Proteoid roots, also called cluster roots, are dense clusters of rootlets of limited growth radiating mostly from the branch roots of members of the Proteaceae as well as some other families (e.g. Purnell, 1960; Gardner, Parbery & Barber, 1981; Lamont, 1982; Louis, Racette & Torrey, 1990). Members of the Proteaceae grow well when available phosphate levels are low and are characteristic of low phosphate habitats. Proteoid roots are usually most abundant when phosphate concentrations are low, and their production is inversely related to phosphate level (Lamont, 1972«; Gardner, Parbery & Barber, 1982 6; Marschner, 1986; Marschner et al., 1986; Grose, 1989; Louis et al., 1990; Handreck, 1991). Phosphate uptake has been shown to be more efficient in proteoid than non-proteoid roots (Jeffrey, 1967; Lamont 1982; Vorster & Jooste 1986«, b). ‘The large surface area increases the secretion of phosphatesolubilizing compounds into localized regions of the rhizosphere (Gardner et al., 1981, 1982«). This is important in the natural environment where phosphorus acquisition is most likely to be limited by the low solubility and diffusion of phosphorus across the soil to the root absorbing surfaces.
Many of the experiments on the physiology of proteoid roots have been done in conditions where sterility was not specified and a potential contribution from the micro-organisms associated with them has not been excluded. The roots are not known to have mycorrhizal or other microsymbiont associations, but a microbial stimulus for their formation has been reported (Lamont & McComb, 1974 ; Malajczuk & Bowen, 1974; Gardner, Barber & Parbery, 1982). Grierson & Attiwill (1989) suggested that micro-organisms may be involved in nutrient uptake in Banksia integrifolia L., but to our knowledge this has not yet been examined. Phosphatesolubilizing bacteria are commonly associated with the roots of a range of species (e.g. Cosgrove, 1977; Stevenson, 1986; Halvorson, Keynan & Kornberg, 1990). We have therefore investigated the phosphatesolubilizing activity of bacteria isolated from proteoid and non-proteoid lateral roots of the Sydney waratah, Telopea speciosissima (Sm.) R.Br. seedlings collected from the wild.
There are several potential mechanisms for phosphate solubilization. These include the modification of pH by the secretion of organic acids or protons. Secretion of H+ from roots into the medium in response to uptake of NH4+ and of OH“ for NO3”
uptake are well documented, and the decrease in pH of the medium in the presence of NH4+ has been correlated with solubilization of inorganic phosphates (Raven & Smith, 1976; Jarvis & Robson, 1983; Marschner et al., 1986; Haynes, 1990; Li, George & Marschner, 1991; Marschner, 1991a). Here we have investigated the effect of various nitrogen sources on the phosphate-solubilizing activity of the bacteria isolated from waratah roots and elsewhere in the soil, and correlated this with their capacity to modify the pH of the medium. This paper does not seek to quantify the contribution of associated bacteria to the phosphate solubilizing activities of roots in their natural environment, but to draw attention to the capacity of the bacteria to solubilize calcium phosphate.
MATERIAL AND METHODS
Plant material used
Telopea speciosissima seedlings (shoot height c. 9 cm) were collected from two native forests in New South Wales, Australia: (1) in the Mittagong state water catchment area (latitude 34° 31’S, longitude 150° 35′ E, altitude 630 m), and (2) from Waratah Ridge, Newnes Plateau, Galah Mountain in the Blue Mountains (latitude 33° 23′ S, longitude 150° 13′ E, altitude 1130 m). The root system and some surrounding soil were excavated and the seedlings transported back to the laboratory.
toluidine blue pH 4-5 in 25 mM Na acetate buffer, rinsed and mounted in water for viewing. Other roots were treated with chlorazol black E by the method of Brundett, Piche & Peterson (1984) to detect fungal hyphae.
