Functional diversity in arbuscular mycorrhizas: exploitation of soil patches with different phosphate enrichment differs among fungal species

Authors

  • T. R. CAVAGNARO,

    Corresponding author
    1. School of Earth and Environmental Sciences and Centre for Soil–Plant Interactions, The University of Adelaide, Adelaide, South Australia 5005, Australia and
    Search for more papers by this author
    • *Department Land, Air & Water Resources, University of California, Davis, One Shields Avenue, Davis, CA 95616–8627, USA

  • F. A. SMITH,

    1. School of Earth and Environmental Sciences and Centre for Soil–Plant Interactions, The University of Adelaide, Adelaide, South Australia 5005, Australia and
    Search for more papers by this author
  • S. E. SMITH,

    1. School of Earth and Environmental Sciences and Centre for Soil–Plant Interactions, The University of Adelaide, Adelaide, South Australia 5005, Australia and
    Search for more papers by this author
  • I. JAKOBSEN

    1. Plant Research Department, Risø National Laboratory, PO Box 49, Roskilde 4000, Denmark
    Search for more papers by this author

Timothy R. Cavagnaro. Fax: +1 530 752 1552; e-mail: trcavagnaro@ucdavis.edu

ABSTRACT

Most terrestrial plant species form associations with arbuscular mycorrhizal fungi (AMF) that transfer soil P to the plant via their external hyphae. The distribution of nutrients in soils is typically patchy (heterogeneous) but little is known about the ability of AMF to exploit P patches in soil. This was studied by growing symbioses of Linum usitatissimum and three AMF (Glomus intraradices, G. mosseae and Gigaspora margarita) in pots with two side-arms, which were accessible to hyphae, but not to roots. Soil in one side-arm was either unamended (P0) or enriched with P; simultaneous labelling of this soil with 32P revealed that G. intraradices responded to P enrichment both in terms of hyphal proliferation and P uptake, whereas the other AMF did not. Labelling with 33P of P0 soil in the other side arm revealed that the increased P uptake by G. intraradices from the P-enriched patch was paralleled by decreased P uptake by other parts of the mycelium. This is the first demonstration of variation in growth and nutrient uptake by an AMF as influenced by a localized P enrichment of the soil. The results are discussed in the context of functional diversity of AMF.

INTRODUCTION

Resources in soil are often heterogeneously distributed and nutrient patches in natural soils can be associated with decomposition of leaf litter and fruits, the faeces and bodies of soil animals, dead seeds, pollen and dead bacteria and fungi (Tibbett 2000). In agricultural soils, nutrient patches can also be created by anthropogenic influences, such as directed placement of fertilizers or incorporation of crop residues or manure. Nutrient patches are particularly significant with respect to nutrients with low mobility in the soil, such as phosphate (P) and zinc (Zn). An ability to locate and utilize heterogeneously distributed resources efficiently can provide a competitive advantage to plants.

Studies of plant responses to heterogeneous nutrient distribution have focused mainly on the roots (see Robinson 1994; Hodge 2004). Their association with hyphae of mycorrhizal fungi provides the plant with a well-distributed and extensive absorbing system in soil (Smith & Read 1997) and therefore with a greater chance of encountering fertile microsites not available to the roots alone. Mycorrhizal mycelia may be even more important than root proliferation in exploring nutrient patches (Tibbett 2000), but this remains to be resolved (see Hodge 2001) and may not be the case for all nutrients (Tibbett 2000; Hodge, Robinson & Fitter 2000).

Exploitation of nutrient patches by ectomycorrhizal fungi has been studied in some detail (see Read 1991; Bending & Read 1995a, b; Tibbett 2000; Leake et al. 2001; Tibbett & Sanders 2002). Our insight is more limited for arbuscular mycorrhizal fungi (AMF), which proliferate in organic patches in the soil in a similar way to plant roots (St John, Coleman & Reid 1983; Joner & Jakobsen 1995; Ravnskov et al. 1999; see also Olsson, Jakobsen & Wallander 2002), although at a smaller scale. AMF can utilize inorganic sources of P from patches and deliver it to the roots (Cui & Caldwell 1996; Gavito & Olsson 2003). Farley & Fitter (1999) and Wijesinghe et al. (2001) demonstrated that different herbaceous plant species varied in their response to heterogeneous nutrient supplies when colonized by AMF. Facelli & Facelli (2002) suggested that the formation of AM might strongly influence plant population structure in patchy soils.

