• We investigated structural and functional diversity in arbuscular mycorrhizal (AM) symbioses involving three plant species and three AM fungi and measured contributions of the fungi to P uptake using compartmented pots and 33P. The plant/fungus combinations varied in growth and P responses. Flax (Linum usitatissimum) responded positively to all fungi, and medic (Medicago truncatula) to Glomus caledonium and G. intraradices, but not Gigaspora rosea. Tomato (Lycopersicon esculentum) showed no positive responses.
• Hyphal growth in soil was very low for Gi. rosea and high for both Glomus spp. Hyphal lengths in root + hyphal compartment (RHC) and hyphal compartment (HC) were similar for G. intraradices, but much higher in HC for G. caledonium.
• Specific activities of 33P in plants and soil indicated that fungal P uptake made substantial contributions to five plant/fungus combinations and significant contributions to a further two. G. intraradices delivered close to 100% of the P in all three plants. G. caledonium and Gi. rosea delivered less P. The amount was not related to colonisation or to growth or P responses.
• We conclude that: AM colonisation can result in complete inactivation of the direct P uptake pathway via root hairs and epidermis; calculations of AM contributions to P uptake from total plant P will often be highly inaccurate; and lack of plant responsiveness does not mean that an AM fungus makes no contribution to P uptake.
Arbuscular mycorrhizas (AM) are found in the vast majority of terrestrial plant species. The associations are normally mutualistic, based functionally on reciprocal transfer of sugars from plant to fungus and soil-derived nutrients (particularly P and Zn) from fungus to plant. Other attributes of the symbioses that may improve plant fitness include increased tolerance of or resistance to some diseases, improved water relations and improved soil structure and structural stability (Newsham et al., 1995; Miller & Jastrow, 2002). This paper will focus on AM effects on plant P nutrition and growth.
It is well established that plants vary in their responsiveness to AM colonisation on the basis of whole plant P uptake and/or growth and that many plant and environmental factors influence the magnitude of the responses (Smith & Gianinazzi-Pearson, 1988; Smith & Read, 1997; Jakobsen et al., 2002). Sometimes there is no positive response and where nonnutritional effects can be discounted, as in controlled experiments, it then appears that the plants receive insignificant amounts of P via the fungal symbiont. Variations in development and function among AM fungi also contribute to the outcome of symbioses with different host plants. Speed and extent of colonisation and development of arbuscules in Arum-type AM are sometimes, but not always, related to responsiveness. The roles in P transfer of intracellular hyphal coils and arbusculate coils typical of Paris-type AM have not been properly established. Studies with compartmented pots have demonstrated that AM fungal species show considerable differences in hyphal growth in soil and in ability to acquire P at a distance from roots and transfer it to the plants (Jakobsen et al., 1992a,b; Pearson & Jakobsen, 1993; Ravnskov & Jakobsen, 1995; Smith et al., 2000), exploit patchy nutrient sources (T. R. Cavagnaro et al., unpublished) and grow through inhospitable soils to acquire P (Drew et al., 2003).
The variations both in plant and fungal capabilities contribute to variations in outcomes of symbioses involving different plant and fungal partners, in terms of growth, nutrient uptake, gene expression or other parameters. This variation has been termed ‘functional diversity’ (Burleigh et al., 2002). The differences in integration of AM fungal and plant processes during P uptake that lead to this diversity have not been clearly elucidated. An AM root has two pathways through which P can be absorbed: first a direct pathway through root epidermal cells and root hairs, absorbing P from the soil solution adjacent to the roots, and second a mycorrhizal pathway in which P is absorbed by fungal mycelium up to at least 10 cm from roots, translocated rapidly to the fungus/plant interfaces within the root and subsequently absorbed by root cortical cells. New work on the changes in expression of plant P transporters in nonmycorrhizal (NM) and AM roots needs to be integrated with physiological information on P fluxes via the two pathways (Smith et al., 2003). Contributions of the mycorrhizal pathway to total plant P uptake have often been calculated from the difference between total P in AM plants and in NM control plants, grown in the same soil (Sanders & Tinker, 1971; Smith, 1982; West et al., 1993; Smith et al., 1994; Graham & Abbott, 2000). This calculation assumes that colonisation itself has no effect on uptake via the direct pathway in an AM root. It has been repeatedly suggested that this assumption is too simplistic, partly because plant P status (which may be increased in AM plants) influences the uptake capacity of roots (Gray & Gerdemann, 1969; Bowen et al., 1975; Jakobsen, 1995). Although several investigations have provided evidence that the operation of the mycorrhizal uptake pathway may lower the P flux through the direct pathway, this has not been accurately quantified for technical reasons (Li et al., 1991; Pearson & Jakobsen, 1993; Ravnskov & Jakobsen, 1995; Schweiger & Jakobsen, 1999).
