• Arbuscular mycorrhizal fungal (AMF) communities were established in pots using fungal isolates from a single field in Switzerland. It was tested whether multispecies mixtures provided more phosphorus and supported greater plant growth than single AMF species.
• Two host plants, medic (Medicago truncatula) and leek (Allium porrum), were inoculated with three AMF species (Glomus mosseae, G. claroideum and G. intraradices), either separately or in mixtures. The composition of the AMF communities in the roots was assessed using real-time PCR to determine the copy number of large ribosomal subunit genes.
• Fungal communities in the roots were usually dominated by one AMF species (G. mosseae). The composition of the communities depended on both plant identity and the time of harvest. Leek colonized by a mixture of G. claroideum and G. intraradices acquired more P than with either of the two AMF separately.
• Direct evidence is provided for functional complementarity among species within the AMF community colonizing a single root system. Competition among the species poses a major challenge in interpreting experiments with mixed inoculations, but this is greatly facilitated by use of real-time PCR.
Arbuscular mycorrhizal fungi (AMF) form symbiotic associations with plant roots, and are involved in plant nutrient uptake, growth and tolerance to environmental stresses (Fitter & Moyersoen, 1996; Smith & Read, 1997). The responses of plants to colonization by AMF vary from negative to positive depending on plant and AMF species, as well as on environmental conditions such as soil nutrient availability, light intensity and temperature (Smith & Smith, 1996; Johnson et al., 1997). This variation has been observed among AMF isolates belonging to different species, as well as among isolates of the same species (van der Heijden et al., 1998; Klironomos, 2003; Munkvold et al., 2004; Smith et al., 2004). (The term ‘isolate’ is used here for a laboratory culture established from one or several spores.) Almost all the data on variability of AMF functions have been obtained from experiments in which the plants have been inoculated with single AMF isolates, and plant growth or total phosphorus uptake have been measured. Such experiments are not fully relevant to field situations, where more than one AMF species is usually present in a single root system (Daft, 1983; Merryweather & Fitter, 1998; Jansa et al., 2003b). The experiments have shown variations in P acquisition strategies by different fungi (Jakobsen et al., 1992; Smith et al., 2000; Jansa et al., 2005). Therefore the current challenge is to establish mixed communities using different AMF species within the roots of host plants to substantiate the theory of functional complementarity suggested by Koide (2000), which arose from the work of Smith et al. (2000).
The consequences of simultaneous colonization of a plant by functionally different AMF have been little explored until very recently (Lekberg et al., 2007; Maherali & Klironomos, 2007). Previously, it has been suggested that if a plant is colonized by AMF species that are complementary in their functions (e.g. uptake of nutrients from different soil pools), they may prove to be more beneficial for the plant as a mixture than any of the species separately (Koide, 2000; Alkan et al., 2006; Gustafson & Casper, 2006). Indeed, the diversity of AMF communities in the roots has been shown to correlate positively with P and nitrogen concentrations in the shoots of Plantago lanceolata (Johnson et al., 2004). However, other data do not support the idea of functional complementarity. Rather, some previous studies indicated that maximum benefits to plants might be achieved with a single, most efficient AMF species, and that increasing mycorrhizal diversity would not bring further benefits (Daft & Hogarth, 1983; Edathil et al., 1996).
Direct experimental evidence for functional complementarity in an AMF community has been difficult to obtain, for several reasons. It requires the availability of inocula of more than one AMF species of comparable infectivity, preferably isolated from the same ecosystem (to ensure the relevance of the results to that ecosystem and to avoid potentially unrealistic competition among AMF of different origin). These conditions are not easy to fulfil. The AMF used for such experiments should be identified clearly and tested for their functions separately, in advance. Detection of different species in an AMF community inside the roots has been very difficult until recent applications of molecular methods (van Tuinen et al., 1998; Husband et al., 2002; Redecker et al., 2003), and the quantification of different AMF species in the roots is still poorly developed. Ordinary PCR-based approaches are all potentially biased because of the differential amplification and cloning efficiencies of the DNA stretches from different AMF species, which provide only limited precision, and they are also laborious and costly (Clapp et al., 2002; Jansa et al., 2003b; Sanders, 2004). Quantitative competitive and real-time PCR approaches have only recently been applied for dissecting the composition of an AMF community composed of two AMF species (Edwards et al., 1997; Alkan et al., 2004; Isayenkov et al., 2004).
