Taxonomic diversity of bacteria associated with the roots of field-grown transgenic Brassica napus cv. Quest, compared to the non-transgenic B. napus cv. Excel and B. rapa cv. Parkland

Authors


*Corresponding author. Tel.: +1 (306) 966-6836; Fax: +1 (306) 966-6881 germida@sask.usask.ca

Abstract

The composition and diversity of the bacterial community associated with plant roots is influenced by a variety of plant factors such as root density and exudation. In turn, these factors are influenced by plant breeding programs. This study assessed the diversity of root-endophytic and rhizosphere bacterial communities associated with three canola cultivars (Parkland, Brassica rapa; Excel, B. napus; and Quest, B. napus) grown at two field sites. Quest, a derivative of Excel developed by the Alberta Wheat Pool, has been genetically engineered to tolerate the herbicide glyphosate. Approximately 2300 bacteria were isolated from roots of plants and identified based on fatty acid methyl ester (FAME) profiles. One third of the isolates were positively identified by FAME analysis (i.e. SIM index ≥0.3) with another third assigned tentative identifications (SIM index <0.3). Fewer Bacillus, Micrococcus and Variovorax isolates, and more Flavobacterium and Pseudomonas isolates were found in the root interior of Quest compared to Excel or Parkland. Furthermore, fewer Arthrobacter and Bacillus isolates were recovered from the rhizosphere of Quest compared to Excel or Parkland. The bacterial root-endophytic community of the transgenic cultivar, Quest, was separated by principal component analysis from the other cultivars, and exhibited a lower diversity compared to Excel or Parkland. The rhizosphere of all cultivars yielded more Arthrobacter, Aureobacterium, and Bacillus isolates, but fewer Micrococcus, Variovorax and Xanthomonas isolates compared to the root interior. The results from this study indicate that the composition of the root-endophytic bacterial community of canola differs between cultivars.

1Introduction

Plants alter the composition and diversity of soil microbial communities in a selective manner [1–3]. For example, the dominant operational taxonomic unit (OTU) in the rhizosphere of the legume, Trifolium repens, matched the theoretical profile of the 16S rRNA gene of Rhizobium leguminosarum. This OTU is found at substantially lower levels in bulk soil, suggesting that T. repens selects for specific bacteria [2]. Non-leguminous plants also exert selective pressure on microbial populations. Flax (Linum usitatissimum L., cv. Opaline) and tomato (Lycopersicon esculentum Mill. cv. H63-5) selectively promote certain fluorescent pseudomonads in their rhizospheres compared to bulk soil [3]. In addition, the selective effect of plants on microbial diversity occurs in field soils [4].

The microbial community resulting from plant selective pressure varies between plant species [4–6]. Rhizoplane communities of wheat and canola grown at the same field site differ in their taxonomic composition [5], and rhizosphere communities of wheat, rye-grass, bent-grass or clover differ in their ability to utilize a variety of carbon substrates [6]. Similarly, Westover et al. [4] found that the structure of rhizosphere communities of field-grown annuals and perennials is dependent upon plant species. These studies suggest that plant types affect which microorganisms colonize their rhizosphere.

Recent works suggest that the selective effect of plants on the rhizosphere community can occur even at the cultivar level [7]. Rengel et al. [7] found that the composition of the root-associated microbial community composition differed between wheat cultivars [7]. Since the organization of the rhizosphere microbial population helps determine the composition of the root-endophytic community [5–8], it is likely that root-endophytic communities are also dependent on cultivar type. Supporting this, our previous study [9] found that the rhizosphere and root-endophytic communities of field-grown canola differed between cultivars as determined by community level physiological profile (CLPP) and fatty acid methyl ester (FAME) analyses. Moreover, the microbial community associated with the non-transgenic Brassica napus cultivar (Excel) is more closely related to a different canola species, B. rapa (Parkland), than a transgenic B. napus cultivar (Quest). The use of CLPP and FAME analyses on soil microbial communities is problematic because it is not clear what fraction of the microbial community responds to CLPP [10,11]. Furthermore, Haack et al. [12] caution that conclusions about community taxonomic composition derived from community level fatty acid profiles should be considered tentative.