The work described in this paper was designed to examine functional diversity among three plant species, Medicago truncatula (medic), Linum usitatissimum (flax) and Lycopersicon esculentum (tomato) colonised by three AM fungi (Gigaspora rosea, Glomus caledonium and G. intraradices). The plant responses had previously been shown to vary in separate experiments that allowed limited comparison (Smith et al., 2000; Burleigh et al., 2002; T. R. Cavagnaro et al., unpublished). Our aim was to extend these earlier findings in a single experiment that measured responsiveness in terms of plant growth and P uptake, the development and morphology of AM colonisation and the extent of development of external mycelium. We included measurement of the AM contribution to P uptake, using 33P as a tracer and a compartmented pot design, which provided AM and NM plants with very nearly the same amounts of soil P (Smith et al., 2003). The design overcomes limitations of some previous compartmented pot designs with large soil compartments which were accessible only to the AM fungi (Li et al., 1991), or which favoured AM fungi able to acquire P at large distances from the roots (Jakobsen et al., 1992a,b; Smith et al., 2000). Measurement of the specific activity of 33P in the hyphal (i.e. root-free) compartment allowed quantitative determination of the P absorbed by the external hyphae from this compartment and hence their contribution to total P uptake from the soil as a whole. The inclusion of tomato, which forms both Arum-type and Paris-type AM, depending on fungal species (Cavagnaro et al., 2001) had the additional advantage that we could examine functional compatibility in the context of these AM morphotypes.
Materials and Methods
Compartmented pots and growth medium
We used a very simple, but novel compartmented pot design (Smith et al., 2003). The main, root + hyphal compartment (RHC) was a plastic pot, lined with a plastic bag, containing 1100 g of growth medium (including AM inoculum or not, see Plants and fungi, below). The hyphal compartment (HC) was a small plastic vial, containing 25.3 g of the same medium (2.3% of the total), capped with 25 µm nylon mesh, which allowed fungal hyphae, but not roots, to penetrate from the RHC and absorb P. The HC was placed horizontally, 5 cm below the surface within the RHC, with the mesh towards the centre of the pot. The small size of the HC meant that the quantities of P and other nutrients in the soil that were available to AM and NM plants were almost equal. The placement of the HC within the main RHC meant that the distance the hyphae had to grow from roots to invade the HC was small. The design also had the advantage that recovery of the HC at harvest was both simple and safe, with radioactive soil localised only within the plastic vial.
The growth medium in both RHC and HC was a 1 : 1 mixture of sand and irradiated soil (10 kGy, 10 MeV electron beam) with basal nutrients minus P added (Pearson & Jakobsen, 1993). This ‘soil’ had a bicarbonate-extractable P content of 9.75 µg P g−1 (Olsen et al., 1960). N was added as 30 mg N kg−1 (NH4NO3) at the start of experiment. Supplemental N was added periodically during plant growth, to provide an additional 70 mg N per pot for flax and medic and 100 mg N per pot for tomato, by the end of the experiment. Soil for the HCs was well mixed with carrier-free H333 PO4 to provide 69.2 kBq mg−1 bicarbonate-extractable P. Where required, soil with high P was prepared by mixing an additional 50 mg P kg−1 into the soil for the RHCs. Sufficient water was added to bring the soil as a whole to a water content of 0.16 g g−1 oven-dry soil.
Plants and fungi
The plants used were Linum usitatissimum L. cv Linetta (flax), Medicago truncatula L. cv Jemalong (medic) and Lycopersicon esculentum Mill. cv Riogrande 76R (tomato). Seed was germinated on moist paper for 2 d and three seedlings planted in each pot. These were thinned to two per pot after establishment (c. 1 wk).
The AM fungi were Gigaspora rosea Nicolson and Schenck (Banque Européen des Glomales, BEG 9), Glomus caledonium (Nicol. and Gerd.) Trappe and Gerdemann (BEG 20) and G. intraradices Schenck and Smith (BEG 87). Inoculum was dry soil, containing spores and colonised root fragments, from pot cultures of the AM fungi grown on Trifolium subterraneum L. in the same soil as described above. This was thoroughly mixed into the soil in the RHC. RHCs of NM pots received an additional 80 g kg−1 sterilised soil mix. There was no inoculum in the HCs.
Experimental design and harvesting
The pots (132) were set up as follows: 10 pots of each plant/fungus combination in low P soil (AM treatments); 10 pots of each plant species without inoculum in low P soil (NM-P treatments); three pots of each plant species without inoculum and with additional P in the RHC (NM + P treatments); and three NM-P pots without plants. Three pots of each AM fungus/plant combination and three NM-P pots of each plant species were harvested at 3, 5 and 6 wk after planting. The NM + P plants were harvested at 5 wk only. The unplanted pots were used to measure the specific activity of the bicarbonate-extractable 33P in the soil in HCs at the end of the experiment. Plants were maintained in a growth room with a 16 : 8 h light : dark cycle with 21 : 16°C temperatures, respectively. Osram daylight lamps provided 550 µmol m−2 s−1 PAR (400–700 nm). Plants were watered to weight to maintain moisture content; towards the end of the experiment watering was done twice per day.
