Introduction

Endophytes are ubiquitous in nature and occupy intercellular spaces of living tissues. These microbes have been isolated from a wide diversity of plants and have been found associated with all plant tissues including roots, stem, leaves, seed, flowers, etc. [26, 32]. Endophytic bacteria are those bacteria which colonize the interiors of plant tissue without having any negative impacts on the host plant [12]. Endophytes generally have a mutualistic interaction with the host plant wherein the host provides a nutrient-rich protective environment for these endophytes, while these microbes produce useful metabolites and signals resulting in plant growth promotion [51]. These microbes are also known to provide tolerance to abiotic stresses, thereby helping the host plant survive and adapt to different environmental conditions [40]. The endophytic bacteria may promote plant growth through various mechanisms including N2 fixation, P solubilization, phytohormone production, production of antifungal metabolites, or induction of resistance to phytopathogens [13].

Bacteria belonging to diverse phylogenetic affiliations, such as α-proteobacteria, β-proteobacteria, γ-proteobacteria, Firmicutes, Actinobacteria, and Bacteroidetes have been isolated from different plant species [16]. However, the abundance and diversity of bacterial endophytes in different plants and different tissues within a host plant varies. Diversity of bacterial endophytes is influenced by many factors such as plant species, developmental stage of the plant, and soil type. Even within the same soil, different plants may harbor different species of bacterial endophytes [6]. Recent reports have indicated that environmental factors and host genotype can also affect the diversity of bacterial endophytes within host plants [1]. In fact, within a single plant species, different genotypes may harbor different endophytic communities [14, 31]. Host plant plays an active role in colonization by different endophytic bacteria [42]. Through defense enzymes, it may also restrict the population of specific endophytic species.

Pearl millet is the sixth most important cereal worldwide and is the main food source in the poorest regions of India. These are small-seeded cereals of potential value particularly in semi-arid regions because of their short-growing season and their adaptability to hot and dry regions. Pearl millet is usually grown in India as a rainfed crop, although in certain regions it is also grown under irrigated conditions. Drought- tolerant and drought-susceptible genotypes suitable for rainfed and irrigated conditions, respectively, have been developed. It is presumed that the microbiota-colonizing interiors of different pearl millet genotypes may vary; drought susceptibility and tissue type may also probably influence the genetic and functional diversity of the endophytes inhabiting different tissues of pearl millet. Further, endophytes are known to provide stress tolerance and also influence plant survival under stress conditions [23, 29]; thus, the diversity of osmotolerant bacterial endophytes inhabiting pearl millet genotypes exhibiting different drought tolerance grown under rainfed conditions may also vary. Keeping this view in mind, the aim of the present investigation was to study the genetic diversity of osmotolerant bacterial endophytes colonizing the various tissues of pearl millet genotypes differing in their drought susceptibility and to functionally characterize these endophytes. Understanding their diversity, distribution, and host preferences will help us to better explore their potential as bioinoculants for drought-stressed regions, which may have immense implications in food security for the poorest regions of India.

Materials and Methods

Endophytic Bacterial Counts and Isolation

Pearl millet (Pennisetum glaucum L.) genotypes, namely PPMI-69 (drought susceptible) and TT-1 (drought tolerant) were used for isolation of endophytic bacteria [43]. These genotypes were grown in the experimental fields of the ICAR-Indian Agricultural Research Institute, New Delhi, India (28° 38′ 23″ N and 77° 09′ 27″ E). The soil characteristics were pH 8.0, organic carbon 0.49%, available nitrogen 0.058%, available phosphorus 0.0014%, available potassium 0.015%, and EC 0.23 dS/m.

