Underground resource allocation between individual networks of mycorrhizal fungi

Authors

  • Bolette L. Mikkelsen,

    1. Biosystems Department, Risø National Laboratory for Sustainable Energy, Technical University of Denmark, P.O. Box 49, DK-4000 Roskilde, Denmark;
    2. Department of Biology, Terrestrial Ecology, University of Copenhagen, Øster Farimagsgade 2D, DK-1353 Copenhagen K, Denmark
    Search for more papers by this author
  • Søren Rosendahl,

    1. Department of Biology, Terrestrial Ecology, University of Copenhagen, Øster Farimagsgade 2D, DK-1353 Copenhagen K, Denmark
    Search for more papers by this author
  • Iver Jakobsen

    1. Biosystems Department, Risø National Laboratory for Sustainable Energy, Technical University of Denmark, P.O. Box 49, DK-4000 Roskilde, Denmark;
    Search for more papers by this author

Author for correspondence:
Iver Jakobsen
Tel:+45 46774154 Fax:+45 46774109
Email: iver.jakobsen@risoe.dk

Summary

  • • Fusions between individual mycelia of arbuscular mycorrhizal (AM) fungi have been observed in two-dimensional systems but never in soil systems. Here, phosphorus (32P) labelling was used to demonstrate nutrient transfer between individual mycelia and to investigate the possible role of anastomosis.
  • • Trifolium subterraneum colonized by Glomus mosseae were grown in root-retaining mesh bags, which were placed 20 cm apart. The mycelium of one plant, the donor, had access to 32P-labelled soil placed adjacent to the mesh bag. Transfer of 32P from the donor mycelium to the receiver plant was measured at three harvests. In a second-harvest control treatment the receiver was colonized by Glomus caledonium in order to determine whether transfer occurred by other means than hyphal fusions.
  • • Significant amounts of P were transferred to the receiver plant at the last harvests when the two mycelia of G. mosseae overlapped. The transfer probably occurred via anastomoses between the mycelia as no transfer of 32P was detected between the mycelia of different fungi at the second harvest.
  • • The indicated ability of AM fungal mycelia to anastomose in soil has implications for the formation of large plant-interlinking functional networks, long-distance nutrient transport and retention of nutrients in readily plant-available pools.

Introduction

Arbuscular mycorrhizal (AM) plant–fungus symbioses are found in most ecosystems. Approximately two-thirds of all plant species can form this mutualistic association with the obligate biotrophic AM fungi, all of which belong to the Glomeromycota (Smith & Read, 1997; Schüßler et al., 2001). The plant provides photosynthetically fixed carbon to the fungus, which in return supplies mineral nutrients taken up and translocated by the extraradical mycelium. The mycorrhiza phosphate uptake pathway can in some cases dominate the total P uptake of the plant even in the absence of mycorrhiza-induced plant growth responses (Smith et al., 2003). The symbiosis is furthermore important for plant productivity and biodiversity as well as ecosystem functioning (van der Heijden et al., 1998).

Although there is increasing evidence for host preference of AM fungi (Vandenkoornhuyse et al., 2002), it is generally assumed that they can connect several different plants of the same or different species in a common mycorrhizal network (Newman, 1988). These underground networks are considered to be a key element in terrestrial ecosystems, but their function and effect on plant communities are not completely understood (Selosse et al., 2006). More effective nutrient cycling is anticipated when plants are linked into a shared mycorrhizal network because nutrients from dying roots would rapidly be taken up by the common mycelium. Accordingly, significant transfers of phosphorus (P) from dying donor roots to receiver plants via AM mycelia have been observed (Newman, 1988; Johansen & Jensen, 1996). Direct AM fungus-mediated nutrient transfer between two living plants has also been reported, but is probably not quantitatively important (Simard et al., 2002; Wilson et al., 2006).

The regulatory role of nutrient distribution in a shared mycelial network and its impact on plant competition and seedling establishment remains to be resolved. A recent study showed that a common mycorrhizal network increased competition by adult plants relative to seedlings but the competition was reduced when the adult plant was defoliated (Pietikäinen & Kytöviita, 2007). It is hence possible that the plants constituting the greatest sink will benefit most from the shared network (Jakobsen, 2004). However, the presence of mycorrhiza promoted seedling establishment in a 1-yr-old grassland microcosm (van der Heijden, 2004) and clear conclusions cannot be drawn.