Isolation and purification of bacteria
Bacteria were isolated from fresh roots within 24 h of their collection (modified from Gochnauer, McCully & Labbe, 1989). A 1 cm section was removed aseptically from each of 10 ‘young’ and 5 ‘old’ proteoid roots (see results for description of roots) and 10 non-proteoid regions of the lateral roots just basal to the proteoid region. Bacteria from the ‘ rhizosphere ’ were isolated by agitating each root in aqueous 0-05% (w/v) agar solution. The root was then removed and ground in fresh agar solution to isolate the ‘rhizoplane’ bacteria. The ‘soil’ bacteria were isolated from loose soil samples taken from around the collected seedlings. The bacteria were plated onto either starch casein agar (Kuster & Williams, 1964), or peptone tryptose yeast glucose (PTYG) medium (Bone & Balkwill, 1986) modified to contain 1-5 mM sucrose instead of glucose, with 50 mg/1 cycloheximide, and incubated for 2 wk at 20 °C. About 300 morphologically distinct bacterial isolates were obtained. Any isolates which produced colonies and/or bacteria of varying morphology were discarded.
Examination of root morphology
Roots were either examined immediately or after storage at 20 °C in 3 % (v/v) glutaraldehyde and 3 % (v/v) formaldehyde fixative solution in 0-025 M potassium phosphate buffer at pH 6-8. Proteoid and non-proteoid regions of lateral roots were examined with a Wild M400 Photomakroskop, and a Zeiss Universal microscope using Nomarski and bright- field optics. Some rootlets were stained with 1-14 mM
Testing of phosphate-solubilizing activity
The standard phosphate-solubilizing medium (PS medium) contained 15 0 mM sucrose, 2-0 mM MgSO4, 7-0 him KC1, 0-04 mM FeSO4, 10-0 mM KNO.t, 20’0 mM CaHPO4, 0-2 mM bromocresol purple dye, and 1-5% (w/v) agar. The final pH was adjusted to about pH 7-0 and 12 ml of medium added to each Petri plate. All isolates
were tested on the standard PS medium with
Figure 1. Unless specifically stated photographs are of chemically fixed roots, (a) Non-proteoid lateral root about 0-5 cm proximal to a proteoid root, stained by the technique of Brundrett et al. (1984) to illustrate fungal hyphae (arrows). Bright field optics. Bar = 0’25 mm. (b) A very young proteoid root cluster showing no adhering soil material. Non-proteoid regions of the lateral root on either side are marked with asterisks. Whole mount of unfixed fresh root. Bar = 0-5 mm. (c) A proteoid root cluster slightly older than in Pig. 1 b with soil material bound to it. Adjacent non-proteoid regions of the lateral roots indicated by asterisks. Whole mount of unfixed fresh root. Bar = 1 mm. (d) An older proteoid root cluster than shown in Figs 1 b, c. Most of the associated soil has been washed ofl except for a few particles (arrows) still attached to rootlet hairs. Many of the rootlets were beginning to lose their epidermis and cortex. Whole mount. Bar = 1 mm. (e) Old proteoid root with no associated soil material and thin rootlets. Whole mount. Bar = 0-5 mm. (f) The tip region of a rootlet from a young proteoid root cluster at about the stage shown in Fig. 16. The root hairs (arrows) are just beginning to form basal to the root tip (T) and are straight. Whole mount. Bar = 0-1 mm. (g). Rootlet (R) taken from a proteoid root cluster slightly younger than in I-ig. 1 d. The rootlet is bent near its tip (T) and is entirely covered with root hairs. Soil particles (asterisk) and fungal hyphae (arrow) were common among the hairs which were usually distorted (inset). Bright field optics. Bar = 0-1 mm. Inset Bar = 0-03 mm. (/;). Rootlet at a similar stage of development as that in Fig. 1#, showing that hairs developed around the tip (T). Soil particles (asterisk) and fungal hyphae (arrows) were interspersed among the hairs including around the tip. Some hairs were branched and multinucleate (inset). Bar = 0-1 mm. Inset Bar = 0-02 mm.
either KNO3 or NH4NO3 as the nitrogen source. Bacteria were inoculated on the PS medium by making a 1×1 cm cross-shaped slash and cultured in the dark at 20 °C for up to 4 wk.