It is important to determine which members in a community are lost in response to environmental perturbations, because the identity of displaced organisms can have a significant influence on the dynamics of the resulting community [13]. Symstad et al. [14] found that the magnitude and direction of change in ecosystem functioning as a result of declining diversity depend on the identity of the species deleted and the composition of the community from which the species is deleted. This study assessed the taxonomic diversity of bacteria associated with the roots of transgenic and non-transgenic canola cultivars which previous CLPP and FAME analyses indicated were significantly different from each other [9]. Our purpose was to compare the bacterial community composition and species abundance of root-endophytic and rhizosphere communities, and to identify specific bacterial genera whose abundance was cultivar dependent.

2Materials and methods

2.1Experimental design

Three canola cultivars: Parkland (B. rapa), Excel (B. napus) and Quest (B. napus) were assessed using a randomized complete block design (RCBD) replicated four times at two different sites, i.e. each site had four replicates of all three cultivars. An individual replicate consisted of four randomly chosen plants from one block that were bulked together such that each replicate was a composite of four plants. Plants were harvested at the flowering stage of development. Quest, developed by the Alberta Wheat Pool, has been genetically engineered to tolerate the herbicide glyphosate, e.g. Round-up™. Quest was derived from Excel (T.E. Nickson, personal communication), and thus allows comparison between related transgenic and non-transgenic genotypes. Field sites were considered random factors, i.e. sites were not chosen for a specific reason but rather as two fields drawn at random from available Saskatchewan sites, with cultivars (n=3) as fixed treatments and replicates (n=4) as the blocks. Field sites were near Watrous and Denholm, Saskatchewan, Canada. Available soil nutrient levels for Watrous and Denholme were respectively (μg g−1), N 63, P 36, K 630, S 14 and N 34, P 23, K 550, and S 14.

2.2Sample processing

Plants and their associated root material were removed from soil with a trowel and placed in a plastic bag. Samples were immediately transported to the laboratory and processed less than 2 h after removal from the ground. The shoot was removed with a scalpel and the roots with adhering soil were sieved (5 mm) for 5 min.

To extract rhizosphere microorganisms, a 5-g portion of root with adhering soil was placed in a 1-l Erlenmeyer flask containing 495 ml phosphate buffered saline (PBS: 1.2 Na2HPO4, 0.18 NaH2PO4, 8.5 NaCl (g l−1); pH 7.6) and placed on a rotary shaker (200 rpm) at 22°C for 20 min [15]. This solution was serially diluted and 0.1 ml of the 10−4, 10−5 and 10−6 dilutions spread plated onto triplicate plates containing 1/10 trypticase soy broth (3 g l−1) solidified with 1.5% agar (1/10 TSA).

Root-endophytic organisms (i.e. isolated from the root interior) were recovered after removing rhizosphere organisms and surface disinfecting. Roots (from rhizosphere sampling) were transferred into a 500-ml Erlenmeyer flask containing 200 ml NaClO (1.05% v/v) in PBS and placed on a rotary shaker (200 rpm) at 22°C for 10 min. Roots were rinsed four times with 200 ml sterile PBS and 0.1 ml of the final wash diluted in 9.9 ml of 1/10 TSB to check for contamination [16]. The roots were chopped into 1-cm sections and then triturated with a sterile mortar and pestle containing 10 ml sterile PBS. The root/PBS mixture was serially diluted in sterile PBS and 0.1 ml of the 10−1, 10−2 and 10−3 dilutions spread plated onto triplicate plates of 1/10 TSA.

