Plant growth depressions in arbuscular mycorrhizal symbioses: not just caused by carbon drain?

Authors

  • Huiying Li,

    1. Soil and Land Systems, School of Earth and Environmental Sciences, Waite Campus, University of Adelaide, Adelaide, South Australia 5005, Australia;
    Search for more papers by this author
  • F. Andrew Smith,

    1. Soil and Land Systems, School of Earth and Environmental Sciences, Waite Campus, University of Adelaide, Adelaide, South Australia 5005, Australia;
    Search for more papers by this author
  • Sandy Dickson,

    1. Soil and Land Systems, School of Earth and Environmental Sciences, Waite Campus, University of Adelaide, Adelaide, South Australia 5005, Australia;
    Search for more papers by this author
  • Robert E. Holloway,

    1. Soil and Land Systems, School of Earth and Environmental Sciences, Waite Campus, University of Adelaide, Adelaide, South Australia 5005, Australia;
    2. Minnipa Agricultural Centre, South Australian Research and Development Institute, PO Box 31, Minnipa, South Australia 5654, Australia
    Search for more papers by this author
  • Sally E. Smith

    1. Soil and Land Systems, School of Earth and Environmental Sciences, Waite Campus, University of Adelaide, Adelaide, South Australia 5005, Australia;
    Search for more papers by this author

Author for correspondence:
F. Andrew Smith
Tel: +61 8 83036517
Fax: +61 8 83036511
Email:andrew.smith@adelaide.edu.au

Summary

  • • This study investigated effects of plant density and arbuscular mycorrhizal (AM) colonization on growth and phosphorus (P) nutrition of a cultivar of wheat (Triticum aestivum) that often shows early AM-induced growth depressions.
  • • Two experiments were conducted. Expt 1 had three plant densities and one soil P concentration. Expt 2 had two plant densities and two P concentrations. Plants were grown in calcareous P-fixing soil, inoculated with Glomus intraradices or Gigaspora margarita, or noninoculated (nonmycorrhizal (NM)). Glomus intraradices colonized well and caused a growth depression only in Expt 1. Gigaspora margarita caused large growth depressions in both experiments even though it colonized poorly.
  • • The results showed that growth depressions were mitigated by changes in relative competition for soil P by NM and AM plants, and probably by decreasing carbon costs of the fungi.
  • • The different effects of the two fungi appear to be attributable to differences in the balance between P uptake by the fungal pathway and direct uptake via the roots. These differences may be important in other AM symbioses that result in growth depressions. The results show that mycorrhizal growth responses of plants grown singly may not apply at the population or community level.

Introduction

Not all plants that form symbioses with arbuscular mycorrhizal (AM) fungi show large positive responses to colonization when grown in soil low in available phosphorus (P). Some plant species show little or no growth increase compared with nonmycorrhizal (NM) controls, and others show growth depressions, at least during vegetative stages. The conventional explanation is that variations in growth responses relate directly to the balance between benefits and costs of the symbiosis. Growth depressions are thus said to be caused by fungal demands for organic carbon (C) from the host plant that outweigh any benefits that might be produced by P transfer via the fungus. The fungus is essentially acting as a weak parasite or a ‘cheater’ in the sense used by Johnson et al. (1997). There is evidence to suggest that differences in growth depressions caused by different AM fungi can be related to differences in C demand, as in wheat (Triticum aestivum; Graham & Abbott, 2000). Similarly, different plant varieties within a species can show differences in responses to colonization by individual AM fungal species or strains: wheat is again a classic example (e.g. Azcón & Ocampo, 1981; Hetrick et al., 1993, 1996, and references therein).

It is now becoming clear that the ‘cost:benefit’ approach is more complex than previously thought. The reason is that, even when there are no positive growth responses to AM formation, or when growth depressions occur, large amounts of plant P can be provided by the AM fungal pathway. This has been shown by transfer of radioactive P from compartments in pots that are accessible to the AM fungi but not to roots. Although not always quantified, such ‘hidden’ P transfer has been shown with the prairie grass Bromus inermis (Hetrick et al., 1994), and a range of cultivated plants: cucumber (Cucumis sativus; Pearson & Jakobsen, 1993), wheat (Ravnskov & Jakobsen, 1995; Hetrick et al., 1996; Li et al., 2006), pea (Pisum sativum; Gavito et al., 2002), barley (Hordeum vulgare; Zhu et al., 2003) and tomato (Solanum lycopersicum; Smith et al., 2003, 2004). Uptake via the AM fungal pathway has also been shown in this way with wheat in the field (Schweiger & Jakobsen, 1999). Our own recent work with pot-grown wheat in a highly calcareous and P-fixing soil type that occurs throughout a major cereal-growing region in South Australia showed that, although AM plants were smaller than the NM controls, they obtained over 50% of their P via the fungus even when P fertilizer was added (Li et al., 2006). Thus, when AM symbioses result in growth depressions the fungi can no longer be assumed to be acting as simple root parasites, and lack of growth ‘responsiveness’ cannot be equated with lack of ‘dependence’ on the AM fungus for P supply (see also Jones & Smith, 2004). Here we use these terms conventionally, rather than as defined by Janos (2007).