The plants were sampled at boot leaf stage of the crop. Three plants from each genotype were uprooted and immediately transferred to the laboratory. Three replicates from each of the genotype were pooled, and then the leaves, roots, and stems were separated. Soil was removed from plant parts by gentle washing under tap water. The plant parts were kept in filter papers for drying, and then 10-g samples were weighed and surface sterilized. The roots and stems were surface sterilized using ethanol (70% v/v) for 30 s, followed by mercuric chloride (0.1% w/v) for 5 min. For the leaves, time interval for surface sterilization using mercury chloride was reduced to 30 s. After sterilization, these were thoroughly washed with sterile water to completely remove mercuric chloride. The plant samples were then macerated aseptically using pestle and mortar, serially diluted and then 100-μl aliquots from suitable dilutions were spread on different media plates. Media used for isolation of bacteria were methyl red nutrient agar medium for Gram positive bacteria [11], crystal violet nutrient agar medium for Gram negative bacteria [8], semi-solid modified Rennie’s combined carbon medium for diazotrophs [41], King’s B agar medium for fluorescent pseudomonads, and R2A medium for oligotrophs [2]. Petri plates were incubated at 28 ± 2 °C in an incubator for 48 h and then counts were taken. Bacterial population was expressed as the number of colony forming units (cfu) per gram fresh weight. Colonies differentiated by color and morphology were picked and purified. The stock cultures were maintained on nutrient agar slants at 4 °C. Subculturing was done as and when required.

Screening for Osmotolerance

Screening of the isolates for osmotolerance was carried out in nutrient broth medium supplemented with 30% PEG 6000 [3]. The nutrient broth (5 ml) was inoculated with 2% inoculum from broth cultures, and the tubes were incubated on an orbital shaker at 28 ± 2 °C for 48 h. After that, the bacterial growth obtained was measured at A600 nm using a spectrophotometer (Model EZ201, Perkin Elmer). Three replicates per treatment were maintained. Percent reduction in bacterial growth in comparison to that obtained under control conditions (without PEG 6000) was calculated. Cultures showing lesser reduction in growth in presence of 30% PEG 6000 were selected for further studies.

Screening for Plant Growth-Promoting Activities

To determine the P solubilization activity, the isolates were inoculated in Pikovyskaya’s broth [39] and incubated at 28 ± 2 °C for 7 days on an orbital shaker. Three replications of each treatment were maintained. Uninoculated broth served as control. Broth cultures were then centrifuged at 8000 rpm for 10 min to pellet the cells. Supernatant was used to estimate the amount of phosphate solubilized by the method of Jackson [18], while pellet was used for protein estimation by the method of Lowry et al. [27]. The amount of P solubilized was expressed as microgram P solubilized per milligram protein.

Nitrogen fixing ability was determined by inoculating the isolates in semi-solid modified Rennie’s combined carbon medium and incubating the tubes at 28 ± 2 °C for 3 days in an incubator. Three replications of each treatment were maintained. Uninoculated medium served as control. After incubation, 10% of the air space in the tubes was replaced with acetylene, and again the tubes were incubated for 24 h. At the end of the incubation period, nitrogen fixing ability of the cultures was determined by the acetylene reduction assay (ARA) by the method described earlier [37]. ARA was expressed as nanomoles of ethylene produced per milligram protein per hour.

IAA production ability of the isolates was determined by inoculating the cultures in Luria broth supplemented with filter sterilized tryptophan solution @ 100 μg/ml. These tubes were incubated at 28 ± 2 °C for 3 days in an orbital shaker. Three replications of each treatment were maintained. Uninoculated broth served as control. At the end of incubation, the broth was centrifuged at 8000 rpm for 10 min to pellet the cells. Supernatant was used for the estimation of IAA by the method of Hartmann et al. [17], and pellet was used for protein estimation as described earlier. IAA production by the cultures was expressed as microgram of IAA produced per milligram protein.

ACC deaminase activity was determined by the method of Penrose and Glick [38]. The isolates were grown in DF salts broth. After 24 h of incubation, these were centrifuged at 8000 rpm for 10 min, pellets washed with sterile water 2–3 times and resuspended in sterile water, and 2 μl of the culture broth was spotted on petri plates containing DF salts minimal medium supplemented with filter sterilized 3.0 mM ACC. The petri plates were incubated at 28 ± 2 °C in an incubator for 48 h, and then culture growth was recorded. Appropriate negative (nitrogen free) and positive (containing ammonium sulphate) controls were also maintained. Three replicates per treatment were maintained.