Growth was monitored regularly throughout the experiment and appearance of 33P in the shoots of the plants was followed (nonquantitatively) using a hand-held monitor. At harvest the shoots were cut from the plants, briefly washed, blotted and weighed and used for determination of d. wt (24 h, 80°C), P content and specific activity of 33P (33P/total P). The HCs were carefully removed from the pot and placed in plastic bags, to be stored frozen for later determination of hyphal length density (HLD). A small weighed sample of soil was also taken from the RHCs for determination of HLD. Roots were carefully washed, blotted and weighed. Two weighed subsamples were prepared, for determination of total root d. wt (from the f. wt : d. wt ratios of the subsamples), P concentration and specific activity of 33P and for AM colonisation. Dried ground material was digested in a solution of nitric and perchloric acids (4 : 1, v : v). Total P content was measured by the molybdate blue method (Murphy & Riley, 1962) on a Technicon Autoanalyser II (Technicon Autoanalysers, Analytical Instruments Recycle, Inc., Golden, CO, USA). 33P in plant and soil extracts was measured in a Packard TR 1900 liquid scintillation counter (Packard Instrument Co., Meriden, CT, USA) using the scintillation cocktail Ultima Gold ™ (Perkin Elmer, Boston, MA, USA) and corrected for isotopic decay.
The per cent contribution of the mycorrhizal pathway to total P uptake was calculated from the specific activity of 33P in the HCs and in the plants and the total bicarbonate-extractable P in the RHCs and the HCs. The calculation of the contribution of the mycorrhizal pathway took the differences in HLD in RHC and HC in the same pot into account. This is an important advance on the calculations in Smith et al. (2003), which were based on the assumption that hyphal development in the two compartments was the same.
Per cent contribution of mycorrhizal pathway = 100 * (SA 33P plant/SA 33P HC) * (P in pot/ P in HC) * (HLD in RHC/HLD in HC)(Eqn 1)
where SA is specific activity, P is bicarbonate-extractable P and HLD is hyphal length density. For the calculations using Eqn 1 the HLD was determined by subtracting mean HLD in NM-P pots, for the appropriate hyphal compartment, plant species and harvest time. The equation assumes that bicarbonate-extractable (Olsen) P is equally available to plants and fungi. This is a reasonable assumption based on work showing that mycorrhizal and nonmycorrhizal plants use the same pools of bicarbonate-extractable P (Hayman & Mosse, 1972).
AM colonisation was determined by the method of McGonigle et al. (1990) following clearing in 10% KOH and staining in Trypan Blue by a modification of the method of Phillips & Hayman (1970), omitting phenol from the reagents and HCl from the rinse. Two hundred intersects per sample were scored for the presence of intercellular hyphae, arbuscules, vesicles, hyphal coils and arbusculate coils. Hyphal length densities in HCs and RHCs were determined on well mixed soil samples (Jakobsen et al., 1992a). Hyphal inflows were calculated from the hyphal length densities in HC and the amount of P reaching the plants from the HC. The equation used to calculate root inflow by Jakobsen (1986) was used, because hyphal length densities increased approximately linearly with time rather than exponentially (see Results).
Where T1 and T2 are the harvest times, P1 and P2 are the amounts of P in the plants taken up via the hyphae at T1 and T2 and HL1 and HL2 are the total hyphal lengths in the HCs at T1 and T2. The HLs were determined after subtraction of background values in equivalent NM-P pots.
Data are presented as means and standard errors of means of three replicates. Data were analysed by analysis of variance (anova) using Genstat Release 6.1, Lawes Agricultural Trust (Rothamsted Experimental Station, UK). Statements of significance are based on α = 0.05, except where otherwise stated.
The three plant species showed different growth responses to AM colonisation (Fig. 1). In low P soil flax grew better, but to different extents when colonised by the three AM fungi, compared with NM-P controls: Gigaspora rosea < Glomus caledonium < G. intraradices, with same order at all three harvests. At 5 wk the NM + P plants had significantly higher d. wts. than all other flax treatments (P < 0.05), except plants colonised by G. intraradices (Fig. 1a). Medic also responded positively and to the same extent with the two Glomus species, but showed a small growth depression (P < 0.05) with Gi. rosea. Growth of NM + P plants at 5 wk was greater than all other medic treatments at that harvest (Fig. 1b). Tomato did not respond positively to any of the AM fungi and showed significant growth depressions (P < 0.05) with Gi. rosea in low P soil at 5 and 6 wk and G. intraradices at 6 wk, compared with NM-P plants. Nonmycorrhizal tomato showed a positive response (P < 0.001) to added P at 5 wk (Fig. 1c).