In this study, synthetic communities (assemblages) were established of three AMF species (Glomus claroideum, G. intraradices, and G. mosseae) isolated from a single field site in Switzerland (Jansa et al., 2002a). These fungi had previously been shown to vary in their strategy to obtain soil P (Jansa et al., 2005). They differed both in the distance from the roots, from which they could take up P (labelled with 33P isotope) and in their efficiency of P uptake from the soil, estimated by measuring 33P uptake of maize via a mycorrhizal pathway from labelled P solution injected into an established mycelial network. The mycelium of G. claroideum took up P from distances less than 6 cm from the roots, whereas G. mosseae and G. intraradices grew 15 cm away from the roots (Jansa et al., 2003a, 2005). Additionally, the efficiency of P uptake from the soil (P taken up per unit hyphal length) was lower for G. mosseae than for G. intraradices (Jansa et al., 2005). We aimed to confirm these effects on plant nutrition in two different host plants with the fungi inoculated separately. We also quantified the AMF community composition in roots using a novel real-time PCR assay that permitted determination of the contribution of the different AMF to overall root colonization. Using this method we sought to establish whether the time course of colonization differed between AMF, and whether different AMF differed in their abilities to colonize roots when inoculated singly or in mixtures. We also aimed to determine whether an AMF community composed of two or three functionally different AMF promoted greater growth and total P uptake of the two plants used (Medicago truncatula and Allium porrum) than single AMF species.
Materials and Methods
Real-time PCR for quantification of AMF colonization
PCR primers (for sequences see Table S1 in Supplementary Material) were designed for specific amplification of large ribosomal subunit (LSU) genes from four different AMF species, based on sequencing data published previously (Jansa et al., 2003b). These primers (synthesized and purified using HPLC at GeneWorks, Hindmarsh, SA, Australia) were each confirmed not to cross-amplify sequences from the other AMF species used in this study by performing cross-amplification tests with spore DNA, extracted and amplified according to established protocols (Jansa et al., 2003b; data not shown). We also established a real-time PCR quantification protocol for LSU copy number, as follows. First, DNA was extracted from single spores of Glomus claroideum Schenck & Smith BEG 155 (BEG, International Bank for the Glomeromycota; http://www.kent.ac.uk/bio/beg); Glomus intraradices Schenck & Smith BEG 158; and Glomus mosseae (Nicol. & Gerd.) Gerd. & Trappe BEG 161, as described previously (Jansa et al., 2003b). Second, these DNA extracts were used as templates for PCR with LR1 and FLR2 primers (Turnau et al., 2001; Jansa et al., 2003b), using cycling conditions identical to those described previously (Jansa et al., 2002a). Third, the concentration of LSU copy numbers (NC, copies per l) in each of the PCR products was calculated by knowing the DNA product fragment length, L (759, 761, and 767 bp for G. claroideum, G. intraradices and G. mosseae, respectively); concentration of DNA in the sample (K, g l−1, determined by UV spectrophotometry at 260 nm using salmon sperm DNA as reference); and molecular weight of DNA (660 Da bp−1) using equation 1, where Na is Avogadro's constant (6.023 × 1023).