Extraradical hyphae of AM fungi can spread over rather large distances from an individual plant (Jakobsen et al., 1992; Jansa et al., 2003) and in this way establish connections to another plant. It has also been demonstrated that individual mycelia can fuse (anastomose) within two-dimensional membrane sandwiches and in this way form connective networks between different plant species (Giovannetti et al., 2004). Successful anastomosis occurred between mycelia of the same isolate of Glomus mosseae, but not between different isolates or species (Giovannetti et al., 1999, 2003). Anastomosis also occurred between different mycelia of Glomus intraradices in monoxenic root cultures (Voets et al., 2006). Protoplasmic flow between the anastomosed hyphae was observed in both studies, but the functionality of the fused mycelia in transfer and cycling of nutrients has not been tested to date.

The fusion of individual mycelia in soil could result in the formation of large common mycorrhizal networks. It was hypothesized that a widely distributed unidentified Glomus species in undisturbed coastal grassland could be one large mycelium (Rosendahl & Stukenbrock, 2004), but in disturbed agricultural systems such networks may also form by hyphal anastomosis (Rosendahl, 2008). The occurrence of hyphal anastomosis in soil has not yet been investigated because of the delicate structure of the hyphae which is easily destroyed when examining the fusions. Furthermore it would be almost impossible to distinguish intra- and inter-network anastomoses. Testing the ability of hyphae to anastomose in soil will therefore require an approach other than direct observations. An indirect way to study anastomoses between mycelia in soil could be to examine the transfer of tracer isotope-labelled nutrients between individual, but overlapping, mycelia. This would also answer the question addressed above: is the network functional in nutrient cycling?

The objective here was to investigate whether functional fusions would form between two individual and overlapping mycelia in soil using 32P as a tracer. The two mycelia originated from plants inoculated with the same isolate of G. mosseae; control treatments with a nonmycorrhizal plant or a plant inoculated with Glomus caledonium were included. It was hypothesized that: anastomoses would only occur between mycelia of the same AM fungal isolate; 32P would be transferred in the network formed between the two mycelia towards the plant constituting the greatest sink; and transfer of 32P would only occur through the anastomosed network and not via soil pathways, for example leakage of 32P from one mycelium and subsequent uptake by another.

Materials and Methods

Plant, fungi and growth medium

The plant material used was subterranean clover (Trifolium subterraneum L.) cv. Woogenellup. The AM fungus Glomus mosseae BEG 84 was used in the main treatment and a mixture of three Glomus caledonium isolates (BEG 20, BEG 86 and H07-1) was used in a control treatment. All fungi were originally isolated from the same agricultural field in Tåstrup, Denmark apart from BEG 20, which originated from the UK. Inoculum was produced with T. subterraneum and consisted of soil containing fungal spores and colonized root pieces.

The growth medium was a 1 : 1 (w/w) mixture of a moderately P-deficient sandy moraine loam and quartz sand that was irradiated (2 × 10 kGy; 10-MeV electron beam) to eliminate native AM fungi. The growth medium (hereafter denoted ‘soil’) had a pH of 6.5 and a bicarbonate-extractable P content of 9 µg P g−1 (Olsen et al., 1954). Basal nutrients (Pearson & Jakobsen, 1993) minus P were mixed into the soil together with 30 mg of NH4NO3-N kg−1 dry soil.

Experimental set-up

The experiment included 21 rectangular pots measuring 30 × 11 × 10 cm (length × height × width). Double-layer bags of 25-µm nylon mesh were filled with 500 g of soil or 425 g of soil and 75 g of inoculum and placed at opposite ends of each pot. This resulted in 20-cm distance between the bags. The rest of the pot, the hyphal compartment (HC), was filled with 2650 g of soil, and two empty test tubes (diameter 2.3 cm) were inserted either 0.5 cm from one donor mesh bags (treatments 1, 3 and 4) or in the middle of the pot (treatment 2) (Fig. 1a). The pots were watered to weight, initially to 60% of the soil water-holding capacity (WHC), and incubated for 1 wk. Then the two test tubes in each pot were removed and replaced by two 36-µm mesh bags, each containing 35 g of 32P-labelled soil. The labelled soil was covered with 4 g of nonlabelled soil. The soil had been uniformly labelled with a carrier-free 32P-orthophosphate solution to obtain 5.5 kBq g−1 soil. A total of 385 kBq was supplied to each pot at initiation of the experiment. Five pre-germinated T.  subterraneum seeds were sown in each mesh bag which retained the roots, but allowed hyphae to pass through. The seedlings were thinned to two plants per bag after emergence. Plants near the labelled soil were denoted ‘donor’ and plants at the other end of the pot were denoted ‘receiver’ (Fig. 1). The pots were later in the experiment watered to 65% of the WHC. The HC was covered with alufoil to reduce evaporation and to avoid growth of algae on the soil surface.