No colonisation was observed in uninoculated plants. In inoculated treatments, AM colonisation (as a percentage of total root length) by the three fungi in all three plant species was well established by the first harvest at 3 wk (Fig. 2). Values in medic were generally greater than in flax or tomato (P < 0.05), particularly for Gi. rosea and G. caledonium. Total colonisation increased over time in tomato, but changed little in flax or medic; there were increases with time in some internal structures. There were also clear differences in the morphology of fungal development (Fig. 2, filled bars and Fig. 3). All three fungi formed Arum-type AM in flax, with intercellular hyphae and intracellular, paired arbuscules (Figs 2a,b,c and 3a). In medic the two Glomus species also formed Arum-type AM (Fig. 2e,f). However, Gi. rosea formed Paris-type AM in this species, with extensive intracellular hyphal coils in the root cortical cells and very small numbers of arbusculate coils (Figs 2d and 3b). In tomato, G. intraradices formed typical Arum-type AM (Figs 2i and 3c) and Gi. rosea formed Paris-type AM with well developed hyphal coils at all three harvests (Figs 2g and 3d); arbusculate coils were not observed until the 6 wk harvest (Fig. 2a). As shown in Fig. 3 (e–g) G. caledonium formed a mixed AM morphology with tomato. Intercellular hyphae, characteristic of Arum-type AM were frequently observed. Hyphal coils were present in the root cortex; c. 12% of the root length at 3 wk, declining at later harvests. Arbusculate coils were also frequent and increased up to 6 wk (Fig. 2h).
Development of external hyphae
Figure 4 shows hyphal length densities (HLD, m g−1 soil) in NM-P pots with all three plants and in all plant/fungus combinations. Note that background values from NM-P pots have not been subtracted from values in equivalent AM pots in this figure. Background levels of hyphal development in NM pots were mostly significantly higher in RHCs than in HCs (P < 0.001). Hyphal development with tomato was less than with flax or medic (P < 0.001) (Fig. 4a,e,i).
External hyphae of Gi. rosea developed very poorly, with values in both RHCs and HCs often equal to or lower than values for NM-P pots with the same plant species (compare Fig. 4b,f,j, with Fig. 4a,e,i). The only exceptions were higher HLD in the RHCs of tomato with Gi. rosea, compared with equivalent NM pots. By contrast the external hyphae of G. caledonium and G. intraradices developed extensively with all three host plants (Fig. 4c,d,g,h,k,l; note 10-fold change in scale). Values for G. caledonium in all plant treatments were low at 3 wk, with HLD in RHCs with medic and tomato significantly (P < 0.01) higher than in HCs. At subsequent harvests HLDs were much higher, remaining steady in RHCs between 5 and 6 wk, but continuing to increase markedly in HCs. In consequence values in HCs were significantly (2–3-fold) higher in HCs than RHCs by 6 wk. G. intraradices showed much smaller differences in development of external hyphae, both between RHCs and HCs and between harvests. There were trends towards increased HLD with time and towards higher lengths in HCs than in RHCs, but none were significant. There were no major effects of plant species on development of external mycelium of either G. caledonium or G. intraradices.
All three AM fungi increased P uptake by flax at all three harvests, relative to NM-P controls, with the ranking the same as for growth (Fig. 5a), although the effect of G. rosea at 3 wk was not significant. The two Glomus species also increased P uptake significantly in medic at all three harvests, with the effect of G. intraradices somewhat higher than that of G. caledonium (Fig. 5b). P uptake by medic colonised by Gi. rosea was similar to NM-P plants throughout the experiment. AM colonisation had no positive effects on total P uptake by tomato. Indeed, plants colonised by Gi. rosea took up significantly less P than NM-P controls at 5 and 6 wk. P uptake by NM + P plants of all plant species at 5 wk was significantly greater than for AM treatments and NM-P controls (Fig. 5).
Uptake of 33P and contribution of the AM and direct uptake pathways to P accumulation by the plants
There was negligible uptake of 33P into NM plants (results not shown). Radioactivity was first detected in the shoots of flax colonised by G. intraradices 10 d after planting, followed by the other plant species grown with the same fungus 16 days after planting. Radioactivity was detectable in plants of all species colonised by G. caledonium 23 d after planting, but was almost undetectable in plants colonised by Gi. rosea up to 26 d, when monitoring ceased (results not shown). At harvests, the specific activity of 33P in the AM plants varied considerably, particularly with fungal species and time (Fig. 6), indicating considerable variations in the proportion of P obtained via the mycorrhizal pathways. Specific activities of 33P in plants colonised by Gi. rosea were consistently low, indicating a low contribution of this fungus. Plants colonised by G. caledonium had received little radioactive P via the fungus at 3 wk, but by 5 and 6 wk specific activities of 33P in plants had increased markedly (P < 0.05). Specific activities of 33P in plants colonised by G. intraradices were high at all three harvests. Values at 3 wk were significantly higher than for the same harvest with the other two fungi. At 5 and 6 wk values were similar to those of G. caledonium and markedly higher than Gi. rosea.
