Structure and activity of the bacterial community in the rhizosphere of different plant species and the effect of arbuscular mycorrhizal colonisation


*Corresponding author. Tel: +46 (46) 222 3760; Fax: +46 (46) 222 4158.


The aim of this study was to determine if arbuscular mycorrhizal (AM) colonisation influences rhizosphere bacteria differently depending on plant species. Thus, the effect of AM colonisation (Glomus intraradices) on rhizosphere bacteria of subterranean clover, cucumber, leek and maize was studied. The bacterial activity was measured as thymidine or leucine incorporation and bacterial numbers as colony forming units and acridine orange direct counts. The phospholipid fatty acid (PLFA) pattern was used to characterise the bacterial community structure and was compared with the community substrate utilisation pattern using Biolog. The bacterial activity and bacterial densities differed between the rhizospheres of the plant species. AM colonisation had a low impact on bacterial activity, but affected bacterial numbers differently depending on the plant species. Only small effects of AM colonisation were detected with the PLFA technique, and no effects were seen with Biolog, while similar effects of the plant species were found for these techniques. Thus, the plant species had greater effects on the bacterial community in the rhizosphere than AM colonisation and the effect of AM differed between plant species.


Microbial growth in soil is usually considered to be limited by the availability of carbon [1–2]. The rhizodeposition of easily available carbon therefore makes the rhizosphere an area of high microbial activity [3–5]. The composition and amount of microorganisms present in the rhizospheres of different plants may differ due to variations in the quantity and quality of compounds exuded by the different plants [6–7]. Consequently, different rhizosphere microbial communities are associated with different plants (e.g. [8–9]). The colonisation of plant roots by arbuscular mycorrhizal (AM) fungi can also affect root exudation as AM formation may alter the root physiology, including the root membrane permeability [3,10–12]. Hence, the microbial community in the rhizosphere may be affected by AM formation.

The extent of the competition between rhizosphere bacteria and AM fungi for root-derived carbon may differ between plants and AM fungal species. Most studies have focused on how different AM fungal species affect the bacterial community in the rhizosphere of a single plant species [13–19], and Marschner [20] concluded that different AM fungal species alter the composition of the bacterial community of the rhizosphere soil in different manners. Little attention has, however, been given to how the interaction between plant species and mycorrhizal status affects the rhizosphere bacterial community. Meyer and Linderman [21], however, investigated how the AM fungus Glomus fasciculatum influenced the bacterial numbers in the rhizosphere of two different plant species. They found a decrease in the bacterial density (colony forming units, cfu) in the rhizosphere of both clover and maize following mycorrhizal colonisation.

We have, in this study, investigated the interactions between AM fungi and roots from different plant species with a combination of old and new techniques for the characterisation of the bacterial community. Techniques such as plate counts of cfu and microscopic direct counts of bacteria have been extensively used to quantify bacteria in rhizosphere soil. One variable that has proven more difficult to assess in the rhizosphere is the bacterial growth rate. The thymidine and leucine incorporation techniques have, however, been used successfully to estimate growth rates in soil bacterial communities [22–23], including rhizosphere bacterial communities [4,24–26].

Another variable that has been difficult to assess is community composition. However, the analysis of ester-linked phospholipid fatty acid (PLFA) patterns has been shown to be a useful tool for studying interacting microorganisms, such as mycorrhizal fungi and bacteria in the rhizosphere [27]. In the present study bacteria-specific PLFAs were used to study the biomass and composition of the bacterial community [28]. Two other methods have also been used to characterise the bacterial community, one extracting lipids for PLFA analysis from only cultivated bacteria (cfu-PLFA) [29–30] and one employing the analysis of the substrate utilisation patterns of the bacterial community using Biolog GN plates [31]. The PLFA and the Biolog techniques have been used previously in rhizosphere studies [32–37].

Our aim was to study the bacterial communities of four different plants, colonised by one strain of the AM fungus Glomus intraradices. Communities associated with colonised and non-colonised plants were compared with regard to overall bacterial activity, numbers and community structure in the rhizosphere. We aimed at testing the hypothesis that the effect of mycorrhizal status on rhizosphere bacteria depends on the plant species. This would be detected statistically by a significant interaction between mycorrhizal status and plant species. The chosen plants (maize, cucumber, subterranean clover and leek) are all extensively colonised by AM, but differ in their growth response due to AM colonisation.

