SEARCH

SEARCH BY CITATION

Keywords:

  • arbuscular mycorrhizal fungi;
  • common mycorrhizal networks;
  • competition for phosphorus;
  • Cucumis sativus (cucumber);
  • seedling growth;
  • Solanum lycopersicon (tomato)

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Common mycorrhizal networks (CMNs) influence competition between plants, but reports regarding their precise effect are conflicting. We studied CMN effects on phosphorus (P) uptake and growth of seedlings as influenced by various disruptions of network components.
  • Tomato (Solanum lycopersicon) seedlings grew into established networks of Rhizophagus irregularis and cucumber (Cucumis sativus) in two experiments. One experiment studied seedling uptake of 32P in the network in response to cutting of cucumber shoots; the other analysed seedling uptake of P and nitrogen (N) in the presence of intact or severed arbuscular mycorrhizal fungus networks and at two soil P concentrations.
  • Pre-established and intact networks suppressed growth of tomato seedlings. Cutting of cucumber shoots mitigated P deficiency symptoms of seedlings, which obtained access to P in the extraradical mycelium and thereby showed improved growth. Solitary seedlings growing in a network patch that had been severed from the CMN also grew much better than seedlings of the corresponding CMN.
  • Interspecific and size-asymmetric competition between plants may be amplified rather than relaxed by CMNs that transfer P to large plants providing most carbon and render small plants P deficient. It is likely that grazing or senescence of the large plants will alleviate the network-induced suppression of seedling growth.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Arbuscular mycorrhizal fungi (AMF) are integral parts of plant ecosystems where individual plants may be linked together in common mycorrhizal networks (CMNs). The connective mycelia receive carbon (C) from the plants and they acquire mineral nutrients from the soil, in particular phosphorus (P). The formation of such networks is easily demonstrated in model systems where extraradical mycelium (ERM) from one plant colonizes another (e.g. Johansen & Jensen, 1996) or where individual conspecific mycelia fuse (Giovannetti et al., 2004). The function of CMNs is potentially important for the shaping of plant communities.

Translocation of C and mineral nutrients can occur over long distances (cm to m) in AMF mycelia in soil (Jakobsen et al., 2002; Olsson et al., 2002; Jansa et al., 2003; Thonar et al., 2011) and translocation of P has been demonstrated between anastomosing mycelia (Mikkelsen et al., 2008). Understanding the impact of such CMN-mediated nutrient transport on competition between individuals remains an important research challenge in plant populations and communities where size asymmetry among individuals is common. In plant populations, larger individuals will in general obtain a disproportionate share of the growth-limiting resources and suppress the growth of smaller individuals (Weiner, 1990). CMNs tend to increase this inequality between individuals (Allsopp & Stock, 1992; Shumway & Koide, 1995; Facelli & Facelli, 2002), which in a recent study on Andropogon geradii was explained by positive feedbacks between mycorrhiza formation, mineral nutrient uptake and host growth (Weremijewicz & Janos, 2013). Seedlings and adult plants represent an extreme degree of asymmetry, and here nutrient availability from CMNs could potentially be crucial to the establishment and growth of the seedlings. However, experimental support for this is ambiguous.

Seedlings become rapidly colonized in the presence of an established mycorrhizal network in grassland turfs (e.g. Read et al., 1976; Birch, 1986) and in nurse-pots (see Smith & Read, 2008), but the implications of the network for seedling growth are not well understood. It has been suggested that a mycorrhizal network constitutes a cheap ‘nutrient adsorption machine’ in which the seedlings do not need to invest resources (van der Heijden & Horton, 2009), and a number of studies have reported beneficial effects of mycorrhizal networks on seedling growth (Grime et al., 1987; Francis & Read, 1995; Marler et al., 1999; Carey et al., 2004; van der Heijden, 2004). However, there have been several reports of suppressed seedling growth in the presence of an established network (Francis & Read, 1995; Moora & Zobel, 1996; Jakobsen, 2004; Nakano-Hylander & Olsson, 2007; Janouskova et al., 2011; Janos et al., 2013). Other studies have reported reduced mycorrhizal growth responses in seedlings grown with a network compared with seedlings grown alone (Ocampo, 1986; Moora & Zobel, 1998; Kytoviita et al., 2003; Pietikainen & Kytoviita, 2007). A review of the literature showed that 42% of the arbuscular mycorrhizal (AM) host seedling species investigated responded positively to the presence of a CMN, while the remainder responded either in a negative or a neutral manner (van der Heijden & Horton, 2009). The apparent lack of agreement between investigations highlights the need to identify the key factors governing the fate of seedlings entering a CMN. Clearly, one key factor is the spatial distribution and movement of nutrients in the networks, and it is interesting that several authors have suggested that pre-emption of nutrients by the ERM might explain the CMN-induced suppression of seedlings (Ocampo, 1986; Moora & Zobel, 1996, 1998; Janouskova et al., 2011).