If hyphal distribution were even throughout the pots (RHC = HC) then (from Eqn 1) the specific activity of 33P in the plants would be directly proportional to the contribution of the fungal pathway to P uptake and a value of 1.65 kBq mg P−1 would indicate that all the P entered the plants via the fungal pathway. Higher values in Fig. 6 are explained by higher development of hyphae in HCs than RHCs in treatments involving G. intraradices and, more particularly G. caledonium (Fig. 4). Where the specific activities of 33P were very low and/or the HLDs were very similar to background values, calculations (from Eqn 1) were subject to large errors. Furthermore, at 3 wk it seems likely that sampling well mixed soil from the HCs of mycorrhizal pots would not reflect uneven distribution of hyphae in the HCs. We therefore only used data for the 5 and 6 wk harvests of plants colonised by G. caledonium and Gi. intraradices for the calculations of the contribution of the mycorrhizal pathway to P uptake (Fig. 7). G. intraradices made major contributions to P uptake of close to 100% at all harvests with flax and tomato and somewhat lower contributions to medic (60–80%). G. caledonium contributed lower amounts, particularly to tomato (c. 25–30%). The highest contribution of this fungus was c. 80% to medic at 5 wk.
P inflows into hyphae for the 5–6 wk harvest period (calculated from Eqn 2) were in the ranges 0.6–4.0 * 10−14 (mean 2.0 * 10−14) and 0.1–2.2 * 10−14 (mean 0.9 * 10−14) mol P m−1 s−1 for G. intraradices and G. caledonium, respectively. Inflows for Gi. rosea were not calculated because of uncertainties about hyphal lengths in HCs (see above).
Growth, total P uptake and colonisation
All NM plants responded to added P at 5 wk, indicating that the low P treatments had provided P-limiting conditions. Comparison of AM and NM-P plants confirmed that there were marked differences in growth responses of the three host species to AM colonisation by the different fungi. Only flax showed a positive response to Gi. rosea as well as to the other fungi, as previously observed (T. R. Cavagnaro et al., unpublished). The growth depressions in medic and tomato colonised by this fungus were also as previously reported for 5 wk-old plants grown under similar conditions (Burleigh et al., 2002). Medic responded positively to G. caledonium and G. intraradices, again as shown previously (Smith et al., 2000; Burleigh et al., 2002), while tomato showed no positive responses to the three fungi used in this experiment. Previous work had indicated a growth depression in this tomato cultivar inoculated with a different isolate of G. caledonium (BEG 15), a small positive AM response to G. intraradices (Burleigh et al., 2002; L.-L. Gao & K. Poulsen, unpublished) and larger responses to G. mosseae and G. versiforme (Burleigh et al., 2002). These inconsistencies can probably be explained by differences between isolates of the fungi used and in soil and growth conditions. The outcome was that our experiment provided material from plant/fungus combinations with markedly different AM growth responses that could be used to investigate the relationships between the responses and the other variables that we measured.
Total P uptake (Fig. 5) showed a similar pattern to growth in response to P supply and AM colonisation, with minor variations accounted for by small differences in tissue P concentrations (results not shown).
There was no clear relationship between total per cent colonisation and growth or P response (compare Figs 1, 2 and 4). All plant species were well colonised at all harvests. There might have been differences in rate of colonisation in the different plant fungus combinations, but an earlier harvest would have been required to show this. Although flax was universally positively responsive, the extent of colonisation was not as high as in other plant species, with a maximum of c. 50% at 6 wk with G. intraradices. Even levels of colonisation as low as 20% were associated with positive growth and P responses to G. caledonium. This is in agreement with results of Ravnskov & Jakobsen (1995) who showed that 26% colonisation by G. caledonium was associated with positive responses in flax and that there was generally lower colonisation in this plant than wheat or cucumber, which are relatively unresponsive. These results emphasise that low colonisation is not necessarily an indication of a small growth increase, as sometimes suggested. Conversely, high colonisation (e.g. medic by Gi. rosea and tomato by G. intraradices) was not associated with positive growth or P responses at the whole plant level.