Figure 1.

Experimental design. (a) Mesh bags with Trifolium subterraneum donor and receiver plants were placed at opposite ends in each pot. Mesh bags with phosphorus (32P)-labelled soil were placed either in the middle of the pot (treatment 2) or adjacent to mesh bags with donor plants (treatments 1, 3 and 4). (b) Diagram showing how far the mycelia of Glomus mosseae (solid lines) had grown at the time of defoliation of donor plants (t0) and at the three subsequent harvest times. The t0 time-point is defined as the time when 32P could be detected in the G. mosseae-colonized plants of treatment 2, that is, when the mycelium had grown half-way through the pot. The dotted line indicates the extension of G. caledonium at 7 d when the treatment 3 control was harvested. The hyphal compartment is drawn to scale.

Plants were maintained in a growth room with a 16 : 8 h light:dark cycle and a temperature of 22 : 15°C, respectively. Osram daylight lamps (Osram GmbH, Munich, Germany) provided 500 µmol m−2 s−1 photosynthetically active radiation (PAR; 400–700 nm). Plants were watered once or twice a day to maintain 65% of the WHC and their position was changed each time. A solution of NH4NO3-N was supplied as required such that plants in each mesh bag had additionally received 135 mg of nitrogen (N) at the first harvest and 165 mg of N at the two last harvests. Supplementary basal nutrients minus P were also added towards the end of the experiment because the plants developed necrotic spots on the older leaflets, which is an indication of potassium deficiency.

Treatments

The experiment included four treatments: the main treatment (1) and three control treatments (2–4). Treatment 1 had both donor and receiver inoculated with G. mosseae and four replicate pots were harvested at each of three time-points. Harvest times were determined on the basis of treatment 2 (three replicates) where the 32P source was located in the middle of the pot and used to define the time when the two mycelia had met. This was determined by daily monitoring of radioactivity in the plants with a hand held G-M tube counter. One plant in treatment 2 was inoculated with G. mosseae, the other with G. caledonium. Donor shoots of treatments 1 and 3 were defoliated 48 d after sowing (referred to as t0) when radioactivity had been detected in G. mosseae-inoculated plants in all three pots of treatment 2. Defoliation was carried out to reduce the sink strength of the donor plants. The three harvests were planned such that the receiver mycelium would not have grown all the way through the pot to the 32P source at the last harvest (see section ‘Analysis of data’).

In treatment 3 with three replicate pots the receiver was colonized by G. caledonium, but otherwise the pots were identical to those of treatment 1. They were harvested at the second harvest, 7 d after defoliation of the donor, because of the higher growth rate of G. caledonium (see section ‘Analysis of data’). This control treatment tested whether transfer of 32P could occur by leakage from one mycelium to the other. Treatment 4, also with three replicate pots, had nonmycorrhizal donor plants, but was otherwise also identical to treatment 1. These pots were harvested at the final harvest, 14 d after defoliation of the donor. This control served to confirm that the receiver mycelium had not grown throughout the pot and absorbed 32P directly from the labelled soil.

Harvest and analyses

Donor plants were defoliated at t0 by cutting shoots, except for the lowest young stems, such that plants were still alive, but much reduced in size. Defoliation material was dried at 70°C for 48 h and weighed. The pots were harvested 3, 7 and 14 d after t0 (Fig. 1b). Shoots were cut from the receiver plants, and treated as above. The dry weight (DW) was also determined for the donor shoot fraction remaining after defoliation. Roots of donor and receiver plants were removed from mesh bags, thoroughly washed, blotted and weighed. A weighed subsample was cleared in 10% KOH and stained in trypan blue (Phillips & Hayman, 1970), omitting phenol from the reagents and HCl from the rinse. The rest of the root sample was dried at 70°C for 48 h for DW determination.

