REGULAR ARTICLE

Evidence of variability in the structure and recruitmentof rhizospheric and endophytic bacterial communitiesassociated with arable sweet sorghum(Sorghum bicolor (L) Moench)

Jean-Baptiste Ramond & Freedom Tshabuse &

Cyprien W. Bopda & Don A. Cowan &

Marla I. Tuffin

Received: 23 November 2012 /Accepted: 16 April 2013# Springer Science+Business Media Dordrecht 2013

AbstractBackground and aims Sorghum is the second mostcultivated crop in Africa and is a staple food sourcein many African communities. Exploiting the associ-ated plant growth-promoting bacteria (PGPB) has po-tential as an agricultural biotechnology strategy toenhance sorghum growth, yield and nutritional prop-erties. Therefore this study aimed to evaluate factorsthat shape bacterial communities associated with sor-ghum farmed in South Africa, and to detect bacteria

consistently associated with sorghum which may im-part PGP activities.Methods Terminal-Restriction Fragment Length Poly-morphism (T-RFLP) was used to assess factors thatpotentially shape rhizospheric (rhizosphere and rhizo-plane) and endophytic (root, shoot, stem) bacterialcommunities associated with South African sorghum,and together with Denaturing Gradient Gel Electro-phoresis (DGGE) to identify consistently sorghum-associated bacterial taxa.Results The sorghum rhizospheric communities wereless variable than the endophytic ones. Geographicallocation was the main driver in describing bacterialcommunity assemblages found in rhizosphericsorghum-linked niches, with total NO3-N, NH4-N,nitrogen, carbon, pH and, to a lesser extent, claycontent identified as the main abiotic factors shapingsorghum-associated soil communities. Endophyticcommunities presented rather stochastic assemblages,with pH being the main variable explaining theirstructures. Despite community variations, specificbacterial taxa were consistently detected in sorghum-created rhizospheric and endophytic environments,irrespective of environmental factor effects.Conclusions Soil structure and composition, whichare influenced by agricultural practices, played majorroles in shaping sorghum-associated edaphic bacterialcommunities. In contrast, endophytic bacterial

Plant SoilDOI 10.1007/s11104-013-1737-6

Responsible Editor: Paul Bodelier.

J.-B. Ramond : F. Tshabuse : C. W. Bopda :D. A. Cowan :M. I. Tuffin (*)Institute for Microbial Biotechnology and Metagenomics(IMBM), Department of Biotechnology, University of theWestern Cape,Bellville 7535 Cape Town, South Africae-mail: ituffin@uwc.ac.za

D. A. CowanCentre for Microbial Ecology and Genomics (CMEG),Department of Genetics, University of Pretoria,Pretoria 0002, South Africa

Present Address:J.-B. RamondCentre for Microbial Ecology and Genomics (CMEG),Department of Genetics, University of Pretoria,Pretoria 0002, South Africa

communities displayed more variation. Nevertheless,potentially agronomically relevant (cyano)bacterial taxaconstantly associated with sorghum were identifiedwhich is suggestive of their deterministic recruitment.

Keywords Bacteria recruitment . Molecularfingerprinting (T-RFLP/DGGE) . Plant-microbeinteractions . Sorghum bicolor

Introduction

The use of natural or genetically modified bacteria toenhance plant productivity has been widely considered(e.g. Berg 2009; Compant et al. 2005; Hafeez et al.2006; Schenk et al. 2012), particularly the use ofplant-associated bacteria as such species have evolveda structured and intimate relationship with the planthost (Berg 2009; Compant et al. 2005; Schenk et al.2012). Such bacteria are typically present in the“plant-created” rhizospheric and endophytic environ-ments. The rhizospheric environment is the soil sur-rounding and influenced by plant roots (Morgan andWhite 2005), and is created by the release of nutrient-and carbon-rich root exudates, making it a “hot-spot”for microbial growth (Hartmann et al. 2009; Morganand White 2005). Endophytic environments corre-spond to micro-environments localized inside plantorgans and tissues (Saito et al. 2007).

In such environments, bacteria have been shown tohave neutral, detrimental or beneficial effects on plants(for reviews, see Berg 2009; Schenk et al. 2012).Beneficial plant-associated bacteria, or plant-growthpromoting bacteria (PGPB), act as biocontrol agentsby inhibiting growth and colonization of phytopatho-genic organisms (generally through the production ofsiderophores or antibiotics) and/or as biofertilizers byproducing phytohormones and/or improving theplants’ nutritional status (by contributing to its nitro-gen fixation process for example) (Bai et al. 2002;Berg 2009; Lugtenberg et al. 1991; Panchal and Ingle2011). PGPBs are therefore attractive vehicles for en-hancing plant productivity since they are considered asnatural, cheap and environmentally friendly fertilizers(Schenk et al. 2012). It can indeed be assumed that suchplant-selected (or plant-adapted) PGPBs are alreadyplaying an active role in the crop’s health and growth,thereby having a mutualistic, symbiotic and/orcommensalistic relationship with its’ plant-host. Once

identified, such PGPB(s) could subsequently beengineered in order to present as much plant growth-promoting activities as desired, and/or be used directlyas biofertilizers.

Sorghum (Sorghum bicolor L.) is the world’s fifthmost cultivated cereal crop after wheat, rice, maizeand barley, with a global production of 60 million tons(Dicko et al. 2006), and is the second most cultivatedcereal grain in Africa after maize (Taylor 2003). It is acritically important food crop, as it is estimated that inAfrica and Asia over 300 million people rely on thiscrop as an essential source of energy (Dicko et al.2006) Agriculture remains the backbone of Africa’seconomy, where 70 % of Africans’ income is relianton agricultural products (FAO/WFP 2010). Due to thesevere famine experienced in this continent, there is ademand for increased sorghum production to meet thecontinents’ needs (FAO 1995). The identification ofbacteria consistently associated with sorghum, i.e. in-dependently of any abiotic factor, agricultural practiceor geographical location, would constitute the firststep in developing a biotechnological and agriculturalstrategy to enhance sorghum growth, yield and nutri-tional properties.

Based on culture-dependent studies, various sorghum-associated bacteria have been isolated and identified(Budi et al. 1999; Pedersen et al. 1978; Zinniel et al.2002), some of which have exhibited PGP activities suchas nitrogen-fixation (members of the Enterobacteriaceaerelated to Klebsiella pneumoniae, Enterobacter cloacae orErwinia herbicola; Pedersen et al. 1978) and the capacityfor biocontrol (Paenibacillus sp. strain B2 and Pseudomo-nas spp.; Budi et al. 1999; Funnel-Harris et al. 2013).Moreover, in sorghum rhizospheres, microbial nifH genesequences (which are indicative of bacterial nitrogen fixa-tion and can be used as phylogenetic markers) ofAzohydromonas spp., Ideonella sp., Rhizobium etli,Bradyrhizobium sp. Delftia tsuruhatensi, Methylocystissp. and Paenibacillus have been detected (Coelho et al.2008, 2009), indicating that a diverse sorghum-associateddiazotrophic microbial community provides bioavailableN to the plant. In this study, molecular fingerprintingmethods (DGGE and T-RFLP) were used to expandthe known range of sorghum-associated bacterial taxa, tocorrelate sorghum-associated bacterial assemblages withabiotic factors and to assess the presence of bacteriaconsistently (or obligately) recruited by sorghum plants.Such taxa would be valid targets for future biofertilizerdevelopment. Moreover, once identified by culture

Plant Soil

independent techniques, strategies to isolate thesespecific bacterial taxa can be designed (Tyson etal. 2005).