2.3Isolation and identification of bacteria

After 72 h incubation at room temperature, one plate containing 50–300 colonies (i.e. typically the 10−2 root-endophytic and 10−5 rhizosphere dilutions) was selected from each replicate and the bacterial colonies numbered. A random number table was consulted and 50 colonies isolated from one plate for each plant replicate. Based on previous work, we estimated that 50 isolates would comprise 70% of the bacterial community capable of growth on 1/10 TSA [5]. Isolates were streaked twice on 1/10 TSA and purified strains stored on 1-ml stabs containing 1/10 TSA, overlaid with sterile mineral oil and stored at 4°C.

Isolates were identified based on whole-cell cellular fatty acids, derivatized to methyl esters (i.e., FAMEs) and analyzed by gas chromatography [5]. Bacterial isolates were analyzed using the MIDI Microbial Identification Software (Sherlock TSBA Library version 3.80; Microbial ID, Inc.). The FAME profile of Xanthomonas maltophilia ATCC 13637 was used as a reference for the MIDI determinations. Strains with a similarity index (SIM) of 0.3 or greater were considered positively identified, whereas strains with a SIM of less than 0.3 were considered tentatively identified.

2.4Statistical analyses

The genus compositions of rhizosphere and root-endophytic communities were compared using the Shannon-Weaver diversity index (H′), Carmago's evenness index (Evar) and genus richness [5]. The Shannon-Weaver index combines measurements of richness with those of evenness, whereas Carmago's evenness index is an estimate of the variance in genus abundance over the number of genera, with 1 being the maximum evenness and 0 the minimum [17]. To estimate the number of genera per sample, a dendrogram analysis (centroid, single linkage) was used to differentiate at the genus level (Euclidean distance of 25) all of the rhizosphere and root-endophytic isolates. This approach allows the use of isolates regardless of their presence in the MIDI library. The diversity and evenness indices were analyzed by ANOVA using the RCBD described above. Consistent with previous studies, there were no significant site or blocking effects [5] and results are presented as means for each cultivar with no separation of site or blocking effects. Rhizosphere and root-endophytic communities of the same plant are not independent of one another [18] and cannot be considered independent samples when comparing rhizosphere and root-endophytic treatments in an ANOVA. Thus, to compare these communities, the difference between the rhizosphere and root-endophytic diversity indices for each replicate was analyzed by the ANOVA RCBD described above.

In addition to diversity indices, we also compared bacterial communities (as represented by the 50 isolates per replicate) by principal component analysis (PCA) using the co-variance matrix [19]. Principal component scores were compared in a manner similar to that suggested by Glimm [20] with the exception that scores were analyzed by the ANOVA RCBD. Means were separated by protected least significant difference and exact probabilities are reported to allow the reader to assess the biological significance of the results.

Numbers of genera or species were compared using the G-test with William's correction when the frequency of occurrence was greater than 3 [21]. We present exact probabilities for mean separation where this is less than 20%, to allow the reader to assess the biological as well as the statistical significance of the result [22].

3Results

3.1Isolation and identification of bacteria obtained from field-grown canola

Approximately 1100 bacteria were obtained from the root interior and 1200 from the rhizosphere of the three canola cultivars. One root-endophytic replicate of Parkland and one root-endophytic replicate of Quest were contaminated. Thus, these replicates were not processed further. The MIDI system identified ca. 30% of isolated bacteria regardless of the cultivar sampled and assigned a tentative identification to another third of the isolates (data not shown). However, approximately 12–17% of the isolates were not present in the MIDI Sherlock TSBA Version 3.80 library. In addition, 21–31% of bacteria isolated on 1/10 TSA were unable to grow on the full strength TSA required for MIDI identification. Interestingly, more (P<0.001) isolates from the rhizosphere (79%) were capable of growth on full strength TSA compared to root-endophytic (69%) isolates.