Dried plant samples were ground up and digested in nitric and perchloric acid (4 : 1 v/v). The activity of 32P in all plant extracts was measured in a Packard TR 1900 Liquid Scintillation counter (Packard Instrument Co., Meriden, CT, USA) using the scintillation cocktail Ultima GoldTM (Perkin Elmer, Boston, MA, USA). The activity was corrected for isotopic decay and background. The P concentration in plant samples was determined by the molybdate blue method (Murphy & Riley, 1962) using AutoAnalyzer 3 (Bran+Luebbe, Norderstedt, Germany).

The soil in the receiver half of the pot was divided into three equally sized vertical slices using a broad knife. Slices were numbered 1–3 starting from the receiver plant end, and the soil of each slice was thoroughly mixed. A small subsample was frozen for later determination of hyphal length (Jakobsen et al., 1992) and a sample was also taken for determination of the dry matter. Hyphae were washed from the rest of each soil slice by repeated aqueous suspension and decanting onto first a 212-µm sieve and then a 63-µm sieve using a modification of the method used by Jakobsen & Rosendahl (1990). The hyphae were dried at 70°C and acid-digested and radioactivity was measured as for the plant samples.

The day before the final harvest, soil cores (diameter 12 mm) were taken at 1, 5, 10, 15 and 19 cm from the receiver mesh bag in treatment 1 pots with G. mosseae at each end and in treatment 4 pots with nonmycorrhizal donor plants. Two soil cores were taken at each distance and mixed well. Hyphal length was measured as above.

Bicarbonate-extractable P (Olsen et al., 1954) in the labelled soil of treatment 4 was measured using the molybdate blue method (Murphy & Riley, 1962) and radioactivity in the extract was determined by scintillation counting. The specific activity in the extract was 133 ± 3 kBq mg−1 P.

Analysis of data

The extension rate for mycelium of G. mosseae was estimated as 3.1 mm d−1. This estimate was obtained using the distance between the 32P soil cores in treatments 1 and 2 in combination with the time lapse between the first detections of 32P in the donor plants of treatment 1 and in the G. mosseae plant of treatment 2. The experimental treatments did not allow a similar estimation of the extension rate of the G. caledonium mycelium. However, this fungus reached the 32P cores more quickly than G. mosseae in treatment 2 and this time difference was used to estimate the mycelial extension rate of G. caledonium as 3.8 mm d−1.

The amount of 32P transferred to receiver plants via the established hyphal network was converted to total P using the specific activity of 32P in bicarbonate extracts of labelled soil: total P transferred via hyphal network =32P in receiver/specific activity in soil extracts. We assumed that added 32P had rapidly equilibrated with the bicarbonate-extractable 31P. The labelled soil volume corresponded to only 3% of the total soil volume in the HC and total transfer is therefore greatly underestimated. A more realistic estimate of the contribution of inter-mycelial transfer to P uptake by the receiver was obtained by multiplying with the weight ratio between 32P-labelled soil and total HC soil. The network P transfer was compared with the total amount of P uptake by the receiver plant during the 14-d harvest period. We assume this uptake to be derived mainly from HC soil because the soil in mesh bags would probably have been depleted for available P towards the end of the experiment as a result of high root densities.

Statistics

Statistical analyses were performed using sas 9.1 enterprise guide 4 (SAS Institute, Cary, NC, USA). Hyphal 32P content, plant P content, root colonization, plant DW, root:shoot ratio and donor plant 32P content in treatment 1 were analysed by a one-way ANOVA with time being the factor. ANOVA was followed by a Tukey's Studentized test to determine which means were significantly different from one another (P < 0.05). Data were tested with Levene's test for homogeneity of variance and transformed when necessary to obtain equal variances. Percentage colonization and root:shoot ratio data were arcsin-transformed before analysis. Two-sample t-tests were used to compare means between treatments. Means are presented with standard errors.