Materials and methods

Study site and sampling procedures

Two mature and healthy sweet sorghum plants (Sor-ghum bicolor (L) Moench; at approx. 100 m spacing)were harvested in three farms from three differentSouth African provinces: Free State (Plant 1: S27°02.975 ′ /E027°31 .405 ′ ; P lan t 2 : S27°03 .665 ′ /E027°31.780′), North West (Plant 1: S26°43.741′/E027°04.870′; Plant 2: S26°44.063′/E027°04.721′),and Limpopo (Plant 1: S24°38.620′/E029°52.484′;Plant 2: S24°39.375′/E029°53.593′) (Table 1). In thethree farms, water supply was primarily throughrainfall, with supplementary irrigation when rainfalllevels were low. The North West academic farm ofthe ARC (Agricultural Research Council) and theFree State commercial farm (i.e. where sorghum iscultivated for commercial purposes) used modernagricultural fertilization procedures when comparedto the Limpopo small household farm which usedcow faeces as fertilizer. In the North West farm, aslow-releasing fertiliser containing Zinc and 32 % ofa mixture of nitrogen (N), phosphorus (P) and po-tassium (K) in a 3:2:1 ratio was applied at a rate of150 kg N/ha. The insecticide Combat® was appliedfor control of stem-borers (~4 kg/ha). Similarly, theFree State farm applied ammonium sulfate at therate of 150 kg/ha with 28 % of zinc sulfate (anherbicide).

For each plant, samples of root, shoot and stemtissues were aseptically excised and stored in sterileplastic bags. Rhizosphere soils were collected byunearthing individual plants and dislodging soilparticles associated with the root structures andcorresponded to a cohesive layer of 3 to 5 mm. Rhi-zoplane soils corresponded to the soil particlesremaining strongly attached (~1 mm in thickness) tothe sorghum root and were collected with the roottissue samples. Open soil (control) samples, from asimilar depth (0 to 5 cm) but not impacted by plantroot systems, were collected at each site. All sampleswere kept on ice and transported to the University ofthe Western Cape (UWC, South Africa) where theywere stored at −80 °C prior to processing.

Soil characterization

The pH, total carbon (C), nitrogen (N), ammonium(NH4-N), nitrate (NO3-N) and phosphorous (P) con-tents of soil samples, and their structure (% silt, clay,fine sand, medium sand and coarse sand) were ana-lyzed by Bemlab (Pty) Ltd (Strand, Western Cape,South Africa) (Table 1).

Briefly, the soil was air dried and sieved through a2 mm sieve. Soil pH was measured in 1 M KCl with asoil/solution ratio of 1:2. Total C and N contents weredetermined through total combustion using a LecoTruspec® CN analyzer. Total NH4-N and NO3-N wereextracted with 1 N KCl, and their concentrations weredetermined calorimetrically on a SEAL AutoAnalyzer3. Total P was extracted with a 1:1 solution of 1 Nnitric acid and hydrochloric acid at 80 °C for 30 minand its concentration in the extract was determinedwith a Varian ICP-OES optical emission spectrometer.

Table 1 Open and rhizospheric soil characteristics

Province % clay/% Silt Soil type pH NH4-N(mg.kg−1)

NO3-N(mg.kg−1)

Total N(%)

Total P(mg.kg−1)

Total C(%)% Sand: fine, medium,

coarse

Free state 22 % Clay/12 % Silt Open soil 4.7 7.88 1.44 0.11 109.94 0.58

Sand: 44 %, 11 %, 12 % Rhizosphere 4.2 8.68 0.52 0.11 115.62 0.40

Limpopo 16 % Clay/8 % Silt Open soil 5.4 9.6 3.72 0.09 63.92 0.19

Sand: 25 %, 19 %, 32 % Rhizosphere 6.3 8.36 4.72 0.10 88.63 0.36

North West 34 % Clay/12 % Silt Open soil 6.2 8.44 11.88 0.14 399.00 0.96

Sand: 39 %, 8 %, 8 % Rhizosphere 6.0 9.60 5.80 0.14 238.01 0.94

C carbon, N nitrogen, NH4-N ammonium, NO3-N nitrate, P phosphorous

Plant Soil

For soil structure analysis, chemical dispersion wasperformed using sodium hexametaphosphate, and thethree sand fractions were determined by sieving asdescribed in the Non-Affiliated Soil Analysis WorkCommittee (1990). Silt and clay contents wereassessed using sedimentation rates at 20 °C, using anASTM E100 (152H-TP) hydrometer.

Plant tissue sterilization

Plant tissues (roots, shoots, stems) were surface steril-ized using a modification of the protocol described byMendes et al. (2007). Each tissue sample was washedfive times with sterile distilled water to remove at-tached soil particles, and placed in 400 mL of 1XPBS buffer and incubated with shaking at room tem-perature for 2 h. Samples were then sequentiallywashed by shaking in (i) a 70 % ethanol solution for10 min, (ii) a 2 % (v/v) sodium perchlorate solution for10 min, (iii) a 70 % ethanol solution for 5 min and (iv)rinsed three times with autoclaved distilled water for1 min. To evaluate the efficiency of the sterilizationprocedure, 100 μL volumes of the final dH2O rinsewere plated on TSA and R2A agar (Merck, Germa-ny), supplemented with the fungicide actidione(100 mg.mL−1) and incubated at 28 °C for 4 days.The plant tissues were stored in the last dH2O rinse at4 °C during these 4 days. Where no colony growthwas observed, the sterilization procedure was consid-ered to be sufficient. Where colonies were observed,the complete sterilization process was repeated. Oncesterilized, the tissue samples were stored at 4 °C forsubsequent molecular analysis. In order to ensure theexclusive study of endophytic bacterial communities,the surface tissue of the stems and shoots were re-moved using sterile surgical blades. This procedurewas not possible for the thinner roots.

Soil and plant organ metagenomic DNA extraction

Total metagenomic DNAwas extracted from 0.5 g soilsamples with the Powersoil DNA isolation kitaccording to the manufacturer’s instructions (MoBiolaboratories, USA). DNA extractions from sorghumtissue samples were performed using a modified ver-sion of the method described by Murray and Thomp-son (1980). Plant tissues were ground to powder formin liquid nitrogen using sterilized mortars and pestles.A pre-heated solution of 700 μL of 2 % CTAB and

1 μL of β-mercaptoethanol was added to each tissuepowder sample, vortexed at maximum speed for 20 sand incubated at 65 °C for 60 min. A chloroform/isoamylalcohol (24:1 v/v) solution (600 μL) was added, mixed byinversion for 5 min and centrifuged (12,000 rpm, 5 min).The supernatant was collected and an equal amount of ice-cold isopropanol was added with RNase A (10 mg.mL−1

final concentration). The mixtures were incubated at roomtemperature for 20 min and centrifuged at 12,000 rpm for5 min. The DNA pellets were washed twice with 250 μLof 70 % ethanol and centrifuged at 12,000 rpm for 5 minprior to drying in a laminar flow cabinet. The DNA wasresuspended in 50 μL of 1X TE buffer and stored at 4 °C.Metagenomic DNA concentrations were measured with aNanoDrop spectrophotometer (NanoDrop Technologies,Montchanin, DE, USA).

