The influences of the arbuscular mycorrhizal fungus Glomus intraradices on the culturable aerobic–heterotrophic bacterial communities in the rhizosphere and hyphosphere of cucumber plants (Cucumis satvius) were investigated. Mycorrhizal and non-mycorrhizal plants were grown in compartmentalised growth units, each with a root compartment and two lateral root-free compartments. Samples representing rhizosphere, root-free soil, root-free sand and washed sand extract were collected 52 days after sowing from treatments both with and without mycorrhiza. No significant difference in total bacterial number was observed between the mycorrhizal and non-mycorrhizal treatment. Fourteen hundred bacterial colonies were isolated and identified by fatty acid methyl ester analysis using the Sherlock system (MIDI Inc.), 87 species within 48 genera were identified with a similarity index >0.30. Pseudomonas, Arthrobacter and Burkholderia were the genera most frequently encountered. Large differences in bacterial community structure were observed between rhizosphere soil, root-free soil/sand and washed sand extract, whereas major differences between mycorrhizal and non-mycorrhizal treatments were observed for a few bacterial species only. Isolates identified as Paenibacillus spp. were more frequently found in the mycorrhizal treatment and especially in the washed sand extract with mycelium of G. intraradices, indicating that bacteria within this genus may live in close association with mycelium of these fungi.
Fatty acid methyl ester
Plant growth and health are strongly influenced by micro-organisms colonising roots and inhabiting the rhizosphere. Arbuscular mycorrhizal (AM) fungi inhabit both plant roots and the surrounding soil and it is well known that these fungi can enhance plant growth and suppress plant diseases . In order to increase our understanding of the role of AM fungi in plant growth and health and make use of their beneficial features in plant production, we need to know more about their interactions with other micro-organisms. The external phase of AM fungi plays a key role in the transport of phosphorus from soil to host plant , in soil aggregation  and perhaps also in the suppression of root pathogens . Bacteria may influence these functions of the external mycelium of AM fungi.
Despite the obligate biotrophic nature of AM fungi, organic matter has been shown to increase growth of their external mycelium, and this growth enhancement has been suggested to be related to bacterial activities [5,6]. Bacteria have been found adhering to the AM hyphae  as well as embedded within the AM spore walls . Bacteria adhering to the AM mycelium may feed on hyphal exudates and/or use the mycelium as a vehicle for colonisation of the rhizosphere . Bacteria from the genus Paenibacillus, which are antagonistic to a broad range of root pathogens and able to stimulate mycorrhiza colonisation, have been isolated from the mycorrhizosphere of sorghum .
Several studies have shown that rhizosphere bacteria are affected by the presence of AM, and negative as well as positive effects of AM on bacteria have been reported. Meyer and Linderman  observed no effect of AM on the total number of bacteria but saw a shift in specific groups of bacteria in the rhizosphere of mycorrhizal plants towards more facultative anaerobic bacteria and less fluorescent pseudomonads. Later studies found an increase in total bacterial population, nitrogen fixers and Gram-negative bacteria in the rhizosphere of mycorrhizal plants . More recently, Ravnskov and Jakobsen  reported an increase in the total number of bacteria in both rhizosphere and hyphosphere by the AM fungus Glomus intraradices. Bacterial activity has likewise been reported to both increase  and decrease  in the rhizosphere of arbuscular and ectomycorrhizal plants, respectively, compared to a non-mycorrhizal control, using the 3H-thymidine incorporation method. Bacterial communities have also been studied using specific phospholipid fatty acids, but here no effects of AM were found .
Few studies have focused on the qualitative influence of mycorrhiza on the bacterial communities in the rhizosphere and these have mainly been based on metabolic functioning [10,11]. However, recently Andrade et al.  used fatty acid methyl ester (FAME) analysis for identification of a relatively small number of bacteria isolated from rhizosphere and hyphosphere of mycorrhizal and non-mycorrhizal plants. The results of Andrade et al.  indicated that the composition of the bacterial population in the rhizosphere and hyphosphere was strongly affected by the presence of AM as well as of the particular species and isolates of AM fungi used.