Results

32P transfer to receiver plants

The most important finding in this study was that activity of 32P increased markedly during the 14-d harvest period in receiver plants of treatment 1 with two overlapping G. mosseae mycelia (Fig. 2, grey bars). At the 3-d harvest, receiver plants contained negligible amounts of 32P (0.16 kBq), which had increased to 0.54 kBq at 7 d. However, this increase was not significant because of the large variation among the four replicates (1.43, 0.44, 0.21 and 0.09 kBq). At the 14-d harvest, significant amounts of 32P were transported to the receiver plants (P < 0.04).

Figure 2.

Activity of phosphorus (32P) in Trifolium subterraneum receiver plants at different times after defoliation of donor shoots. Grey bars, treatment 1 (both donor and receiver colonized by Glomus mosseae); white bar, treatment 3 (donor and receiver colonized by G. mosseae and G. caledonium, respectively); black bar, treatment 4 (donor nonmycorrhizal; receiver colonized by G. mosseae). Treatment 1 bars with the same small letter are not significantly different (P < 0.05). Different capital letters indicate that bars of control treatments 3 and 4 differ from treatment 1 bars at the relevant time-point. Controls at 7 and 14 d did not differ from each other or from treatment 1 at 3 d (P < 0.05). Bars are means of four (treatment 1) or three (treatments 3 and 4) replicates ± standard error of the mean.

The rate of extension of the mycelia of the two AM fungi measured in treatment 2 allowed the harvest times to be planned such that the mycelia of receiver plants would not reach the 32P soil patches. This goal was fulfilled at the 14-d harvest, as shown by the low background levels of 32P in receiver plants of treatment 4 where donor plants were nonmycorrhizal (Fig. 2, black bar). This means that 32P measured in receivers of treatment 1 could not have originated from direct access of receiver plant mycelium to the 32P soil patches or from soil-mediated transport of 32P away from the patches. The transfer observed in treatment 1 must therefore have occurred from one mycelium to the other.

The faster extension rate of the mycelium of G. caledonium than that of G. mosseae forced us to harvest this important control (treatment 3) at 7 d instead of 14 d as originally planned. In consequence, the mechanism of inter-mycelium 32P transfer was investigated at the 7-d harvest only. The G. caledonium receiver plants contained low levels of 32P (0.115, 0.145 and 0.126 kBq for the three replicate pots), but the mean value (Fig. 2, white bar) did not differ significantly from the corresponding treatment 1 value with the large error bar (Fig. 2). Therefore, the 7-d data set for 32P in receiver plants could not be used to draw final conclusions on the mechanism of transfer between mycelia, but contents of 32P in mycelium extracted from the soil provided more information.

Mycelium washed from adjacent soil layers in the receiver half of treatment 1 pots contained 32P at all harvest times (Fig. 3a–c). Activities of the mycelium did not differ between soil layers at 3 d, but levels increased and differences appeared over time, indicating that 32P was transferred towards the receiver plant during the 14-d window of measurement. Activity at 7 d was thus higher in the layer at the pot centre than in the layer near the receiver plant (Fig. 3b) and at 14 d also mycelium from the middle layer had higher activity than that from the receiver plant layer (Fig. 3c). The presence of most activity in mycelium from the soil layer at the pot centre reflects the fact that this layer was closest to the 32P source and contained donor and receiver mycelium as well as the highest hyphal length densities (Fig. 4). The content of 32P in mycelium extracted from treatment 3 with G. caledonium exceeded background levels only in the layer closest to the pot centre where G. mosseae mycelium would also have been present. Importantly, the 32P content in mycelium from the middle soil layer of treatment 1 was significantly higher than the corresponding background level in treatment 3 (P < 0.05). This means that 32P was not transferred from G. mosseae to G. caledonium to any significant extent in treatment 3, although the period of overlap between mycelia would have been longer because of the faster extension rate of G. caledonium. We conclude that 32P was not transferred between mycelia via indirect release-uptake pathways. In consequence, the transfer of 32P to receiver plants in treatment 1 must have occurred via anastomoses.

Figure 3.

The activity of phosphorus (32P) in mycelium washed from adjacent soil layers and expressed on a soil dry weight basis. The soil layers covered the distance from the receiver plants (0 cm) to half the distance between receiver and donor plants (10 cm). (a–c) Treatment 1 at 3 d (a), 7 d (b) and 14 d (c) after defoliation of the donor. (d) Treatment 3 at 7 d after defoliation of the donor (Glomus caledonium control). Bars are means of four (treatment 1) or three (treatments 3) replicates ± standard error of the mean. Columns for each harvest time with the same letter are not significantly different (P < 0.05).