PCR amplification, purification and restrictiondigestion

All polymerase chain reactions (PCRs) were carriedout in a Bio-Rad Thermocycler (T100™ Thermal Cy-cler). Bacterial 16S rRNA encoding genes were am-plified using the universal primers E9F (5 ′-GAGTTTGATCCTGGCTCAG-3′) and U1510R (5′-GGTTACCTTGTTACGACTT-3′) (Marchesi et al.1998; Reysenbach and Pace 1995). This primer setwas specifically chosen as it would not amplify Sor-ghum bicolor chloroplast DNA [Genbank accessionnumber: NC_008602.1]. PCR was carried out in 50 μlreaction volumes. Each reaction contained 1X PCRbuffer, 0.2 U DreamTaq™ polymerase (Fermentas,USA), 200 μM of each dNTP, 0.5 μM of each primer,0.1 % BSA and between 10 and 20 ng of metagenomicDNA. PCR amplification was carried out as follows:4 min at 94 °C for denaturation; 30 cycles of 30 s at94 °C, 30 s annealing at 52 °C and 105 s at 72 °C; anda final elongation step of 10 min at 72 °C.

To perform T-RFLP, the E9F primer was 5′-endFAM-labelled and the PCR products were purified usingthe GFX™ PCR DNA and gel band purification kit asdirected by the supplier (GE Healthcare, UK). PurifiedPCR products (200 ng) were digested with the restric-tion enzyme HaeIII at 37 °C overnight.

For DGGE, a nested-PCR was performed with thesame 50 μL reaction mixture described above, usingthe primer set 341f-GC (5′-CCTACGGGAGGCAGCAG-3′)/534r (5′-ATTACCGCGGCTGCTG-3′)(Muyzer et al. 1993), 1 μL of the amplicon obtained

Plant Soil

with the primer set E9F/U1510R as template DNA andas follows: 94 °C for 4 min; 20 cycles – 94 °C for 45 s;65 °C for 45 s; 72 °C for 60 s; additional 20 cycles –94 °C for 30 s; 55 °C for 30 s; 72 °C for 60 s; and afinal elongation step at 72 °C for 10 min. A 40mer GCclamp was added to the 5′ ends of the forward primers341f-GC: GC clamp – CGCCCGCCGCGCGCGGCGGGCGGGGCGGGGGCACGGGGGG.

T-RFLP analysis

Bacterial community structures were assessed byT-RFLP fingerprinting using the 16S rRNA gene asa marker. The precise lengths of the T-RFs weredetermined by capillary electrophoresis using theApplied Biosystems DNA Sequencer 3130 (AppliedBiosystems, Foster City, California, USA) andaccording to the molecular weight standard Rox1.1(with an acceptable error of ±1 bp). T-RFLP patternsand quality were analyzed using the freewarePeakScanner™ (version 1.0) (Applied Biosystems,https://products.appliedbiosystems.com). Peak heightwas used to characterize each unique T-RF, and validT-RF peaks (between 35 and 1,000 bp) from triplicateT-RFLP profiles were identified, compiled and alignedto produce large data matrices using the online soft-ware T-REX (http://trex.biohpc.org/; Culman et al.2009). T-RFs with intensities lower than 0.5 %, whichmay have originated from background interference,were excluded from the matrices. The term OTU (Op-erational Taxonomic Unit) is used to refer to individ-ual terminal restriction fragments (T-RF) in T-RFLPpatterns, with recognition that each OTU may com-prise more than one distinct bacterial ribotype (Nockeret al. 2007). The web-based tool MiCA (MicrobialCommunity Analysis; Shyu et al. 2007), with the“RDP (R10,U27) 700,829 Good Quality (>1200) Bac-terial” database, was used for the in silico affiliation ofT-RFs. A±3 bp size margin was implemented to takeinto account potential differences between real andpredicted T-RFs (Sercu et al. 2011).

DGGE analysis

PCR amplicons obtained with the nested primer sets(341f-GC/534r) were analyzed by DGGE as describedpreviously (Rodriguez-Caballero et al. 2012). SelectedDGGE bands, i.e. which migrated at the same distanceon the DGGE gels, were excised using sterile surgical

blades and eluted in 50 μL of filter-sterilized water at4 °C overnight. One microlitre of the supernatant wasthen analyzed again by PCR and DGGE to eliminateany residual contamination by ‘parasite’ bands. Theremaining PCR products (~25 μL) were purified usingthe GFX™ PCR DNA and gel band purification kit asdirected by the supplier (GE Healthcare, UK). Thepurified PCR products from DGGE bands were se-quenced with a Hitachi 3730xl DNA Analyzer (Ap-plied Biosystems).

Statistical analysis

The community structures obtained by T-RFLPwere analyzed by ordination using non-metricmultidimensional scaling (nMDS) of Bray-Curtis sim-ilarity matrices of square-root transformed data withthe software Primer 6 (Primer-E Ltd, UK). In nMDSplots, the distance between points reflects the degreeof similarity between the microbial community pro-files in the samples. An analysis of similarity(ANOSIM), performed on the resemblance matrix,was used to test for differences in bacterial communitystructure between predefined groups (Clarke 1993).BEST (Biota Environment STepwise matching;Clarke and Gorley 2006) analysis was performed todetermine correlations between the soil and endophy-tic bacterial T-RFLP profiles and normalised abioticvariables (presented in Table 1) (Carson et al. 2007).The rhizosphere data were used to perform BESTanalysis on the endophytic communities, as the latter(and notably the root endophytic communities) aregenerally considered to be a subset of the rhizosphericcommunities (Cocking 2003). BEST determines therank correlation between the underlying similarity ma-trices for bacterial community data and environmentalvariables using the Spearman coefficient (ρ). As ρincreases, the correlation between the bacterial com-munity data and environmental variables increasesfrom no correlation (0) to complete correlation (1).

Phylogenetic analysis

Unrooted phylogenetic trees were constructed withMEGA5 (Tamura et al. 2011) using the Maximum Par-simony (MP) method. The bootstrap consensus treeinferred from 1,000 replicates is taken to represent theevolutionary history of the taxa analyzed (Felsenstein1985). The MP tree was obtained using the Subtree-

Plant Soil

Pruning-Regrafting (SPR) algorithm (Nei and Kumar2000) with search level 1 in which the initial trees wereobtained by the random addition of sequences (10 rep-licates). Sequences obtained in this study were deposit-ed in the NCBI GenBank database under accessionnumbers KC570917-KC570934.

