extent of colonization in three cultivars at 14 dai showed production of conidia in the most susceptible cultivar.
Initial hyphal growth on leaf sheath tissues was observed to involve frequent interactions with trichome bases. While the clearing and fixation procedures may have resulted in hyphae wrapping around trichomes, the close proximity to and tight formation of hyphae against the trichomes suggest that hyphal wrapping is a method by which the fungus attaches itself to the host surface. Even though trichome penetration events were not clearly observed, the formation of lesions at the base of trichomes provides evidence that trichomes can be initial points of interaction. Similar observations of hyphal wrapping and penetration of trichomes have also been reported for F. solani f. sp. phaseoli, F. lini and F. graminearum (Nair & Kommedahl, 1957; Mulligan et al., 1990; Stephens et al., 2008).
Leaf sheaths were most frequently penetrated via stomata, indicated by initial lesion formation at the guard cells and hyphal penetration of the stoma. Stomatal penetration has been reported for several Fusarium phytopathogens. Initial flecks of discoloration reported during F. solani infection of pea plants have been shown to be stomata penetrated by hyphae (Christou & Snyder, 1962). Similarly, F. graminearum, F. culmorum and F. nivale have been reported to penetrate tissues of cereals via stomata (Malalasekera et al., 1973; Pritsch et al., 2000; Rittenour & Harris, 2010; Beccari et al., 2011). In addition to stomatal penetration of leaf sheath tissues by F. pseudograminearum, re-emergence from stomata by multiple hyphae was also a common occurrence. Stomata on the abaxial surface provided a site for hyphal emergence and subsequent spread across the tissue surface and also a site for conidial production and dispersal in Puseas, the most susceptible cultivar examined, while the adaxial stomata appeared to provide a route for hyphae to spread inwards and across to subtended younger leaf sheaths.
The infection edge of the hyphal mycelium within a leaf sheath progressed through the epidermal cell layer of the host and was located in close proximity to the leading edge of symptom development. Comparisons of fungal growth within the host cultivars examined failed to reveal differences in hyphal structures, infected tissue anatomy, host responses or the types of tissue colonized. A detailed comparison between genotypes involving lesion measurements was not undertaken due to significant variation in lesion development on different plants of the same genotype. In general there was earlier and more extensive colonization in Puseas leaf sheath tissues compared to the other cultivars, leading, within the time course of the experiment, to prolific production of conidia at the surface of some infected stomata on the abaxial surface of leaf sheath 1 of this host genotype. This conidial production was not observed in the other host genotypes. These differences in sporulation require examination in the field to determine their significance in crop resistance.
While this initial study of the histopathology of F. pseudograminearum infection in seedlings has been descriptive in focus, the observation that lesion development is more rapid in more susceptible host genotypes is supported by a parallel quantitative PCR study which has reported significantly higher levels of F. pseudograminearum biomass in Puseas compared to 2-49 at equivalent time periods after seedling infection (Knight et al., 2012). Percy et al. (2012) similarly described variation in the rate of F. pseudograminearum infection spread in different genotypes. Both Knight et al. (2012) and Percy et al. (2012) reported a strong correlation between symptom development and fungal biomass in leaf sheath tissues.
The study by Percy et al. (2012) also provides strong evidence that the lateral growth of the pathogen from the adaxial surface of older, infected leaf sheaths directly into the abaxial tissues of subtended younger leaf sheaths is a major route of host tissue colonization. This is in addition to any infection via the basal crown tissues. The further observation of this current study, that F. pseudograminearum does not penetrate the vascular tissues of seedling leaf sheaths, thus excluding systemic spread by the vascular pathway, is consistent with the importance of the more direct lateral route.
Within leaf sheath tissues both intra- and inter-cellular hyphal growth occurred. Only silica cells and vascular bundles in the leaf sheath displayed an absence of hyphal penetration, perhaps due to a strengthened cell wall structure. The observed colonization of vascular bundles within coleoptiles may reflect the less differentiated cell wall structures of these tissues. Where penetration of cell walls did occur in other cell types, it was facilitated by swollen hyphal tips with a range of morphologies consisting of slightly swollen hyphal tips (Fig. 3a), hooked hyphal swellings (Fig. 1e,f), swollen oblong-shaped hyphae (Fig. 3j) and well-defined foot-shaped septate appressoria producing penetration pegs (Emmett & Parbery, 1975; Fig. 3b). This has not been previously reported in F. pseudograminearum.
Penetration of cell walls appears to be a primarily mechanical process, as shown by the production of cell wall indentations under some hyphal tips (Fig. 3a; McKeen, 1974; Koga, 1994). Differences in morphology of both hyphae and infection structures appear to be related to the tissue type being colonized. The formation of swollen, oblong hyphae predominantly near vascular tissues suggests that these hyphae may have a role as feeding structures.
Silica cells successfully resisted penetration in all host genotypes observed (Fig. 3h). Increased levels of silica in cell walls has been reported to enhance resistance to attack by pathogenic fungi in various plant–pathogen interactions, including rice blast disease (Pyricularia oryzae; Miyake & Ikeda, 1932), rice leaf spot disease (Helminthosporium oryzae; Jones & Handreck, 1967), bean rust of French bean (Uromyces phaseoli var. typical; Heath, 1981) and spot blotch of wheat (Bipolaris sorokiniana; Domiciano et al., 2010). Silicification in grasses is a normal feature of development (Lewin & Reimann, 1969). Silica occurs as a component of the cell wall, where it is strongly bound to cellulose or lignin, and also forms solid bodies in the lumen of silica cells (Lewin & Reimann, 1969; Ponzi & Pizzolongo, 2003). Earlier studies have failed to discover the role of silica cells (Kaufman et al., 1985) and this remains an open question, particularly with regard to interactions with fungal hyphae. One recent study by Guenther & Trail (2005) has reported that perithecium formation by Gibberella zeae (anamorph F. graminearum) was initiated in association with both stomata and silica cells. Perithecia formed from epidermal cells adjacent to silica cells located in nodes, but hyphae did not colonize the silica cells.
In conclusion, this investigation provides the first comprehensive histopathological overview of F. pseudograminearum–wheat interactions in seedling leaf sheaths. The observations demonstrate that tissue browning only occurs at sites of infection, an observation consistent with other evidence (Knight et al., 2012; Percy et al., 2012). This characterization of hyphal penetration events and leaf sheath colonization has revealed several morphological hyphal adaptations of F. pseudograminearum occurring in response to varying host tissue types. While specific differences between infection patterns of partially resistant and susceptible cultivars were not observed, the rate of fungal spread was faster in the more susceptible host genotypes and conidia were produced only on the most susceptible genotype. These results will greatly assist future studies of host–pathogen interactions and ultimately the development of strategies to control crown rot disease in commercial cereal crops.
The authors would like to thank their colleague Dr Anke Martin for her critical reading of the manuscript. Financial support for this project was in part provided by the Grains Research and Development Corporation, Australia. Noel Knight acknowledges the support of an Australian Postgraduate Award.