Positive growth responses of AM plants compared with NM controls have been shown to decrease with increasing plant density, that is, increasing numbers of plants per unit soil volume (Bååth & Hayman, 1984; Koide, 1991; Allsopp & Stock, 1992; Facelli et al., 1999; Schroeder & Janos, 2004, 2005). The generally accepted explanation is that with increasing plant density competition for P among the AM fungal hyphae reduces the benefits of the AM fungal pathway, so that competition for P among the AM plants becomes the same as that among NM plants. The costs of the symbiosis might then result in growth depressions at high density. The same reasoning suggests that growth depressions that are sometimes found in AM plants grown in high-P soils are likely to be exacerbated at high density. Schroeder & Janos (2004, 2005) examined the interactions among AM colonization, intraspecific plant density and P fertilization in a range of agricultural plant species and concluded that increased density does indeed influence plant responses, by affecting whether or not C (photosynthate) or available soil P limits plant growth. However, there were large differences among the plants that reflected factors such as per cent colonization and root morphology.

So far, very little attention has been given to the effects of plant density on AM plants that are nonresponsive or are negatively responsive (in terms of growth or P uptake) at low soil P. Allsopp & Stock (1992) showed that Aspalathus linearis (Fabaceae) was nonresponsive at low density but negatively responsive at high density. Schroeder & Janos (2004) found transitory growth depressions in maize grown at two densities over a large range of P additions but (perhaps surprisingly) little difference between the treatments. Here we have investigated effects of increasing plant density on wheat grown in the P-fixing soil previously used (Li et al., 2006). In the second of the two experiments there were two soil P concentrations. We do not claim that the densities can be equated with those in the field. The original aim was simply to ascertain effects of density on negative growth responses of AM wheat, and the roles played by two AM fungi that produce different amounts of colonization. The results have led us to reappraise possible explanations for the mitigation of the negative growth responses at high plant densities. More importantly, they have led us to challenge the assumption that growth depressions always arise primarily from the costs of C drain to the fungus exceeding the benefits resulting from P uptake via the AM fungal pathway. We suggest that, at least in plants that do not show large positive responses when mycorrhizal, the balance between P uptake via the AM fungal pathway and directly via the roots may have an important role in determining the plant growth responses to increasing plant density.

Materials and Methods

Experimental design

Two glasshouse experiments were conducted, both with a complete factorial design. Expt 1 had three plant densities (one, four and eight plants per pot) and three AM fungal treatments (two AM fungi and uninoculated). Expt 2 had two plant densities (one and four plants per pot), three AM fungal treatments (as in Expt 1) and two soil P concentrations: no additional P (as in Expt 1) or 20 mg kg−1 P added as CaHPO4. There were four replicates of each treatment in both experiments. Plants were grown under semicontrolled glasshouse conditions. The diurnal air temperatures mostly ranged between 16 and 22°C, but occasionally reached 30°C during summer. Plants were grown under natural light that varied with different seasons and weather conditions in the range 200–1800 µmol m−2 s−1. Expt 1 was conducted between 24 January and 3 March 2004 (39 d; Australian summer) and Expt 2 was conducted between 13 September and 25 October 2005 (42 d; Australian spring). Thus, plants in Expt 1 were exposed to longer day-lengths.

Plant material and AM fungi

Spring wheat (Triticum aestivum L.) cv. Brookton was used, as previously (Li et al., 2005, 2006). Seeds were surface-sterilized in 3% sodium hypochlorite solution for 15 min, rinsed with reverse osmosis (RO) water and germinated on moist paper for 3 d. Germinated seeds were then transplanted into pots. These were thinned to produce different plant densities after emergence.

The AM fungi were Glomus intraradices Schenck and Smith (DAOM 181602) and Gigaspora margarita Becker and Hall (WFVAM21), grown in pot cultures with Allium porrum L. Neither isolate is indigenous to the original soil. Ten per cent of dry inoculum was incorporated into soil in AM treatments. Inoculum consisted of colonized root fragments and soil:sand mix containing spores and external hyphae of the AM fungi. Uninoculated pots with 10% sterilized soil added served as nonmycorrhizal (NM) controls. With these soil treatments such controls have shown no significant differences in terms of plant growth and P uptake, compared with controls grown with added filtrate of the inoculum (Li, 2005).

Growth medium

The growth medium was a 70:30 mix of soil and sand, contained in undrained plastic pots (1.4 kg per pot). It had a pH of 8.0 (0.01 m CaCl2) and 24 and 4 mg kg−1 available P determined, respectively, by bicarbonate extraction (Colwell, 1963) and with anion exchange resin strips (McLaughlin et al., 1994). The soil, collected from a cereal-growing area (Cungena, on the Eyre Peninsula, South Australia), is a highly calcareous grey sandy loam (Li et al., 2005). It was air-dried and passed through a 2-mm sieve. Soil and sand were autoclaved separately at 121°C for 1 h twice over a 3-d period before mixing. The soil:sand mix is referred to as ‘soil’ hereafter.