3.2Cultivar dependence of root-associated communities

PCA separated the bacterial communities associated with the root interior of Quest from Parkland or Excel along the PC1 axis (Fig. 1). Genera with the largest loadings on this axis were Xanthomonas (loading factor, 0.681), Pseudomonas (loading factor, 0.449) and Variovorax (loading factor, −0.561). The average PC1 score for Quest was 5.6 compared to 0.49 for Parkland and 0.55 for Excel. However, the principal component scores were highly variable with an average 21% coefficient of variation. There was little difference in the rhizosphere communities of different cultivars as determined by PCA (data not shown). The diversity and evenness of the bacterial communities associated with the root interior were similar for Quest (H′ of 1.19, Evar of 0.43, 6.4 genera), Excel (H′ of 1.44, Evar of 0.49, 7.5 genera) and Parkland (H′ of 1.34, Evar of 0.48, 6.0 genera).

Figure 1.

Principal component analysis of bacterial communities associated with the root interior of three cultivars of canola grown at two field sites. Each symbol represents 50 isolates obtained from the root interior and identified using the MIDI Sherlock system. The proportion of variance accounted for by each component is indicated in parentheses. Open circles, Parkland cultivar replicates. Open squares, Excel cultivar replicates. Open triangles, Quest cultivar replicates. Closed symbols are the means of corresponding cultivars.

3.3Cultivar dependence of root-associated bacterial genera

Rare genera (i.e. those present only in one cultivar) were more prevalent in the root interior compared to the rhizosphere of canola. Of the 22 root-endophytic genera, 11 (50%) were present in only one of the three cultivars. Six of these 11 rare genera were represented by only one isolate. Six (27%) genera were present in all three cultivars. In contrast to the root interior, seven (32%) rhizosphere genera were present in only one cultivar and 10 (45%) genera present in all three cultivars.

The abundance of five bacterial genera in the root interior differed between cultivars (Fig. 2). There were fewer Micrococcus (P<0.18) and Variovorax (P<0.04) isolates in the root interior of Quest compared to Excel, but more Pseudomonas (P<0.03) isolates in the root interior of Quest compared to Excel or Parkland. Furthermore, there were fewer (P<0.073) Bacillus isolates, and more (P<0.054) Flavobacterium isolates in Quest compared to Excel. No Bacillus or Flavobacterium isolates were obtained from the root-interior of Parkland.

Figure 2.

Abundance of genera obtained from the root interior and rhizosphere of three field-grown canola cultivars. Root colonization by genera marked with asterisks was significantly (P<0.20) different between the Quest and Excel cultivars.

The decreased abundance of bacilli in the Quest root interior compared to the rhizosphere was largely due to the absence of Bacillus megaterium, thereby reducing the Bacillus species richness from 4 to 3 (Table 1). In contrast, the increased numbers of Flavobacterium isolates in Quest was spread across four species. Decreases in Micrococcus and Variovorax abundance were limited to only a single species in each case, M. kristinae and V. paradoxus respectively. The increase in the number of Pseudomonas isolates was limited to three of the eight Pseudomonas species found in the root interior of canola. Specifically, there were more Pseudomonas corrugata, P. putida and P. savastanoi isolates in Quest compared to Excel and Parkland. Furthermore, more P. chlororaphis isolates were found in Quest and Excel compared to Parkland. Consequently, the diversity and evenness of the pseudomonad population was higher in Quest (H′ of 1.61; Evar of 0.729) compared to Excel (H′ of 1.44; Evar of 0.626) or Parkland (H′ of 1.55; Evar of 0.707).

Table 1.  Species abundance of identified (i.e. SIM>0.3) genera that differ significantly between canola cultivarsa
SpeciesNumber of isolates foundTotal
 Root interiorRhizosphere 
 ExcelQuestParklandExcelQuestParkland 
  1. aThe genera Micrococcus and Variovorax are not included because only one species in each genus differed between cultivars (i.e. M. kristinae and V. paradoxus) with higher levels of M. kristinae found in the root interior of Excel compared to the rhizosphere and lower levels of V. paradoxus found in the root interior of all three plant cultivars compared to the rhizosphere.