FAME analysis has been used in a number of studies to characterise the diversity of the rhizosphere microbial communities through identification of large numbers of isolated bacteria [17–19]. To our knowledge, the present study is the first in which this approach has been employed to characterise the bacterial communities of mycorrhizal plants. The main objective of the study was to characterise bacterial communities associated with the external mycelium of the AM fungus G. intraradices. A compartmentalised growth system with root-free compartments allowed us to study the hyphosphere without the interfering effects from the roots.
2 Materials and methods
2.1 Experimental design
Mycorrhizal and non-mycorrhizal cucumber plants were grown in compartmentalised growth units (Fig. 1) constructed from 50 mm polyvinyl chloride tubes with an internal diameter of 4.5 cm. The growth units consisted of a 35 cm long vertical central root compartment separated from two 7 cm long root-free lateral compartments by a 38 μm pore size nylon mesh. The two treatments (AM and non-AM) had five replicates giving a total of 10 plants.
Design of the compartmentalised growth system with a central root compartment and two lateral root-free compartments separated by 38 μm nylon mesh. Bacteria were isolated from four sample fractions: (1) rhizosphere, (2) root-free soil, (3) root-free sand and (4) washed sand extract from the root-free sand compartment. These fractions were sampled from both mycorrhizal and non-mycorrhizal treatments.
2.2 Soil and biological materials
The plants were grown in a soil consisting of a 1:3 (w/w) mixture of a sandy loam and quartz sand with a pH of 6.1, which was partially sterilised by irradiation (10 kGy) to eliminate the indigenous mycorrhizal fungi, so a known isolate could be introduced and studied. Inoculum of the AM fungus G. intraradices Schenck and Smith (isolate BEG87) consisting of soil:sand (1:1 w/w), root fragments, spores and mycelium was obtained from a 6-week-old cucumber pot culture. Non-AM inoculum was obtained from a similar but non-mycorrhizal cucumber pot culture and added to the control plants. The reasoning for using both AM and non-AM soil inoculum was to try to create identical growth media with respect to the biological, chemical and physical conditions, in both AM and non-AM treatments, except for the introduced AM fungus. Basal nutrients (mg kg−1) were mixed into the soil in the following amounts: K2SO2 (75), CaCl2·2H2O (75), CuSO4·5H2O (2.1), ZnSO4·7H2O (5.4), MnSO4·7H2O (10,5), CoSO4·7H2O (0.39), Na2MoO4·2H2O (0.18) and MgSO4·H2O (45). The central root compartment contained a bottom, middle and top soil layer where the bottom and top layer consisted of 300 g and 130 g soil respectively. The middle layer consisted of 250 g soil mixed with either 80 g AM or non-AM inoculum. One lateral compartment of each unit received 60 g of soil, whereas the other lateral compartment received 60 g of quartz sand to facilitate extraction of external mycelium of G. intraradices from the AM treatment. In order to promote establishment of a similar initial microflora in the two treatments, all units received 50 ml of a soil filtrate obtained by suspending 100 g AM inoculum, 100 g non-AM inoculum and 100 g fresh field soil in 1000 ml water and filtering through a 38 μm mesh, which allowed passage of common soil microbes, but retained spores of indigenous mycorrhizal fungi. The growth units were watered to 65% of water holding capacity and placed at room temperature for 4 days. Finally, two pre-germinated seeds of cucumber (Cucumis sativus, c.v. Aminex, Novartis Seeds, Denmark) were sown in the central compartment of each growth unit and thinned to one after seedling emergence and moved to a growth chamber.
2.3 Growth conditions
The growth units were randomly positioned in a growth chamber with day/night temperature of 21/16°C and provided with supplementary light equivalent to 250 μmol m−2 s−1 of photosynthetically active radiation for 16 h per day. The units were repositioned and watered daily to 65% of water holding capacity. The plants were supplied with 10 mg N (as NH4NO3) 10, 14, 18 and 22 days after sowing and with 20 mg N 26, 30, 34 and 38 days after sowing, in total 120 mg N.