Results

Factors shaping sorghum-associated bacterialcommunities

The clustering of the samples in the 3D-MDS plotpresented in Fig. 1 clearly demonstrates that there aresignificant differences (ANOSIM, Global R=0.437,P<0.001) between the soil (rhizosphere, rhizoplaneand open soil) and the endophytic (root, shoot andstem) communities. Notably, rhizosphere and rhizo-plane communities presented higher bacterial speciesrichness (~ 60 OTUs) than the endophytic communi-ties (ranging from 4 to 37 OTUs) (Fig. 2). These

results also indicated that the sterilization procedureremoved the rhizospheric community from the sor-ghum root tissues. Moreover, soil communities aremore clustered (indicated by triangles) compared withthe endophytic ones (indicated by squares) which

3D Stress: 0,1

Fig. 1 3D-Nonmetric Multi-Dimensional Scaling plot of BrayCurtis similarity plot of soil (triangles) and endophytic(squares) bacterial community profiles associated with SouthAfrican sorghum (stress=0.1). The soil bacterial communitiesoriginate from open (white triangle), rhizospheric (grey trian-gle), and rhizoplane (black triangle) soils and the endophyticcommunities from shoot (white square), roots (black square)and stems (grey square) of healthy sorghum plants harvested inLimpopo, North West and Free State provinces

A

B

Total:Rhizosphere: N = 87Rhizoplane: N = 88

Limpopo:N = 61 (37)N = 62 (27)

North West:N = 63 (40)N = 63 (28)

Free State:N = 68 (23)N = 52 (20)

39 (12)24(4)

44 (22)40 (9)

50 (15) 32 (9)

50 (17)41 (5)

6 (0)5 (0)

8 (1)15 (3)

7 (1)3 (1)

Fig. 2 Venn Diagrams showing the distribution of T-RFs pres-ent in the different sorghum-associated environments. a: Endo-phytic environment. The numbers in italic, bold, or underlinedindicate the number of T-RFs observed in sorghum root, shootor stem respectively. The numbers in bracket indicate the num-ber of OTUs present in the tissues of all the sorghum plantsrespectively compared. b: Rhizospheric environment. The num-bers in bold or underlined indicate the numbers of T-RFs ob-served in rhizospheric or rhizoplanic soils respectively. Thenumbers in bracket indicate the number of OTUs present inthe specific rhizospheric environments of all the sorghum plantsrespectively compared

Plant Soil

display greater scattering and hence community vari-ation (Fig. 1).

Depending on the province, the open and rhizosphericsoil characteristics varied, with notably lower pH values inFree State soils, elevated total C, P and NO3-N concentra-tions in North West soils and low total C and total N inLimpopo soils (Table 1). Apart from the rhizosphere com-munities of the Free State plant 2, a clear clustering of thesoil (open, rhizosphere and rhizoplane) bacterial commu-nities is observed in the nMDS plot presented in Fig. 3abased on their geographical origin (ANOSIM, GlobalR=0.5, P=0.022), and not on their soil type (i.e. open soil,rhizosphere or rhizoplane; ANOSIM, Global R=−0.083,P=0.649). Multiple rank correlations (BEST analysis) ofthe abiotic factors (Table 1) and soil community diversity

(measured by T-RFLP) demonstrated that NO3-N(ρ=0.381), pH (ρ=0.328) and clay content (ρ=0.270)were the principal individual abiotic factors defining thedifferent soil community structures, and that the best com-bination of edaphic variables included pH, NO3-N, NH4-N,% N and % C (ρ=0.381). The nMDS representationexplaining (dis)similarities in South African sorghum en-dophytic communities was ambiguous, with no clear clus-tering of the communities as a function of plant tissue typeor on the geographical origin (Fig. 3b). One-wayANOSIM analysis confirmed that the sorghum endophyticcommunities were not plant-tissue specific (GlobalR=0.109; P>0.05) and that the province of origin wasnot a significant factor (Global R=0.241; P=0.029)(Clarke and Gorley 2006); with ANOSIM pairwise tests

A

B

2 2

1

1

1

1

2

1

1

2

2

2

2D Stress: 0,1

1

1

2

12

12

1

1

2

2

21

1

2

1

2

2

2D Stress: 0,15

Fig. 3 2D-Nonmetric Multi-Dimensional Scaling plot ofBray Curtis similarity of bac-terial community structuresdetermined by T-RFLP anal-ysis of 16S rRNA genes as-sociated with sorghum inthree South African prov-inces. Triangles representbacterial communities fromFree State, squares fromNorth West and circles fromLimpopo. Numbers (1, 2) in-dicate sorghum plant. a: Soilbacterial communities (stress=0.1). Black: Open Soil/Grey: Rhizosphere/White:Rhizoplane. b: Endophyticbacterial communities (stress=0.15). Black: Root/Grey:Shoot/White: Stem

Plant Soil

showing that only the Limpopo and Free State endophyticcommunities were significantly different (R=0.417;P=0.009). BEST analysis, using the rhizospheric abioticdata in Table 1, showed that pHwas themain abiotic factorshaping sorghum endophytic communities (ρ=0.246).

Identification of consistently sorghum-associatedbacteria

Using T-RFLP, no consistently sorghum-associatedendophytic bacteria were detected (Fig. 2a). In con-trast, definite OTUs were constantly observed in itsrhizosphere and rhizoplane, i.e. independent of any ofthe environment-associated factors (Fig. 2b). Thirtynine OTUs were observed in the sorghum rhizospherefrom the three provinces sampled (Fig. 2b), 12 ofwhich (with respective sizes of 35, 71, 76, 77, 120,123, 192, 211, 227, 280, 290 and 374 bp; data notshown) were consistently detected in the rhizospheresof the plants sampled. Similarly, 24 OTUs were iden-tified in the sorghum rhizoplane of samples from thethree provinces (Fig. 2b), with 4 (with sizes of 71, 192,195 and 211 bp; data not shown) observed in all therhizoplane samples. Amongst these 13 OTUs, three(71, 192 and 211 bp), were repeatedly detected in boththe rhizosphere and rhizoplane of sorghum. Predictivephylogenetic affiliations (by in silico identification)matched the 71 bp OTU mainly to Acetobacterand Azospirillum species, the 195 bp OTU to α-

Proteobacteria (mostly to Bradyrhizobium species)and the 290 bp OTU essentially to cultured anduncultured cyanobacteria and notably Synechococcusspecies. In contrast, the 35, 76, 77, 120, 123, 192, 211,227, 280 and 374 bp OTUs could not be related to asingle class or phylum.

To identify bacterial taxa which were ubiquitouslyassociated with sorghum, DGGE coupled with post-electrophoretic phylogenetic analysis was also used(Fig. 4, Table 2). Since endophytic microbial commu-nities are typically characterized by a low taxonomicdiversity (Gottel et al. 2011; Fig. 2), and since DGGEis less sensitive than T-RFLP (Nocker et al. 2007), thePCR products from similar plant-tissue samples werepooled prior to loading on DGGE gels (Fig. 4). Thecomposite endophytic bacterial community finger-prints from each province for each plant-tissue typeis presented in Fig. 4a, b and c, while rhizospheric andrhizoplanic fingerprints for each plant sampled areshown in Fig. 4d and e.

Except in the stem samples, co-migrating DGGE-bands (i.e. which migrated at the same distance onthe DGGE gels, and indicated by arrows in Fig. 4)were observed in all the sorghum-associated micro-environments studied and were sequenced (Fig. 5,Table 2). The bands A1, A2 and A3 detected insorghum root metagenomic DNA (Fig. 4a) wererelated to bacterial sequences (Fig. 5a) and showedhigh sequence identities (100 %) with numerous

A CBFS L NWFS L NW

A1 A2 A3

NW L FS

B1 B2 B3

C1 C2 C3 C4 C5 C6

1 2 1 2 1 2

FS L NW D E

1 2 1 2 1 2

FS L NW

D1 D2 D3 D4 D5 D6

Fig. 4 DGGE profiles of the sorghum-associated bacterial com-munities in South Africa. a: Root endophytic communities. b:Shoot endophytic communities. c: Stem endophytic communi-ties. d: Rhizosphere communities. e: Rhizoplane communities.