2Materials and methods

2.1Soil and plant growth conditions

The experiment was performed twice, using the same set-up. The soil was a mixture of moraine clay loam and river sand (1:1). The clay loam was a top soil from Tåstrup, Denmark, with properties as described in [38]. The soil mixture was irradiated (10 kGy, 10 MeV electron beam) to eliminate indigenous AM fungi. The soil had been sieved through a 4-mm sieve. The following nutrients were mixed into the irradiated soil (pH(H20)=6.1) (mg kg−1 dry soil): K2SO4 (70), CaCl2 (70), MgSO4·7H2O (20), MnSO4·7H2O (10), CuSO4·5H2O (2.2), ZnSO4·7H2O (5), CoSO4·7H2O (0.33), NaMoO4·2H2O (0.2) and KH2PO4 (44).

Maize (Zea mays L.) and cucumber (Cucumis sativus L.) were grown for 6 weeks while subterranean clover (Trifolium subterraneum L.) and leek (Allium porrum L.) were grown for 7 weeks. Seeds were sown in pots, and 5 days later the seedlings were thinned to one or three seedlings per pot, depending on plant species. The pots were filled with 300 g soil for leek (three seedlings per pot) and clover (three seedlings per pot in experiment 1 and one in experiment 2) and 500 g per pot for cucumber and maize (one seedling per pot). Different amounts of nitrogen was given to different plants adjusted after earlier experience in order to avoid nitrogen deficiency in plants. In all cases the nitrogen given was much lower than levels which have a direct effect on bacterial activity and growth (Söderberg and Bååth, unpublished). The cucumber and maize seedlings were supplied with 30 mg N (as NH4NO3) on four different occasions during the growth period in the two experiments. The clover seedlings were supplied with a total of 120 mg N (as NH4NO3) in the first experiment (three plants per pot) and with 40 mg N in the second experiment (one plant per pot) to assure the same amount of added nitrogen per plant in each experiment. The plants were inoculated with the AM fungus G. intraradices Schenk and Smith (isolate BEG 87) in half the number of pots by adding soil and roots from a pot culture underneath the seed. For each plant species and mycorrhizal treatment, four replicate pots were used. The plants were grown in a greenhouse with a 16/8 h light/dark cycle at 22/20°C with Osram daylight lamps providing a photosynthetic photon flux density of approximately 150 μmol m−2 s−1. The pots were watered to 60% of water-holding capacity every other day.

2.2Harvesting and extraction of bacteria

Plants were harvested 6 (cucumber and maize) or 7 (clover and leek) weeks after sowing and the shoots were dried (60°C for 24 h) and weighed. Root colonisation by AM fungi was measured as described by [39]. Extraction of bacteria was performed as described by [4]. Briefly, the contents of the pots were gently shaken out onto a plastic sheet. Rhizosphere soil was defined as soil attached to the roots. Small root samples (approximately 1 g) were taken from the whole root system of the plants in each pot and bulked into one composite sample. The soil in the rhizosphere was then removed from the root samples by allowing the roots to soak in 20 ml of distilled water. Four to six root samples were required to obtain 1 g of rhizosphere soil. One gram of soil was homogenised with 20 ml distilled water in a knife blender (Bühlermixer) at 70% of maximum speed for 30 s. The soil suspension was then centrifuged at 1000×g for 10 min at 5°C, and the supernatant containing the bacterial suspension was collected for analysis.

2.3[3H]Thymidine and [14C]leucine incorporation

Leucine incorporation was measured on 2 ml of the bacterial suspension at the same time as thymidine incorporation using a dual-labelling approach as described by [4]. Methyl[3H]thymidine (Amersham, 25 Ci mmol−1, 925 GBq mmol−1) and [14C]leucine (Amersham, 59 mCi mmol−1, 218 GBq mmol−1) were added to the bacterial suspension. The final concentration of thymidine was 100 nM and of leucine 387 nM. Incorporation at room temperature (22°C) was stopped after 2 h.