2.4 Harvest and sampling
Plants were harvested 52 days after sowing. The content of the root-free sand compartment was removed and thoroughly mixed in a plastic bag. A subsample of 4 g sand wet weight was suspended in 100 ml sterile phosphate buffer solution (PBS, pH 7.2) and vortex mixed for 5 min. After 30 s, 10 ml of the suspension were transferred to a centrifuge tube as representing the root-free sand fraction. The remaining part of the sand compartment (approximately 75 g wet weight) was placed in a 500 ml flask and washed by repeatedly adding 100 ml sterile PBS followed by gentle shaking and wet sieving of the supernatant through a sterile 38 μm nylon mesh. This step was repeated five times. The mesh was finally washed with four times 50 ml of sterile PBS before being transferred to a 50 ml centrifuge tube in 10 ml of sterile PBS. Five ceramic mill balls were added and the tube was vortex mixed for 5 min. The resulting suspension represented the washed sand extract fraction, which was expected to contain AM mycelium in the AM treatment and no mycelium in the non-AM treatment. The content of the root-free soil compartment was removed and thoroughly mixed in a plastic bag. A subsample of 4 g was suspended in 100 ml sterile PBS, vortex mixed for 4 min and then considered to be the root-free soil fraction sample. The content of the central root compartment was removed and loose (non-rhizosphere) soil was shaken off the roots. The remaining soil on the roots, a layer of 2–3 mm in thickness, was released by vortex mixing of 10 g of root fragments in 100 ml of sterile PBS buffer. After sedimentation for 30 s, 10 ml of the suspension was transferred to a centrifuge tube as representing the rhizosphere soil fraction. The remaining part of the suspension was wet sieved (mesh size 100 μm) for root fragments, the roots thoroughly washed, dried with tissue paper and weighed. The root samples were stored at −20°C for later determination of mycorrhiza colonisation.
2.5 Dilution plating, enumeration and isolation
Aliquots (25 µl) of three 10-fold dilutions of each fraction suspension were plated on 1/10 strength TSBA (Sigma) plates for enumerating total bacteria. Plates were incubated at 21°C and colony forming units were counted after 2 days. Random samples of 35 colonies were picked from the TSBA plates from each sample fraction and pure cultured on full strength TSBA plates. The pure cultures were transferred to cryo tubes containing 500 μl of a peptone–glycerol medium (per 1000 ml: 150.0 ml glycerol, 20.0 g peptone, 1.5 g K2HPO4; pH 7.2) and stored at −80°C until further processing.
2.6 FAME analysis of isolates
FAMEs were extracted from each isolate using standard and recommended procedures for gas chromatographic FAME analysis as described by Sasser . Analysis was performed with a Agilent 6890 Plus Chromatograph and the Sherlock system software version 3.1 using the method TSBA41 and the library TSBA41 (MIDI Inc, Delaware, USA). A similarity Index threshold of 0.30 was applied for acceptance of identifications as recommended by the MIDI manual. Isolates with a similarity index below 0.3 were considered ‘no match’ and grouped together. A minimum separation to second best match was not adopted. Bacterial diversity indices in the non-AM and the AM treatments of the different fractions were measured in terms of the Shannon index as described in Magurran .
2.7 Plant growth and mycorrhiza development
Shoots were dried (80°C for 20 h) and weighed. Roots were washed free of the soil and cut in to 5–10 mm segments. Root samples (0.5 g) were cleared and stained as described by Kormanick and McGraw  except that trypan blue was used instead of acid fuchsin. The stained roots were analysed for mycorrhiza colonisation using the line-intercept method as described by Giovanetti and Mosse . Just before harvesting mycelial colonisation of the root-free compartments was checked by examining the soil/sand surface at the distal end of the root-free compartments using a compound microscope.
2.8 Statistical analysis
Data on the total bacterial population were subjected to two-way analysis of variance. In the data from the fatty acid analysis the observed values were normalised to give equal numbers of isolates analysed for each treatment per fraction. The two treatments within each fraction were then compared using CHI square test. The fractions were compared pairwise by the same procedure. SAS version 6.12 was used to perform the described statistics. Test for significant difference between Shannon diversity indices of non-AM and AM treatments was done by calculating ‘t’ as described by Hutcheson .
Plants from the non-AM and the AM treatments developed similarly. The average shoot dry weight of all plants was 2.65 g. Root colonisation of G. intraradices-inoculated plants was in the range of 49–60% with a mean of 63% whereas roots of uninoculated plants remained non-mycorrhizal. Similar observations were made in the root-free compartments of the inoculated plants, which were extensively permeated by mycelium, whilst no mycelium was observed in the treatment without mycorrhiza. This was confirmed by visually inspecting the washed sand extract fractions where AM mycelium was recovered in the sand washing of the AM treatment but not from the non-AM treatment.