From the results of the initial tests using NH4NO3, five bacterial isolates with ‘strong’ phosphatesolubilizing activity on the sucrose medium (group A, isolates 1-5) and five that did not grow well nor solubilize phosphate on this medium (group B, isolates 6-10) were selected for the following experiments. To test the effect of the C and N compounds on phosphate-solubilizing activity, 15-0 mM glucose was substituted for 15-0mM sucrose and 10-0 mM Ca(NO3)2, NH4C1, or (NH4)2SO4 for 10-0 mM KNO3 in the standard PS medium, resulting in eight different PS media (2Cx4NxlP). Solubilization of the following different phosphate compounds was also tested (all 20-0 mM): CaHPO4, Ca3(PO4)2, FePO., A1PO4, or K2HPO4. These latter media all contained either 15-0 mM sucrose or glucose and 10 0 mM NH4C1 instead of KNO3 making a further 10 different PS media (2C x IN x 5P). The bacteria were inoculated with a single central cross per plate. After 0-1, 1, 7 and 14 d in the dark at 20 °C the growth of the bacteria, extent of phosphate solubilization and colour change of the pH indicator dye were noted. The pl I of the inoculated and control .(uninoculated) media was measured using a surface pH electrode.
Further characterization of the ten selected bacterial isolates
The following tests were performed on each bacterial isolate: growth on Pseudomonas F and P agar, and on MacConkey agar; gram staining; morphology after culture on all media; motility and ability to form spores (modified from Gochnauer et al., 1989); results with Biolog Gram Negative Bacteria Identification kit (Biolog Co., Hayward, CA, USA).
Effect of HCl on PS media
The change in pH of the PS media was measured 2 h after adding 10 //I of 01 M HCl to the centre of each plate. To test the capacity of HCl to solubilize the different phosphates, 01 ml of 10m HCl was added (10/d 15 min-1) to the centre of each plate.
RESULTS
General root morphology
Non-proteoid lateral roots had no obvious proteoid rootlets and had hyphae of unknown fungi on their surface (Fig. I a). Soil particles did not adhere to newly emerged proteoid rootlets (Fig. 1 b), but did adhere strongly to slightly older proteoid roots even after agitation in water (Fig. 1 c). In yet older
proteoid roots most of the associated soil material was lost on agitation in water except for a few particles attached to rootlet hairs (Figs ld,g,h). Fungal hyphae (species not known) were usually interspersed among the rootlet hairs (Figs \g,h). Proteoid roots up to and including this stage of development were defined as being ‘young’. All seedlings also had ‘old’ proteoid roots with thin reddish-brown rootlets which had very little bound soil material and appeared to be dead, although we did not verify this (Fig. 1 <?).
In proteoid roots such as in Figs i b, c the rootlets had short, straight root hairs (Fig. 1 f). At later stages (e.g. Fig. Id) the root hairs were much longer, distorted, and often branched and multinucleate (Figs lg, h). Some of the rootlets in these proteoid roots had dead epidermal cells (Fig. 2«) but the cortical cells were alive and had nuclei (Fig. 26). In ‘old’ proteoid roots, the epidermis and usually all cortical cells of the rootlets were dead and sloughing off (Fig. 2c) or had been completely lost (Fig. 2d).
Phosphate-solubilizing activity of all bacteria and selection of isolates for detailed study
Two hundred and ninety two bacterial isolates were collected. Of these only 12 weakly solubilized the phosphate on the standard PS medium containing sucrose and KNO3 as indicated by a clear region extending from the edge of the bacterial growth zone (compare Figs 2e,f). However, when NH4NO3 was substituted for the KN03 in the medium, 137 of the isolates from proteoid and non-proteoid roots and from the soil showed phosphate-solubilizing activity (Fig. 2/).