Figure 4.

Hyphal length density at different distances from Trifolium subterraneum receiver plants. Soil cores were taken 13 d after defoliation of donor shoots. White bars, treatment 1 (both donor and receiver colonized by Glomus mosseae); black bars, treatment 4 (donor nonmycorrhizal; receiver colonized by G. mosseae). Bars are means of four (treatment 1) or three (treatment 4) replicates ± standard error of the mean.

The total amount of P in the receiver plants increased significantly during the 2-wk harvesting period (Table 1). The amount of P transfer to the receiver from the 32P-labelled soil was 0.028 mg P as estimated from the specific activity in bicarbonate soil extracts (see Materials and Methods). This corresponds to 0.58 ± 0.16% of the total P uptake by the receiver plant during the time when fusion between mycelia could have occurred (4.85 mg P in 14 d). Extrapolating this estimated transfer to transfer occurring throughout the HC, as much as 20.2 ± 5.7% of P uptake by the receiver plant over the 14-d harvest period could have originated from mycelia of donor plants.

Table 1.  Total dry weight (DW), root:shoot ratio, root colonization and total phosphorus (P) content of Trifolium subterraneum receiver plants and nonmycorrhizal ‘donor’ plants
Harvest (d)Total plant DW (g )Root:shoot ratioColonization (%)Total P (mg)
  1. Values are means of four replicates (treatment 1) or three replicates (treatments 3 and 4). Values with the same letters in columns within treatments are not significantly different (P < 0.05).

Treatment 1, Glomus mosseae receiver
36.89 a0.17 a69 a10.94 a
77.39 a0.15 ab70 a12.37 a
1410.08 b0.13 b63 a14.75 b
Treatment 3, G. caledonium receiver
79.390.166313.14
Treatment 4, G. mosseae receiver
149.62 a0.12 a75 a15.22 a
Treatment 4, nonmycorrhizal ‘donor’
148.43 a0.19 b0 b6.45 b

32P uptake by donor plants

At the time when donor plants of treatments 1 and 3 were defoliated, the shoot material removed contained as much as 78.9 ± 3.2 kBq or 20.5% of the amount of 32P initially added to the pot. This 32P had been delivered via the AM fungal uptake pathway and not via diffusion and mass flow because nonmycorrhizal plants of treatment 4 contained only negligible amounts of 32P (data not shown). Although defoliated donor plants had no net uptake of P or 32P over the 14-d harvest period, contents of P and 32P increased in shoots but decreased in roots during that period (Table 2). This indicates that relocation of resources took place and that donor plants remained viable.

Table 2.  Root colonization, dry weight (DW) and phosphorus (P) content of defoliated Trifolium subterraneum donor plants at three harvest times
Harvest (d)Colonization (%)DW after defoliation (g)P in shoot partsP in roots
Shoot partsRootsTotal (mg)32P (kBq)Total (mg)32P (kBq)
  1. Values are means of four replicates. Values with the same letters in each column are not significantly different (P < 0.05).

374 a0.43 a0.98 a0.97 a8.1 a1.83 a14.8 a
767 a0.41 a1.05 a1.33 b9.5 a1.91 a13.5 a
1475 a0.37 a0.92 a1.88 c10.3 a1.33 b8.6 b

Root colonization, plant growth and external hyphal length

All donor plants inoculated with G. mosseae were well colonized, and the colonization of the donor roots did not decrease after defoliation of the donor shoots (Table 2). Similarly, colonization of receiver plants by G. mosseae did not change during the 14-d harvest period (Table 1). Root colonization of receiver plants colonized by G. caledonium (treatment 3) did not differ significantly from that of the corresponding G. mosseae-colonized plants, whereas plant DW was higher with G. caledonium than with G. mosseae (P < 0.01).

Plant DW of receiver plants in treatment 1 increased during the 14-d harvest period and was enhanced by mycorrhiza (see Fig. 1a, treatment 4). At the final harvest, however, total DW did not differ significantly between AM and NM plants in spite of a marked mycorrhiza enhancement of plant P content (Table 1). The influence of mycorrhiza was also reflected in a lower root:shoot ratio in G. mosseae-colonized than in nonmycorrhizal plants (Table 1).