NC = (K × Na)/(660 × L) (Eqn 1)
Fourth, the PCR products were serially diluted with sterile water so as to obtain billions to thousands of copies per µl. Additionally, we used a pGEM-T Easy plasmid (Promega, Annandale, NSW, Australia) carrying an LSU fragment of Scutellospora pellucida (Nicol. & Schenck) Walker & Sanders BEG 163, delimited by LR1 and FLR2 primers, as an internal standard for quantification of AMF in root DNA extracts (see below). These samples were used as templates for a real-time PCR assay, employing the LightCycler 1.0 System (Roche, Castle Hill, NSW, Australia) and Qiagen chemistry (QuantiTect SYBR Green PCR Kit, Qiagen, Doncaster, VIC, Australia). Cycling conditions were as follows: initial denaturation at 95°C for 15 min, then 45 cycles with denaturation at 95°C for 15 s, annealing at 58°C for 60 s (for S. pellucida, G. mosseae and G. claroideum) or 90 s (for S. pellucida and G. intraradices), followed by elongation at 72°C for 1 min. The cycling was finalized by elongation at 72°C for 10 min. Melting curves were then analysed in the LightCycler to confirm the length of amplified DNA fragments. The cross points were recorded, where real-time PCR curves reached their second derivation maxima. The numbers of cycles at which these points were reached were used as the DNA quantity estimates correlating with the DNA copy numbers. This way, quantification of target LSU copies in the samples was possible and highly reproducible within the entire range tested (Fig. S1).
To ensure the robustness of the real-time PCR procedure for identification of a specific AMF species in the presence of another AMF species, we performed a dilution assay. The DNA extracted from roots of M. truncatula colonized by a target AMF species (e.g. G. mosseae) was serially diluted with DNA from roots colonized by different AMF species, or with DNA from nonmycorrhizal roots. We confirmed that the presence of DNA from a plant or from nontarget AMF species did not interfere with the real-time PCR assay (Fig. S2).
Plant growth substrate
The substrate used in the pot experiment consisted of autoclaved soil (from Mallala, South Australia) and heat-sterilized sand, mixed in the ratio 1 : 9. It had the following properties: total P (HNO3 : HClO4 digest) 32 mg kg−1; available P (Colwell, 1963) 0.83 mg kg−1; immediately available P (E1min) assessed by isotope exchange kinetics approach (Frossard & Sinaj, 1997) 0.045 mg kg−1; P available within 8 wk extrapolated from a short-term (60 min) isotope-exchange kinetics 0.31 mg kg−1, pH(CaCl2) 7.88.
Plants and AMF
Seeds of Medicago truncatula Gaertn. (medic) cv. Jemalong and Allium porrum L. (leek) cv. Vertina were sterilized in 1% active chlorine solution (diluted commercial bleach) for 5 min, then washed with sterile water and germinated for 3 or 9 d (for medic and leek, respectively) on moist sand at 25°C in darkness. Single seedlings were planted in each pot at the start of the experiment (here referred to as sowing time).
Three AMF isolates belonging to different species of the genus Glomus, all isolated from a single field site in Switzerland (Jansa et al., 2002a), were included. These were G. mosseae BEG 161, G. claroideum BEG 155 and G. intraradices BEG 158. The inoculum was produced in 1.4-kg pots for 6 months between April and September 2003, and consisted of colonized Mallala soil and sand (1 : 9) planted with medic and leek mixtures. Colonization of roots in the inoculum pots was 98–100% of root length for all three AMF species. The pots with G. intraradices inoculum contained many vesicles inside the roots, as well as an extensive mycelial network in the substrate, on which no spores were observed. Glomus mosseae inoculum contained 18 spores g−1 and much mycelium. Inoculum of G. claroideum contained 114 spores g−1, as well as colonized roots and loose mycelium fragments. Inoculum was harvested just before the start of the pot experiment; the roots were cut to 1-cm fragments and mixed into the substrate of the pot culture, and the moist inoculum was used for establishment of the pot experiment.