2.4Bacterial enumeration

Bacterial cfu were determined on 0.2% tryptic soy agar plates, incubated for 14 days at 20°C. Acridine orange direct counts (AODC) of total numbers of bacteria were measured as described earlier [4].

2.5Analysis of PLFAs and substrate utilisation patterns

The method of lipid extraction followed that of [40], which is a modification of the Bligh and Dyer procedure [41]. Lipids were extracted directly from washed root samples that had been freeze-dried and grinded (direct-PLFA) or from cultivated bacteria (cfu-PLFA) for PLFA analysis. In the latter case, agar plates which had been incubated for 3 weeks at 22°C with rhizosphere soil solution were used. Agar plates on which the number of cfu was high (about 100–300) were flooded with citrate buffer, and a 1.5-ml portion of the bacterial suspension was collected for phospholipid extraction. Extracted lipids were fractionated into neutral lipids, glycolipids and polar lipids (mainly phospholipids) on silicic acid (100–200 mesh, Unisil) columns by eluting with chloroform, acetone and methanol, in that order. The phospholipids were subjected to mild alkaline methanolysis to transform the lipid fatty acids into free fatty acid methyl esters. These were analysed on a Hewlett Packard 5890 gas chromatograph with a flame ionisation detector and a 50 m HP5 capillary column. The following PLFAs were considered to be of mainly bacterial origin: i15:0, a15:0, i16:0, a17:0, cy17:0 and cy19:0 [26,28], and the sum of these was used as a measure of bacterial biomass. Changes in bacterial community structure in the soils were examined using principal component analysis (PCA) of the six bacteria-specific PLFAs (relative amounts). The fatty acid 16:1ω5 was used as indicator of the AM fungal phospholipids [39,42].

The substrate utilisation patterns of the bacterial community were examined using Biolog GN plates [31]. The microtitre plates were incubated with 150 μl of a bacterial suspension in each well containing around 105 cells ml−1. The plates were incubated for 5 days at 20°C and colour development was measured using a microtitre plate reader at 550 nm.

2.6Statistical methods

Results are given as the means and the standard error of the mean (S.E.M.). The effects of mycorrhizal treatment and plant species were tested with ANOVA (F-ratio) on each experiment separately, except when differences between experiments were explicity tested. In that case, experiments were included as blocks in the ANOVA. The bacterial PLFA composition and the substrate utilisation pattern were analysed using PCA with the computer program SIRIUS (Pattern Recognition Systems, Bergen, Norway). For the PLFA data, concentrations (expressed as mol% of total amount of PLFAs) of all the individual PLFAs were subjected to PCA. For the Biolog analysis, the optical density of the bacterial suspension in each well was divided by the mean of the values for all wells on the plate. The resulting values were standardised by dividing by the standard deviation for each substrate, to achieve unit variance before being subjected to PCA.


3.1Shoot biomass and mycorrhizal colonisation

Different plant species showed different growth responses when inoculated with G. intraradices (Table 1), indicated by significant (P<0.001) statistical interaction between plant species and mycorrhizal colonisation. A positive effect of AM colonisation on plant growth was found for clover and leek, while no effect was observed for cucumber and a negative effect was found on maize in the second experiment (Fig. 1A).

Table 1.  Shoot weight, shoot weight/root weight (only experiment 2), cfu and AODC of bacteria from the rhizospheres and bacterial activity measured as thymidine (TdR) and leucine (Leu) incorporation in the two experiments
  1. All data represent means±S.E.M. (n=4).

  2. aStatistical analysis performed on data for clover, cucumber and maize, leek not included.

Plant speciesMycorrhizal statusShoot weight (g)Shoot weight/root weight (g g−1)cfu (×107) (cells g−1 soil)AODC (×108) (cells g−1 soil)TdR (×10−13)a(mol ml−1 h−1)Leu (×10−11)a(mol ml−1 h−1)
  Exp. 1Exp. 2Exp. 2Exp. 1Exp. 2Exp. 1Exp. 2Exp. 1Exp. 2Exp. 1Exp. 2
ANOVA, P-value
Species <0.001<0.001<0.001<0.0010.267<0.001<0.001<0.05<0.001<0.001<0.001
Mycorrhiza <0.01<0.01<0.001<0.0010.156<0.010.0980.7840.2200.3970.188
Species×mycorrhiza <0.01<0.001<0.001<0.0010.064<0.01<0.05<0.0010.191<0.0010.427
Figure 1.