The hyphae length density measured in soil cores taken at different distances from the receiver plants at 13 d after defoliation was highest and symmetric around the middle of the pots in treatment 1 (Fig. 4). In the control pots where only the receiver was colonized (treatment 4), the distribution was skewed such that hyphal length densities were lower at both 15 and 19 cm, the sampling points closest to the nonmycorrhizal plant. The hyphal length of 2.3 m g−1 soil measured at 19 cm in treatment 4 was considered to be background, because G. mosseae would have grown only c. 15.5 cm away from the receiver. Comparison of treatments 1 and 4 reveals that there was a rather limited additive effect of two overlapping mycelia on hyphal length density (Fig. 4). Hyphal length densities in the soil layers showed the same pattern as in the soil core samples such that hyphal length was highest in the layer at the pot centre, except for the first harvest when similar length densities were found in the layer at the pot centre and in the middle layer. Hyphal lengths at the pot centre in treatment 1 were 20.5 ± 1.9, 25.2 ± 1.4 and 21.6 ± 1.3 m g−1 soil at the 3-, 7- and 14-d harvests, respectively. The hyphal length density at the pot centre in treatment 3 was 23.7 ± 0.6 m g−1 soil and not significantly different from the corresponding data for treatment 1 measured at the 7-d harvest.

Discussion

This work is the first to demonstrate that ecologically significant amounts of P can move between overlapping individual mycelia of an AM fungus. It also shows that no transfer takes place between mycelia of two different AM fungi. These results suggest that transfer had occurred via fusions or anastomoses between the hyphal networks.

Anastomoses in AM fungi were described half a century ago (Mosse, 1959) and fusions between individual mycelia were already demonstrated in two-dimensional systems (Giovannetti et al., 2004). Still, it is difficult to directly observe such fusions in three-dimensional soil systems because AM hyphae are fragile and because intra- and intermycelial anastomoses cannot be distinguished. Our indirect approach of measuring the transfer of a radioactive tracer isotope from one mycelium to another (Fig. 1) provided a solution to this problem. In designing the experiment we met two challenges. One was to define the time window of measurements to ensure that mycelia overlapped and the other was to ensure that the receiver mycelium did not grow all the way to the 32P-labelled soil placed at the mesh bag of donor plants. Treatment 2 successfully resolved the first challenge by defining when the mycelia had reached half the distance between donor and receiver plants. The estimated extension rate of the G. mosseae mycelium (3.1 mm d−1) was in the range of previous estimates (Jakobsen et al., 1992) and allowed for a 14-d window of measurement starting at the time of initiation of overlap between the two mycelia. The corresponding time window for G. caledonium in treatment 3 was limited to 7 d because the mycelium of that fungus extended at a higher rate (3.8 mm d−1). The second challenge was also met as the receiver plants of treatment 4 contained only negligible 32P amounts.

The hyphal length density required for the formation of anastomoses in soil is not known. However, the 20.5 m g−1 soil measured at the 3-d sampling in the zone of overlap by far exceeded the 0.25 m hyphae cm−3 media giving rise to 9–19 anastomoses m−1 hyphae of two Glomus species in monoxenic root cultures (Voets et al., 2006). Assuming that conditions for formation of anastomoses were similar in gel- and soil-based systems, at least 200 anastomoses g−1 soil would have been formed in the present study. The frequency of anastomoses measured in two-dimensional systems is much higher and in the range of 150–620 m−1 hyphae (Giovannetti et al., 2006).

The donor plants were defoliated at the start of the 14-d harvest period because we wanted to prevent the possibility that sink strength effects of the donor plant would limit transfer between individual mycelia. In other words, we wanted to make sure that the 32P contained in the donor mycelium was available for transfer to the receiver mycelium. We have previously shown that the host plant is a strong sink for nutrients in the AM fungal mycelium. Thus, 32P supplied locally to an AM fungal mycelium was directed predominantly towards the plant, while more distal parts of the mycelium contained only small amounts of 32P (Johansen et al., 1993). A similar sink regulation of nutrient flow may explain why a growth limitation of seedlings was offset by defoliation of their interlinked donor plants in two other studies (Jakobsen, 2004; Pietikäinen & Kytöviita, 2007). Defoliation of donors in the present work appeared to be justified as contents of 32P in hyphae from soil layers of the receiver half of the pots increased over time, indicating a predominant transport of 32P towards the largest sink.