Nutrients were thoroughly incorporated at the following rates (mg kg−1 dry soil): K2SO4, 75; CaCl2Þ2H2O, 75; MgSO4Þ7H2O, 45; CuSO4Þ5H2O, 2.1; ZnSO4Þ7H2O, 5.4; MnSO4Þ7H2O, 10.5; CoSO4Þ7H2O, 0.39; Na2MoO4Þ2H2O, 0.18; H3BO3, 0.3; Fe EDTA, 0.4. Nitrogen was added as 50 mg N kg−1 (NH4NO3) at the start of the experiment, and 50 and 20 mg N kg−1 after 20 and 30 d, respectively. Soil water content was maintained at 12.5% (w/w) by regular watering with reverse-osmosis RO water. Pots were randomly placed on a single bench in the glasshouse and rearranged when watered.

Harvests and analysis

At harvest, all plants were at the vegetative stage. The shoot of each plant in a pot was cut off, washed, dried (for 24 h at 80°C) and weighed separately. Shoots from each pot were then mixed. Roots were washed and weighed but could not be effectively separated. A subsample of known weight was taken from each pot for measuring AM fungal colonization. The remaining roots were dried and root dry weight (DW) per pot was measured. Soil in each pot was well mixed and subsamples were taken for determination of hyphal length density (HLD) and water content (g g−1 oven-dry soil).

Dry weight per plant was calculated based on the shoot DW of individual plants and means of root DW per plant in each pot, corrected from known fresh:dry weight ratios to take into account removal of subsamples. AM fungal colonization 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. Colonized and noncolonized intersects were scored, but details of arbuscule and vesicle development were not recorded. The HLD was determined by the method of Jakobsen et al. (1992) and calculated as m g−1 oven-dry soil. Plant tissue P concentration was determined following digestion in a solution of nitric-perchloric acid (6 : 1, v/v), and analysed by the phosphovanado-molydate method (Hanson, 1950).

Mycorrhizal growth responses (MGRs) were calculated in terms of plant growth as 100[(AM – NM)/NM], using dry weights of individual AM plants and mean dry weights of corresponding NM plants (Cavagnaro et al., 2003). Plant dry weights were obtained by adding weights of roots and shoots of individual plants. Mycorrhizal responses in terms of P content (MPR) were calculated similarly. Relative competition intensity (RCI) was calculated in terms of dry weight (RCGI) as 100[(W1 − W>1)/W1], where W1 is mean dry weight per plant with one plant per pot, and W>1 is mean dry weight per plant in individual pots with more than one plant per pot. This equation is modified from that in Facelli et al. (1999), who used individual values for W1; the modification reflects different experimental design. Relative competition was calculated similarly in terms of P content (RCPI). High values of these indices indicate high competition.

Data were calculated as means and standard errors of means (SEM) of the four replicates and were analysed by analysis of variance (ANOVA) using GenStat Release 6.1, Lawes Agricultural Trust (Rothamsted Experimental Station, UK). Comparisons between means are based on the least significant differences at the 0.05 probability level. The analysis was checked using permutational multivariate analysis of variance (PERMANOVA: Anderson, 2001; McArdle & Anderson, 2001; see http://www.stat.auckland.ac.nz/~mja/), in particular for pair-wise a posteriori comparisons between values of mycorrhizal responses, RCIs, and shoot:root ratios.

Results

Expt 1. Plant growth

Both AM fungal treatment and density affected plant growth, and there was significant interaction between the two factors (P < 0.001). Individual NM plants weighed significantly more than AM plants at the same density (P < 0.001; Fig. 1a). Weights per plant decreased greatly with increasing density of NM plants, but the decrease was not so marked for plants inoculated with G. intraradices, for which there was no difference between weights with four and eight plants per pot. There was no decrease with the plants colonized by G. margarita, and their highest dry weight per plant was with four plants per pot, although variability was also high. It follows from the data illustrated in Fig. 1(a) that total dry weights per pot increased with increasing density in all treatments (P < 0.001). Thus, the mean total dry weight per pot with eight plants was 13.2 g for NM plants, 7.5 g for plants with G. intraradices, and 8.1 g for plants with G. margarita. The last two values are not significantly different. As indicated in Fig. 1(a), growth depressions caused by the AM fungi became smaller with increasing plant density. This is illustrated in Table 1 by the less negative MGR at eight plants per pot than at one plant per pot for plants colonized by G. intraradices or G. margarita.

Figure 1.

Expt 1: (a) dry weight and (b) phosphorus (P) content per wheat (Triticum aestivum) plant for shoots (above the line) and roots (below the line), for noninoculated (nonmycorrhizal (NM)) plants and plants inoculated with Glomus intraradices (G. intr) or Gigaspora margarita (G. marg), grown at densities of one, four and eight plants per pot. Bars are means of four replicates ± SE. Bars for shoots or roots with the same letters are not significantly different (P < 0.05).

Table 1.  Expt 1: mycorrhizal growth response (MGR), mycorrhizal phosphorus (P) response (MPR) and relative competition index based on plant dry weight (RCGI) or P content (RCPI) of noninoculated wheat (Triticum aestivum; nonmycorrhizal (NM)) or wheat inoculated with Glomus intraradices or Gigaspora margarita, grown at densities of one, four and eight plants per pot
 NMG. intraradicesG. margarita
148148148
  • Values are means of four replicates ± SE. Values with the same letters in each row are not significantly different across treatments by ANOVA (P < 0.05).

  • *

    Pairs of values in a row that are significantly different by PERMANOVA. n/a, data not applicable.