Ratio between AM and non-mycorrhizal (NM) plants with regard to (A) shoot weight, (B) bacterial cfu, (C) leucine (Leu) incorporation and (D) thymidine (TdR) incorporation. Unshaded bars represent the first experiment and hatched bars the second experiment. The thin line indicates the no effect level of AM (ratio AM/NM=1).

The ratio of shoot weight to root weight in experiment 2 (not measured in experiment 1) was higher for mycorrhizal maize and clover than in the corresponding non-mycorrhizal plants. For leek the opposite was true and for cucumber no effect was found (Table 1).

AM colonisation was between 40 and 75% of the root length for clover, cucumber and maize in both experiments (Fig. 2A). AM colonisation in leek roots was not estimated microscopically, since all root material had to be used for lipid extraction.

Figure 2.

Different methods of indicating the degree of AM fungal root colonisation. (A) Microscopical estimation of AM fungal root colonisation (Pspec=0.563, Pexp=0.211 and Pspec·exp=0.376), (B) the amount of PLFA 16:1ω5 (Pspec=0.324, Pexp=0.031 and Pspec·exp=0.567). Background values of the fatty acid signature 16:1ω5 in non-colonised plants have been subtracted. AM colonisation in leek roots was not estimated microscopically since all root material had to be used for lipid extraction. Unshaded bars represent the first experiment and hatched bars the second experiment. Error bars indicate the standard error of the mean (n=4).

AM colonisation, as estimated by the amount of PLFA 16:1ω5 in the the roots, was higher in the second experiment (P<0.05 for the different experiments), while no significant differences between plant species were found (Fig. 2B). The level of 16:1ω5 for leek was similar as for the other plant species, indicating that the level of AM colonisation in leek was similar to that found for the other plant species.

3.2Bacterial numbers and activity

Bacterial numbers (cfu) and the activity of bacteria (measured as both thymidine and leucine incorporation rates) were generally twice as high in the rhizosphere as in the bulk soil (data not shown). This comparison between rhizosphere and bulk soil was performed in experiment 1 only.

The highest number of cfu was found in the rhizosphere of cucumber in both experiments, while for AODC, the highest amount was found in the rhizosphere of leek and clover (Table 1). We found a significant statistical interaction (P<0.001) between plant species and mycorrhizal status for bacterial cfu in the rhizosphere for experiment 1, indicating that the presence of AM affected cfu differently depending on the plant species (Table 1). A similar effect was found in experiment 2 with an almost significant interaction (P=0.064). When comparing the number of cfu in the rhizosphere of mycorrhizal and non-mycorrhizal plants, higher amounts of bacteria were found in the rhizosphere of mycorrhizal clover and leek than in non-colonised plants (Fig. 1B) with the exception of leek, experiment 2. In both experiments, the number of cfu tended to be less in the rhizosphere of mycorrhizal maize and cucumber than in the corresponding non-mycorrhizal plants with the exception of cucumber, experiment 1 (Fig. 1B).

AODC was not affected in this way, although in both experiments a significant statistical interaction between plant species and mycorrhizal status was found (Table 1). However, the direction of the mycorrhizal effect for the different plant species differed between the two experiments.

The thymidine and leucine incorporation rates were much lower in the leek rhizosphere than for the rhizospheres of the other plant species. This was later shown to be due to substances leaking from injured leek roots inhibiting thymidine and leucine incorporation (see also Section 4). The bacterial activity values for leek were therefore not included in the statistical analysis to assess the effects of mycorrhizal status. We found significant differences between the different plant species both for thymidine (P<0.05 and P<0.001) and leucine (P<0.001) incorporation rates for both experiments (Table 1). We also found a significant interaction between plant species and mycorrhiza (P<0.001) for thymidine and leucine incorporation in experiment 1 but not in experiment 2 (Table 1). The incorporation rates, both when using the thymidine and the leucine incorporation technique, were higher in the rhizosphere of mycorrhizal clover plants than in non-mycorrhizal ones (P<0.001) (Fig. 1C,D), while no or even negative effects of mycorrhizal colonisation were found for cucumber and maize.