Arthrobacter atrocyaneus00040610
A. crystallopoietes0001001
A. globiformis10142715
A. ilicis2001205
A. oxydans0003036
A. pascens0002125
A. protophormiae/ramosus0002013
A. viscosus0000101
Arthrobacter total3011761946
Bacillus atrophaeus0000011
B. brevis1001125
B. chitinosporus0000202
B. gordonae0001012
B. licheniformis0101002
B. longisporus0001001
B. macerans0100001
B. megaterium50031716
B. mycoides1000001
B. pabuli0000011
B. pumilus30050210
B. thuringiensis0100001
Bacillus total10301241443
Flavobacterium aquatile14011613
F. balustinum0300036
F. esteraromaticum0101204
F. indologenes2101015
Flavobacterium total390331028
Pseudomonas chlororaphis171831726990
P. cichorii2000013
P. corrugata081541331
P. fluorescens3001138
P. marginalis25151115
P. putida310335428
P. savastanoi17140215
P. syringae22634320
Pseudomonas total305015384136210

In the rhizosphere, the abundance of two genera differed between cultivars (Fig. 2). There were fewer Arthrobacter (P<0.004) and Bacillus (P<0.008) isolates in the rhizosphere of Quest compared to Excel or Parkland. The decrease in Arthrobacter numbers in the Quest rhizosphere was spread across several species with A. atrocyaneus, A. crystallopoietes, A. oxydans and A. protophormiae/ramosus present in Excel but not in Quest. This results in a higher diversity index for Excel (H′ of 1.82) compared to Quest (1.33) for Arthrobacter. Similar to that seen for the root interior, the decrease in Bacillus abundance was largely limited to Bacillus pumilus. B. pumilus was absent from the Quest rhizosphere despite making up the majority of the Bacillus species in the Excel rhizosphere and being present in the Parkland rhizosphere.

3.4Differences between root interior and rhizosphere bacterial communities

The root interior and rhizosphere communities were clearly separated (P<0.001) by PCA (Fig. 3). PC1 loadings were Xanthomonas (0.81), Phyllobacterium (−0.32) and Sphingobacterium (−0.30). In addition, the diversity and evenness of the bacterial community was lower (P<0.001) in the root interior compared to the rhizosphere. The difference between rhizosphere and root-endophytic diversity displayed no cultivar or site dependence with lower diversity and evenness indices present at all sites for all cultivars.

Figure 3.

Principal component analysis of bacterial communities associated with the root interior and rhizosphere of field-grown canola. Each symbol represents 50 isolates obtained from the root interior or rhizosphere and identified using the MIDI Sherlock system. The proportion of variance accounted for by each component is indicated in parentheses. Open triangles, root interior communities. Open circles, rhizosphere communities. Closed symbols are the means of corresponding treatments.

Among the 20 genera found in both environments, six were found in either greater (Micrococcus, Variovorax and Xanthomonas spp.) or lower (Arthrobacter, Aureobacterium and Bacillus spp.) abundance in the root interior compared to the rhizosphere (Table 2). Furthermore, only three genera found in root interior were not found in the rhizosphere and these genera were very rare, individually comprising <0.26% of the root-endophytic population. The increased number of Arthrobacter isolates in the rhizosphere compared to the root interior was spread across five species found only in the rhizosphere. Furthermore the abundance of A. globiformis was greater in the rhizosphere (n=13) compared to the root interior (n=2). There were more Aureobacterium isolates in the rhizosphere compared to the root interior because Aureobacterium was only found in the root interior of Quest despite being present in the rhizosphere of all three cultivars (data not shown). The increase in Bacillus abundance was largely limited to increases in the abundance of species also present in the root interior and not to additional species being found in the rhizosphere (nine species) compared to the root interior (seven species). The decrease in Micrococcus isolates paradoxically corresponded to an increased number of species found in the rhizosphere (four species) compared to the root interior (one species). M. luteus, M. roseus, and M. varians were found in the rhizosphere but not the root interior. Numbers of Variovorax and Xanthomonas isolates were lower in the rhizosphere but this decrease was limited to a single species in each case, i.e. V. paradoxus, X. campestris.