MGR (%)n/an/an/a−64 ± 6b−64 ± 3b−43 ± 13c−82 ± 4a−42 ± 15c−39 ± 7c
RCGI (%)n/a37 ± 1c*60 ± 3c*n/a38 ± 6c37 ± 15cn/a−102 ± 51a−36 ± 15b
MPR (%)n/an/an/a−66 ± 6b−54 ± 4b−31 ± 19c−91 ± 2a−26 ± 20c−25 ± 11c
RCPI (%)n/a67 ± 1c*82 ± 2c*n/a56 ± 4c63 ± 10cn/a−171 ± 73a−53 ± 22b

The decreases in dry weight per plant with increasing plant density (Fig. 1a), when these occurred, resulted from plant competition. This is illustrated by the positive RCGI values in Table 1 for NM plants (where competition was high with eight plants per pot) and for plants colonized by G. intraradices. By contrast, there was no competition with increasing density among plants colonized by G. margarita, where the RCGI values were negative (Table 1). However, the RCGI with eight plants per pot was not significantly different from zero, as can be seen from the dry weights in Fig. 1(a).

As also indicated in Fig. 1(a), there were effects on shoot:root weight ratios, which were highest with one and four NM plants per pot (2.3 ± 0.3 and 1.9 ± 0.2, respectively), decreasing to 1.4 ± 0.3 with eight NM plants per pot. This trend did not occur in the AM plants, which had quite similar ratios (means of 1.4–1.7) at all plant densities.

Phosphorus uptake

Effects of density and AM fungi on P uptake again varied with different plant density, fungus and plant tissue. Decreases in P content per plant with increasing plant density for NM plants and plants with G. intraradices were steeper than decreases in DW (Fig. 1b; cf. Fig. 1a), as a result of decreasing tissue P concentrations (mg P g−1 DW) with increasing plant density (Table 2). With eight plants per pot, the P concentrations in both shoots and roots of these plants were c. 50% of those with one plant per pot. The values at high density matched the values in plants colonized by G. margarita, which were low even at the lowest plant density. The only significant difference between P concentrations in NM plants and plants colonized by G. intraradices was a higher value in shoots of the latter at four plants per pot. Concentrations in plants colonized by G. margarita were the same with one and eight plants per pot, but with four plants per pot there was also a significantly higher value in shoots than in the NM plants, associated with the relatively high plant growth of plants colonized by G. margarita, as shown in Fig. 1(a).

Table 2.  Expt 1: phosphorus (P) concentration in shoots and roots of noninoculated wheat (Triticum aestivum; nonmycorrhizal (NM)) or wheat inoculated with Glomus intraradices or Gigaspora margarita, grown at densities of one, four and eight plants per pot
 NMG. intraradicesG. margarita
148148148
  1. Values are means of four replicates ± SE. Values with the same letters in each row are not significantly different (P < 0.05).

P concentration (mg g−1)Shoot1.2 ± 0.1c0.6 ± 0.0a0.5 ± 0.1a1.1 ± 0.1c0.9 ± 0.1b0.6 ± 0.1a0.6 ± 0.0a0.8 ± 0.1b0.6 ± 0.1a
Root0.8 ± 0.1d0.4 ± 0.0a0.4 ± 0.0a0.9 ± 0.1d0.5 ± 0.1ab0.5 ± 0.0abc0.5 ± 0.1ab0.6 ± 0.1c0.6 ± 0.0bc

The data for total P per plant (Fig. 1b) were used to calculate MPRs (Table 1); as with MGRs, values became less negative with increasing plant density. Values of RCI on the basis of plant P content (RCPI), also shown in Table 1, followed the pattern of RCGI, taking into account the differences in P concentrations (Table 2); the strongest competition was again with eight NM plants per pot. There was again large variance in the negative values of RCPI in plants colonized by G. margarita, with the value at the highest density again not significantly different from zero (cf. P content in Fig. 1b).

AM colonization and hyphal length densities

All uninoculated plants remained nonmycorrhizal (results not shown). All inoculated plants became mycorrhizal, with colonization much higher for G. intraradices than for G. margarita (Fig. 2a). Arbuscules and vesicles were found in the roots with G. intraradices, but not in the roots with G. margarita, which contained only hyphae, including coils. Increasing plant density decreased colonization by G. intraradices but not by G. margarita. HLDs in NM pots were ~0.4 m g−1 dry soil, which provided a background value (of dead hyphae or soil saprophytes), with no significant differences between different plant densities (results not shown). These values have not been subtracted from those shown in Fig. 2(b). Trends in HLDs for the two AM fungi and different densities were generally similar to those of colonization (cf. Fig. 2a,b).

Figure 2.

Expt 1: (a) per cent arbuscular mycorrhizal (AM) colonization, and (b) hyphal length densities for wheat (Triticum aestivum) inoculated with Glomus intraradices (G. intr) or Gigaspora margarita (G. marg), grown with densities of one, four and eight plants per pot. Bars are means of four replicates ± SE. Bars with the same letters are not significantly different (P < 0.05).