Arrows and their associate reference indicate the co-migratedDGGE-bands sequenced and presented in Table 2. In therhizospheric environments, numbers (1, 2) indicate the sorghumplant. NW: North West province/L: Limpopo/FS: Free State

Plant Soil

cyanobacterial species and notably Synechococcus sp.(Table 2). The sorghum shoot sample B1, B2 and B3bands (Fig. 4b) were closely related to Pantoea spe-cies with 96 % to 100 % sequence identity (Fig. 5b,Table 2). The co-migrating bands observed in therhizospheric metagenomic DNA samples presentedhigh sequence homology to different taxonomicgroups (bands C1 to C6, Figs. 4d and 5c, Table 2), adisadvantage associated with DGGE where differentphylotypes may possess similar electrophoretic mobil-ities (Nocker et al. 2007). Finally, the rhizoplanicbands D1 to D6 possessed high sequence identities

(99 % to 100 %) with Bacillus megaterium strains(Figs. 4e and 5d, Table 2).

Discussion

Plant-associated microbial communities are suitabletargets to improve the production yields of food cropssuch as sorghum as, in such communities, PGPBmembers already play key roles in the plants’ healthand growth (Berg 2009). This has particular signifi-cance in prone-to-food-shortage countries because a

Fig. 5 Maximum parsimony analysis of the sequenced DGGE-bands (graphically shown in Fig. 4). a. Sorghum root. b. Sor-ghum shoot. c: Sorghum rhizosphere. d: Sorghum rhizoplane.

Only bootstrap values ≥40 % are shown. There were a total of133, 120, 117 and 132 positions in the final data sets,respectively

Plant Soil

Tab

le2

Sequencesimilaritiesof

excisedDGGE-bands

show

nin

Fig.5

Ecologicalniche

Province

oforigin

DGGEband

[Accession

number]

Mostcloselyrelatedsequ

ence

[Accession

number]

%of

Identity

(num

berof

bases)a

Origin

Taxon

omic

grou

p

Sorgh

umroot

Freestate

A1[K

C57

0917

]Various

cyanob

acterial

16SrRNA

gene,includ

ing

Synechococcussp.clon

eR4C

P3R

1F09

[HQ01

8568

.1]

100%

(182

)Sug

arcane

rhizosph

ere

(Brazil)

Cyano

bacteria

Lim

popo

A2[K

C57

0918

]10

0%

(180

)

North

West

A3[K

C57

0919

]10

0%

(171

)

Sorgh

umshoo

tFreestate

B1[K

C57

0920

]Various

Pan

toea

strains,includ

ingPan

toea

dispersa

strain

BH10

3[JQ76

5428

.1]

100%

(188

)Jasm

inepetal

γ-Proteob

acteria

Lim

popo

B2[K

C57

0921

]Various

Enterob

acteriaceae,includ

ingPan

toea

agglom

eran

sstrainsCECRI-IO

C29

[HM75

6482

],IG

CAR-17/07

[EF52

3431

.1]andIG

CAR-18/07

[EF52

3432

]

96%

(190

)Sea

water

γ-Proteob

acteria

Freestate

B3[K

C57

0922

]Various

Pan

toea

strains,includ

ingPan

toea

anan

atis

strain

JB1/KB-105

11[JQ51

3929

.1]

99%

(170

)Rainwater

(Ind

onesia)

γ-Proteob

acteria

Sorgh

umrhizosph

ere

Freestate

C1[K

C57

0923

]Unculturedacidob

acterium

[HM06

2411.1]

92%

(155

)Agriculturalsoil

Firmicutes

C2[K

C57

0924

]Escherichia

colistrain

G4M

80[G

U64

6119

.1]

86%

(161

)Small-scalefarm

γ-Proteob

acteria

Lim

popo

C3[K

C57

0925

]Escherichia

ferguson

iistrain

BAN86

[JX41

5362

.1]

99%

(158

)BarrenIsland

γ-Proteob

acteria

C4[K

C57

0926

]Unculturedactin

obacterium

[EU30

0221

.1]and

Firmicutes

[EF65

1750

.1]clon

es10

0%

(151

)Soil

Actinob

acteria/

Firmicutes

North

West

C5[K

C57

0927

]Unculturedactin

obacterium

[EU30

0221

.1]and

Firmicutes

[EF65

1750

.1]clon

es10

0%

(158

)Soil

Actinob

acteria/

Firmicutes

C6[K

C57

0928

]Various

γ-proteob

acteria,includ

ingVibrio

sp.U32

[AY86

4627

.1]

100%

(158

)Sea

water

γ-Proteob

acteria

Sorgh

umrhizop

lane

Freestate

D1[K

C57

0929

]Bacillus

megaterium

strainsAIM

ST3.24

.2[H

Q69

4028

.1]andAIM

ST1.Hb.20

[HQ67

0443

.1]

100%

(155

)Leaftissue

Firmicutes

D2[K

C57

0930

]Bacillus

megaterium

strain

JL35

-9[JN1184

34.1]

99%

(160

)Paddy

soil

Firmicutes

Lim

popo

D3[K

C57

0931

]Various

Bacillus

megaterium

strainsinclud

ingthe

pesticidedegradingstrain

APDB9[JX27

4543

.1]

andstrain

GMC50

01-b

[AB74

1472

.1]

99%

(155

)Grassland

soil

Firmicutes

D4[K

C57

0932

]Bacillus

megaterium

strain

JL35

-9[JN1184

34.1]

99%

(169

)Paddy

soil

Firmicutes

North

West

D5[K

C57

0933

]Various

Bacillus

megaterium

strainsJL35

-9[JN1184

34.1],APDB9[JX27

4543

.1]and

GMC50

01-b

[AB74

1472

.1]

100%

(169

)Paddy

Soil/G

rassland

soil

Firmicutes

D6[K

C57

0934

]Various

Bacillus

megaterium

strainsJL35

-9[JN1184

34.1],APDB9[JX27

4543

.1]and

GMC50

01-b

[AB74

1472

.1]

99%

(168

)Paddy

soil/grasslandsoil

Firmicutes

aThe

numbers

inparenthesescorrespo

ndto

thenu

mberof

basedused

tocalculatethelevelsof

sequ

ence

identity

Plant Soil

strategy employing beneficial plant-associated bacte-ria represents a natural, inexpensive and environmen-tally friendly solution to fertilization (Schenk et al.2012). The requirement of such PGPB-biofertilizer(s)would be for the target organism to be consistentlypresent in the target plant’s endophytic and/orrhizospheric environments, i.e. independently of any(a)biotic factor. This indeed indicates that such taxaare plant-selected and/or pre-adapted to thrive in suchenvironments.