Table 2.  Percent composition of the bacterial communities associated with the root interior or rhizosphere of field-grown canolaa
GenusbRoot interiorRhizospherecCombined
  1. aOf the 1100 isolates from the root interior, only 758 could grow on full strength TSA and be processed by the MIDI Sherlock system. Similarly, only 953 of the 1200 isolates from the rhizosphere were processed by the MIDI Sherlock system.

  2. bNo match, isolates with no match in Sherlock TSBA Library Version 3.80. Tentatively identified, isolates identified with a SIM <0.3.

  3. cAsterisks indicate genera with significantly different composition in the root interior compared to the rhizosphere. ***P<0.001; **P<0.05; *P<0.10.

Acidovorax0.260.100.18
Agrobacterium0.400.520.47
Alcaligenes0.530.210.35
Arthrobacter0.534.41***2.69
Aureobacterium0.401.26**0.88
Bacillus1.723.25**2.57
Bradyrhizobium0.000.100.06
Cellulomonas0.260.000.12
Clavibacter0.920.630.76
Comamonas0.130.100.12
Corynebacterium0.130.100.12
Curtobacterium1.191.151.17
Cytophaga0.921.361.17
Enterobacter0.130.310.23
Enterococcus0.260.000.12
Flavobacterium1.581.681.64
Methylobacterium0.130.210.18
Micrococcus3.171.78*2.40
Nocardia0.000.420.23
Pseudomonas12.5312.1712.33
Rathayibacter0.130.310.23
Rhodococcus0.130.000.06
Salmonella0.000.100.06
Variovorax13.984.20***8.53
Xanthomonas2.901.47**2.10
Tentatively identified40.5042.1841.44
No match17.1521.93***19.81

4Discussion

Our results demonstrate that the composition of the root-associated microbial community differs between transgenic and non-transgenic canola cultivars. Furthermore, this cultivar effect was more pronounced in the root interior compared to the rhizosphere community. These results support our previous CLPP and FAME analyses [9] where we found that the microbial community differed between these same canola cultivars and that root-endophytic communities differed to the greatest extent. Furthermore, we found that the root-associated microbial communities of non-transgenic Excel and Parkland cultivars were related to one another to a greater extent, despite being different species, than either one was to transgenic Quest. Our present taxonomic analysis of the community associated with canola roots indicates that Quest was significantly different from Excel or Parkland (Figs. 1 and 2), and extends our conclusions to specific taxonomic units.

The basis for the differences between cultivars is not clear. Root exudates are widely postulated to control rhizosphere populations [4–10] and differences between plant species are well known [23]. However, there have been very few reports of differences in root exudates between canola cultivars. Cieslinski et al. [24] found that exudation of low molecular mass organic acids in the rhizosphere of wheat and flax differed significantly between cultivars. To the best of our knowledge, there have been no reports of differences in the root exudation between the canola cultivars we studied. In any case, it is not clear if root exudates are the mechanism by which plants apply selective pressure on the root interior community. Bacteria enter the root interior by either hydrolyzing wall-bound cellulose, entering through auxin induced tumors, with water flow, wounds or at lateral root branching [8]. However, in some instances certain plant phenotypic characteristics such as lateral roots or environmental conditions such as low pO2[25], must be present for bacteria to enter the root interior. It is possible that differences between cultivars in root morphology or exudation affect the ability of certain bacteria to colonize the root interior but this awaits experimental verification.