HCN production by the isolates was detected as described earlier [28]. Briefly, King’s B medium amended with 4-g/l of glycine was poured in the wells of sterile tissue culture plates and allowed to solidify. The isolates were spot-inoculated individually in these wells. The lid of these tissue culture plates was lined with sterile Whatman filter paper No. 1, soaked in picric acid (0.5% in 2% sodium carbonate), then the tissue culture plates were sealed with parafilm and these were incubated at 28 ± 2 °C for 10 days. At the end of incubation, filter paper was observed for the development of orange color, and the intensity of the color was recorded. Three replications for each treatment were maintained.

Siderophore production by the isolates was detected by using chrome azurol assay (CAS) developed by Schwyn and Neilands [45]. Solutions A and B of CAS were autoclaved separately and mixed before pouring in petri plates. The isolates were then spotted on the petri plates and incubated at 28 ± 2 °C for 7 days. Siderophore production was confirmed by the development of halo around the colony upon incubation.

DNA Extraction and 16S rDNA Amplification

The selected osmotolerant cultures were grown by inoculating a single colony from a freshly streaked nutrient agar medium petri plate in 5 ml of LB broth and incubating it at 28 ± 2 °C for 24 h. Log phase cultures were used for isolation of total genomic DNA using the ZR Genomic DNA II isolation kit (Prolab) as per the manufacturer’s instructions. The 16S rRNA gene from the bacterial genomic DNA was amplified using the universal bacterial primers 27f (5′ AGAGTTTGATCCTGGCTC 3′) and 1492r (5′ TACGGTACCTTGTTACGACTT 3′) [24] using standard PCR conditions. In a reaction mixture, 50–100 ng of template DNA primer, 10 pmol each dNTP (200 μM each), and 2.5-U Taq DNA polymerase (MbI Fermentas) were used. Amplification was carried out in a thermal cycler (Model Q Cycler II, Quanta Biotech) under standard conditions with initial denaturation at 94 °C for 2 min, followed by 30 cycles of amplification (94 °C for 1 min, 56 °C for 1 min, and 72 °C for 1 min and 30 s) and final extension at 72 °C for 6 min. PCR product was resolved in 1.2% of agarose gel in 1× TAE buffer incorporated with (10 mg/ml) ethidium bromide, run at 60 V for 1.5 h, and visualized on a gel documentation system (Alpha Imager). Amplified product was purified by using the Nucleo spin gel (Macherey-Nagel) PCR cleanup kit according to the instructions given in the manual.

Restriction Fragment Length Polymorphism Analysis and Sequencing

Purified amplified product was digested with the restriction endonucleases Alu-1, Msp-1, and Mbo-1 (MBI, Fermentas) in a 20 μl of reaction volume using recommended buffer respectively, at 37 °C for 3 h. Restricted PCR product was resolved by electrophoresis at 45 V for 2 h in 1.5% agarose gel in 1× TAE buffer and visualized on a gel documentation system.

Strong and clear bands were scored as binary data. Numerical taxonomy analysis program (NTYSIS) package (Exerter software 2.02e package, USA) was used to score similarity and clustering analysis using the binary data. Jaccard’s coefficient was used to calculate the similarity among the isolates, and dendrogram was constructed using the UPGMA method [34]. Isolates were grouped based on their restriction patterns in the 16S rRNA gene-ARDRA analysis.

Amplified 16S rDNA of the representative isolates from each cluster/group was partially sequenced. Sequencing of the purified DNA was done using automated fluorescent sequencer. The sequence was compared from BLASTn search with the known and identified cultures of NCBI database (http://ncbi.nlm.nih.gov/blast) and then submitted to the GenBank.

Phylogenetic Analysis

The nucleotide sequences of the representative isolates were aligned using the Clustal W program, and the phylogenetic tree was constructed using neighbor-joining algorithm with the Poisson correction and 1000 bootstrap replicates in the MEGA 6 program.

Data Analysis

Statistical analysis of the data was performed using the OPSTAT statistical software (http://14.139.232.166/opstat/default.asp). The bacterial counts were transformed using Log10 transformations before statistically analyzing the data.