The pot experiment was carried out in pots filled with 400 g substrate. There were two host plants (medic and leek), two harvest times (4 and 8 wk after sowing), and eight inoculation treatments. Four replicate pots were established for each treatment combination. The inoculation treatments were three single AMF species treatments, three double species mixtures, one triple species mixture, and one nonmycorrhizal control. The amount of inoculum for single species inoculation was 18 g per pot, the amounts of each AMF species for double species treatments were 9 g per pot, and the amounts for triple species treatment were 6 g per pot. Inoculum was mixed into the whole volume of the pot before sowing. Nonmycorrhizal pots were inoculated with 5 ml per pot of aqueous filtrate of the triple species inoculum mixture (20% suspension, w/v) filtered twice through Whatman no. 1 paper. Extra pots inoculated with single AMF species or with inoculum filtrate (already described) were established for destructive sampling at weekly intervals to determine the progress of AM colonization (one pot for each sampling time and each inoculation treatment). Pots were watered daily to weight with deionized water to maintain 75% water-holding capacity of the substrate. Each pot received 10 ml Long Ashton nutrient solution without P, weekly (Cavagnaro et al., 2001). Pots were completely randomized in the glasshouse and grown at 25 : 22°C (day : night) under mixed natural and supplemental fluorescent light (16 h photoperiod).
Harvest and measurements
Four replicate pots for each treatment were harvested 4 and 8 wk after sowing. The extra pots were harvested at weekly intervals. Plant shoots were dried at 105°C for 48 h and weighed. Roots were washed from the substrate with deionized water and split into three parts. One part (approx. 100 mg) was frozen in liquid N and stored at –20°C for DNA extraction; one part was dried at 105°C for 48 h and weighed; and one part was stained according to a protocol based on Phillips & Hayman (1970) and Brundrett et al. (1984). Briefly, roots were digested in 10% KOH at 80°C for 30 min, rinsed with water, incubated in 3% HCl at room temperature for 30 min, then transferred (with no further rinsing) to 0.05% Trypan blue in lactic acid : glycerol : water (1 : 1 : 1, v/v/v) and stained at 80°C for 4 h in a water bath. Finally, the roots were incubated overnight in water at room temperature. The extent of root length colonized by hyphae, arbuscules and vesicles was determined on stained root samples according to the method of McGonigle et al. (1990), recording 100 root intersects per sample. Subsamples of dry biomass (100–300 mg) of both shoots and roots were digested with 7 ml HNO3 : HClO4 (6 : 1, v/v) at 150°C for 10 h, evaporated to dryness, and made up to 20 ml with 1% HCl. Phosphorus concentration was measured in these extracts by the malachite green method (Ohno & Zibilske, 1991). Five grams of substrate were used for estimation of hyphal length density as described previously (Jansa et al., 2003a). The composition of AMF communities in the roots was estimated by real-time PCR as described above. DNA was extracted by DNeasy Plant Mini Kit (Qiagen) from the frozen root samples after homogenization in liquid N, following the manufacturer's recommendations. All samples were spiked with a known number (5 × 106) of gene copies of S. pellucida LSU clone (internal standard) before homogenization. Specific amplification of that LSU clone from the purified DNA extract allowed for correction for both (1) DNA lost during extraction and (2) variable PCR amplification efficiency across the samples (e.g. Weiss et al., 2004; Bustin et al., 2005). Real-time PCR was performed with specific primers using crude DNA extract as template.
Calculations and statistics
The percentage of root length colonized by AMF hyphae, arbuscules and vesicles is given as the ratio of intersects with these structures to all root intersects per sample × 100. Phosphorus concentrations in plant digests were used for calculation of plant P content. Plant P uptake from the substrate was determined by subtraction of P contained in the seed (11.0 and 19.5 µg P per seed for leek and medic, respectively) from total plant P content at harvest. The LSU copy numbers of each of the AMF species in each sample were calculated using the regressions given in Fig. S1. anova and correlation analyses were performed in statgraphics plus for windows ver. 3.1 (Manugistics, Inc., Rockville, MD, USA). Regression analyses (linear and sigmoid models) were performed in SigmaPlot 2002 ver. 8.01 (SPSS, Inc., Chicago, IL, USA). Data for percentage root length colonized by the AMF were arcsin- and square root-transformed (Linder & Berchtold, 1976) for statistical analysis so as not to violate anova assumptions.