Phosphate solubilization was accompanied by a lowering of pH in the medium. Initially in all cultures the pH indicator transiently turned slightly yellow at the point of inoculation. However, in the standard PS medium containing KNO3 the indicator reverted back to its original colour and thereafter remained purple. The time taken for this was correlated with phosphate-solubilizing activity, from a few hours where there was no solubilization to about 2d for the. 12 isolates that showed some solubilization on this medium. In contrast, with all isolates when NH4NO3 was substituted for KNO3 the entire plate became progressively more yellow, especially with isolates that produced a large clear zone of phosphate-solubilization.
Ten isolates were selected for more detailed examination according to their phosphate-solubilizing activity on the standard PS medium containing NH4NO3 and sucrose. Group A isolates (1-5) had strong phosphate-solubilizing activity and produced a large clear region around their growth zone after 1-2 wk, while group B isolates (6-10) had little or no activity on this medium.
Figure 2. Unless specifically stated, photographs are of chemically fixed roots, (a) Rootlet taken from a proteoid root cluster about the stage shown in Fig. Id. The root hairs are dead and broken. Stele is at left. Nomarski optics. Bar = 01 mm. (b) Rootlet taken from a proteoid root cluster at the stage shown in Fig. \d. The epidermis is lost (at right beyond field of view) and the remaining cortical cells are alive and have nuclei (arrows). Nomarski optics. Bar = 01 mm. (c) Rootlet from ‘old’ proteoid root cluster. Most of the epidermis and cortex were dead and the remaining cells were sloughing off (arrows) to leave only the stele (S). Nomarski optics. Bar = 0 05 mm. (d) Rootlet from ‘old’ proteoid root. Only the lignified stele which extended to the tip (T) remained. Nomarski optics. Bar = 0 02 mm. (e) Bacterial isolate (B) grown on PS medium containing sucrose, KNO3 and CaHPO4 shows no detectable phosphate-solubilization. There is no apparent loss of particulate matter from the agar. Dark field optics. Bar = 0’2 mm. (f) Another bacterial isolate (B) grown on PS medium containing sucrose, NH4NO3 and CaHPO4 has solubilized the phosphate, as indicated by the clear region where CaHPO4 particles have been lost from around the bacterial growth zone. Dark field optics. Bar = 0-2 mm.
Correlation between phosphate-solubilizing activity and modification of pH in the medium by the ten selected bacterial isolates
Effect of nitrogen source. The relationship between solubilization of CaHPO4 and lowering of pH held
for a wide range of conditions. None of the isolates showed any detectable phosphate-solubilization on the PS media containing NO3~ and sucrose (Figs 3 a-c), nor did they acidify the medium apart from an initial brief yellowing around the edges of the inoculum which became purple again within a few
Figure 3a-l. For legend see opposite.
hours (Fig. 3 a). The pH progressively increased and by 1-2 wk the plates were deep purple (Figs 3b, c). Group A isolates changed the media pH more than group B isolates.
The isolates acidified PS media containing NH4+ and sucrose. For group A, a yellow zone 2-3 cm in diameter developed around each cross within a few hours of inoculation (Fig. 3d). After 1-2 wk the entire plate was yellow with pH < 4’5 and a large clear region around each bacterial growth zone (Fig. 3f). In contrast, with group B isolates initially only a very small yellow zone developed around the inoculation slash (Fig. 3e). Even after 2 wk the medium had turned only slightly yellowish (c. pH 6)’ and there was very little or no detectable phosphatesolubilization (Fig. 3g). Results obtained with the pH indicator agreed with values obtained by measuring the surface agar with pH electrode. This lack of reaction was correlated with relatively poor growth of group B isolates on all the sucrose-containing media (compare Fig. 3 b with 3 c and Fig. 3f with 3g).