As for many other plants (Gottel et al. 2011;Hartmann et al. 2009; Roesch et al. 2008; Rosenbluethand Martinez-Romero 2006; Seghers et al. 2004),sweet sorghum (Sorghum bicolor (L) Moench) farmedin South Africa was found to contain higher prokary-ote diversities in its rhizospheric environments (rhizo-sphere and rhizoplane) than in its endophytic ones(root, shoot, stem). Despite the fact that this representsthe different nutrient availabilities due to the release ofcarbon- and nutrient-rich compounds by the plantsinto the surrounding soil which fosters more microbialproductivity (Morgan and White 2005), it is suggestedthat endophytic diversities may often be underestimatedin metagenomic studies as the co-extracted plant DNArepresents the majority of the total extracted DNA.

As previously shown in other different environ-ments (for review, see Martiny et al. 2006), strongcorrelations between sorghum-associated bacterialcommunity assemblages and geographical originwere observed for the rhizospheric samples. Thisindicated that soil structure and composition (af-fected by agricultural practices), play a more sig-nificant role in determining the sorghum-associatedrhizospheric microbial communities than the sor-ghum plant itself (Hinsinger et al. 2009). Differ-ences in agricultural practices (as in this study)have been shown to have an effect on the physicaland the chemical compositions of the soil, as wellas on soil microbial community structures (Girvanet al. 2003). Similarly, soil structure is, with pH,one of the main parameters influencing soil micro-bial communities (Carson et al. 2007; Lauber et al.2009). The edaphic variables defining the SouthAfrican sorghum rhizospheric communities wereessentially related to the N-cycle (NO3-N, NH4-Nand % N) and pH (Fierer and Jackson 2005;Lauber et al. 2009). Based on these observations,and since one of the major roles of PGPBs is thefacilitation of plant N-uptake (Berg 2009), it is

hypothesized that the soil N-status may influencethe N-fixing capacities and/or the compositions ofdifferent plant-associated rhizosphere microbialcommunities.

In contrast, the sorghum endophytic communitystructures were highly variable, i.e. neither tissue-nor region-specific. The rhizosphere and endospherebacterial communities of Populus deltoides studied by454 pyrosequencing also presented a similar observa-tion (Gottel et al. 2011). This suggests that sorghumendophytic communities are generally constituted byopportunistic endophytes (Hardoim et al. 2008) andthat the bacterial colonization of sorghum internaltissues is a stochastic rather than a deterministic pro-cess. Nevertheless, the consistent detection of specificbacterial taxa in sorghum root (cyanobacteria) andshoot (Pantoea sp.) indicates that a minority of sor-ghum endophytes could be specifically recruited. Suchendophytes are expected to possess essential coloniza-tion traits to be competent endophytes and a fortioriPGPBs (e.g. motility, chemotaxis, capacities to com-municate with the plant and to produce cell-walldegrading enzymes) (Hardoim et al. 2008). The per-sistent detection in sorghum roots of cyanobacteriarelated to the motile and N-fixing Synechococcus sp.supports this view. A symbiotic relationship betweenSynechococcus sp. and S. bicolor can indeed be hy-pothesized, with the plant providing shelter to thecyanobyont which in return conveys fixed nitrogento its host. This hypothesis is validated by the fact thatSynechococcus sp. was detected in both the rhizo-sphere and the root of sorghum. Indeed, for successfulendosymbiosis, specific chemical signals must be re-leased (generally in the rhizosphere) by the plant host(S. bicolor) to attract its future endosymbyont(Synechococcus sp.) (Adams and Duggan 2012). Thisalso suggested that sorghum roots do constitute entriesfor endophytic bacterial colonization (Cocking 2003).However, this remains a speculative hypothesis of therole of the cyanobacterial endophyte identified in thisstudy, and further research is required to validate thisrelationship.

Despite the fact that geographical location and soilchemistry significantly impacts the structuring ofsorghum rhizosphere communities (as previouslydiscussed), such chemical signaling could also explainwhy sorghum-associated rhizospheric communitiesare less variable and with more reliably detected OTUsthan that of sorghum endosphere. This could indeed

Plant Soil

suggest that sorghum plants partly ‘control’ the com-position of their rhizosphere communities via deter-ministic recruitment processes (Gottel et al. 2011).

The two different methods (T-RFLP and DGGE)used in this study were not corroborative in the iden-tification of persistent bacterial taxa. Only cyanobacteria(Synechococcus sp.) were detected by DGGE and T-RFLP (in the roots and the rhizosphere, respectively),but not in the same niche. These results confirm that asynergistic approach is important in microbial ecologystudies (van Felten et al. 2011) since it is recognized thatthere are inherent (dis)advantages of all molecular fin-gerprinting tools: T-RFLP is highly sensitive but one T-RF may comprise more than one distinct bacterialribotype (as indicated by the fact that 10/13 consistentlydetected T-RFs could not even be assigned to a singleclass or phylum) and DGGE is limited by the resolutionof the gel but DGGE-bands of interest may (as in thisstudy) be sequenced and identified (Blackwood et al.2007; Nocker et al. 2007). This also raises the possibilityof other persistent bacterial taxa that were not detectedin this study, and a pyrosequencing approach is current-ly underway to identify other organisms at a muchhigher resolution.

The consistently detected sorghum-associatedrhizospheric and endophytic (cyano)bacterial taxa(Acetobacter sp., Azospirillum sp., Bradyrhizobiumsp., Pantoea sp., Bacillus sp. and Synechococcus sp.)represent members with known PGP capacities, par-ticularly their N-fixation capacity (Franche et al. 2009;Kevin Vessey 2003; Loiret et al. 2004; Terakado-Tonooka et al. 2008). Notably the diazotrophicAzospirillum sp. and Bradyrhizobium sp. have alreadybeen found to be associated with sorghum (Franche etal. 2009; Coelho et al. 2008). Rout and Chrzanowski(2009) also reported the presence of an endophyticdiazotrophe community in other members of the Sor-ghum genus (S. halepense). Sorghum has important N-uptake capacities, particularly in N-starved conditionswhen compared to maize, and the mechanisms in-volved in conferring this physiological trait remainunknown (Hirel et al. 2007). The role of the PGPBsymbiosis in providing bioavailable N to plants beingwell documented (e.g. Kevin Vessey 2003; Berg 2009;Franche et al. 2009), together with the identification ofconsistently sorghum-associated bacteria taxadisplaying potential N-fixation capabilities could there-fore suggest a mechanism which contributes to sor-ghum’s elevated N-uptake capacity: the consistent

recruitment of specific endophytic and rhizosphericPGPBs with N-fixing capacities.

Finally, identifying Bacillus sp. (B. megaterium) asa consistent sorghum rhizoplanic community memberis of agronomical importance, as (i) members of thisgenus have shown numerous PGP activities (N-fixa-tion, metabolite and phytohormone production, im-provement of root performances; Bai et al. 2002;Lugtenberg et al. 1991; Saharan and Nehra 2011),and (ii) transformation protocols and genetic toolsfor this genus have been established (e.g. Nijland etal. 2010), and would therefore be an ideal target for theengineering of PGP properties.