Inocula of all three fungi were highly infective. Data from the weekly harvests of single pots showed that > 50% of root length of both medic and leek became colonized within the first 4 wk by the individual AMF species (Trypan blue staining, Fig. 1; Table 1). Glomus mosseae colonized roots of both plant species very rapidly, resulting in approx. 80% of root length colonized 2 wk after sowing (Fig. 1). Values remained high for the remainder of the experiment (8 wk). The development of G. intraradices was initially slower, but the values for percentage root length colonized were similar to or higher than for G. mosseae by 4 wk, and again remained high up to 8 wk (Fig. 1; Table 1). The development of G. claroideum was similar to that of G. intraradices, but the percentage of root length colonized saturated at a lower value than for the other two AMF (Fig. 1; Table 1). The percentage of root length containing arbuscules (Table 1) was high for all fungi at all harvests, and similar to the percentage of root length containing hyphae. Glomus mosseae did not produce any vesicles in the roots of either host plant, and G. intraradices produced substantially more vesicles than G. claroideum (Table 1). Hyphal length density (HLD) in the growth substrate was close to zero in noninoculated pots and ranged between 3.5 and 14 m g−1 substrate in the different inoculated treatments (Fig. 2). The HLDs were significantly higher for medic inoculated with G. mosseae than with other AMF species or their mixtures at 4 wk after sowing (Fig. 2). At 8 wk after sowing, HLDs were higher for medic or leek inoculated with G. mosseae or G. intraradices compared with G. claroideum. The HLDs in the mixed inoculation treatments at 8 wk after sowing were mostly intermediate between the highest and lowest values observed for the single AMF species (Fig. 2).
Table 1. Colonization of plant roots by arbuscular mycorrhizal fungi (AMF) assessed by the magnified intersection method following Trypan blue staining
Colonization parameter (%)
AMF treatment (M, G. mosseae; C, G. claroideum; I, G. intraradices)
Only the AMF species introduced to the pots by inoculation were detected by the real-time PCR assay in the pots after harvest, indicating that no cross-contamination occurred and no PCR cross-amplification was encountered (Table 2). The number of G. mosseae LSU copies per unit weight of roots in the single inoculation treatment was very high at 4 wk compared with the other fungi, and decreased substantially between 4 and 8 wk. By contrast, values were much lower for G. intraradices and G. claroideum, and were similar at both 4 and 8 wk (Table 2). When G. mosseae was present in the inoculum mixture, it was always a large component of the AMF community in the roots of both medic and leek (Fig. 3), but was relatively more abundant in the former (Table 3). The decrease in abundance of G. mosseae LSU copies between the two harvests was more marked in leek than in medic (see interaction between host plant and harvest time in Table 3). Inoculation with a mixture of G. intraradices and G. claroideum resulted in almost equal colonization of roots by the two AMF species 4 wk after sowing, but G. intraradices became dominant by 8 wk (Fig. 3; Table 3). When summing all the LSU copy numbers per unit of root biomass across all AMF species in the mixtures, no increased total colonization by AMF mixtures could be substantiated as compared with the single species treatments (analysis not shown). Further, no significant correlations were found between traditional root colonization estimates and the real-time PCR assays (checked for each AMF species at each harvest).
Table 2. Copy numbers of large ribosomal subunit (LSU, thousands mg−1 FW roots) of three arbuscular mycorrhizal fungal (AMF) species in roots of medic and leek
LSU of AMF species
AMF treatment (M, G. mosseae; C, G. claroideum; I, G. intraradices)
, 0.001 ≤ P < 0.01;
, P < 0.001; different letters following treatment means in each row indicate significant differences (P < 0.05).
Table 3. Three-way anova for relative proportions of large ribosomal subunit copy numbers of the three arbuscular mycorrhizal fungal (AMF) species in roots of medic and leek inoculated with AMF mixtures 4 and 8 wk after sowing
ns, Not significant; (*), 0.05 ≤ P < 0.1; *, 0.01 ≤ P < 0.05; **, 0.001 ≤ P < 0.01; ***, P < 0.001.