Principles of nutrient transport across the symbiotic interface in individual mycorrhizas extend to the function of CMNs: there is neither significant transport of C from fungus to plant (Pfeffer et al., 2004; Voets et al., 2008) nor significant transport of mineral nutrients from plant to fungus (Newman, 1988; Johansen & Jensen, 1996). This means that the connective mycelium does not mediate any exchange of nutrients between plants. The key issue of nutrient dynamics in CMNs is therefore about sharing: which plants will deliver C to the fungus and which plants will have access to the mineral nutrients in the mycelium? Several studies with mycorrhizal root cultures have suggested that the fungus-to-plant P transfer is stimulated by a high carbohydrate status of the roots (Bücking & Shachar-Hill, 2005; Lekberg et al., 2010; Kiers et al., 2011). Similarly, transfer of nitrogen (N) from fungus to plant also depends on the C status of the roots (Fellbaum et al., 2012). Root cultures can be assumed to have a low sink strength for P as compared with established plants, and it is crucial to extend these physiological studies to CMNs with intact plants. Here, P would be expected to be directed predominantly to large plants representing the highest C source strength rather than to seedlings. A preliminary CMN study with cucumber (Cucumis sativus) plants and tomato (Solanum lycopersicum) seedlings indeed suggested that the seedlings were poor competitors for nutrient pools in the common mycelium (Jakobsen, 2004).

The aim of the present work was to investigate the mechanisms underlying the performance of tomato seedlings grown in CMNs with large cucumber plants. We tested two hypotheses: (1) that such CMNs would suppress the growth of tomato seedlings; and (2) that the assumed suppression was caused by intensified competition for P in the mycelium as a result of predominant transfer of P to the large plant representing the major C source.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Cucumber (Cucumis sativus L. cv Aminex F1 hybrid) plants were grown in pots and rhizoboxes together with seedlings of tomato (Solanum lycopersicum L. cv 76R). One experiment also included the reduced mycorrhiza colonization (rmc) mutant that was selected from a mutagenized population of 76R tomato plants (Barker et al., 1998). Cucumber plants grew in association with Rhizophagus irregularis (syn. Glomus intraradices Schenck and Smith) BEG87 before planting of tomato seedlings. The AMF inoculum consisted of spores, root pieces and soil from pot cultures of R. irregularis and Trifolium subterraneum. Soil compartments for tomato contained no inoculum. The growth medium (hereafter referred to as soil) was a mixture (1 : 1 w/w) of sandy moraine loam and quartz sand that had been partially sterilized by irradiation (2 × 10 kGy; 10-keV electron beam) to eliminate native AMF. The pH was 7.5 and plant available P was 7 mg P kg−1 soil (0.5 M NaHCO3 extraction method; Olsen et al., 1954). Mineral nutrients were mixed into the soil in the following amounts (mg kg−1): NH4NO3, 86; K2SO4, 75; CaCl2·2H2O, 75; CuSO4·5H2O, 2.1; ZnSO4·7H2O, 5.4; MnSO4· H2O, 10.5; CoSO4·7 H2O, 0.39; MgSO4·7 H2O, 45.0; Na2MoO4·2H2O, 0.18.

Experiment 1: cutting the shoot of the large plant

Experiment 1 was designed to study how seedlings of the tomato genotype 76R respond to cutting the shoot of a large cucumber plant and thereby reducing its C source strength. In addition to the main treatments (intact or cut cucumber shoots), four control treatments were included: (1) concurrent planting of tomato and cucumber seeds to test the assumption that tomato seedlings will not be suppressed in this situation; (2) planting of not only 76R seedlings but also seedlings of the rmc mutant in all pots to confirm that suppression occurs only when seedlings are linked to the ERM of the large plant; (3) addition of 100 μg P g−1 soil to tomato seedling compartments to investigate whether P deficiency was actually responsible for the suppression; and (4) growth of plants in soil without AMF inoculum to test whether nonmycorrhizal tomato seedlings would grow better than CMN seedlings (Table 1). Roots of tomato seedlings were confined by hypha-permeable mesh bags and therefore made no direct contact with the cucumber root system. Movement of P in the mycelium was traced using 32P, which was absorbed from labeled soil in a third root-exclusion mesh bag (Figs 1a, S1).

Table 1. Treatments and main events in Experiment 1
TreatmentsDays after sowing of cucumber
Sowing of tomatoCutting of cucumber shootHarvests
  1. All time-points are with reference to sowing of all cucumber plants at day 0. The monitoring pots contained one cucumber plant and three mesh bags with a 76R tomato plant in each bag. All other pots contained one cucumber plant, one 76R tomato seedling and one reduced mycorrhiza colonization (rmc) tomato seedling (see Fig. 1a). The rmc seedlings provided a root colonization control in all pots. The harvests at 45, 50 and 57 d correspond to ages of tomato seedlings of 20, 25 and 32 d as used in Fig. 3. AMF, arbuscular mycorrhizal fungi.

  2. a

    Not applicable.

  3. b

    Three replicate pots.

  4. c

    Four replicate pots.

  5. d

    One 76R seedling from each of three replicate pots.

Intact cucumber shoot25n.a.a45b50b57b
Cut cucumber shoot2545n.a.50b57c
Planting time control0n.a. 50c 
P addition control25n.a. 50c 
No AMF inoculation control25n.a. 50c 
Monitoring of colonization of 76R tomato roots0n.a.17d24d46d
image

Figure 1. Outline of growth systems used in Experiment 1 (a) and Experiment 2 (b). Plant roots are represented by brown lines, while the overall shading illustrates common mycorrhizal network (CMN) hyphae growing throughout arbuscular mycorrhizal fungus (AMF)-inoculated pots and rhizoboxes. The dashed circles (a) and the dashed line (b) represent bags and a sheet, respectively, of 25-μm nylon mesh that separated root systems of cucumber plants and tomato seedlings and confined the 32P-labeled soil patch in (a).