Conclusion

In this study, differences in bacterial communitystructures in micro-environments (rhizoplane/rhizosphere/root/shoot/stem) associated with sorghumfarmed in different South African provinces (Limpo-po, North West and Free State) were observed, withthe endophytic communities being more variable thanrhizospheric ones. Geographical location (hereencompassing, agricultural practices) and soil charac-teristics (soil pH, N-content [NO3-N, NH4-N and totalN] and to a lesser extent clay content) were the maindetermining factors in shaping sorghum rhizospherecommunities. In contrast, no clear factor could explainthe sorghum endophytic community structures. It wasconcluded that sorghum endosphere was essentiallycolonized by opportunistic bacteria as a result of sto-chastic bacterial assemblages. Nevertheless, the detec-tion of consistently sorghum-associated bacteria both inrhizospheric and endophytic environments was indica-tive of a deterministic mode of recruitment. Future workincluding pyrosequencing of a larger sampling set(i.e. more sorghum plants) and in-depth rhizosphericsoil chemistry to unveil potential sorghum-releasedchemotaxic compounds are suggested to further under-stand the recruitment of sorghum-associated communi-ties. The identification of (cyano)bacterial taxa(Acetobacter sp., Azospirillum sp., Bradyrhizobiumsp., Pantoea sp., Bacillus sp. and Synechococcus sp.)reliably associated with sorghum in South Africa issignificant as it lays the groundwork for their futureisolation, characterization and application. Theycould be directly used as bio-inoculants, and pos-sibly engineered with enhanced PGP activities to

Plant Soil

introduce various crop improvements. For exam-ple, antibiotic production capacities targetingsorghum-pathogens such as Fusarium sp. couldbe valuable objectives. With a worldwide annualgrowth rate of 10 % in the “microbial inoculant”market (Berg 2009), the sorghum-specific microor-ganisms identified in this study have significantpotential in agricultural biotechnology as crop im-provement tools.

Acknowledgments We thank the South African NationalResearch Foundation (NRF) for funding the study. Wewould also like to acknowledge Dr Shagi from the NorthWest academic farm of the ARC (Agricultural ResearchCouncil), Mr Osterhaizen (Free State farmer) and the com-munity farm at Jane Furst (Limpopo) for allowing us tosample sorghum. We thank anonymous reviewers whosuggested modifications that greatly improved the manu-script. J-BR holds a NRF Free-standing Postdoctoral Fel-lowship from the National Research Foundation of SouthAfrica.

References

Adams DG, Duggan PS (2012) Signaling in cyanobacteria–plant symbioses. In: Perotto S, Baluška F (eds) Signalingand communication in plant symbiosis. Springer, BerlinHeidelberg, pp 93–121

Bai Y, Aoust FD, Smith DL, Driscoll BT (2002) Isolation ofplant growth promoting Bacillus strains from soybean rootnodules. Can J Microbiol 48:230–238

Berg G (2009) Plant–microbe interactions promoting plant growthand health: perspectives for controlled use of microorganismsin agriculture. Appl Microbiol Biotechnol 84:11–18

Blackwood CB, Hudleston D, Zak DR, Buyer JS (2007)Interpreting ecological diversity indices applied to terminalrestriction fragment length polymorphism data: insightsfrom simulated microbial communities. Appl EnvironMicrobiol 73:5276–5283

Budi SW, Van Tuinen D, Martinotti G, Gianinazzi S (1999)Isolation from the Sorghum bicolor mycorrhizosphere ofa bacterium compatible with arbuscular mycorrhiza devel-opment and antagonistic towards soilborne fungal patho-gens. Appl Environ Microbiol 65:5148–5150

Carson JK, Rooney D, Gleeson DB, Clipson N (2007) Alteringthe mineral composition of soil causes a shift in microbialcommunity structure. FEMS Microbiol Ecol 61:414–423

Clarke K (1993) Non-parametric multivariate analysis of changesin community structure. Aust J Ecol 18:117–143

Clarke KR, Gorley RN (2006) PRIMER v6: user manual. Plym-outh Marine Laboratory, Plymouth

Cocking EC (2003) Endophytic colonization of plant roots bynitrogen-fixing bacteria. Plant Soil 252:169–175

Coelho MRR, de Vos M, Carneiro NP, Marriel IE, Paiva E, SeldiL (2008) Diversity of nifH gene pools in the rhizosphere of

two cultivars of sorghum (Sorghum bicolor) treated withcontrasting levels of nitrogen fertilizer. FEMS MicrobiolLett 279:15–22

Coelho MRR, Carneiro NP, Marriel IE, Seldin L (2009) Molec-ular detection of nifH gene-containing Paenibacillus in therhizosphere of sorghum (Sorghum bicolor) sown inCerrado soil. Lett Appl Microbiol 48:611–617

Compant S, Duffy B, Nowak J, Clemet C, Barka EA (2005) Useof plant-growth promoting bacteria for biocontrol of plantdiseases: principles, mechanisms of action, and future pros-pects. Appl Environ Microbiol 71:4951–4959

Culman SW, Bukowski R, Gauch HG, Cadillo-Quiroz H, Buck-ley DH (2009) T-REX: software for the processing andanalysis of T-RFLP data. BMC Bioinforma 10:171

Dicko MH, Gruppen H, Traore AS, Voragen AGJ, van BerkelWJH (2006) Sorghum grain as human food in Africa:relevance of content of starch and amylase activities. AfrJ Biotechnol 5:384–395

Felsenstein J (1985) Confidence limits on phylogenies: an ap-proach using the bootstrap. Evolution 39:783–791

Fierer N, Jackson RB (2005) The diversity and biogeography ofsoil bacterial communities. PNAS 103:626–631

Food and Agricultural Organization (FAO) (1995) FAO food andnutrition series, no. 27 Sorghum and millet in human nutri-tion. http://www.fao.org/docrep/T0818E/T0818E00.htm

Food and Agricultural Organization/World Food Production(FAO/WFP) (2010) Crop and food security assessmentmission to Mozambique. http://www.fao.org/docrep/012/ak350e/ak350e00.htm

Franche C, Lindström K, Elmerich C (2009) Nitrogen-fixingbacteria associated with leguminous and non-leguminousplants. Plant Soil 321:35–59

Funnel-Harris DL, Sattler SE, Pedersen JF (2013) Characteriza-tion of fluorescent Pseudomonas spp. Associated withroots and soil of two sorghum genotypes. Eur J PlantPathol. doi:10.1007/s10658-013-0179-6

Girvan MS, Bullimore J, Pretty JN, Osborn MA, Ball AS (2003)Soil type is the primary determinant of the composition of thetotal and active bacterial communities in arable soils. ApplEnviron Microbiol 69:1800–1809

Gottel NR, Castro HF, Kerley M et al (2011) Distinct microbialcommunities within the endosphere and rhizosphere ofPopulus deltoides roots across contrasting soil types. ApplEnviron Microbiol 77:5934–5944

Hafeez FY, Yasmin S, Ariani D, Mehboob-ur-Rahman Zafar Y,Malik KA (2006) Plant growth-promoting bacteria asbiofertilizer. Agron Sustain Dev 26:143–150

Hardoim PR, van Overbeek LS, van Elsas JD (2008) Propertiesof bacterial endophytes and their proposed role in plantgrowth. Trends Microbiol 16:463–471

Hartmann A, Schimd M, van Tuinen D, Berg G (2009) Plant-driven selection of microbes. Plant Soil 321:235–257