Mycorrhizal plants were consistently larger than nonmycorrhizal plants, with the exception of leek inoculated only with G. claroideum 4 wk after sowing (Fig. 4). The differences among inoculation treatments for both host plants at both harvest times were highly significant (P < 0.001). Glomus mosseae inoculated singly promoted greatest biomass production in both medic and leek at both harvests. When inoculated singly, G. intraradices promoted greater biomass of both host plants than G. claroideum, at 8 wk after sowing only. The biomass of plants in mixed inoculation treatments never exceeded the range of growth promotion by the respective single-species inoculations (Fig. 4). The biomass of medic plants 8 wk after sowing was similar, and high in all treatments that included inoculation with G. mosseae, either alone or in a mixture, whereas there was significant variation among the same treatments in the case of leek (Fig. 4).
Uptake of P showed a similar pattern to growth. Mycorrhizal plants always took up more P from the substrate than nonmycorrhizal plants, with the single exception of leek inoculated with G. claroideum 4 wk after sowing (Fig. 5). Glomus mosseae inoculated singly promoted greater total P uptake by medic than the other two AMF species 4 wk after sowing. However, by 8 wk values for plants inoculated with G. intraradices had reached the same total P content as those inoculated with G. mosseae. Medic inoculated with G. claroideum took up less P during 8 wk of growth than those inoculated with either G. intraradices or G. mosseae. As with biomass production, total P uptake by medic inoculated with AMF species mixtures never exceeded the range of P uptake by plants inoculated with the respective single AMF species (Fig. 5). Colonization by G. claroideum was associated with lower P uptake by leek from the substrate compared with G. mosseae, but not G. intraradices at both harvests (Fig. 5). Phosphorus uptake by leek inoculated with AMF species mixtures did not exceed the range of P uptake by plants inoculated with the respective single AMF species, with the notable exception of a mixed inoculation with G. intraradices and G. claroideum. This combination supported greater P uptake by leek than either of the two AMF species separately 8 wk after sowing (Fig. 5).
Root colonization assessed by traditional (non-vital) Trypan blue staining revealed typical sigmoid colonization patterns by all three AMF species (Fig. 1). When inoculated singly, the three AMF varied in development, including length of lag, slope of the rapid growth phase, and plateau values of percentage colonization (Fig. 1; Table 1). Glomus mosseae was the fastest colonizer and G. claroideum the slowest, with G. intraradices intermediate, but eventually reaching the same plateau value as G. mosseae. Importantly, all three AMF had essentially reached plateau values of colonization by 4 wk when the first destructive harvest took place. Real-time PCR assay at 4 and 8 wk, however, gave a different picture. Glomus mosseae showed values an order of magnitude higher for LSU copy numbers in the roots at 4 wk, compared with the other fungi. This declined dramatically (to about 1/4) between 4 and 8 wk, so that at 8 wk all the AMF inoculated singly showed similar values in roots of both plant species (Table 2). The other fungi showed LSU copy numbers at both harvests that more closely mirrored the percentage colonization. If LSU copy number is to be regarded as a proxy for living fungal tissue, then these results require careful consideration. High initial values and subsequent decline in LSU copy number for G. mosseae could be explained by a rapid spread of active mycelium, followed by a decrease in vitality of the colonization. Similar decrease in AMF activity with time has been documented previously using vital stains and measurement of arbuscular development (Smith & Dickson, 1991). What is not clear is why G. mosseae behaved differently from the other two AMF. It could be that this fungus both developed very rapidly in roots and then declined in activity very rapidly. We can speculate that multiplication of nuclei within the hyphae of (some) AMF may occur at times of peak metabolic activities, leading to a discrepancy between DNA quantification and staining as measures of fungal biomass. Formation of syncytial (multinucleated) cells, which would lead to an increase in LSU copy number, has been described in many different organisms (Beer & Arber, 1919; Comoglio et al., 1969; Lamb & Laird, 1976; Cantalejo et al., 2004) and is usually accompanied by intense metabolic activity. However, it is not known whether similar processes occur in the AMF. Furthermore, traditional estimates of the extent of root colonization include both living and dead fungus, and are related to root length rather than root weight. They also provide only semiquantitative measures of fungal development, because colonization intensities (numbers of hyphae, arbuscules and vesicles per colonized intersection) are normally not recorded. These factors, together with the fact that only a narrow range of colonization percentage for each AMF species at each harvest was obtained, resulted in nonsignificant correlation between root colonization estimates by staining and microscopy and real-time PCR.