One important question is whether the 32P transferred to donor plants could have been mediated by processes unlinked to anastomosis. Our results with G. caledonium-inoculated receiver plants suggest that this was not the case. A previous study showed that fusions established between individual mycelia originating from the same G. mosseae isolate, but did not occur between mycelia originating from different isolates (Giovannetti et al., 2003). In consequence, we assumed that anastomoses would not form between G. mosseae and G. caledonium. Our assumption could not be fully justified from the 32P contents in receiver plants as the difference observed between treatments 1 (G. mosseae receiver) and 3 (G. caledonium receiver) was nonsignificant. However, the parallel measurements of 32P content in mycelia extracted from the soil exclude the possibility that transfer could have originated from turnover of donor mycelium or leakage from living hyphae. As the mycelia of receiver plants did not reach the 32P patches we conclude that transport via anastomoses was the most likely mechanism of transfer.

The quite variable quantities of 32P transferred at the second harvest (7 d) probably reflect the variability between pots in time of establishment of functional inter-mycelium fusions. However, correlation analysis revealed no relationship to root colonization, plant DWs or hyphal length densities in the middle of the pot. Other variables could have contributed to the variation in 32P transfer, such as spread and 32P content of donor mycelium as well as time delay in transfer through anastomoses and receiver mycelium.

Our extrapolation that 20% of the P uptake of the receiver during the 14-d harvest period would have originated from the donor mycelium is probably an overestimation as the degree of overlap between mycelia increased over time. However, the corresponding measured transfer of 0.6% originating from the small labelling compartments is high enough to have potential impact on plant nutrient uptake and growth. Here we assume that all 32P transferred originated directly from the labelled soil patch only and not also from 32P contained in the living donor plant. This is based on previous studies showing that direct interplant transfer of P via a shared mycelium is insignificant (Johansen & Jensen, 1996; Simard et al., 2002; Yao et al., 2003).

It appears that the length density of the G. mosseae mycelium has a maximum value in our soil-based system and that this upper limit was reached. Thus, we observed no additive effect on hyphal length when two mycelia were overlapping. This indicates that neighbour-sensing and negative autotrophism (Meskauskas et al., 2004) could be present in AM fungi. The mechanism behind negative autotrophism is not known, but could involve depletion of nutrients, inhibitory signalling molecules or interaction of electric fields (Robson, 1999).

Individual mycelia of the same AM fungal genotype in cultivated soils have been predicted to fuse and form common mycelial networks and such ability to quickly form interconnecting hyphal networks may be an important feature of AM fungi in disturbed soils (Rosendahl, 2008). Glomus mosseae is common in agricultural fields and the ability to anastomose and form common mycelia networks could explain its adaptation to disturbance. Spatial genetic heterogeneity of G. mosseae was similar in fallow and cultivated fields, indicating that the fungus was able maintain its population structure in spite of the disturbance (Rosendahl & Matzen, 2008). It is therefore likely that extensive mycorrhizal networks can rapidly form by anastomosing of mycelia and the present work suggests that an anastomosed mycorrhizal network in the field can mediate P transport over a distance of at least 20 cm. The large networks may have important impacts on nutrient retention because nutrients released from decomposing roots will be effectively taken up by the hyphal network before entering the soil pool (Newman, 1988). This would help to maintain a tight nutrient cycle, keeping nutrients more directly available to the plants. The plants constituting the largest sinks will probably benefit the most from this mobile and biologically confined P pool and competition between the individuals linked into the network is likely to be altered (Pietikäinen & Kytöviita, 2007). Because of the high functional diversity in the AM symbiosis (Ravnskov & Jakobsen, 1995; Helgason et al., 2002; van der Heijden et al., 2003; Munkvold et al., 2004) the identity of the plant and fungal partners would probably also play a role in the nutrient distribution in the network.

Acknowledgements

We thank Anne Olsen for dedicated and excellent technical assistance.

Ancillary