Download figure to PowerPoint

Cucumber seeds were planted in 28 plastic bag-lined 12-cm pots containing a mixture of 900 g of soil and 75 g of inoculum of R. irregularis; seedlings were thinned to one per pot. In each pot there were three nylon mesh bags (37-μm mesh; bags 3 cm in diameter), two of which contained 65 g of inoculum-free soil and the third of which contained an empty test tube. Four of the 28 pots had 76R and rmc tomato seeds planted at the same time as cucumber seeds and served as the concurrent planting control. Three additional pots with mycorrhizal cucumber plants were used to monitor in-growth and establishment of ERM in the soil-filled mesh bags. These monitoring pots contained three soil-filled mesh bag compartments, and one 76R tomato seed was sown in each mesh bag at the same time as cucumber (Table 1). One seedling per pot was sampled at three time-points, and root staining revealed a rapid establishment of mycorrhiza: the proportion of root length colonized was 43, 71 and 92% in 16-, 23- and 46-d-old plants, respectively.

Having confirmed the growth of ERM into soil-filled mesh bags, we planted tomato seeds 25 d after planting of cucumber in the main treatment (16 pots) and in four replicate pots for control treatments with additional P or without AMF inoculum (Table 1). Germinated tomato seeds of the 76R and rmc genotypes were sown in the two mesh bag compartments in each of the 24 pots. Emerged seedlings were thinned to one per mesh bag and vertically placed plastic meshes were used to prevent cucumber leaves from shading the tomato seedlings (Fig. S1). In parallel with seedling establishment, a 32P soil patch was established by replacing the empty test tube with 45 g of soil labeled with 175-kBq carrier-free orthophosphate (Perkin Elmer, Waltham, MA, USA). Pots were watered daily to 70% of water-holding capacity and received 30 mg of additional N as NH4NO3 solution 18, 38 and 47 d after sowing of cucumber. Plants were maintained in a growth room with a 16 h : 8 h light : dark cycle at 20°C : 15°C. Osram daylight lamps (Osram GmbH, Munich, Germany) provided 500 μmol m−2 s−1 photosynthetically active radiation (400–700 nm).

Three pots of the main treatment were harvested when tomato seedlings were 20 d old and at the same time cucumber shoots were cut just above the soil surface in seven of the remaining 13 pots. Harvests comprised three replicate pots of both treatments at 25 and 32 d seedling age and four replicate pots of the cutting treatment at 32 d (Table 1). Cucumber plants that had their shoots cut had no aboveground re-growth during the subsequent 12 d. Cucumber plants remained intact in the control treatments which were harvested when tomato seedlings were 50 d old (concurrent planting control) or 25 d old (P addition and nil inoculation controls; Table 1).

Fresh weights were recorded for all shoots and for roots washed free of soil. When root fresh weights were greater than 500 mg, subsamples were stored in 50% EtOH for later mycorrhiza analysis. The remaining root parts were dried together with the shoots at 70°C and dry weights were determined.

Experiment 2: severing of ERM connections before seedling establishment

Experiment 2 was designed to compare the performance of 76R tomato seedlings that were linked to the large cucumber plant by an AMF mycelium with that of seedlings where the AMF links to the large plant had been severed. Experiment 2 also comprised a nonmycorrhizal control treatment. Plants were grown in a rhizobox system with a central mesh-enclosed cucumber compartment that was sandwiched between two side units. When ERM from cucumber roots had colonized the soil in both side units, one of these was removed and seedlings were planted (Figs 1b and S2). Side units had two soil P concentrations (Fig. 1b) in order to study how the expected seedling suppression depended on soil P availability.

Each side unit comprised six compartments (length × width × height = 40 × 30 × 150 mm) with removable outer walls and was separated from the central compartment (length × width × height = 240 × 30 × 150 mm) by 25-μm nylon mesh. The three units were held together by bolts and nuts (Fig. 1b). Three replicate rhizoboxes contained a mixture of 1650 g of soil and 200 g of inoculum of R. irregularis in the central unit and 260 g of soil in each of the 12 side unit compartments. Half of the side unit compartments received soil that had been thoroughly mixed with KH2PO4 at 30 mg P kg−1 soil (see Fig. 1b). A fourth box was prepared similarly but with 200 g of soil instead of inoculum. Water was added to 70% of water-holding capacity and two germinated cucumber seeds were planted in the central compartment of each box. Additional N was added twice as NH4NO3 solution: 100 mg at 30 d and 60 mg at 50 d after planting. The outer wall was temporarily removed from one side unit compartment of two boxes after 60 d and thin layers of soil were sampled from the soil surface at 3 cm distance from the central cucumber compartment. Microscopic examination of the fresh soil samples revealed the presence of fresh AMF mycelium that had grown across the side unit compartments. Soil samples taken in a similar manner from the uninoculated unit contained no visible AMF mycelium. One side unit was then removed from each box system, avoiding disturbance of the position of the other side unit. The exposed surfaces of the central unit and the side unit were now covered by separate PVC plates that were fastened by bolts and nuts. This resulted in four disconnected side units and four side units that were still connected to the central compartment; that is, the ERM connections between the central compartment and side units were severed or not. Two germinated 76R tomato seeds were then planted in each side unit compartment, 60 d after planting of cucumber. Upon emergence, seedlings were thinned to one per compartment and plastic meshes were used as in Experiment 1 to prevent cucumber leaves from shading the tomato seedlings (Fig. S2). The eight growth containers were not sealed at the bottom, but drainage of water was avoided by watering to only 70% of water-holding capacity.

Tomato seedlings were harvested at 17, 24 and 31 d after sowing. At each harvest, the outer wall and the soil column was removed from each of two side unit compartments in each unit: one with nil P (P0) soil and one with 30 mg P kg−1 soil (P30). Shoots were cut at the soil surface and dried at 70°C, and their dry weight was determined. A soil sample composed of several subsamples without roots was taken and stored frozen. Roots were washed free of the remaining soil, a sample was stored in 50% EtOH and the remainder was dried at 80°C. The 24-d harvest included all 12 seedlings from the two mycorrhiza-free side units.