Consistent with the fast rate of root colonization and early peak in colonization vitality, G. mosseae was the most successful competitor, as shown by the relatively high LSU copy numbers of this species in roots containing mixtures of fungi. Although conclusions at 4 wk might have been biased by the very high LSU values for G. mosseae, this fungus also dominated at 8 wk, providing some confidence in the finding. Inoculation with a mixture of G. intraradices and G. claroideum resulted in almost equal colonization of roots by the two AMF species at 4 wk, but G. intraradices became dominant by 8 wk (Fig. 3; Table 3). When the three fungi were inoculated together, G. mosseae again predominated. Glomus mosseae was also faster in colonizing the substrate than the other AMF species, but by 8 wk G. intraradices had similar HLDs. Glomus claroideum mycelium developed poorly by comparison (Fig. 2). All the data are consistent with G. mosseae being the most effective competitor and G. claroideum the weakest, with G. intraradices intermediate. This observation is in accordance with previous results showing that other isolates of G. mosseae and G. caledonium were faster colonizers of leek or clover roots compared with other Glomus species (Wilson & Trinick, 1983; Wilson, 1984; Hepper et al., 1988). However, Alkan et al. (2006) showed that G. intraradices BEG141 occupied a higher proportion of the roots of medic at 4 wk than G. mosseae BEG 12, despite the fact that the infectivity of G. mosseae inoculum was higher. These inconsistencies are probably caused by differences between fungal isolates, as well as in growth conditions, P availability, and many other factors shown to influence colonization (Daft, 1983; Boddington & Dodd, 2000; Alkan et al., 2006).
The mechanisms of competition among AMF within a community are not clear and need further elucidation with respect to competition for carbon supply, exchange of signals and other factors. Nevertheless, a competitive advantage of early root colonization has been proposed (Hepper et al., 1988), and this is also suggested by some of our results. However, it is also clear that the composition of (active) AMF communities can change with time because of a drop in activity of the very fast colonizers such as G. mosseae and steady increase in colonization and activity of G. intraradices over a longer time period (Table 2). This is in agreement with suggestions of some previous studies (e.g. Wilson & Trinick, 1983).
The identity of the host plant also appeared to be an important determinant of AMF community composition in our experiment (Table 3). This is consistent with accumulating evidence for selectivity of AMF symbionts by the host plants (Bever et al., 1996; Helgason et al., 2002) and, in some cases, complete failure of particular AMF to colonize one host species, although others did become colonized (Alkan et al., 2006). However, the most common situation appears to be the one we describe here, which is that colonization by the individual AMF in a community varies quantitatively (not as presence or absence), and it is crucial to develop methods that will allow the variations to be measured.
All in all, real-time PCR proved a very useful tool for dissecting AMF community composition in our study and extends the research of Alkan et al. (2006), who employed a similar approach. Use of ribosomal RNA genes is the only currently viable molecular approach to targeting different AMF species, as too little is known about other parts of AMF genomes. Single-copy genes should be targeted in future, overcoming problems likely to be associated with the substantial sequence polymorphism within ribosomal RNA genes sharing the same cytoplasm (Jansa et al., 2002b; Sanders, 2004). Preliminary work also indicates that real-time PCR could be used to analyse AMF communities in soil (J.J., unpublished data), possibly overcoming both the PCR and cloning biases associated with the current methods (Hempel et al., 2007). Our approach may also be useful in assessing vitality of the AMF colonization, although assessment of fungal RNA would be preferable when this becomes realistic.