Effect of carbon source. Glucose enhanced the growth of group B isolates to a level equivalent to that of group A when cultured on either sucrose or glucose. In this case all 10 isolates produced a large clear zone of phosphate solubilization on the media containing NH4+ and significantly acidified the medium (Tigs 3h,f Fig. 4). ‘rhe decrease in pH was about the same for all 10 isolates with glucose present, and was also similar to that obtained with group A isolates on PS media with sucrose and NH4+.
In general no phosphate-solubilization was detected on PS media containing glucose and NO., . However, all group A isolates and isolates 6 and 8 of group B sometimes produced a small zone of phosphate-solubilization during the first day after inoculation when the pH decreased (Fig. 4). I his eflect was transient and by 24 h the pH indicator became purple again, starting adjacent to the inoculum and spreading progressively outwards (Fig. 3/) as the entire plate gradually became more alkaline
(Fig. 3k and 4). Phosphate-solubilization was detected only at the time when the medium was demonstrably acid.
Solubilization of other phosphates
All isolates partially solubilized the Ca3(PO4)2 with glucose as a carbon source, but only group A isolates solubilized this phosphate on sucrose-containing media. None of the isolates caused any detectable solubilization of the FePO4 or A1PO4 under any of the conditions used. All isolates acidified the medium of all PS media tested, although this occurred to a lesser extent with K2HPO4 (Fig. 5) and when group B isolates were tested with sucrose as a carbon source.
Effect of HCI on PS media
Addition of 0’1 M HCI resulted in a significant acidification of all PS media except those containing K2HPO4 which was acting as a buffer (Fig. 5). Addition of 10m HCI resulted in solubilization of CaHPO4, (Fig. 3/), Ca3(PO4)2 less well, FePO4 only slightly and A1PO4 not at all (Fig. 6). Even addition of 1’0 ml of 10-0 M HCI did not solubilize the A1PO4.
Characteristics of the ten bacterial isolates
All 10 isolates were gram-negative, motile and not spore-formers. They were rod-shaped with dimensions 0’25-0-5 /¿m x 1-2 /mi, but became more cocco- bacilloid after 1-2 wk culture on most media. Group B isolates remained filamentous and/or swelled up to 4 /zm in diameter after 2 wk culture on PS media containing NH4C1, glucose and A1PO4 and sometimes also with FePO4. All isolates except group B isolates 9 and 10 grew well on MacConkey agar. All isolates grew on the Pseudomonas F and P agar but no fluorescence was seen under UV irradiation.
‘The Biolog results indicated the following: only group A isolates metabolized quinic acid ; none of the
Figures 3a-l. Phosphate solubilization and colour change of pH indicator with various isolates on standard PS media containing CaHPO4 with the following variations: (i) sucrose and KNO., (Figs a-c), (ii) sucrose and NH4C1 (Figs d-g, I), (iii) glucose and NH4C1 (Figs, g, i), and (iv) glucose and KNO., (Figs;, k). Figures 3a-k have been inoculated with either group A isolate (1) or group B isolate (6) for various time periods. Fig. 3/ has had 60 //I of 10 M HCI added to the centre of the plate at a rate of 10//I 15 min”1. Bars = 2-0 cm. (a) Isolate 1 at 0-1 d. There is little change in pH indicator from its original purple colour and no evidence of solubilization. (b) Isolate 1 at 2 wk. ‘The medium is bright purple and there is no evidence of solubilization, (c) Isolate 6 at 2 wk. The medium is slightly brighter purple. There is less bacterial growth than in Fig. 3b and no solubilization is apparent, (d) Isolate 1 at 0-1 d. 1 he pH indicator is bright yellow in a large zone around the bacteria and there is no detectable solubilization, (e) Isolate 6 at 0’1 d. Only a very narrow zone of pH indicator is yellow around the bacteria and there is no solubilization, (f) Isolate 1 at 2 wk. The entire plate is bright yellow and a clear ring around the bacteria indicates solubilization of the phosphate, (g) Isolate 6 at 2 wk. The medium is more yellowish, especially around the bacteria. There is little bacterial growth or apparent solubilization, (h) Isolate 6 at 01 d. A large zone around the bacteria is yellow and no solubilization is detected. (i) Isolate 6 at 2 wk. The entire plate is yellow and a large clear zone around the bacteria indicates solubilization of the phosphate, (jj Isolate 6 at 1-0 d. The pH indicator adjacent to the bacteria is purple surrounded by a yellow halo. (Ze) Isolate 6 at 2 wk. The entire plate is bright purple and no solubilization is apparent. (/) The HCI has turned the pH indicator yellow and a large zone of the phosphate is solubilized.