The total number of culturable bacteria was unaffected by inoculation with G. intraradices both in the rhizosphere and the root-free soil/sand fractions. However, markedly more bacteria were found in the washed sand extract fraction of the mycorrhizal treatment (Fig. 2).
Total number of culturable aerobic heterotrophic bacteria on tryptic soy agar in the rhizosphere, root-free soil, root-free sand and washed sand extract from the root-free sand compartment in treatments with (black columns) and without mycorrhiza (grey columns). Error bars represent standard errors of the different means.
Fourteen hundred colonies were isolated of which 1367 were successfully analysed by FAME analysis. In total, 87 species were identified of which 29 were represented by a single isolate only (data not shown). The six most frequently identified species were the Gram-negative species Burkholderia glathei, Pseudomonas putida, Pseudomonas chlororaphis, Pseudomonas fluorescens, Variovorax paradoxus and Chromobacterium violaceum which comprised from 33 to 71% of identified isolates in the different fractions. Between 25 and 36 species were identified in each of the eight different fractions. More Gram-positive species and fewer Gram-negative species were identified in the rhizosphere soil and hyphosphere soil fractions compared to the hyphosphere sand and the washed sand extract fractions (Table 1). While virtually absent in the non-rhizosphere fractions, the species B. glathei was the most frequently identified species in the rhizosphere fraction of both AM and non-AM treatment, where it constituted approximately 30% of the isolates identified. Numbers of B. glathei were unaffected by mycorrhiza in all fractions. The species P. putida, P. chlororaphis and P. fluorescens made up a large part of isolates identified in the root-free soil/sand fractions in both treatments with and without mycorrhiza (total 30–60%). The three species were found in much lower number in the rhizosphere fractions (total 10–15%). The influence of G. intraradices on these pseudomonads varied between species and fractions. In terms of numbers, P. putida was unaffected, whereas P. chlororaphis was increased by mycorrhiza. The increase of P. chlororaphis was mainly observed in the root-free soil and the washed sand extract fraction. Similarly, numbers of P. fluorescens were increased in the root-free soil fraction by mycorrhiza. However, fewer isolates of P. fluorescens were recovered from the root-free sand compartments in the presence of mycorrhiza mycelium. Though found in relatively low numbers the genus Paenibacillus was more frequently encountered in the washed sand extract of the AM treatment than in any other fraction. No Paenibacillus was identified in the washed sand extract fraction of the non-AM treatment. The genus Arthrobacter, which is very similar in fatty acid contents to Paenibacillus, was found in all fractions though more frequently in the washed sand extract fraction of the AM treatment than in any other fraction. The species Agrobacterium radiobacter was almost exclusively found in the washed sand extract and root-free fractions of the AM treatment. All six isolates of A. radiobacter in the washed sand extract fraction were however found in the same replicate (Table 1). Bacterial diversity in terms of both species richness (number of bacterial species) and Shannon diversity index seemed to be unaffected by mycorrhiza in all fractions, except for the washed sand extract fraction where the AM treatment had a significant higher diversity index than the non-AM treatment (Fig. 3).
Number of bacterial strains per species isolated from the fractions (1) washed sand extract from the root-free sand, (2) root-free sand, (3) root-free soil and (4) rhizosphere of cucumber plants inoculated with G. intraradices (+AM) or uninoculated (−AM)
Bacterial diversity in terms of species richness (A) and Shannon index (B) of culturable aerobic heterotrophic bacteria recovered on tryptic soy agar from the rhizosphere, root-free soil, root-free sand and mycelium extracted from the root-free sand compartments in treatments with (black columns) and without mycorrhiza (grey columns). Results of statistical tests comparing AM and non-AM results are given above columns in panel B; ns=no significant difference, **=significant difference at P≤0.01.