We observed strong positive effects of all inoculation treatments on biomass production of, and P uptake by, both leek and medic. When inoculated singly, G. claroideum was consistently least effective in increasing P uptake and growth compared with G. mosseae and G. intraradices. These differences were related to differences in the rate and extent of colonization, and particularly to development of extraradical mycelium. The HLDs of the fungi were significantly positively correlated with plant P uptake (F1,126 = 175.1, P < 0.001, R2 = 0.58). This finding extends previous results showing that the same isolates of G. mosseae and G. intraradices were able to take up P more quickly and from greater distances from the roots than G. claroideum (Jansa et al., 2005). Interestingly, plant P uptake was not always mirrored in plant biomass. For example, medic biomass was greater when inoculated with G. mosseae than with G. intraradices (Fig. 4), but total P uptake was greater with G. intraradices (Fig. 5). It is possible that G. intraradices utilized more carbon from the medic plants, resulting in lower growth but higher tissue P concentrations (results not shown) in the short term. This variation is in accord with many previous studies showing that mycorrhizal benefits are strongly dependent on the plant–fungus combination (Smith et al., 2004). Interestingly, the mycorrhizal plants took several times more P from the soil than predicted by Colwell (1963) and isotopic exchange kinetic approaches (see above), meaning that these P-availability indicators worked particularly poorly for our substrate, where possibly organic P might present a significant reserve of slowly plant-available P (Oberson et al., 2001).
The effects of AMF mixtures on plant growth and P uptake were mostly within the range of the effects exerted by the respective single AMF species. In the main, there was little evidence for increased P uptake and/or growth of plants colonized by several AMF species compared with a single species. When G. mosseae was included, the community in the roots was dominated by this species (Fig. 3), which was highly effective when inoculated singly, and clearly had a major influence on symbiotic performance of plants colonized by mixtures.
In one case we did observe a synergistic effect of dual inoculation. Phosphorus uptake by leek inoculated with a mixture of G. intraradices and G. claroideum was greater at 8 wk than when inoculated with either of the two AMF species separately. This is the first direct evidence for functional complementarity with respect to P acquisition between AMF species colonizing roots, as suggested by Koide (2000). These two fungi showed similar progress of colonization when inoculated singly; when inoculated together, this was one of the few AMF communities that was not dominated by a single fungal species at some time point during development (Fig. 3). Synergy between the fungi can therefore probably be explained on the basis of differences in P uptake by the fungi, which in this experiment had different HLDs in the soil. Previous work has also shown that G. intraradices could bridge P-poor soil volumes close to the roots without extensive branching in them (Jansa et al., 2003a). These features would predispose G. intraradices for successful cooperation with G. claroideum over the same period of development, as we have shown. Previously, the benefits of colonization of plants by several AMF species were suggested to be in temporal partitioning of activities (different AMF active at different periods of time); as a buffer against change (different AMF adapted to different environmental conditions); and as a means of minimizing growth depressions that can arise if single, inefficient fungi are inoculated singly (Daft & Hogarth, 1983; Abbott & Gazey, 1994; Pringle & Bever, 2002; Sanders, 2003). The question remains why such effects of multiple inoculations are not observed more often. Early experiments generally lacked the technology to estimate precisely the composition of AMF communities developing in roots. In consequence, lack of synergistic effects among AMF species with respect to plant growth (Daft, 1983; Daft & Hogarth, 1983; Pearson et al., 1994) and/or plant P uptake may have been caused by the fact that one AMF species became dominant, as shown here with G. mosseae. The quantitative molecular tools now available will enable links between the composition and function of AMF communities to be unravelled. It will be important to link the community studies to direct assessment of nutrient fluxes via hyphae and roots by using radio- and stable isotopes and compartmented cultivation systems.
We are grateful to Stephen Rogers (then at CSIRO Land and Water, Adelaide, Australia) for sharing real-time PCR equipment and for valuable advice and discussions. We would like to express our gratitude to Colin Rivers and Debbie Miller for their excellent technical assistance, and Sandy Dickson for her help with maintaining the pot experiments. Financial support of the Swiss National Science Foundation for a postdoctoral fellowship to J.J. and the Australian Research Council for project support is gratefully acknowledged.