Results

Bacterial Population and Morphotypes in Different Tissues of Pearl Millet

Highest endophytic bacterial population was observed on semi-solid modified Rennie’s medium in the root tissues of the drought-tolerant and the drought-susceptible genotypes of pearl millet (log10 5.93 cfu/g fr wt and log10 5.44 cfu/g fr wt), respectively, indicating presence of a large population of putative diazotrophs (Fig. 1). Second highest population (probably of heterotrophic bacteria) in the root tissues of the drought-tolerant genotype was observed on King’s B medium (log10 4.36 cfu/g fr wt). Although, bacterial growth (heterotrophs) was obtained on King’s B medium, no fluorescent pseudomonads population was recorded. Second highest population of endophytes, in the root tissues of the drought susceptible genotype of pearl millet, was obtained on methyl red medium (log10 3.83 cfu/g fr wt), indicating a large population of Gram positive bacteria. Very few bacterial colonies were obtained on crystal violet nutrient agar medium, indicating a very low population of Gram negative bacteria in the root tissues of both the genotypes of pearl millet.

Fig. 1
figure 1

Population of specific groups of endophytic bacteria obtained from the roots, the stem, and the leaf tissues of the drought-tolerant (TT-1) and the drought-susceptible (PPMI-69) genotypes of pearl millet. Bacterial population was expressed as log cfu/g

Full size image

Highest endophytic populations obtained in the stems and the leaves of the drought-tolerant (log10 5.60 cfu/g fr wt and log10 5.62 cfu/g fr wt, respectively) and the drought-susceptible (log10 5.42 cfu/g fr wt and log10 5.05 cfu/g fr wt, respectively) genotypes of pearl millet were of putative diazotrophs. Second highest group was of Gram positive bacteria in the stems of the drought-tolerant and the leaves of the drought-susceptible genotypes, and oligotrophs in the leaves of the drought-susceptible genotype. However, Gram negative bacteria formed the second largest group in the stem tissues of the drought-susceptible genotype. Nearly similar population of Gram positive bacteria was obtained in the stems and the leaves of the drought-susceptible (log10 3.44 cfu/g fr wt and log10 3.56 cfu/g fr wt, respectively) and the leaves of the drought-tolerant (log10 3.94 cfu/g fr wt) genotypes of pearl millet. In these tissues also, no fluorescent pseudomonads could be obtained on the medium specific for these bacteria.

A total of 113 and 105 isolates were obtained on various media from the different tissues of the drought-tolerant and the drought-susceptible genotypes of pearl millet, respectively. A very high number of morphologically diverse bacterial cultures were obtained on R2A medium from the root, the stem, and the leaf tissues of the drought-tolerant genotype of pearl millet (Fig. 2). Diversity of putative diazotrophs was high in the stem tissues (11); however, in spite of maintaining a very high population, these were the least diverse group in the leaves (2). Highest morphological diversity of the bacterial cultures was obtained on King’s B medium from the root tissues of the drought-susceptible genotype of pearl millet, followed by R2A medium. Although highest population among different bacterial endophytes was recorded for putative diazotrophs in the root tissues, these were among the least diverse group, as indicated by the number of morphologically different colonies obtained from the root tissues (5). Gram positive bacteria were the most diverse group among the stem endophytes. Highest diversity for the leaf bacterial endophytes was obtained on R2A medium, followed by modified Rennie’s medium.

Fig. 2
figure 2

Number of morphotypes of the different specific groups of endophytic bacteria obtained from the roots, the stem, and the leaf tissues of the drought-tolerant (TT-1) and the drought-susceptible (PPMI-69) genotypes of pearl millet

Full size image

Osmotolerant Endophytes and Their Tissue Specificity

All the isolates obtained from the tissues of both the genotypes of pearl millet were screened for osmotolerance using 30% PEG 6000 (Fig. 3). Isolates obtained from the drought-tolerant genotype had a higher proportion of osmotolerant isolates. Around 51.4% (55 out of 107) and 57.8% (59 out of 102) of the isolates from the drought-tolerant and the drought-susceptible genotypes of pearl millet, respectively, showed 50% or lower growth as compared to no stress control treatment. Thirty one cultures comprising of 29% and 19 cultures comprising of 18.6% of the total isolates obtained from the drought-tolerant and the drought-susceptible genotypes of pearl millet, respectively, were selected for further studies. These cultures showed less than 40% reduction in growth. Tissue specific distribution of the selected osmotolerant isolates in both the genotypes of pearl millet was studied. Out of the 31 selected cultures from the drought-tolerant genotype of pearl millet, 12 were from the root and the stem tissues while five were isolated from the leaves. Out of the 19 selected cultures from the drought-susceptible genotype of pearl millet, four, five, and ten cultures were isolated from the root, the stem, and the leaves, respectively.