Expt 2. Plant growth

All three experimental factors (P application, fungus and density) affected plant growth and there were significant two- and three-factor interactions (P < 0.001). With no added P (P0), NM plants were ~20–25% smaller than in the corresponding treatments in Expt 1 (cf. Figs 1a and 3a). Overall, application of P (P20) significantly increased plant growth (Fig. 3a). As in Expt 1, both NM plants and plants colonized by G. intraradices were smaller with four plants per pot than with one plant per pot, but with four plants per pot growth with G. intraradices at P0 (~1.8 g per plant) was greater than in Expt 1 (~0.9 g per plant). The net effect of the different growth rates at P0 in this experiment was that mean MGR values for G. intraradices were close to zero (Table 3; see also Fig. 3a). There were again large growth depressions (negative MGR) with G. margarita at P0, although these were slightly ameliorated with four plants per pot (Table 3).

Figure 3.

Expt 2: (a) dry weight and (b) phosphorus (P) content per wheat (Triticum aestivum) plant for shoots (above the line) and roots (below the line), for noninoculated (nonmycorrhizal (NM)) plants and plants inoculated with Glomus intraradices (G. intr) or Gigaspora margarita (G. marg), grown at densities of one and four plants per pot in soil with no additional P (P0) or 20 mg kg−1 P added as CaHPO4 (P20). Bars are means of four replicates ± SE. Bars for shoots or roots with the same letters are not significantly different (P < 0.05).

Table 3.  Expt 2: mycorrhizal growth response (MGR), mycorrhizal phosphorus (P) response (MPR) and relative competition index based on plant dry weight (RCGI) or P content (RCPI) of noninoculated wheat (Triticum aestivum; nonmycorrhizal (NM)) or wheat inoculated with Glomus intraradices or Gigaspora margarita, grown in soil with no additional P (P0) or 20 mg kg−1 P added as CaHPO4 (P20), and at densities of one and four plants per pot
 NMG. intraradices  G. margarita
P0P20 P0 P20P0P20
141414141414
  1. Values are means of four replicates ± SE. Values with the same letters in each row are not significantly different (P < 0.05). n/a, data not applicable.

MGR (%)n/an/an/an/a7 ± 12d−16 ± 3c−14 ± 6c3 ± 4d−55 ± 3a−43 ± 6b−64 ± 3a−21 ± 5c
RCGI (%)n/a33 ± 3bn/a59 ± 2dn/a48 ± 2cn/a51 ± 2cn/a14 ± 10an/a10 ± 6a
MPR (%)n/an/an/an/a−8 ± 4d−27 ± 4bc−30 ± 5b−14 ± 5cd−54 ± 2a−24 ± 11bc−61 ± 3a−21 ± 3c
RCPI (%)n/a61 ± 3bn/a71 ± 1bn/a69 ± 2bn/a65 ± 2bn/a36 ± 9an/a41 ± 3a

At P20, mean values of MGR with G. intraradices were again small, and with four plants per pot there was no significant difference from zero (Table 3, Fig. 3a). Mean MGR for G. margarita with four plants per pot at P20 was much less negative than with one plant per pot, and was less negative than at P0 (Table 3).

In this experiment, RCGI values related only to four plants per pot (Table 3). Mean RCGIs at P0 were higher for plants colonized by G. intraradices than for NM plants, and plants colonized by G. margarita had a small positive value. The RCGIs were higher at P20 than at P0 for NM plants but similar for plants colonized by G. intraradices and for those colonized by G. margarita (Table 3).

As indicated in Fig. 3(a), there were again differences in shoot:root weight ratios. At P0 these were higher in NM plants with one plant per pot (1.8 ± 0.4) than with four plants per pot (1.4 ± 0.1). The same trend was evident in plants colonized by G. intraradices (2.0 ± 0.3 and 1.4 ± 0.2) or by G. margarita (2.3 ± 0.4 and 1.6 ± 0.2), again with one or four plants per pot, respectively. At P20 the shoot:root ratios of NM plants were the same with one and four plants per pot (1.4 ± 0.2 and 1.5 ± 0.2), but values decreased from 1.9 ± 0.1 to 1.5 ± 0 with G. intraradices, and from 2.1 ± 0.2 to 1.7 ± 0.2 with G. margarita.

Phosphorus uptake

As with growth, the three experimental factors affected P uptake and there were significant two- and three-factor interactions (P < 0.001). The trends of P uptake per plant with increased density and the AM fungal treatments were generally similar to those for weight (Fig. 3b), with differences in detail reflecting changes in P concentrations in both roots and shoots (Table 4). These concentrations were much higher at P0 than in Expt 1. They were generally similar across the NM and AM fungal treatments and lower at the higher plant density. Concentrations of P in plants colonized by G. margarita with one plant per pot were ∼2.5 mg g−1 in shoots and ∼1.2 mg g−1 in roots, compared, respectively, with only ∼0.6 and ∼0.5 mg g−1 in Expt 1 (Table 2). Application of P generally increased shoot and root P concentrations of plants (Table 4). Plants colonized by G. intraradices had shoot P concentrations ∼10–20% lower at both soil P and density levels compared with NM plants, but there were no differences between root P concentrations overall. The only significant difference between plants colonized by G. margarita and NM plants was a higher shoot P concentration at P0 in the former with four plants per pot.