Figure 4. The effect of inoculation of a group A isolate (1) or a group B isolate (6) on the pH of PS media containing CaHPO4, glucose and KNO3 or NH4C1 as measured by pH electrode. The measurements are taken directly over the bacterial isolate in the centre of the plate (n = 3 ; + se). With KNO3, initially both isolates acidified the medium but thereafter the medium became more alkaline. Both isolates acidified the medium containing NH4C1 to the same extent, aside from the slight increase in pH at 2 wk with isolate 1. Treatments are as follows: (□) control with no inoculation, NH4C1 medium; (△) isolate 1, NH4C1 medium; (O) isolate 6, NH4C1 medium; (■) control with no inoculation, KNO3 medium; (A) isolate 1, KNO3 medium; (•) isolate 6, KNO3 medium.
Figure 5. Comparison of the effect of group A isolates (□) after 2 wk growth and HC1 (S) 2 h after application, on the pH of media containing sucrose, NH4C1 and various phosphates. A volume of 10 //I of 01 M HC1 was added. Uninoculated control plates (■) are also shown (n = 3; ±se). All measurements were taken directly over the centre of the plate (w = 5; + se). For each medium, one plate of each of the five isolates of group A was measured, giving n = 5, except for CaHPO4 when n = 15. The bacteria and HC1 significantly acidified all PS media except when K2HPO4 was present.
isolates metabolized //-lactose, even though growth on MacConkey agar indicated positive lactose fermentation ; none metabolized sucrose but all metabolized glucose. The Biolog data base provided a tentative identification of 6 isolates. Isolates 2 and 6
Figure 6. Comparison of the effect of HC1 on solubilization of various phosphates. Change in diameter of solubilized spot in PS media containing sucrose, NH4C1 and different phosphates after addition of 100/¿I of 1-0 m HO at a rate of 10 /d 15 min 1 (n = 5; ±se). The CaHPO4 (■) was readily solubilized, Ca3(PO4)2 (A) to a lesser extent, the FePO4 (#) only slightly and A1PO4 (□) not at all.
were closest to Pseudomonas mucilodens even though on PTYGa medium isolate 2 had translucent, flat, ruffled colonies and isolate 6 had round, elevated colonies with a yellow diffusing pigment. Isolate 1 was closest to Acidovorax sp., 3 to Pseudomonas solanacearum B., 4 to Deleya aesta, and 8 to Acidovorax delafieldii.
DISCUSSION
Mechanism of bacterial phosphate solubilization
Solubilization of CaHPO4 was positively correlated with bacterial acidification of the medium below about pH 5-0. There are many reports of secretion of organic acids such as oxalic, malic or citric from bacteria or roots which solubilize phosphate (Cosgrove, 1977; Mengel & Kirkby, 1978; Gardner, Barber & Parbery, 1983; Marschner, 1986; Marschner, Romheld & Cakmak, 1987). Our results suggest that the acidification may have been caused by the secretion of H+ which would also solubilize phosphate by exchange with Ca2+. The secretion of H+ in exchange for NH4+ ions is a well documented phenomenon in both bacteria and roots (e.g. Raven & Smith, 1976). The transient initial acidification of the medium in the presence of NO3 ions arose probably because of a high initial cation exchange with the K+ or Ca2+ in the medium, but the ultimate pH increase was probably due to the extrusion of OH during uptake of NO./ . Furthermore phosphate solubilization by HC1 was essentially similar to that of the bacteria, agreeing with reports that ferric and aluminium phosphates are less soluble than calcium phosphates at acid pH (Bohn, McNeal & O’Connor, 1979; Stevenson, 1986).