Analyses

Ground plant material was digested in a 4 : 1 (v/v) mixture of nitric acid and perchloric acid and P contents in extracts were determined by the molybdate blue method (Murphy & Riley, 1962) on an AutoAnalyzer3 (Bran + Lubbe, Norderstedt, Germany). Extracts from Experiment 1 were analyzed for content of 32P by liquid scintillation counting (Packard TR1900; Perkin Elmer, Boston, MA, USA) and counts were corrected for decay and background. Total N in shoot samples from Experiment 2 was determined in 2 mg of finely homogenized samples on an EA 1110 elemental analyzer (Thermo Scientific, Bremen, Germany).

Roots from Experiments 1 and 2 were stained with black ink (Vierheilig & Piche, 1998) and trypan blue (Kormanik & McGraw, 1982), respectively, and analyzed for mycorrhiza colonization using a grid-line intersect method (Giovannetti & Mosse, 1980). Hyphal length density in soil was determined on well-mixed soil samples from the 31-d harvest in Experiment 2 (Jakobsen et al., 1992).

Soil samples were extracted with 0.5 M NaHCO3 for P analysis (Olsen et al., 1954) and with 0.5 M K2SO4 for analysis of ammonium and nitrate. Extracts were analyzed using AutoAnalyzer3.

Statistics

Statistical analyses were conducted in R (2.15.2; R Core Team, 2012). Data from Experiment 1 were analyzed using a three-way ANOVA to assess the effects of cucumber shoot cutting, time of harvest and tomato seedling genotype. The data set for 32P content was not normally distributed, but the subset for 76R seedlings met the requirements for two-way ANOVA. The corresponding rmc data were analyzed by Kruskal–Wallis tests. Control treatments (additional P; no mycorrhizas) were compared separately with 25-d-old intact plants by two-way ANOVAs to check for interactions of either P addition or mycorrhiza with tomato genotype. Data for 32P content were not normally distributed and were analyzed by Kruskal–Wallis tests for each tomato genotype.

Data from Experiment 2 were analyzed by three-way ANOVAs using connectedness, harvest time, and P addition as factors. The experimental design could have introduced random variation: side units of the rhizoboxes were not fully independent and rhizoboxes were therefore treated as blocks. Pairs of neighboring plants in each side unit compartment were also treated as blocks each including high and low P amendment (Fig. 1). The model was therefore extended by a random statement that accounted for nesting of rhizobox within time, block within rhizobox and connectedness within rhizobox. The three-way ANOVAs were then conducted using the lmer function in the LMERConvenienceFunctions package of R.

Effects of R. irregularis inoculum at day 24 were analyzed by three-way ANOVAs with mycorrhiza (+/− inoculum), connectedness and P addition as factors. This involved a random statement accounting for nesting of block within rhizobox and connectedness within rhizobox.

Data were log, arcsine or rank transformed as needed to obtain normality and homogeneity of variances. A Kruskal–Wallis test was used if assumptions were not met. P-values < 0.05 were considered significant.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Experiment 1

The 76R tomato genotype grew well in the concurrent planting time control and its shoot dry weight was more than twice that of the co-cultivated rmc mutant (Table 2). This difference was in agreement with the contrasting levels of root colonization in the two genotypes. The importance of mycorrhiza for growth of 76R was confirmed by the shoot contents of P and 32P, which were much lower in rmc than in 76R (Table 2). The 32P was available to the ERM but not to the roots, and the 76R-to-cucumber ratios for 32P content and P content were both 0.19. The relative importance of mycorrhiza for P uptake was therefore similar in 76R and in cucumber in the concurrent planting control.

Table 2. Shoot variables and root colonization of co-cultivated tomato and cucumber plants when planted at the same time (control #1)
PlantDry weight (g)P content (mg)32P content (103 cpm)Colonization (%)
  1. The 76R and reduced mycorrhiza colonization (rmc) tomato genotypes grew in mesh bags (Fig. 1). Pots were inoculated with Rhizophagus irregularis BEG87. All plants were harvested 50 d after sowing. Data are mean ± SE; = 4.

Tomato, 76R0.54 ± 0.020.49 ± 0.03246 ± 14387 ± 3
Tomato, rmc0.22 ± 0.010.16 ± 0.010.7 ± 0.211 ± 2
Cucumber2.61 ± 0.082.58 ± 0.071262 ± 18790 ± 1