Arum-type mycorrhizas were formed by all three fungi in flax, with paired arbuscules as shown previously (Dickson et al., 2003). Medic had more than 50% of the root length colonised by all the fungi and, as shown previously, formed Arum-type AM with the two Glomus species (Smith et al., 2000; Burleigh et al., 2002). Surprisingly, we found that the association with Gi. rosea was characterised by extensive development of intracellular hyphal coils, with few arbusculate coils and intracellular hyphae. This contrasts with the results of Burleigh et al. (2002) who reported Arum-type AM in this plant/fungus combination. We have no explanation for the differences in AM development, as the plants were grown in similar soil and the same growth conditions in the two experiments. Colonisation in tomato was relatively low for Gi. rosea and G. caledonium (< 40%), but higher for G. intraradices. The production of Paris-AM colonisation by Gi. rosea, with relatively few arbusculate coils, confirms previous work (Burleigh et al., 2002). G. caledonium produced a mixed AM morphology in tomato, which included a strong development of arbusculate coils (again as found by Burleigh et al., 2002). The significance of different AM morphologies in P uptake is discussed later.
Development of external mycelium
Hyphal length density (HLD) in the uninoculated pots indicated that the ‘background’ values included hyphal growth during the experiment. The higher values in RHCs than in HCs (Fig. 4) suggest that the background included saprophytic fungi that used exudates and sloughed or dead cells produced in RHCs and grew more there than in the HCs. Furthermore, the abilities of the plant species to support these saprophytes differed, possibly relating to differences in rhizodeposition. HLD in NM pots did not increase between 3 and 6 wk, suggesting that the turnover of hyphae was at a steady state or (less likely) that all the hyphae were dead.
The HLDs in Gi. rosea-inoculated pots were higher than values in NM pots in only two cases (RHCs in tomato pots at 5 and 6 wk), suggesting that this fungus did not colonise the soil in either RHC or HC (Fig. 4). However, the presence of 33P in the plants (Fig. 6) indicated that hyphae of Gi. rosea must have reached the HCs. HLD was determined on soil samples that had been thoroughly mixed, so values would not reflect any localised hyphal proliferation just inside the HC. These findings call into question the subtraction of values from NM pots from those of inoculated pots to obtain values for HLD of AM fungi, especially when the HLD in inoculated pots is low; errors will be relatively small with high HLDs. Our comparisons of HLD in NM and Gi. rosea pots also suggest that AM fungal hyphae may suppress saprophytes (as has been found for ectomycorrhizal fungi (Leake et al., 2002)) but further work would be required to confirm this.
Previous work has focused largely on hyphal development in HCs. We found that Gi. rosea developed very poorly, as also found by T. R. Cavagnaro et al. (unpublished). G. caledonium and G. intraradices developed much more extensively, again as shown previously (Pearson & Jakobsen, 1993; Ravnskov & Jakobsen, 1995; Drew et al., 2003; T. R. Cavagnaro et al. unpublished), with G. caledonium relatively slower at colonising the soil but reaching much higher values at 5 and 6 wk. Our results comparing HLD in HCs and RHCs also showed some consistent trends. With G. caledonium values were generally lower in HCs than RHCs at 3 wk, presumably reflecting slow growth into the HCs and possibly also uneven distribution of hyphae within them. There have been rather few previous comparisons of hyphal development in RHCs. Using G. caledonium, Pearson & Jakobsen (1993) found no significant differences in HLD in RHCs and HCs associated with cucumber at either 17 or 27 d after planting, consistent with our findings for flax at 3 wk. Similarly, Ravnskov & Jakobsen (1995) found no differences in HLD between HCs and RHCs at 28 d with flax as the host, but values for HCs were greater than RHCs with both wheat and cucumber at the same harvest time. No reported experiments have provided information at longer times. Our results for both G. caledonium and G. intraradices at 3 wk and at 5/6 wk indicate that the timing of harvests can make a very large difference to the values obtained, presumably because of the progressive spread of the hyphae from the roots and some delay in accessing the HCs, even though these were located within the main RHC in our experiment. Our findings, particularly for G. caledonium, indicate that there can be considerable proliferation of hyphae in HCs, which are effectively root-free patches in the soil. The reasons for this proliferation are not clear, but deserve further investigation in the context of patch exploitation as well as usefulness in calculating the per cent contribution of hyphae to P uptake (see next section).