Spatial differences in P uptake patterns between species of AMF have been reported (Jakobsen, Abbott & Robson 1992b; Smith, Jakobsen & Smith 2000; Drew et al. 2003) and may have consequences for exploitation of nutrient patches, although this is yet to be demonstrated (Olsson et al. 2002). Understanding the interactions between nutrient patches, hyphal proliferation and hyphal nutrient uptake will improve our understanding of functional diversity of AMF and the ecological function of mycorrhizal symbioses in natural and agricultural soils.

The present experiment used pots with two side-arms to assess the ability of three AMF (Glomus intraradices, G. mosseae and Gigaspora margarita) in symbiosis with Linum usitatissimum to exploit P-enriched patches of soil by increasing hyphal proliferation and P uptake. Side arms were accessible to the AMF, but not to roots, and one side-arm contained 32P-labelled soil, which was either unamended (P0) or enriched with P at two concentrations. The second (reference) side arm of all pots contained 33P-labelled soil, which was unamended as soil in the root compartment. Labelling of P in the ‘reference’ regions of the pots made it possible, for the first time, to study whether increased hyphal P uptake from a P-enriched patch would influence P uptake by other parts of the mycelium.

MATERIALS AND METHODS

Containers and growth medium

The experiment was set up in ‘cross-pots’ constructed from 55 mm diameter PVC tubing, with two side-arms (Fig. 1); see also Pearson & Jakobsen (1993). The side-arms were separated from the root compartment (RC) with 25 µm mesh, which allowed passage of AMF hyphae while preventing passage of the roots, thus providing two hyphal compartments (HC1 and HC2). 32P- and 33P-labelled soil was added to HC1 and HC2, respectively. The growth medium was a 1 : 1 mixture of sand and irradiated soil (15 kGy, 10 MeV electron beam), with basal nutrients (minus P) added (Pearson & Jakobsen 1993); it had a bicarbonate-extractable P content of 9 µg P g−1 (Olsen et al. 1954) and a pH(H2O) of 6.7. The RC was filled in three zones that were not physically separated: RC1 and RC3 contained 195 g and 375 g of soil, respectively; RC2, located directly between HC1 and HC2, contained either 75 g inoculum + 75 g soil (see later) or 150 g of soil in non-mycorrhizal controls.

Figure 1.

Schematic diagram of cross-pot with two side-arms. RC1 + 2 + 3 = root compartment; HC1 and HC2 = hyphal compartments; soil treatments: P0 (9 mg kg−1),  P1  (29 mg kg−1),  P2  (68 mg kg−1). Diagram not drawn to scale.

The side-arms were each filled in two zones. A buffer zone adjacent to the mesh was 8 mm thick (20 g of soil) whereas the remainder of each HC was filled with 80 g of soil with different P treatments. The soil in HC1, the ‘patch side-arm’, received KH2PO4 to obtain three concentrations of bicarbonate-extractable P: 9 (no added P), 29 and 68 µg g−1 soil, referred to as P0 (the same as in the RC and the buffer zones in the HCs), P1 and P2, respectively (see Fig. 1). The P0, P1 and P2 patches were mixed with carrier-free 32P-labelled orthophosphate (Amersham Biosciences, Cardiff, UK) at 5 kBq g−1 soil. The specific activities of the bicarbonate-extractable P in HC1 were 113 (P0), 38 (P1) and 16 (P2) Bq µg−1 31P, respectively. The P contents of 32P-labelled soil in HC1 were 9% (P0), 23% (P1) and 42% (P2) of the total soil P in the cross-pots. The soil in HC2, the reference side-arm, contained P0 soil mixed with carrier-free 33P-labelled orthophosphate (Amersham Biosciences) at 5 kBq g soil−1. The specific activity of the bicarbonate-extractable P in HC2 was 122 Bq µg−1 31P.

Plants and fungi

Seeds of Linum usitatissimum L. (cv. Linetta) were germinated overnight in aerated distilled water. Three AM fungi were used: Glomus mosseae (Nicol. & Gerd.) Gerd. and Trappe (BEG 161), G. intraradices Schenck & Smith (BEG 87) and Gigaspora rosea Nicolson and Schenck (BEG 9). Inoculum was dried soil containing root pieces and spores from pot cultures with Trifolium subterraneum L. grown in the same growth medium described above.

Experimental design

The experiment was a completely randomized design with two main factors, fungi (four levels: G. mosseae, G. intraradices, Gi. rosea and non-mycorrhizal controls) and patch concentration (three levels: P0, P1 and P2). There were four replicates for each treatment except for the NM controls, for which there were three replicates, giving a total of 45 pots.