The concurrently planted 76R tomato seedlings grew markedly better than seedlings established 25 d after the cucumber plants (Fig. 2). This delay in planting resulted in tomato seedlings that, at a given plant age, had only about half the biomass of seedlings that were sown concurrently with cucumber. However, this negative effect of a pre-established cucumber plant on growth of 76R seedlings was significantly mitigated by cutting the cucumber shoots (Fig. 3). Table S1 reports ANOVA results for the variables measured on the tomato seedlings. Shoot dry weights were significantly influenced not only by cutting but also by tomato genotype and harvest time (< 0.001). Furthermore, all two- and three-way interactions were significant (see Table S1 for statistical details). The cutting × genotype interaction was caused by the absence of a cutting effect in the rmc tomato genotype, the cutting × time interaction shows that the effect of cutting increased over time and the genotype × time interaction reveals that the 76R seedlings grew faster than the rmc seedlings (Table S1). The reason for the cutting × genotype × time interaction is that a strong concurrent effect of cutting and genotype was observed only at the last harvest time. The growth response to cutting was accompanied by a change in the leaf phenotype of 76R seedlings from strong anthocyanin pigmentation towards green (Fig. S1). Accordingly, already after 5 d, the cutting resulted in a 51% increase in the shoot P concentration of the seedlings (< 0.001). Shoot P contents also increased after cutting, and effects of treatment and their interactions were statistically similar to those for dry weights, except that the effects were stronger (Fig. 3, Table S1). The strongest effect of cutting was observed in 32P content, which in 76R seedlings of the cutting treatment reached a value that was 6.5-fold the value in seedlings of the undisturbed CMN (Fig. 3). The 32P content of seedling shoots was also clearly affected by genotype and time. However, a three-way ANOVA was not possible because the 32P data for the rmc genotype included a number of zero values and data were not normally distributed. Therefore, the ANOVA was restricted to the 76R seedlings. Here, the 32P in the ERM was more available to the seedlings when the shoot of the cucumber plant had been removed (< 0.001; Table S1). Kruskal–Wallis tests of the rmc data showed that 32P content was influenced by time only (< 0.05). The cucumber roots contained 2.3 × 106 cpm at the time of shoot excision and this content decreased significantly to 1.6 × 106 cpm after 12 d (< 0.05).

image

Figure 2. Growth of 76R tomato seedlings sown concurrently with cucumber (upper dashed line) or when cucumber plants were 25 d old (lower dashed line). Data originate from monitoring of mycorrhiza formation in 76R roots (diamonds; = 3), from the concurrent planting control (X; n = 4) and from the treatment with pre-established and intact cucumber plants (black circles; = 3). 76R seedlings grew in mesh bags (37-μm mesh; bags 3 cm in diameter) and all pots were inoculated with Rhizophagus irregularis BEG87. Fresh weights are presented because dry weight data were not obtained for plants used to monitor mycorrhiza formation in 76R. Data points are mean ± SE (Experiment 1).

Download figure to PowerPoint

image

Figure 3. Dry weight and phosphorus (P) and 32P contents in shoots of 76R and reduced mycorrhiza colonization (rmc) tomato seedlings linking into a common mycorrhizal network (CMN) with a large cucumber plant that remained intact (black circles) or had its shoot cut at 20 d seedling age (open circles). Data for control treatments with 25-d-old seedlings grown with intact cucumber plants are also shown: additional P supplied to tomato seedlings (black squares) and absence of mycorrhizal inoculum (black triangles). Data points are mean ± SE;= 3 or n = 4 (cutting treatment harvested at 32 d and controls harvested at 25 d). See Tables S1 and S2 for F statistics and probabilities of significance (Experiment 1).

Download figure to PowerPoint

Two-way ANOVAs were applied to data from the 25-d harvest to separately analyze the effects of the additional P and the no AMF inoculum controls, and the results are summarized in Table S2. The addition of P to soil in the seedling mesh bags increased shoot dry weight and P content significantly (< 0.01) as compared with the corresponding intact seedlings of the same age (Fig. 3, Table S2). This effect of P addition was strikingly similar in the two tomato genotypes. A Kruskal–Wallis test revealed that P addition significantly inhibited 32P uptake in 76R seedlings (< 0.05) whereas 32P uptake by rmc seedlings was nil in both the main and additional P treatments (Fig. 3). Tomato seedlings in pots without AMF inoculum also had higher dry weights and P content than the corresponding intact seedlings of the same age (Fig. 3, Table S2). Again, this increased growth of nonmycorrhizal plants was unaffected by genotype. As expected, 32P uptake was completely inhibited in shoots of nonmycorrhizal 76R seedlings (< 0.05; Kruskal–Wallis test) (Fig. 3). The 50-d-old cucumber plants in the nonmycorrhizal treatment were much smaller than corresponding plants in the mycorrhizal treatment (0.39 ± 0.07 versus 4.20 ± 0.24 g shoot DW, respectively; mean ± SE).

Experiment 2

In Experiment 1 the possibility could not be fully excluded that the strongly increased content of 32P in 76R tomato seedlings in response to cutting of the cucumber shoot was partly derived from decomposing cucumber roots. A second experiment was therefore carried out where undisturbed CMN seedlings were compared with solitary seedlings that grew in disconnected side units where the ERM connection to the central compartment had been severed. More than 95% of the root length of cucumber plants was densely colonized by R. irregularis at the final harvest (31 d after planting of seedlings), when high length densities of ERM were also recorded in soil taken from CMN tomato seedlings (14.4 ± 1.1 and 17.0 ± 2.7 m g−1 (mean ± SE) in P0 and P30 soil, respectively). Tomato seedlings became highly colonized when grown both in undisturbed and in disconnected side units; however, additional P resulted in significantly reduced colonization, in particular in solitary seedlings in disconnected side units at 24 d (Fig. S3).