Contribution of the mycorrhizal pathway to P uptake
As discussed thus far, the data for total P uptake are completely consistent with the traditional interpretation that in those plant/fungus combinations that showed a positive AM growth response the mycorrhizal pathway of P uptake was operational, but that in the nonresponsive fungus/plant combinations it was not. However, the specific activities of 33P in the plants indicate that this interpretation is incorrect. If the mycorrhizal pathway in tomato were completely inoperative and all the P reached the plants via the direct uptake pathway through root epidermis and root hairs then no 33P would appear in the AM plants. In fact all AM plants contained some 33P, regardless of responsiveness. All plants inoculated with Gi. rosea, and especially medic, which showed a growth depression, also obtained some 33P from the HCs (Fig. 6). Some previous work is in line with our finding that P transfer via the mycorrhizal pathway is not related to responsiveness, but perhaps because the contribution of the mycorrhizal pathway was not accurately quantified, its significance has not been widely appreciated (see Introduction). Considerable 32P or 33P uptake via hyphae from HCs has been observed in nonresponsive cucumber (Pearson & Jakobsen, 1993; Ravnskov & Jakobsen, 1995), wheat (Ravnskov & Jakobsen, 1995; Schweiger & Jakobsen, 1999) and barley (Zhu et al., 2003). The present work quantified the contribution of the mycorrhizal pathway by utilising the specific activity of 33P in the HC as well as in the plants. This has apparently only been done before in one field study where the P available to the plants and fungi outside the HC (buried in field soil in this case) could not be quantified (Schweiger & Jakobsen, 1999). We made the assumption that external AM hyphae and roots accessed the same pools of soil P, as shown for a number of 32P labelled soils based on similar specific activities of 32P in AM and NM plants grown in them (Hayman & Mosse, 1972). Importantly, we also took different hyphal distributions in RHC and HCs into account, thereby adjusting the values for the extent of hyphal development in the two compartments. Comparison with calculated values in Smith et al. (2003) indicates that this is essential in determining more realistically the per cent contributions of the mycorrhizal pathway for different AM fungi. As suggested previously, we now show that values for G. caledonium in excess of the theoretical maximum are the result of higher development of external mycelium in HC than RHC. The calculation still makes assumptions that all the hyphae were active at all harvests and that they were evenly distributed within the compartments (see previous section); both are likely to be oversimplifications. Nevertheless, there was a general positive relationship between HLDs for the different fungi and the extent of transfer of 33P to the plants (but not percent contribution of the AM pathway).
Relationship of AM morphotype to P transfer
The specific activities of 33P in the plants and per cent contributions of the fungi to total P uptake (Figs 6 and 7), show that G. intraradices, which consistently formed Arum-type AM, was highly efficient in delivery of P to all three plant species, regardless of their responsiveness in terms of total plant dry weight or P uptake. Values at 5 wk (presented for the first time here) and 6 wk were close to 100% in both the highly responsive flax and the unresponsive tomato. This finding is fully in accord with our previous conclusions from the data collected at 6 wk (Smith et al., 2003). The mycorrhizal pathway was apparently slower to become operational in plants colonised by G. caledonium, as shown by the low values for specific activity of 33P in plant species at 3 wk. Per cent colonisation of the roots was near maximal at this stage in all three plants, and the results are consistent with relatively slow development of the external hyphae. Values for HLD were so close to those in uninoculated pots that calculation of the per cent contribution of the mycorrhizal pathway was not possible. By 5 and 6 wk hyphal development was very much higher and considerably greater in HCs than RHCs. The maximum calculated per cent contribution of this fungus to P uptake was 80% in medic (5 wk), and was generally lower than this, with trends for the different species increasing in the order tomato < flax < medic. The findings with G. caledonium also reinforce our earlier conclusions (Smith et al., 2003), while emphasising how crucial it is to measure HLD in both RHCs and HCs, not only for the types of calculation we have used here, but also in investigations of patch exploitation.
Our calculations of hyphal P inflow appear to be the first to be based on direct measurements of hyphal P uptake and length. Inflow was slightly higher for G. intraradices than G. caledonium, and hence generally related to the per cent contributions of the fungi to P uptake. Previous reports of hyphal inflows are in the same range (0.13–2.0 mol * 10−14 P m−1 s−1 between 0 and 28 d; 0.06–0.28 mol P * 10−14 m−1 s−1 between 28 and 47 d) (Jakobsen et al., 1992a). Our values were obtained for the 5–6-wk period and are therefore rather higher than the comparable values for the second harvest period of Jakobsen et al. (1992a). The latter are likely to be underestimates because they were calculated from total P uptake by AM and NM plants, assuming that colonisation had no effect on direct uptake of P from soil, which we now confirm is not the case.
Our findings that AM uptake can make a highly significant contribution to total P uptake (100% in some symbioses involving G. intraradices) implies that the direct uptake pathway via root hairs and root epidermal cells must cease to function in some plant/fungus combinations and that this change in the relative contributions of the two pathways is not related to plant responsiveness or AM colonisation. This extends the earlier findings of Pearson & Jakobsen (1993) and Ravnskov & Jakobsen (1995), who provided results suggesting that direct uptake pathways in flax and cucumber ceased to function in roots colonised by Glomus caledonium.