In the present study a compartmentalised growth system with root-free compartments allowed us, for the first time, to perform a qualitative characterisation of the bacterial communities in a hyphosphere of an AM fungus with no direct interference from roots. Furthermore, extracting the external mycelium of G. intraradices from the sand containing root-free compartments made it possible to characterise bacteria closely associated with G. intraradices mycelium. In general, the influence of G. intraradices on the frequency of the different bacterial species isolated from the rhizosphere was limited, whereas mycelium of G. intraradices either reduced or increased the population density of 14 bacteria species. Bacteria from the genus Paenibacillus were almost exclusively found associated with external mycelium of G. intraradices. Whether these bacteria are living in the proximity of the mycelium, on the surface of the mycelium or inside the mycelium cannot be clarified from the present study. Two species from the genus Paenibacillus were isolated from both the root-free soil/sand and the washed sand extract fraction and were identified as Paenibacillus macerans and Paenibacillus polymyxa. These two species have previously been isolated from the rhizosphere of various plants [25,26]. Furthermore, they have been suggested to be involved in soil aggregation , plant growth promotion  and suppression of pathogens . Other Paenibacillus species have been found to be associated with both arbuscular and ectomycorrhizal fungi [9,30]. Paenibacillus amylolyticus was isolated from root tips of Pinus sylvestris–Lactarius rufus ectomycorrhizas and were found to stimulate mycorrhiza formation . Similarly, another Paenibacillus isolate, which stimulated mycorrhiza formation, was isolated from the mycorrhizosphere of a Sorghum bicolor–Glomus mosseae symbiosis .
Our findings that G. intraradices—cucumber symbiosis has little effect on the bacterial community in the rhizosphere are in contrast to Andrade et al.  who found qualitative changes in bacterial communities as affected by different AM fungi in both the rhizosphere and the hyphosphere. In their study, however, the hyphosphere was simply defined as soil not adhering to roots, which is different from our root-free hyphosphere fraction, thus actual comparisons between the two experiments are difficult to make.
The subtle qualitative differences between the communities in the AM and non-AM treatments observed in the present experiment contradict the findings of Meyer and Linderman  as well as the findings of Andrade et al. , who reported a profound impact of AM on the composition of the bacterial communities in the rhizosphere as well as the hyphosphere. The study of Andrade et al.  however, employed colony morphological characters for determining taxonomic relationships, which may have biased the results .
The bacterial community structure found in the present study is based on the heterotrophic aerobe bacteria that are culturable on TSA. This medium was chosen as a conventional medium for the isolation of bacteria. However, in future studies we plan to include other media which perhaps will add new species to the list of bacteria presumed to be associated with arbuscular mycorrhiza.
The reliability of the MIDI system for identification of soil bacteria has been questioned as, e.g. identification of Ralstonia solanacearum based on MIDI was not confirmed using PCR . Probably too many isolates are identified as plant pathogenic or clinical species as these are overly represented in the database compared to environmental species. However, in the present experiment we chose this method, which allowed us to include a higher number of isolates than would have been possible using most other identification methods.
In conclusion, the present Glomus–cucumber symbiosis seems to have little influence on bacterial diversity, but the mycelium of G. intraradices may alter the population density of different bacteria at the species level. In particular isolates of the spore forming Gram-positive bacteria P. macerans and P. polymyxa were found to be closely associated with the external mycelium of the AM fungus G. intraradices.
Tina Tønnersen is thanked for excellent technical assistance.
(1994) Hyphal phosphorus transport — a keystone to mycorrhizal enhancement of plant growth In: Impact of Arbuscular Mycorrhizas on Sustainable Agriculture and Natural Ecosystems (GianinazziS., SchueppH. Eds.), pp. 133–146BirkhauserBasel
(1997) Inhibition of Fusarium oxysporum f.sp. dianthi in non-VAM species Dianthus caryophyllus by co-culture with Tagetes patula companion plants colonized by Glomus intraradices. Can. J. Bot. 75, 998–1005.
(1999) Isolation from the Sorghum bicolor mycorrhizosphere of a bacterium compatible with arbuscular mycorrhiza development and antagonistic towards soilborne fungal pathogens. Appl. Environ. Microbiol. 65, 5148–5150.
(1997) Bacterial communities of the rhizosphere and endorhiza associated with field-grown cucumber plants inoculated with a plant growth-promoting rhizobacterium or its genetically modified derivative. Can. J. Microbiol. 43, 344–353.
(1993) Differentiation of Pseudomonas solanacearum, Pseudomonas syzygii, Pseudomonas picketti and the blood disease bacterium by partial 16S-ribosomal-RNA sequencing construction of oligonucleotide primers for sensitive detection of polymerase chain reaction. J. Gen. Microbiol. 139, 1587–1594.