Shoot dry weight of tomato seedlings was markedly higher in disconnected than in undisturbed side units where ERM remained connected to the pre-established large cucumber plants. Seedlings also grew better in P-amended soil and growth increased over time (Fig. 4). Effects of connectedness, P addition and time were all highly significant (< 0.001, 0.01 and 0.001, respectively; see Table S3 for details). The significant connectedness × time interaction (< 0.05) was a consequence of an increased effect of connectedness over time. Shoot P content in tomato seedlings was influenced by treatments in a similar way to plant growth, such that P content was clearly highest in seedlings grown in disconnected side units and in seedlings given additional P (Fig. 4, Table S3). The P addition × time interaction was significant (< 0.05) as a consequence of an increase in the effect of P addition over time, irrespective of connectedness. Shoot N content was significantly higher in seedlings grown in disconnected side units and increased significantly over time, but the effect of P addition was nonsignificant (Fig. 4, Table S3). A significant connectedness × time interaction (< 0.05) was attributable to an increased effect of connectedness over time. The shoot N : P ratios were in most cases lowest for CMN tomato seedlings (Fig. 4). N : P ratios decreased over time (< 0.01) and when plants had received additional P (< 0.001) (Table S3). However, all three pairs of treatment factors showed significant interactions (< 0.05). The connectedness × P addition interaction was caused by a larger magnitude of the connectedness effect at the higher soil P concentration, and the connectedness × time interaction was attributable to a time-dependent effect of connectedness at P0 in particular. Finally, the P addition × time interaction was attributable to a stronger decrease in the N : P ratio with time at P30 than at P0.

image

Figure 4. Dry weight, phosphorus (P) content, nitrogen (N) content and N:P ratio in shoots of tomato seedlings as influenced by a Rhizophagus irregularis BEG 87 network with undisturbed (black circles; common mycorrhizal network (CMN) seedlings) or disconnected (open circles; solitary seedlings) links to a large cucumber plant. Seedlings were grown in soil supplied with nil (P0) or 30 mg P kg−1 soil (P30). Data for 24-d-old nonmycorrhizal (NM) control seedlings in undisturbed (black triangles) or disconnected (open triangles) rhizobox side units are also shown. Data points are mean ± SE;= 3; see Tables S3 and S4 for F statistics and proba-bilities of significance (Experiment 2).

Download figure to PowerPoint

The 24-d harvest included a set of nonmycorrhizal control plants grown in an identical rhizobox with an undisturbed and a disconnected side unit. Here, roots of these uninoculated ‘donor’ cucumber plants and tomato seedlings remained uncolonized. The nonmycorrhizal control seedlings were larger than mycorrhizal seedlings of the same age (< 0.05; Fig. 4). This mycorrhiza effect was accompanied by significant effects of connectedness (< 0.05) and P addition (< 0.001), as also observed in the time course data set. Details of the statistical analyses are shown in Table S4. A significant mycorrhiza × connectedness interaction (< 0.05) was attributable to a larger suppressive effect of mycorrhiza in seedlings grown in undisturbed than in disconnected side units. Shoot P content was also higher in nonmycorrhizal than in mycorrhizal plants and P addition had the expected stimulating effect (Fig. 4, Table S4). The mycorrhiza × connectedness interaction for shoot P content was significant at < 0.01; this shows that mycorrhiza suppressed P uptake much more in seedlings grown in undisturbed than in disconnected side units. Shoot N content was influenced by mycorrhiza in a similar way to P content, including a significant mycorrhiza × connectedness interaction (Fig. 4, Table S4). However, N content was also influenced by connectedness, in particular with added P (the connectedness × P addition interaction was significant at < 0.001). Shoot N : P ratios were unaffected by mycorrhizas, but were decreased by P addition to the soil in side unit compartments (< 0.001).

The effect of connectedness, P addition and time on measured soil pools of ‘plant-available P’, mineral inline image and inline image is shown in Fig. 5. As expected, P concentrations were higher in soil given extra P (< 0.001) and concentrations decreased over time (< 0.05) (Fig. 5; statistical details are shown in Table S3). Concentrations of inline image remained low throughout the experiment (2–4 ppm) and treatments had no significant effects. Concentrations of inline image in soil from 17-d-old CMN seedlings were 4-fold the corresponding concentrations of inline image and furthermore responded significantly to the disconnection treatment (Fig. 5, Table S3). The disconnection treatment thus resulted in 5–10-fold higher inline image concentrations, which were 50–76 μg g−1 soil at the first two harvests. inline image concentrations decreased significantly over time (< 0.05). At 24 d, soil nutrient pools were also measured in soil from side unit compartments of the nonmycorrhizal rhizobox (see Fig. 5 and details of the statistical analysis in Table S4). The effect of P addition on soil P concentrations was obvious, but lack of variance homogeneity prevented further statistical analysis of this set of soil P data. However, concentrations of inline image at 24 d were significantly higher in soil from nonmycorrhizal than in soil from mycorrhizal seedlings (< 0.001) and were furthermore higher in soil from disconnected than in soil from undisturbed side unit compartments (< 0.05) (Fig. 5, Table S4). Concentrations of inline image revealed an opposite pattern, being significantly higher in soil from mycorrhizal than in soil from nonmycorrhizal seedlings (< 0.05) (Fig. 5, Table S4) and connectedness had a strong effect at 24 d (< 0.001), as also observed in the time course data set. Soil from plants harvested at 24 d contained less inline image in P-amended than in P0 treatments (P < 0.05).