Table 4.  Expt 2: phosphorus (P) concentration in shoots and roots of noninoculated wheat (Triticum aestivum; nonmycorrhizal (NM)) and wheat inoculated with Glomus intraradices or Gigaspora margarita, grown at densities of one and four plants per pot in soil with no additional P (P0) or 20 mg kg−1 P added as CaHPO4 (P20)
 NMG. intraradicesG. margarita
141414
  1. Values are means of four replicates ± SE. Values with the same letters in each row are not significantly different (P < 0.05).

P concentration (mg g−1)ShootP02.5 ± 0.1e1.4 ± 0.1b2.1 ± 0.1d1.2 ± 0.0a2.5 ± 0.0e1.9 ± 0.1c
P203.3 ± 0.2c2.2 ± 0.0b2.4 ± 0.1b1.7 ± 0.1a3.5 ± 0.2c2.2 ± 0.2b
RootP01.3 ± 0.1d0.8 ± 0.1a1.1 ± 0.1bc0.8 ± 0.0a1.2 ± 0.0c1.0 ± 0.1b
P201.7 ± 0.2c1.3 ± 0.0a1.7 ± 0.1bc1.3 ± 0.1a1.4 ± 0.2ab1.2 ± 0.2a

In terms of P content per plant (Fig. 3b), MPRs generally followed the trends in MGRs, allowing for differences in P concentrations (Table 3). This was also the case with RCPIs. With four plants per pot at P0, RCPIs were high for NM plants and for plants colonized by G. intraradices, and lower (but positive) for plants colonized by G. margarita (Table 3). Values of RCPI at P20 followed the same pattern. Differences between RCI values based on weight and P content again reflect differences in plant P concentrations between treatments, this time at P0 or P20.

AM colonization and hyphal length densities

There was again no colonization in uninoculated plants (results not shown). Colonization in inoculated plants was again much higher for G. intraradices than for G. margarita (Fig. 4a). Both application of P and higher plant densities decreased colonization by G. intraradices, but this was not always so for G. margarita. Hyphal length densities in NM pots were ∼0.3 m g−1, with no significant differences between treatments (results not shown). Increased plant density decreased HLDs in G. intraradices treatments only at P0 (Fig. 4b). With G. margarita, HLDs were lower than with G. intraradices and were independent of plant density, as in Expt 1.

Figure 4.

Expt 2: (a) per cent arbuscular mycorrhizal (AM) colonization, and (b) hyphal length densities for wheat (Triticum aestivum) inoculated with Glomus intraradices (G. intr) or Gigaspora margarita (G. marg), at densities of one and four plants per pot in soil with no additional P (P0) or 20 mg kg−1 P added as CaHPO4 (P20). Bars are means of four replicates ± SE. Bars with the same letters are not significantly different (P < 0.05).

Discussion

Nonmycorrhizal plants

The results overall showed that plant growth and P uptake were limited by the low available P in the calcareous soil that was used as the P0 treatment. This agrees with previous findings that the soil is very low in plant-available P (Holloway et al., 2001; Li et al., 2005, 2006). The NM plants competed more strongly for P at increased plant density, as shown by decreased growth and P uptake per plant (leading to lower tissue P concentrations) and increasing RCIs in terms of both dry weight and P content. There were also lower shoot:root ratios at higher density, a typical plant response to low-P soil. Growth limitation by soil P at P0 was also shown in Expt 2 by growth increases of the NM plants at P20. However, at P20 there was still considerable competition for soil P at the higher plant density, as again shown by decreased growth and P uptake per plant and the RCI values. Large differences in plant P concentration between the two experiments at P0 (also found in AM plants) are unexplained but presumably relate to different environmental conditions in the glasshouse. They have been found previously in other experiments at about the same growth stage (Li et al., 2005, 2006).

These findings show that mitigation of growth depressions (less negative MGR) with increasing plant density relates to higher competition for P by NM plants compared with smaller AM plants. The argument is analogous to that used to explain decreasing MGR in positively responsive AM plants with increasing density, although in that case the cause is the stronger competition for soil P by the larger AM plants (Bååth & Hayman, 1984; Koide, 1991; Facelli et al., 1999). However, the results obtained here with the two AM fungi were different overall, and these have to be considered separately, as follows.

Symbiosis with Glomus intraradices

As with the NM plants, plants colonized by G. intraradices showed considerable competition with increasing density, that is, there were decreases in weight and P uptake per plant, the latter again reflecting lower tissue P concentrations. However, large growth depressions were only found in Expt 1. There is no reason to believe that fungal C demand was higher in Expt 1, as there was high fungal biomass in both experiments, so the difference between experiments was presumably also a result of differences in growth conditions in the glasshouse that clearly affected P uptake. Growth depressions are frequently associated with relatively low light intensities (see Johnson et al., 1997). In this case the depressions caused by G. intraradices occurred in the Australian summer when the growth conditions ought to have ameliorated any such effects. Large growth depressions in wheat colonized by G. intraradices have been found previously in experiments carried out over the period April–September (winter), but not in October–December (late spring) (Li et al., 2005, 2006). It is possible that higher soil temperature under semicontrolled ‘late spring’ or ‘summer’ conditions might have increased C costs via increased fungal respiration and possibly turnover. These possibilities need further study, for example under fully controlled conditions and with ‘typical’ field plant densities.