There was a greater abundance and diversity of pseudomonads found in the root interior of the B. napus cv. Quest (50 isolates) compared to B. napus cv. Excel (30 isolates) or B. rapa cv. Parkland (15 isolates). This may explain why the previous CLPP resulted in greater separation between the cultivars compared to taxonomic analyses [9]. Grayston et al. [6] found that pseudomonads influence CLPP out of proportion to their numbers in the microbial community. However, the importance of this bias to an accurate interpretation of CLPP is unclear. Pseudomonads are thought to be important members of the root-associated microbial community due to their aggressive colonization of the root surface, importance in plant disease interactions and plant growth promoting abilities [26–28]. Thus, analyses sensitive to pseudomonad populations may be appropriate for root-associated microbial community analysis.

The results from this study demonstrate that plants apply selective pressure on microorganisms colonizing their root interior. Consequently, there is lower diversity and evenness seen in the root interior compared to the rhizosphere, as well as changes in the abundance of certain genera in the root interior compared to the rhizosphere. Furthermore, this selective pressure is cultivar dependent. For example, P. corrugata was often found in the rhizospheres of all three cultivars, with five isolates identified in the Excel rhizosphere, 13 in Parkland and four in Quest. Despite this, P. corrugata was not found in the root interior of Excel and only found once in Parkland. However, eight isolates of P. corrugata were found in the root interior of Quest. Similarly, numbers of Aureobacterium spp., P. savastanoi and Flavobacterium aquatile were higher in the Quest root interior despite being present in the rhizosphere of all three canola cultivars. These results suggest that Quest selectively recruited and promoted the growth of certain Pseudomonas, Aureobacter and Flavobacter species in its root interior. The evolutionary basis for this is unclear, but mutualistic associations between plants and bacteria may be an important ecological advantage [1]. In this study there was little difference seen between the root-associated microbial communities at different field sites. However, only two sites were considered and thus, the experimental design cannot adequately test the influence of field sites on root-associated microbial communities. Others (see e.g. [3]) have suggested that field sites play an important role and that site factors influence the composition of root-associated microbial communities. This concept is currently being assessed in our lab in a multi-year, multi-site study.

Germida et al. [5] found four genera, Bacillus, Flavobacterium, Micrococcus and Rathayibacter, present in higher numbers in the root interior compared to the rhizosphere. Our study only found more Micrococcus, Variovorax and Xanthomonas isolates. Furthermore, Germida et al. [5] found very low numbers of pseudomonads in the rhizosphere and root interior (ca. 2.5%), whereas here we found that pseudomonads accounted for ca. 12% of the bacterial community. These differences may be attributed to differences in the isolation technique. Germida et al. [5] soaked roots for 10 min, followed by chopping up the roots and sterilizing with mercuric chloride. In this study, we avoided breaking up roots. Furthermore, we surface sterilized roots with bleach before triturating. Despite these differences, the diversity and evenness indices were comparable with a H′=1.32 and Evar=0.463 for the present study and a H′=1.35 and Evar=0.538 for Germida et al. [5].

We found that there were cultivar-dependent differences in the root interior microbial community. Furthermore, we found that specific members of the root interior community associated with a transgenic canola variety were present at different levels compared to a non-transgenic variety of the same species. It is possible that the insertion of foreign genetic material into canola has provoked unintended alterations in the root-endophytic community. This hypothesis awaits further experimental verification. Chapin et al. [29] suggest that altering the diversity and composition of ecological communities can reduce the sustainability or the resilience of communities to environmental stresses, such as drought or climate change. This may occur by altering the composition of a community and perhaps interrupting an important ecological function such as nutrient cycling. Alternatively, reducing the diversity of organisms that provide a specific ecological function can decrease the resilience of an ecosystem to stress. Due to the important role of bacteria in nutrient transformations, determining the impact of new crop cultivars on the root-associated community deserves additional attention.

Acknowledgements

The authors would like to thank Jim Ferrie and Garry Hnatowhich of the Saskatchewan Wheat Pool for access to canola sites. The help of J.R. de Freitas, A. Mason, A. Seib, S. Ersali and K. Dunfield is appreciated. This research was supported by NSERC. Contribution R837 Saskatchewan Centre for Soil Research.

Ancillary