Growth conditions, watering and nutrient addition

Plants were grown in a controlled environment chamber with 16 h/8 h and 21 °C/16 °C day/night. Osram daylight lamps provided a photosynthetic photon flux density of 550 µmol m−2 s−1 (400–700 nm). Plants were watered by weight to 55% of water-holding capacity over the course of the experiment as required. Nitrogen (as NH4NO3) was added to the pots 14, 20, 24 and 29 d after planting to provide a total of 140 mg N per pot.

Appearance of radioactivity (32P and/or 33P) in the shoots was monitored every second day over the course of the experiment using a hand-held monitor, which helped to track the uptake of P from the side-arms and determine timing of the single destructive harvest at 34 d. Due to the difference in rates of isotopic decay of 32P and 33P and the inability to distinguish between 32P and 33P using the hand-held monitor, these data were not corrected for isotopic decay over time.

Harvest and sample analysis

Shoots were cut off at the soil surface, washed, blotted dry and used for determination of fresh weights, dry weights, P content and for uptake and specific activities of 32P and 33P. The side-arms were detached from the pots, the buffer soil sliced off and discarded, and the remaining soil frozen and stored for subsequent determination of hyphal length (Jakobsen, Abbott & Robson 1992a). The soil and roots were harvested from the RCs in three sections corresponding to RC1, RC2 and RC3. Roots were washed from the soil, blotted dry and divided into weighed subsamples for determination of root length and mycorrhizal colonization, root fresh and dry weights, root P contents and uptake and specific activities of 32P and 33P.

Dried plant tissue was ground and digested in a nitric/perchloric acid solution (4 : 1, v/v). Total P content was determined colorimetrically using the molybdate blue method (Murphy & Riley 1962), on a Technicon Autoanalyser II (Technicon Autoanalysers, Analytical Instrument Recycle, Inc., Golden, CO, USA). 32P and 33P were measured separately in the same extracts in a Packard TR 1900 liquid scintillation counter (Packard Instrument Co, Meriden, CT, USA), with automatic correction for isotopic decay. This allowed the amounts of P taken up from HC1 and HC2 in a single sample to be calculated separately, by expressing the ratio of shoot or root 32P- or 33P-specific activity to the specific activity of available P in the side arms, as a percentage of total shoot or root P content. It has been shown previously that AMF do not discriminate between 32P and 33P (Pearson & Jakobsen 1993).

A sample of roots was cleared in 10% KOH and stained with trypan blue using a modification (omitting phenol from all reagents) of the method of Phillips & Hayman (1970). Total root length and AMF-colonized root length were determined using the gridline intersect method (Newman 1966).

Statistical analysis

Analysis of variance (anova) was performed on all data using Genstat 5 Release 4.1 (1998; VSN International Ltd, UK). Where significant differences were found (P < 0.05), differences between treatment means were calculated using the least significant differences (LSD) method.

RESULTS

Colonization of roots and side-arm soil by the AMF

The three AMF colonized roots to significantly different extents but patch treatments had no significant effects on colonization by any of the fungi. Combining the patch treatments, the greatest percentage of the root length was colonized by G. mosseae (46%), followed by G. intraradices (35%) and Gi. rosea (26%). Non-inoculated (NM) plants remained uncolonized.

There were no significant differences in the length of hyphae in the patch and reference side-arms at P0 for any treatments (three-way anova, P > 0.05, data not shown). There were some hyphae in the side-arms of the pots containing NM plants (Fig. 2). Their length was not significantly influenced by patch concentration or by presence of 32P or 33P. These hyphae were probably dead AMF background material in the soil or saprophytic hyphae and it is assumed that they were also present in the mycorrhizal treatments. In general, the length of the hyphae in the side-arms increased in the order NM < Gi. rosea < G. mosseae < G. intraradices. There was no significant ‘patch effect’ on the amount of hyphae of Gi. rosea or G. mosseae in either of the side-arms (Fig. 2), but hyphae of G. intraradices increased in the patch side-arm at P1 compared with P0 and P2 (when patch side arm was considered alone, see Fig. 2). There were no differences in length of hyphae in the corresponding reference side-arms.

Figure 2.

Length of hyphae (m g−1 soil) in the patch (HC1) and reference (HC2) side-arms, in the different fungal and ‘patch’ treatments, P0 (no added P, white bars), P1 (stippled bars) and P2 (black bars). Means followed by the same letter are not significantly different. n = 4 except in NM controls where n = 3.