Hinsinger P, Bengough AG, Vetterlein D, Young IM (2009)Rhizosphere: biophysics, biogeochemistry and ecologicalrelevance. Plant Soil 321:117–152

Hirel B, Le Gouis J, Ney B, Gallais A (2007) The challenge ofimproving nitrogen use efficiency in crop plants: towards amorecentral role for genetic variability and quantitative genetics with-in integrated approaches. J Exp Bot 58:2369–2387

Kevin Vessey J (2003) Plant growth promoting rhizobacteria asbiofertilizers. Plant Soil 255:571–586

Plant Soil

Lauber C, Knight R, Hamady M, Fierer N (2009) Soil pH as apredictor of soil bacterial community structure at the con-tinental scale: a pyrosequencing-based assessment. ApplEnviron Microbiol 75:5111–5120

Loiret F, Ortega E, Kleiner D, Ortega-Rodes P, Rodes R, DongZ (2004) A putative new endophytic nitrogen-fixing bac-terium Pantoea sp. from sugarcane. J Appl Microbiol97:504–511

Lugtenberg B, De Weger L, Bennett J (1991) Microbial stimu-lation of plant growth and protection from disease. CurrOpin Biotechnol 2:457–464

Marchesi JR, Sato T, Weightman AJ, Martin TA, Fry JC, HiomSJ, Wade WG (1998) Design and evaluation of usefulbacterial primers that amplify genes coding for bacterial16S rRNA. Appl Environ Microbiol 64:795–799

Martiny JBH, Bohannan BJM, Brown JH et al (2006) Microbialbiogeography: putting microorganisms on the map. Nat Rev4:102–112

Mendes R, Pizzirani-Kleiner A, Araujo W, Raaijmarkers J(2007) Diversity of cultivated endophytic bacteria fromsugarcane: genetic and biochemical characterization ofBurkholderia cepacia complex isolates. Appl EnvironMicrobiol 73:7259–7267

Morgan JAW, White PJ (2005) Biological costs and benefits toplant-microbe interactions in the rhizosphere. J Exp Bot56:1729–1739

Murray M, Thompson W (1980) Rapid isolation of high molec-ular weight plant DNA. Nucleic Acids Res 8:4321–4325

Muyzer G, de Waal EC, Uitterlinden AG (1993) Profiling ofcomplex microbial populations by denaturing gradient gelelectrophoresis analysis of polymerase chain reaction-amplified genes coding for 16S rRNA. Appl EnvironMicrobiol 59:695–700

Nei M, Kumar S (2000) Molecular evolution and phylogenetics.Oxford University Press, New York

Nijland R, Burgess JG, Errington J, Veening J-W (2010) Trans-formation of environmental Bacillus subtilis isolates bytransiently inducing genetic competence. PLoS One5:e9724. doi:10.1371/journal.pone.0009724

Nocker A, Burr M, Camper AK (2007) Genotypic microbialcommunity profiling: a critical technical review. MicrobEcol 54:276–289

Non-Affiliated Soil Analysis Work Committee (1990) Hand-book of standard soil testing methods for advisory pur-poses, vol 160. Soil Science Society of South Africa,Pretoria

Panchal H, Ingle S (2011) Isolation and characterization of endo-phytes from the root of the medicinal plant Chlorophytumborivilianum (Safed musli). J Adv Dev Res 2:205–209

Pedersen WL, Chakrabarty K, Klucas RV, Vidave AK (1978)Nitrogen fixation (acetylene reduction) associated withroots of winter wheat and sorghum in Nebraska. ApplEnviron Microbiol 35:129–135

Reysenbach A-L, Pace NR (1995) In: Robb FT, Place AR (eds)Archaea: a laboratory. Manual – thermophiles. Cold SpringHarbour Laboratory Press, New York, pp 101–107

Rodriguez-Caballero A, Ramond J-B, Welz PJ, Cowan DA,Odlare M, Burton SG (2012) Treatment of high ethanol

concentration wastewater by biological sand filters: en-hanced COD removal and bacterial community dynamics.J Environ Manag 109:54–60

Roesch LFW, Camargo FAO, Bento FM, Triplett EW (2008)Biodiversity of diazotrophic bacteria within the soil, rootand stem of field-grown maize. Plant Soil 302:91–104

Rosenblueth M, Martinez-Romero E (2006) A review. Bacterialendophytes and their interactions with hosts. Mol PlantMicrobe Interact 19:827–837

Rout ME, Chrzanowski TH (2009) The invasive Sorghumhalepense harbors endophytic N2-fixing bacteria and alterssoil biogeochemistry. Plant Soil 315:163–172

Saharan BS, Nehra V (2011) Plant growth promotingrhizobacteria: a critical review. Life Sci Med Res 21:1–30

Saito A, Ikeda S, Ezura H, Minamisawa K (2007) A minireview.Microbial community analysis of the phytosphere usingculture-independent methodologies. Microbes Environ22:93–105

Schenk PM, Carvalhais LC, Kazan K (2012) Unraveling plant–microbe interactions: Can multi-species transcriptomicshelp? Trends Biotechnol 30:177–184

Seghers D, Wittebolle L, Top EM, Verstraete W, Siciliano SD (2004)Impact of agricultural practices on the Zea mays L. endophyticcommunity. Appl Environ Microbiol 70:1475–1482

Sercu B, van de Werfhorst LC, Murray JLS, Holden PA (2011)Cultivation-independent analysis of bacteria in IDEXXquanti-tray/2000 fecal indicator assays. Appl EnvironMicrobiol 77:627–633

Shyu C, Soule T, Bent SJ, Foster JA, Forney LJ (2007) MiCA: aweb-based tool for the analysis of microbial communitiesbased on terminal-restriction fragment length polymor-phisms of 16S and 18S rRNA genes. J Microb Ecol53:562–570

Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S(2011) MEGA5: molecular evolutionary genetics analysisusing maximum likelihood, evolutionary distance, andmaximum parsimony methods. Mol Biol Evol 28:2731–2739

Taylor JRN (2003) Overview: importance of sorghum in Africa.http://www.afripro.org.uk/papers/Paper01Taylor.pdf

Terakado-Tonooka J, Ohwaki Y, Yamakawa H, Tanaka F,Yoneyama T, Fujihara S (2008) Expressed nifH genes ofendophytic bacteria detected in field-grown sweet patatoes(Ipoema batatas L.). Microbes Environ 23:89–93

Tyson GW, Baker BJ, Alle EE, Hugenholtz P, Banfield JF(2005) Genome-directed isolation of the key nitrogen fixerLeptospirillum ferrodiazotrophum sp. nov. from an acidicmicrobial community. Appl Environ Microbiol 71:6319–6324

van Felten A, Meyer JB, Défago G, Maurhofer M (2011) NovelT-RFLP method to investigate six main groups of 2,4-diacetylphloroglucinol-producing pseudomonads in envi-ronmental samples. J Microbiol Methods 84:379–387

Zinniel DK, Lambrecht P, Beth Harris N, Feng Z, Kuczmarski D,Higley P, Ishimaru CA, Arunakumari A, Barletta RG,Vidaver AK (2002) Isolation and characterization of endo-phytic colonizing bacteria from agronomic crops and prairieplants. Appl Environ Microbiol 68:2198–2208

Plant Soil