3.3Characterisation of bacterial community structure

Two approaches based on PLFA analysis were used to identify differences in the bacterial community structure; PLFAs extracted directly from bacteria on roots (direct-PLFA) and PLFAs from cultivated bacteria (cfu-PLFA) (Fig. 3A,B). When the data were analysed using PCA, significant (P<0.01) differences were found between the plants for the two methods along the first principal component 1 (PC1) for direct-PLFA and PC2 for cfu-PLFA in both experiments. There was also a consistent difference between plant species for the two methods, in that in both cases clover and leek had more negative values compared to maize and cucumber, which had more positive values. The first component of the cfu-PLFA data mainly differentiated between the two experiments (P<0.001, data not shown), while a mycorrhizal effect was found in PC3 (Fig. 3B, P<0.01 in both experiments). For direct-PLFA no mycorrhizal effect was found.

Figure 3.

PCA ordination of (A) the bacterial PLFA patterns on root (direct-PLFA; PC1 and 2), (B) the bacterial cfu-PLFA patterns (PC2 and 3) and (C) community substrate utilisation pattern using Biolog plates (PC1 and 2). The scores of the different plant species colonised with mycorrhiza are shown as filled symbols and non-mycorrhizal as unfilled. The variation explained by each component is indicated for each axis. Error bars indicate the standard error of the mean (n=4).

The community analysis using Biolog GN plates (experiment 1 only, Fig. 3C) only differentiated between the bacterial communities on roots of different plant species in PC1 (P<0.001), while no effect of the mycorrhizal status of the plants was found. The order in which the plant species appeared in the PCA ordination was similar to that of the two PLFA techniques, with the main differences between clover and leek on the left side, and maize and cucumber on the right side.

The differences between plant species seen in PC1 were due to similar changes for individual PLFAs both for direct-PLFA and cfu-PLFA. This was seen when loading values for individual PLFAs from PC1 for direct-PLFA and PC2 from cfu-PLFA (the main components differentiating between plant species) were plotted against each other (Fig. 4). A significant correlation (r=0.82, P<0.05) was found between the loading values of these two PCs.

Figure 4.

Relationship between loadings of individual PLFAs along the PCs differentiating between plant species (PC1 for the direct-PLFA and PC2 for the cfu-PLFA technique).


We found that different plant species differed in their rhizosphere microflora, with regard to both bacterial activity, as indicated by thymidine and leucine incorporation rate, and bacterial community composition, as indicated by differences found for the PLFA and Biolog measurements. This was expected, since different plant species show different exudation patterns, both quantitatively and qualitatively [7], and several studies have earlier shown that different plants affect the rhizosphere community differently [8–9,35]. More important, we also found that rhizosphere bacteria of different plant species responded differently to AM fungal colonisation (indicated by significant interaction between plant species and mycorrhizal status, Table 1). In general, indicators of bacterial activity (cfu, thymidine and leucine incorporation rates (the latter two are not applicable for leek)) in the rhizosphere increased due to AM colonisation in leek and clover (Fig. 1B–D), the two plant species that responded to AM colonisation with increased growth (Fig. 1A). In the rhizosphere of the other two species, maize and cucumber, the plant growth response was slightly negative or growth was not affected at all by AM colonisation, and the bacterial activity was negatively or not affected.

Thus, the effect of AM fungi on the rhizosphere bacteria varied with the plant species. This indicates that the main AM fungal effect is not caused by a direct interaction between the fungal mycelia of G. intraradices and the bacteria. In that case, similar effects would have been expected, irrespective of plant species, unless the plant could influence the fungal competitive ability. Instead, different plant responses to AM fungal colonisation most probably cause these different effects. The nutrient status of plants influences root exudate composition [43–44]. A plant growing well, photosynthesises more and so may exude a high quantity of carbon. If AM colonisation had a positive effect on the plant growth this might have resulted in increased bacterial activity. On the other hand, when AM colonisation had no or a negative effect on plant growth, this might have resulted in less carbon being exuded from the roots.