image

Figure 5. Phosphorus (P), ammonium and nitrate in extracts of soil from side unit compartments as influenced by a Rhizophagus irregularis BEG 87 network with undisturbed (black circles) or disconnected (open circles) links to a large cucumber plant. Soil was supplied with nil (P0) or 30 mg P kg−1 soil (P30). Soil data for 24-d-old nonmycorrhizal (NM) control seedlings in undisturbed (black triangles) or disconnected (open triangles) rhizobox side units are also shown. Data points are mean ± SE;= 3; see Tables S3 and S4 for F statistics and probabilities of significance (Experiment 2).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Cucumber plants and tomato seedlings shared the same ERM, because roots of the tomato seedlings were colonized by R. irregularis when inoculum was added solely to the cucumber plants. In agreement with our hypothesis, seedlings were indeed suppressed by linking into a CMN with a large plant. The suppression was caused by P deficiency, and the CMN obviously amplifies competition for P between the large plant and the seedling. We also accepted the second hypothesis that the P in the connective ERM is only poorly accessible to seedlings and is preferentially transported to large plants, which represent the major C source. These whole-plant results accord with observations in mycorrhizal root cultures (Lekberg et al., 2010; Kiers et al., 2011) and offer a likely explanation of the causal relationships underlying similar CMN-induced seedling suppression in previous studies (e.g. Nakano-Hylander & Olsson, 2007; Janouskova et al., 2011). To our knowledge, the present work represents the first evidence in whole-plant CMNs that net transfer of P from an AMF mycelium to its connected hosts favors the larger and older plants over smaller and younger seedlings. This finding represents one answer to the list of key questions raised by van der Heijden & Horton (2009) to provide explanations for the observed variation in CMN effects on seedlings.

Mechanisms underlying CMN effects on seedling growth

Phosphorus was the major growth-limiting nutrient in both experiments, as P amendment increased seedling growth (Figs 3, 4). In Experiment 2, however, the growth of CMN plants given high P was limited by N deficiency, in accordance with the fact that N : P ratios < 14 (Fig. 4) are indicative of N limitation (Koerselman & Meuleman, 1996). In this case, the possibility cannot be excluded that the growth suppression of CMN seedlings was associated with ERM-mediated transfer of N from the side unit compartment to the cucumber compartment. However, it is more likely that transfer via mass flow played the major role, as inline image concentrations were low not only in the soil of CMN seedlings, but also in the soil of nonmycorrhizal seedlings (Fig. 5).

When P is growth-limiting, mycorrhizas usually increase P uptake and growth. In this work, however, 32P in the mycelium was less available, not only to a nonhost plant such as the rmc mutant, but also to the CMN seedlings (Experiment 1; Fig. 3); in Experiment 2, P uptake was much higher in solitary than in CMN seedlings (Fig. 4). Assessment of growth of CMN seedlings against solitary mycorrhizal seedlings excludes the possibility that CMN-induced suppression was caused by a general negative growth response to mycorrhiza. The neutral growth response to mycorrhiza in the solitary seedlings in Experiment 2 (Fig. 4; disconnected, mycorrhizal versus nonmycorrhizal) confirmed that the suppression in the CMN plants was solely attributable to the CMN effect. In previous studies, solitary mycorrhizal control plants also grew much better than corresponding CMN plants (Ocampo, 1986; Moora & Zobel, 1998; Kytoviita et al., 2003; Pietikainen & Kytoviita, 2007). Using uncolonized plants as reference treatments, the CMN responses in the same four studies were calculated as being positive, negative, neutral and neutral, respectively (van der Heijden & Horton, 2009).

Our study demonstrates that the seedling-to-‘donor’ ratio for age or biomass has a great impact on CMN effects on seedling performance (Fig. 2). The importance of seedling age accords with the fact that CMN-induced growth suppression has mostly been reported for relatively young seedlings grown with larger plants (this study and e.g. Nakano-Hylander & Olsson, 2007; Pietikainen & Kytoviita, 2007). In contrast, in reports of CMN-induced benefits, seedlings have mainly been older and associated with a more complex mixture of more mature plant neighbors (Grime et al., 1987; Read & Birch, 1988; van der Heijden, 2004). The latter studies represent plant communities where some plants are grazed or becoming senescent. The importance of this age or biomass ratio for seedlings and ‘donors’ was demonstrated by Carey et al. (2004), who reported that Centauria maculosa seedlings grew better with a Festuca idahoensis companion plant, but not with Bouteloua gracilis, which had a plant dry weight more than threefold that of F. idahoensis. The inherent characteristics of a plant species, for example, growth rate, size and root:shoot ratio, are likely to influence its capacity to supply C to associated AMF. Such differences may also have influenced the relative growth of cucumber and tomato plants in this study. Still, suppression of seedlings has previously been demonstrated in other heterospecific combinations (see Moora & Zobel, 2010; Janos et al., 2013) and was similar in magnitude in conspecific and heterospecific combinations of two species (Nakano-Hylander & Olsson, 2007). Information on exchange of C and P at the level of individual arbuscules will be important for further untangling the mechanisms underlying the observed seedling suppression. Interestingly, a root culture study showed that the abundance of arbuscules was indeed lower in C-starved roots than in roots with an ample C supply (Lekberg et al., 2010).

As the growth suppression of tomato seedlings was mainly caused by P deficiency, the P pools in the ERM must have been less available to the tomato seedlings than to the large cucumber plant. The rapid relief of P deficiency of 76R tomato seedlings upon cutting of ‘donor’ shoots suggests that 32P in the ERM was initially directed toward the large ‘donor’ plant of the CMN, but that the direction of transport shifted toward the small seedlings when the donor no longer represented an active C source. The dominant P transport toward the largest C source prohibited the seedling from receiving significant amounts of P via its mycorrhizal uptake pathway, which in previous studies with tomato accounted for 78–100% of total P uptake (Smith et al., 2003; Nagy et al., 2009). In Experiment 1 of the present study, seedling P uptake would have been more or less confined to Pi uptake at the root epidermis, the direct pathway. The maintained P deficiency of CMN plants revealed that such direct root uptake was small, and two explanations of this are possible: (1) the Pi transporter genes of the direct pathway may have been down-regulated by the presence of mycorrhiza in the tomato roots (see Javot et al., 2007; Grønlund et al., 2013); or (2) available P may have been pre-empted by the ERM colonizing the soil of seedling compartments. The increased P uptake in seedlings in P-amended soil suggests that pre-emption was the causal factor.