Mitigation of the growth depression at the highest plant density in Expt 1 is clearly shown by the less negative MGR (Table 1: –43% compared with –64% with one plant per pot). This finding is also consistent with an explanation in terms of differences in competition for soil P between NM and AM plants (Schroeder & Janos, 2004). Nevertheless, the results indicate that there is another contributory factor. This relates to decreasing per cent root colonization and HLDs with increasing plant density (Figs 2, 4). Responses of AM colonization to increased plant density in plants that are positively AM-responsive have been shown to be very variable. Per cent colonization decreased in Allium cepa (onion) with increasing density (Bååth & Hayman, 1984) and in one experiment with Trifolium subterraneum, but not in another (Facelli et al., 1999). There was no change in Abutilon theophrasti (Koide, 1991), and the five species used by Schroeder & Janos (2004, 2005) showed a range of responses that included density-dependent increases in tomato. A likely cause of decreases, when they occur, is that with a constant inoculum source there would be a lower number of AM ‘infection units’ per metre of root at higher plant densities, at least initially. The decreasing HLDs with increasing plant density found here were unexpected, because with increasing plant biomass per pot (and hence per g soil) some increase in HLD over the time-span of the experiment would have been expected. As this was not found, it is clear that overall AM fungal biomass decreased enormously per unit plant biomass. We conclude that decreasing C costs per unit plant biomass with increasing plant density may be a factor in mitigating the plant growth depressions with G. intraradices, when these occur. However, an entirely contrary argument can also be advanced in the light of the knowledge that G. intraradices can transport a considerable proportion of P to wheat plants (one plant per pot) grown under similar experimental conditions to those used here (Li et al., 2006). Thus, depending on the (unknown) effect of increasing plant density on P uptake directly by the roots, the decreasing AM fungal biomass (especially HLDs) per unit plant biomass might actually result in a decrease in total P uptake per plant and hence plant growth. This would act against any mitigation resulting from increasing competition for P among the NM plants.

Symbiosis with Gigaspora margarita

Inoculation with G. margarita resulted in very large growth depressions in both experiments. In Expt 1 these were associated with low tissue P concentrations even at low plant density, but this was not the case in Expt 2, so the explanation cannot be simply plant P deficiency. In Expt 2 there was no growth increase with four plants per pot compared with one plant per pot, as had been found in Expt 1. In Expt 1, the depression was much smaller with eight plants per pot than with one plant per pot. Depressions were also smaller at the higher density in Expt 2 at both P0 and P20, particularly in the latter case.

Because plants colonized by G. margarita were so small it is not surprising that plant competition was lacking or low on a dry weight basis compared with NM plants and plants colonized with G. intraradices. The negative values of RCI found in Expt 1 with four plants per pot are unexplained but the high variance should be noted. In Expt 2 the total P uptake by plants colonized by G. margarita was much higher than in Expt 1, and this would account for the positive RCIs calculated on the basis of P content (Table 3).

Unlike G. intraradices, G. margarita colonized plants very poorly and HLDs were in general also lower in both experiments, except at P20 with four plants per pot in Expt 2. In other words, C demands of G. margarita must have been much lower than those of G. intraradices, but there was a similar growth depression in Expt 1, and a large depression in Expt 2 where G. intraradices had little effect. It follows that the large growth depressions caused by G. margarita cannot be explained simply in terms of high C cost to the plant. Also, it is very unlikely that decreasing C cost can be a factor in the mitigation of the depressions at high plant density. The simplest explanation – alluded to as a possibility with G. intraradices– is that colonization by G. margarita, although quite low, greatly decreased P uptake directly into the roots, but that in this case G. margarita acted as a ‘cheater’ and supplied little or no P to the roots via the AM pathway. The nature of the signalling between the symbionts is not known, and it is interesting that the plants did not respond (as they would to very low soil P availability) by increased root biomass relative to shoots, that is, decreased shoot:root ratio. However, increases in root length:weight ratio cannot be ruled out.

Conclusions

As summarized in the Introduction, there is now good evidence that AM plants that do not show large positive growth responses and those prone to growth depressions can, depending on the individual AM fungus, obtain large amounts of P via the AM fungal pathway. Unless quantified by use of 32P or 33P this can be ‘hidden’ in terms of total P uptake and changes (or not) in growth. Potential operation of the AM fungal pathway is also indicated by AM-induced expression of P transporter genes in some nonresponsive plants or plants that show only small positive responses, including tomato (Poulsen et al., 2005). Such genes exist in wheat, and although tissue expression patterns have not been confirmed there is an expectation that the AM-inducible P transporter would be expressed in colonized cortical cells (Glassop et al., 2005). However, such expression does not quantify the P flux through the AM fungal pathway.