Dry matter production and P content of plants

Linum usitatissimum inoculated with G. intraradices, G. mosseae, or Gi. rosea, grew markedly better than the non-inoculated (NM) control plants (Fig. 3). The growth responses depended on fungal species and were reflected in both shoot and root dry weights (SDW and RDW). Plants colonized by G. intraradices, but not by the other two AMF, responded significantly to increased P in HC1, with increases in SDW between P0 and P1 and between P1 and P2, and in RDW between P0 and P1 only.

Figure 3.

Shoot (above X axis) and root (below X axis) dry weights of L. usitatissimum colonized by Gi. rosea, G. mosseae and G. intraradices and non-mycorrhizal controls, in pots with three different P concentrations in HC1: P0 (no added P, white bars), P1 (stippled bars) and P2 (black bars). Means followed by the same letter are not significantly different. n = 4 except in NM controls where n = 3.

Phosphorus contents in shoots and roots were significantly greater in all inoculated plants compared with NM plants (Fig. 4), but did not respond to P enrichment of HC1 soil patch in the case of Gi. rosea- and G. mosseae- colonized plants. In contrast, the P1 enrichment, but not the P2 enrichment, led to increased shoot P content in plants colonized by G. intraradices. Root P contents followed a similar trend as shoots. Soil P enrichment of the HC1 patch had little effect on shoot and root P concentrations.

Figure 4.

Shoot (above X axis) and root (below X axis) P contents of L. usitatissimum colonized by Gi. rosea, G. mosseae and G. intraradices and non-mycorrhizal controls, in pots with three different P concentrations in HC1: P0 (no added P, white bars), P1 (stippled bars) and P2 (black bars). Means followed by the same letter are not significantly different. n = 4 except in NM controls where n = 3. Values in brackets are mean tissue P concentrations.

Uptake of P from side-arms

During the plant growth period, radioactivity was first detected consistently in the shoots of Gi. rosea 14 d after planting, followed by G. intraradices and G. mosseae after 16 d (Fig. 5). Data for the P1 and P2 patch treatments followed the same trend as P0 (data not shown). This monitoring of radioactivity did not allow for the separate detection of 32P and 33P and it was therefore not possible to normalize for isotopic decay or to make conclusions on the uptake of P from the patch versus reference side-arms from these data.

Figure 5.

Appearance of detectable radioactivity (32P and 33P) in the shoots of L. usitatissimum (counts per second) colonized by Gi. rosea (•), G. mosseae (◆) and G. intraradices (▪) in P0 (no added P) treatment, over time.

Uptake of P from the patch and reference side-arms at harvest was calculated separately from specific activities of 32P or 33P (see Methods). The NM plants contained negligible 32P or 33P, confirming that diffusion or mass flow of P from labelled soil patches was very small (Figs 6a & b). In all P0 treatments (P0 in both reference and patch side arms), there were no differences in the amount of P accessed from either side-arm (P > 0.05), verifying that there was no discrimination in uptake of 32P and 33P by the AMF.

Figure 6.

P accessed from (a) the patch side-arm and (b) the reference side-arm in the shoots (above X axis) and roots (below X axis) of L. usitatissimum colonized by Gi. rosea, G. mosseae and G. intraradices and non-mycorrhizal controls, in pots with three different P concentrations in HC1: P0 (no added P, white bars), P1 (stippled bars) and P2 (black bars). Means followed by the same letter are not significantly different. n = 4 except in NM controls where n = 3.

The shoots and roots of plants colonized by Gi. rosea took up very little P from either of the side-arms (not significantly different from NM plants), irrespective of patch P concentration (Figs 6a.b). Plants colonized by G. mosseae took up slightly more P, but did not show a significant response to increased patch P concentration. Plants colonized by G. intraradices accessed the greatest quantities of P from the side-arms. Between P0 and P1 there was a more than two-fold increase in P in the shoots derived from the patch side-arm, but there was no further increase between P1 and P2 (Fig. 6a). At the same time P accessed from the reference side-arm decreased between P0 and P1 but not between P1 and P2 (Fig. 6b). Root P accessed from the patch side-arm also increased with increased soil P. There was a corresponding decrease in the amount of P taken up from the reference side-arm when the patch side arm contained P1 and P2 soil. The total P (shoot + root P content) taken up from the patch was significantly greater than taken up from the reference side arm in the P1 and P2 (but not P0) treatments (direct one way anova comparison for each patch treatment).