Although we cannot make satisfactory calculations of the per cent contribution of Gi. rosea to total P uptake, this fungus does make a positive contribution, particularly in medic. Specific activity of 33P increased consistently with time and was associated with a small increase in total P uptake, but no growth response. The finding suggests that Gi. rosea requires a relatively large amount of carbon from its host, which might be associated with the development of hyphal coils, which have a higher biomass per cell than arbuscules (Dickson & Kolesik, 1999). Importantly, the Paris-type AM formed in this plant/fungus combination was capable of P transfer, even though there were very few arbusculate coils. This is consistent with mycorrhizal effects on P uptake via Paris-type AM in Asphodelus (Cavagnaro et al., 2003) and with the finding that the mycorrhiza-specific P transporter in potato (StPT3) is expressed in cortical cells containing coils as well as in arbuscule-containing cells (V. Karandashov, pers. comm.). However, correlation of the relative contribution of the mycorrhizal pathway of P uptake with AM morphology is not straightforward. Development of Arum-type AM in flax (all fungi), medic (G. caledonium and G. intraradices) and tomato (G. intraradices) was consistently associated with operation of the mycorrhizal pathway, but to highly variable extents. Paris-type morphology was associated with AM P delivery in medic (Gi. rosea) in the absence of arbusculate coils. Similar colonisation in tomato, by the same fungus, was associated with the lowest specific activities of 33P (Fig. 6c). The mixed colonisation pattern formed by G. caledonium in tomato was moderately effective, delivering c. 30% of the plant P by 6 wk. Further experiments are required, focusing on Paris-type AM and combining both physiological and molecular techniques.
Implications for P transporter expression
Our findings have significant implications for hypotheses relating to the expression of plant P transporters in roots, including the recently discovered ‘mycorrhiza-specific’ transporters in potato, medic and rice (Rausch et al., 2001; Harrison et al., 2002; Paszkowski et al., 2002). The specific transporters, together with fungal transporters in the external mycelium (Harrison & van Buuren, 1995; Maldonado-Mendoza et al., 2002), must be crucial in the operation of the mycorrhizal uptake pathway. Differences in the amount of P delivered via this pathway in different plant/fungus combinations are likely to reflect different expression of fungal and mycorrhiza-specific plant P transporters, relative to transporters in root hairs and epidermis, as well as the abilities of the fungi to spread in soil and to absorb and translocate P. The functional diversity that we have demonstrated may well be related to fungal species-specific induction/suppression of the pathway. The association of P transporter expression with different AM structures in Arum- and Paris-type AM will help to throw light on the functions of the different structures.
In summary, we have established that plant P can be acquired by plants exclusively via the mycorrhizal pathway. We showed this particularly with G. intraradices in flax and tomato. Contributions of > 60% were also observed in medic with both G. intraradices (5 and 6 wk) and G. caledonium (5 wk). These large contributions of the mycorrhizal pathway were not related to plant growth response, P uptake or extent of AM colonisation, but may be related to the development and effectiveness of the fungal hyphae in absorbing P and their influence on transporter expression (Liu et al., 1998; Burleigh et al., 2002). Much previous work has indicated that in responsive hosts the extent of development of external mycelium may be related to increases in P uptake by the plants and hence to mycorrhizal P responses. However, similar data relating HLD to operation of the mycorrhizal P uptake pathway in nonresponsive plants is, with the exception of our results for tomato, simply not available. Until more work with nonresponsive plants has been done it is premature to suggest that high HLD is necessarily a good indicator of operation of the mycorrhizal uptake pathway.
Although tomato grown alone (as here) is nonresponsive, we have shown that it does gain an advantage over a nonmycorrhizal competitor (nonmycorrhizal mutant tomato rmc; Barker et al., 1998) because of increased P acquisition via an effective fungus (Cavagnaro et al., 2004). This finding provides a rationale for the persistence of the mycorrhizal condition in supposedly nonresponsive plants when growing in mixed plant communities. It is premature to speculate on the costs and benefits to a plant of transferring the site of P uptake from the root epidermis to the cortical cells. The net effect will depend on the carbon demand of developing and maintaining the symbiosis and any compensatory changes in root growth. In our experiment colonisation was fully developed by 3 wk, so that relative costs of the two pathways seem unlikely to be very different.
Lastly, our results are relevant to interpretation of AM experiments at scales ranging from the molecular (e.g. the expression of P transporter genes), through the physiological (including investigation of uptake of nutrients other than P via the mycorrhizal pathway), to the ecological (e.g. competitive interactions between plants of different responsiveness in soils with diverse communities of AM fungi). We suggest that no one scale is more relevant than any other in aiming for an understanding of the success of AM associations world-wide (Read, 2002).
The work described in this paper was carried out at Risø National Laboratory, Roskilde, Denmark. SES is grateful to the University of Adelaide for Special Studies Program, which enabled her participation. We owe special thanks to Anne and Anette Olsen at Risø and Greg Hay at Adelaide for dedicated and excellent technical assistance. We also thank our research groups for stimulating discussions. Some data for the 6-wk harvest in this experiment have been reported in brief previously (Smith et al., 2003). Our research is funded by the Australian Research Council and the Danish National Research Foundation.