The cutting of cucumber shoots had no significant effect on 32P uptake by rmc seedlings, and this indicates that P release from decomposing roots was not responsible for the marked effect of ‘donor’ cutting on the growth of 76R seedlings. This conclusion is supported by the enhanced nutrient uptake and growth of solitary seedlings in Experiment 2 as compared with CMN seedlings.

The role of pre-emption of soil nutrients by the ERM of common mycorrhizal networks

Pre-emption of soil nutrients by ERM was suggested to be responsible for CMN-induced growth depression of seedlings in several studies (Ocampo, 1986; Nakano-Hylander & Olsson, 2007; Janouskova et al., 2011), but actual changes in soil nutrients were not reported. Pool sizes of soil P and N did indeed change with treatment and time in the present study (Fig. 5). Plant available P concentrations decreased over time, even in P0 soil with CMN seedlings, which absorbed only small amounts of P over the 31-d growth period. This confirms that the ERM of the CMN had rather efficient uptake and transfer of P from the seedling soil compartments to the large cucumber plants. The rather low and constant inline image pools over the 17–31-d period may be explained by high nitrification rates and/or high uptake of N by the ERM. The ERM has a high inline image uptake capacity (Govindarajulu et al., 2005; Cruz et al., 2007) and is capable of depleting root-free soil compartments of mineral N (Johansen et al., 1992). Furthermore, the ERM has a stronger affinity for inline image than for inline image (Tanaka & Yano, 2005). The significantly higher inline image concentrations in soil from nonmycorrhizal than from CMN seedlings (and at both soil P concentrations) (Fig. 5) actually suggest that the ERM was partly responsible for the low inline image concentrations.

The decrease in inline image concentrations over time would in the case of CMN seedlings have been caused primarily by mass flow to the large cucumber plants, as seedling shoot N content remained low at 31 d (0.6 and 0.9 mg N at P0 and P30, respectively). The much stronger decrease in concentrations of inline image in soil of solitary seedlings was closely related to the seedling biomass production. Pre-emption by the ERM could partly explain the low inline image concentrations in the CMN treatment as compared with the much higher inline image concentrations in the solitary treatment. However, the maintenance of 11–13 μg inline image g−1 in soil from 17-d-old CMN seedlings suggests that mass flow transport of inline image to the ‘donor’ root compartment was in part replaced by high nitrification activity. This explanation is supported by the observation that soil of solitary seedlings accumulated high inline image concentrations, which may have been derived from a combination of high mineralization/nitrification rates and disruption of mass flow to the ‘donor’ plant sink. Interestingly, this disruption also led to inline image accumulation in mycorrhiza-free soil, but only at 20–30% of the concentration measured in soil from the corresponding solitary mycorrhizal seedlings (24-d harvest). This suggests that the AMF mycelium provided a substrate to stimulate N mineralization, an intriguing finding that suggests a possible key process underlying the reported capacity of AMF to increase N uptake from organic matter patches in soil (Hodge et al., 2001; Leigh et al., 2009). This needs further investigation. A safe conclusion regarding the direct role of ERM in soil N transformation is confounded by the possible difference in composition of the general soil microflora between AMF-inoculated and noninoculated soil. Autotrophic nitrifiers are sensitive to partial soil sterilization (Jakobsen & Andersen, 1982) and introduction of nitrifiers with the inoculum could have been responsible for the increased nitrification rates in soil with mycorrhizal as compared with nonmycorrhizal seedlings. This could also have contributed to the lower concentrations of inline image in soil from mycorrhizal than from nonmycorrhizal plants. It appears that suppression of seedling growth was primarily caused by initial low soil P concentrations and subsequent pre-emption of P (and to some degree N) by the ERM of the CMN. Mycorrhizal colonization did not help to mitigate the suppression of the seedlings, as these had poor access to nutrients in the ERM.

Conclusions

Our simple CMN model systems represent an extreme size asymmetry between two plant individuals and demonstrate the need to qualify the widespread perception that AM networks exert a general positive effect on growth of individuals in a plant community. The AM benefit appears to be life stage dependent and in low P soil the CMN-interlinking seedling first becomes P starved and growth suppressed. Accordingly, mycorrhizal networks do not relax but rather amplify competition between seedling and adult plant and thereby increase inequality between individuals. However, there is little doubt that the CMN will improve seedling performance in the longer term: larger neighbor plants will be grazed or become senescent and P in the CMN will become available to the seedling individual. The CMN connection could thus be considered as insurance for longer term nutrient availability and thereby sustained growth of the seedling.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank Bente Andersen and Mette Flodgaard for technical assistance. We also thank three anonymous referees for their valuable comments on a previous version. This work was supported by The Danish Council for Independent Research | Technology and Production Sciences, grants 09-061126 and 10-082459.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Please note: Wiley Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

FilenameFormatSizeDescription
nph12351-sup-0001-FigS1-S3_TableS1-S4.pdfapplication/PDF330K

Fig. S1 Effect on tomato seedlings of cutting of cucumber shoots in Experiment 1.

Fig. S2 Rhizobox set-up in Experiment 2.

Fig. S3 Root colonization of tomato seedlings in Experiment 2.

Table S1 Statistical details for the time course study in Experiment 1

Table S2 Statistical details for the effects of additional P and absence of AMF inoculum at 25 d in Experiment 1

Table S3 Statistical details for the time course study in Experiment 2

Table S4 Statistical details for mycorrhiza effects at 24 d in Experiment 2