There are of course a number of caveats. One arises from the poor ‘fitness’ of the wheat/G. margarita association in the soil used here. Nevertheless, we do not believe that the conclusions overall can be considered irrelevant to the understanding of AM function generally. It may be that the growth depressions that have been found experimentally in many other AM symbioses (Johnson et al., 1997) arise not solely from C costs of an individual symbiosis but also from replacement of direct P uptake by the root epidermis and root-hairs by less efficient uptake (per unit plant biomass) via the AM fungal pathway. This possibility is strongest when growth depressions are very large when compared with likely C costs of the symbiosis. It may explain the finding by Graham & Abbott (2000) that early growth of wheat was depressed by a group of ‘nonaggressive’ AM fungi (low colonization in both low- and high-P soil) as well as by ‘aggressive’ AM fungi (high colonization in low-P soil). Other examples in the literature – although considered at the time in terms of C costs – include work with soybean (Glycine max; Bethlenfalvay et al., 1982a,b) and, especially, tobacco (Nicotiana tabacum; Modjo & Hendrix, 1986). Inactivation of the direct P uptake pathway was suggested as one possible explanation for the large growth depressions in the AM-defective tomato mutant reduced mycorrhizal colonization (rmc) that were associated with surface colonization (only) by a mixed inoculum of Glomus mosseae and G. intraradices (Neumann & George, 2005). This effect apparently relates to the AM fungal taxa that were used. Thus, there was no such depression in rmc inoculated with Glomus coronatum (Cavagnaro et al., 2004). Also, there was none in rmc with another isolate of G. intraradices, but there was a large depression in rmc after inoculation with Glomus versiforme, although not in the wild-type tomato (Poulsen et al., 2005). Poulsen et al. (2005) also showed that 32P uptake into rmc via the fungal pathway was (not surprisingly) negligible and, interestingly, that after inoculation with G. versiforme not only was there no gene expression for two AM-inducible P transporters (LePT3 and LePT4), but one of the epidermal P transporters, LePT2, was also very low. As intraradical colonization was negligible and HLD was very low, the rmc/G. versiforme association seems a very strong case for suppression of the direct P uptake pathway as a major contribution to the growth depression. Apart from the suggestion by Neumann & George (2005), changes in operation of the two pathways for P uptake by AM plants have not previously been considered in the ‘benefits vs costs’ approach to AM-mediated growth depressions, as far as we are aware.

Clearly, low efficiency of P uptake will aggravate the C costs of individual AM symbioses that become well established. The present results suggest that lower efficiency may arise from any or all of (1) the individual AM fungus, (2) the individual plant host, and (3) the growth conditions. Whether such low P uptake efficiencies can be considered as artefacts compared with those in the field is of course debatable. There are certainly convincing demonstrations that individual AM fungi isolated from the field give no or negative growth responses in some of their co-occurring plants (grown singly), while they show positive responses in others (e.g. Klironomos, 2003).

Another caveat is that a relationship between AM fungi and plants that shows net benefits in terms of P nutrition may occur only during periods of high P demand. Periods of high P demand are likely to be associated with periods of high respiration and photosynthetic rates such as during flowering or seed production, and this is when AM fungi may be particularly important to maintain high P uptake rates (McGonigle & Fitter, 1988; Bryla & Koide, 1990a,b; Koide & Dickie, 2002). In a previous study with the same soil, AM fungi suppressed early vegetative growth of wheat but increased seed P content and grain yield, especially with moderate additions of P (Li et al., 2005). These results show that functional strategies in AM associations in wheat, as in other plants, will also depend on plant growth stages.

Despite the complexities – including the varying responses to G. intraradices– this study adds to the increasing evidence that AM growth responses (whether positive or negative) of plants grown singly may not apply when plants of the same or different species are grown together. In the absence of knowledge of fungal biomass and amounts of P taken up via the individual AM fungi, and the AM fungal assemblage within a plant root system, conclusions about the role of AM fungi in plant community and population ecology are likely to remain hazardous. This is particularly a problem where natural ecosystems involve AM plants that are constitutively not positively responsive when grown alone, or are in combination with AM fungi that are constitutively likely to cheat potential hosts. In this context, it is worth noting the argument by Kiers & van der Heijden (2006) that cheating should be common in nature, and the opposing argument by Fitter (2006) that it should be rare. The arguments depend very much on assumptions about the functional interactions between fungus and plant that we have shown require further examination.

In summary, we have shown that plant growth depressions in wheat caused by AM associations decreased with increasing plant density. Factors likely to have determined the magnitude of the depressions were the balance among: increasing competition for soil P by the larger NM plants compared with the AM plants; decreasing C costs to the plants resulting from decreasing AM fungal biomass per unit plant biomass; decreasing uptake of P via the AM fungi; and (especially for G. margarita) decreased P uptake by the root irrespective of plant density. The last factor will depend on signalling between the symbionts that is unresolved at the molecular level. We believe that such factors are likely to be involved in AM fungus/plant interactions in plant ecosystems generally, especially at the population and community levels, and that they should not be overlooked in future. However, teasing out the importance of the individual factors will be a challenging task in terms of the range of experimental techniques that will be required.

Acknowledgements

Our research is funded by the South Australia Grain Industry Trust (Project no. 05SAGIT_UA1/05) and the Australian Research Council. Huiying Li is grateful for an International Postgraduate Research Scholarship from the University of Adelaide. We thank Debbie Miller and Colin Rivers (University of Adelaide) for technical support, and Dot Brace (Minnipa Station, South Australia Research and Development Institute: SARDI) for providing soil. We are also grateful to Nigel Wilhelm (SARDI) for valuable comments on a draft of this paper, and Dr Evelina Facelli and Jean-Patrick Toussaint (University of Adelaide) for advice on statistics.

Ancillary