Uptake of P per unit hyphal length at P0 (patch and reference side-arms combined) allows comparison of the fungi in terms of spatial patterns of P uptake. Assuming that background hyphae in non-mycorrhizal controls were also present in the mycorrhizal treatments and that AMF growth rates were linear, Gi. rosea, G. mosseae and G. intraradices took up, on a whole plant basis, 0.06, 0.2 and 0.3 µg P m−1 hyphae, respectively. These differences between fungi were significant.

DISCUSSION

The present work is the first to use compartmented pots and dual isotopic labelling of P to study the P uptake by different AMF in relation to P enrichment of radio-labelled soil patches. The work is also the first to show that hyphal P uptake from P-enriched patches can influence the concurrent uptake from non-enriched regions of the soil, in one of the AMF tested here. This approach is distinct from previous mycorrhiza experiments using similar model systems where the aims were to investigate the separate contributions of roots and hyphae to P uptake (Pearson & Jakobsen 1993), spatial patterns of P uptake by different AMF (Smith et al. 2000) or P uptake by AMF growing through different media (Drew et al. 2003).

Linum usitatissimum was very responsive to its AMF symbionts with respect to growth. The difference between the three AMF in their ability to increase growth and P content is in accordance with a previous report on differing responsiveness of L. usitatissimum to AM fungi (Ravnskov & Jakobsen 1995). P uptake by some AMF can even account for the total P uptake of mycorrhizal L. usitatissimum plants (Smith, Smith & Jakobsen 2003). Such functional diversity between the AMF used here has been demonstrated with Lycopersicon esculentum and Medicago truncatula (Burleigh, Cavagnaro & Jakobsen 2002) and also with other plant/fungal combinations (e.g. Ravnskov & Jakobsen 1995; Schweiger & Jakobsen 2000; Smith et al. 2000). The data also show a correlation between P uptake from side-arms and hyphal growth there.

The early 32P uptake by Gi. rosea suggests that this fungus extended most rapidly away from the roots, but this was not reflected in the amount of P transferred from the labelled soil to the plants at the end of the experiment. A maximum of 2.6% of the P in the shoots was derived from the P2 patch even though about 42% of the total P in the pot was located there. As plants colonized by Gi. rosea contained much more P than the NM controls this fungus clearly accessed most of its P from close to the roots. A similar pattern of P uptake has been found for M. truncatula colonized by Scutellospora calospora (Smith et al. 2000) and for Trifolium subterraneum colonized by G. mosseae (Drew et al. 2003). Burleigh et al. (2002) found that Gi. rosea was a very poor symbiont with L. esculentum and M. truncatula compared with the other AMF that they tested. The reason for this remains unknown, but the low growth response and higher percentage colonization suggests that the fungus may be utilizing relatively large amounts of C from its hosts. It may also be a reflection of taxonomic (plant) and environmental specificity; this remains to be resolved.

Glomus mosseae readily grew into the side-arms, but the P accessed from the patch as a proportion of the total P absorbed was again small (up to 12.2% at P2). Furthermore, G. mosseae did not respond to increased patch P concentration in terms of P uptake from either side-arm. Therefore, G. mosseae can be considered an intermediate between the other two fungi as it increased plant growth and P uptake by accessing P from near and at a distance from the roots. However, the proportion of P obtained at a distance from the roots may increase as P in the central compartment is depleted. The failure of G. mosseae, and also Gi. rosea, to exploit the patch may reflect the function of hyphae that grew into the patch. For example, these hyphae may be ‘searching’ for another host, rather than foraging for P, or they may simply be inefficient at perceiving, taking up and/or translocating P at a distance from the roots. Given the range of functions that hyphae of AMF serve (Bago, Azcón-Aguilar & Piché 1998), estimation of hyphal length/biomass alone may not be a good indicator of potential nutrient acquisition (see below).