Our results are partly contradictory to those of an earlier study by Meyer and Linderman [21], which showed that colonisation by the AM fungus Glomus mossae decreased the bacterial numbers in the rhizosphere of both subterranean clover and maize. We found an increase in bacterial numbers (cfu and AODCs) for clover and unchanged or decreasing numbers of bacteria for maize following mycorrhizal colonisation (Table 1). The differences between our results and those of [21] could, however, not be due to different growth effects of AM fungal inoculation, since Meyer and Linderman [21] found, in accordance with us, an increase in shoot weight for clover when colonised by AM, whilst for maize they found no effect on shoot weight when colonised by AM.

PCA of the cfu-PLFA pattern revealed changes in the bacterial community structure due to the presence of mycorrhizal fungi in PC3 (Fig. 3B), while the effect of plant species was found in PC2. For the direct-PLFA no effect of mycorrhizal inoculation was found, while the most prominent differences were due to plant species. Thus, the effect of different plant species on the microbial community was greater than the effect of AM colonisation. This may be due to a larger qualitative or quantitative difference between exudates from different plant species than between mycorrhizal and non-mycorrhizal plants.

The order in which the plant species appeared in the PCA ordination for the PC that differentiated between plant species was the same for both PLFA community analyses, direct-PLFA (PC1) and cfu-PLFA (PC2), in spite of the fact that the cfu-PLFA technique only measures the part of the bacterial community that is culturable, and direct-PLFA should represent the total bacterial community. Furthermore, loadings for the individual PLFAs along PC1 and PC2 showed a correlation between the direct-PLFA technique and cfu-PLFA technique, indicating that overall the similar results for both techniques were due to the same changes in PLFA patterns (Fig. 4). Hence the cfu-PLFA might be a good indicator of the total bacterial community in the rhizosphere, and yield results similar to those obtained by direct-PFLA. This might not be the case in other environments, although similar results have been found for direct-PLFA and cfu-PLFA following acidification of soil [29].

We also used the Biolog technique, which is a technique that is supposed to indicate functional diversity of the bacterial community [45], to study effects on the rhizosphere bacterial community structure (Fig. 3C). The three methods indicated similar patterns in the distribution of the rhizosphere samples with leek and clover being different from the other plants. However, the Biolog technique appeared to be inferior to the two techniques based on PLFA measurements in differentiating between different treatments. Less of the variation was explained by the first PCs differentiating between plant species compared to direct-PLFA (40.5 and 30.8% for direct-PLFA and Biolog, respectively). Furthermore, the Biolog technique failed to indicate any effect on the bacterial community due to AM, an effect that was indicated for cfu-PLFA (Fig. 3B). The standard deviations of treatment means for the Biolog technique also indicated larger variations compared with the two PLFA-based techniques.

Söderberg and Bååth [4] suggested that the culturable fraction of the bacteria (cfu/AODC) often represents highly active bacteria and subsequently a high incorporation rate of labelled substrates would be expected in this fraction. This was exemplified by the correlation between the cell-specific thymidine incorporation rate and the culturable fraction [4,46]. Such a correlation was also found in the present study, except for leek, where up to 10 times lower bacterial activity than expected from the cfu data was found (Table 1). This indicates that the low thymidine incorporation was an artefact. Separate studies showed that substances released from leek roots, injured during sampling of the rhizosphere soil, inhibited thymidine and leucine incorporation (unpublished results), which thus did not reflect the actual bacterial activity of the rhizosphere community.

We conclude that the impact of plant species on the bacterial community in the rhizosphere was greater than the effect of AM colonisation, as shown by the community analysis methods. We further conclude that the effects of AM colonisation were not due to a direct interaction with the fungal mycelia, since the same AM fungal species gave different effects when growing with different host plants. Neither was there any consistent indication that the carbon demand of the fungus competed with that of the rhizosphere bacteria, as has been suggested earlier [47]. On the contrary, the AM fungal colonisation, leading to a positive growth response of the plant, may positively influence the quality and quantity of root exudates for microbial growth, thus increasing rhizosphere bacterial activity.


This study was supported by grants from the Swedish Council for Forestry and Agricultural Research.