Ecological significance
There are many reports that proteoid roots acidify the rhizosphere, in some cases due to organic acid secretion, and solubilize Ca, Fe and Al phosphates (Gardner et al., 1981, 1982a, 1983; Marschner, 1991a, 6; Marschner et al., 1986; Dinkelaker, Romheld & Marschner, 1989; Grierson & Attiwill, 1989). The work here suggests that the bacteria associated with proteoid and non-proteoid roots of the waratah have the potential to acidify the rhizosphere and solubilize Ca phosphates. Most bacteria, including nitrifying bacteria, secrete H+ in the presence of NH4+ (Mengel & Kirkby, 1978; van Breemen, Mulder & Driscoll, 1983; Salsac et al., 1987; Raven, Wollenweber & Handley, 1992).
We have not attempted to quantify the actual contribution of these bacteria to phosphate solubilizing activities associated with waratah roots in their natural environment. We have only focused here on the activities of 10 of the approximately 300 bacterial isolates we obtained. Hence we do not know if all isolates would be capable of solubilizing Ca phosphate in the presence of glucose and NH4+, nor if any could solubilize Fe or Al phosphates. The solubilizing activities of all isolates and their relative abundance on waratah roots needs to be determined, as well as the local soil conditions such as the type of phosphates and concentration and availability of C sources and NH4+. The concentrations of C and N used in our experiments were probably high compared to those in the native environment. Finally the phosphate-solubilizing activities of the total root- associated bacterial population would need to be quantified and compared to that of the roots themselves in order to assess the ecological significance of bacterial phosphate-solubilization.
The developmental stages reported here for the proteoid roots of Telopea speciosissima are similar to those reported for other proteoid roots. The aggregation of soil particles around the young developing proteoid roots and their loss as the rootlets matured (Figs 1 b-e) have also been reported for proteoid roots of other Proteaceae (Lamont, 19726). Interestingly, this aggregation bears a resemblance to the soil sheath found in corn and other cereal roots which, in these cereals, is indicative of immature xylem vessels (McCully & Canny, 1989). The proteoid root-hair distortions have been reported for other Proteaceae (Purnell, 1960; Jeffrey, 1967; Lamont, 19726). There are many reports showing bacteria associated with the distorted regions of root hairs of other species (e.g. Gochnauer et al., 1989), and such distortions are characteristic of an interaction between root hairs and micro-organisms (e.g. Faucher et al., 1989; Lerouge et al., 1990). We do not know if the root-hair distortions reported here are also caused by microbial activity. Branch roots of Zea mays can also have root hairs dif-
ferentiated over the root tip (Varney & McCully, 1991) as was found with the proteoid rootlets (Figs lg, h). Finally, Lamont (19726, 1983) also noted the absence of root hairs and cortex from old proteoid rootlets of numerous Proteaceae. Bacteria capable of solubilizing calcium phosphates were isolated from all developmental stages of proteoid roots. However, presumably the bacterial phosphate-solubilizing activity in the vicinity of proteoid roots will only enhance uptake while the uptake and transport systems in proteoid rootlets remain intact. This is unlikely to be the case when the epidermis and cortex have been lost.
The actual contribution of associated bacteria to the overall phosphate solubilizing activities of the waratah roots in their natural environment remains to be determined. However, we have provided evidence that some of the associated bacteria possess the ability to solubilize calcium phosphates. We need to determine if there is a role for bacteria and possibly other micro-organisms such as fungi (Figs 1 a, g, h) in the mineral nutrition of waratahs growing in non-sterile environments.
ACKNOWLEDGEMENTS
We thank W. G. Allaway and D. G. Keerthisinghe for helpful comments, and B. L. Elliott, D. L. Montgomery, B. J. Rees, C. A. Rugg and E. Steinke for their technical assistance.