Glomus intraradices was not only able to access the greatest amount of P from the side-arms (up to 35.1% of total plant P at P2) but also increased uptake (compared to other AMF) in response to increased patch concentration. The non-linear increase in P taken up from the patch side arm (i.e. no increase in total plant P between P1 and P2) may simply be a reflection of a plateau in P uptake via the AMF P transporters at P1. The hyphal length increased between P0 and P1. The increase in hyphal length taken together with increases in biomass and P uptake could be explained as a simple dose–response or as foraging for P in the patches by G. intraradices. Interpretation of these results, in the context of foraging, is complicated by the observed decrease in hyphal length in the patch side-arms between P1 and P2. This decrease did not translate into a decrease in P acquisition. However, foraging may not only occur in the form of hyphal growth and proliferation; there may also be physiological responses. Despite the uncertainties, it is probably overly simplistic to only consider hyphal foraging and P uptake in terms of hyphal length. There may be a simple substrate effect (increased rate of P uptake in response to higher soil P) or changes in the abundance and activity of fungal P transporters. This last possibility is supported by an increase in the expression of a G. intraradices P transporter (GiPT) in response to increased P supply (in the range of 1–35 µm phosphate) reported by Maldonado-Mendoza, Dewbre & Harrison (2001). In terms of cost to the fungus it would be more economical to increase the density of transporters to better acquire P than to construct additional hyphae, especially as P depletion around hyphae is probably not a limiting factor in hyphal P uptake (see Barber 1995).

A key result of this experiment was that G. intraradices decreased uptake of P from the reference side-arm concurrent with increased uptake from the patch side-arm. One explanation is that G. intraradices can somehow perceive nutrient heterogeneity and alter its P uptake accordingly. This would be consistent with the conclusion that this AMF can perceive and respond to P levels surrounding the hyphae in monoxenic cultures (Maldonado-Mendoza et al. 2001). The response seems analogous to the plastic response by roots of some plant species that allows them to respond to and utilize nutrient patches (Jackson & Caldwell 1989; van Vuuren, Robinson & Griffiths 1996). Similarly, parallels can be drawn with mechanisms involved in split root systems where the P status of the plant is an important driving force. The mechanisms involved in the exploitation of a P-enriched patch and, importantly, the concurrent decrease in P acquisition from a non-enriched reference area of soil (and presumably from the rest of the pot), cannot yet be explained, but are worthy of further investigation. It is well established that the P status of the plant can influence the uptake of P from the soil by the fungus (Smith & Read 1997) and it may be that increased transfer of P along part of a hyphal network (involving changed gradients of P concentration and of energy expenditure) changes the translocation pattern elsewhere in the network. Possible signal molecules have yet to be identified (as is the case with many changes in nutrient translocation in plants). Maldonado-Mendoza et al. (2001), who found that regulation of expression of GiPT in axenic cultures was a dynamic process, as transcript levels changed in response to changes in the P concentration in the environment surrounding the external hyphae and was modulated by the overall P status of the colonized root. They suggested that AMF may have a mechanism of P sensing (in addition to other driving factors); this is also the case with yeast (Cutler et al. 2001; Gagiano, Bauer & Pretorius 2002). In order to unravel the mechanisms involved in AMF, an approach such as that used in the present experiment combined with gene expression studies (Maldonado-Mendoza et al. 2001; see above) would certainly yield further insight into the response of AMF to nutrient patches and test the suggested ability of AMF to perceive P.

In the current experiment the aim was to investigate the ability of AMF to exploit inorganic nutrient patches; however, in field situations (and non-compartmented pots) both the roots and hyphae would be able to access some of the same patches (depending on patch scale). Plants colonized by AMF such as Gi. rosea and G. mosseae would have to rely upon their own ability to exploit patches to a greater extent than plants colonized by an AMF such as G. intraradices. In the field there may be a range of AMF colonizing the same root system, thus a combination of AMF with a range of strategies (and occupation of potentially different niches) may best benefit the plant, and in turn, themselves (Koide 2000); dual inoculation experiments are required to investigate this.

In summary, the three AMF differed in their exploitation of the nutrient patches. Furthermore, G. intraradices exhibited a differential response to increased patch P concentration in terms of plant growth and P nutrition. The lack of responses by Gi. rosea and G. mosseae may be constitutive or due to inability to respond over the time period assessed. This may have important ecological consequences; for example, a plant colonized by an AMF such as G. intraradices might have a competitive advantage in a patchy environment over plants colonized by AMF such as Gi. rosea or G. mosseae. Our results suggest the existence of a considerable biodiversity in AMF function, which may have wide ecological implications.

ACKNOWLEDGMENTS

This work was conducted while T.R.C. was a visitor in I.J.'s laboratory at Risø National Laboratory. We especially wish to thank the Olsen ladies (Anette and Anne) for their excellent technical assistance in Denmark. We also thank Dr E. Drew for valuable discussions. Our research is funded by the Australian Research Council and the Danish National Research Foundation.

Ancillary