Fig. 3
figure 3

Effect of osmotic stress on the growth of the endophytic bacterial isolates. Bacterial growth is expressed as percentage of growth obtained as compared to that obtained under control conditions

Full size image

Functional Characterization of the Osmotolerant Endophytes for PGP Traits

All the osmotolerant bacterial isolates were observed to possess multiple plant growth-promoting activities. Endophytic bacterial isolates from the root and the stem tissues of the drought-tolerant genotype; and the stem tissues of the drought-susceptible genotype had moderate incidence of P solubilizers, while isolates from the leaves of the drought-tolerant genotype and from the roots and the leaves of the drought-susceptible genotype had a low incidence of P solubilizers. Endophytic bacterial isolates from all the tissues of the drought tolerant; and the stem and the leaves of the drought-susceptible genotypes had a very high incidence of IAA producers (Table 1). Most of the isolates possessed IAA production ability. In contrast, only 50% of the isolates from the roots of the drought-susceptible genotype possessed IAA production ability. None of the isolates from the drought-tolerant genotype possessed N2 fixation ability, while 25% of the isolates from the drought-susceptible genotype possessed N2 fixation ability.

Table 1 Incidence of plant growth-promoting activities in the bacterial endophytes from different genotypes of pearl millet
Full size table

Endophytic isolates from the roots and the stems of the drought susceptible genotype had a very high incidence (80–100%) of HCN producers while the leaves had a moderate incidence (50%). In contrast, the roots and the stems of the drought tolerant genotype had a moderate incidence of HCN producers (58.3–66.7%), while the leaves had a very high incidence of HCN producers (71.4%). Isolates from all the tissues of both the genotypes of pearl millet showed a very high incidence of siderophore production ability (70–100%). Surprisingly, isolates from all the tissues of the drought-susceptible genotype showed a moderate incidence of ACC deaminase activity (40–60%). Isolates from the roots and the leaves of the drought-tolerant genotype showed a low incidence of ACC deaminase activity (16.7–28.6%) while those from the stem showed a moderate incidence of this activity (50%).

Genetic Diversity of Osmotolerant Endophytes

The 50 osmotolerant bacterial endophytes were grouped according to their ARDRA pattern and were found to be grouped into 16 clusters at 93% similarity (Fig. 4). Sixty percent of the isolates representing nine clusters had only one isolate, indicating the uniqueness of the isolate and high diversity among the osmotolerant endophytic bacteria. Highest number of isolates (21) could be grouped together in a single cluster indicating the ubiquitous nature of this group of endophytes. The isolates belonging to this cluster were obtained from all the plant tissues, namely root, stem, and leaves of both the genotypes of pearl millet. Rest all of the clusters were represented by isolates ranging from 2 to 4 in number.

Fig. 4
figure 4

Cluster analysis of 16S rDNA restriction pattern for determination of percent similarity between the isolates

Full size image

The representative isolates from all the clusters were identified, and it was observed that in spite of the osmotolerant isolates being grouped into 16 clusters, these isolates belonged to only four bacterial genera (Table 2). Bacillus sp. was the predominant genus. Isolates from ten clusters out of the 16 belonged to different species of Bacillus. The Bacillus isolates belonging to six different species, namely B. cereus, B. endophyticus, B. anthracis, B. axarquiensis, B. thuringiensis, and B. licheniformis were recovered from the different tissues of pearl millet. Second most abundant genus was Pseudomonas sp. Isolates identified as B. cereus belonged to more than one cluster (3), indicating that there was a diversity among the different isolates of this species. Other endophytic bacterial genera present were Macrococcus caseolyticus, Pseudomonas sp., Stenotrophomonas sp., and Stenotrophomonas maltophila. Thus, Gram positive bacteria were the predominant osmotolerant endophytes colonizing the different tissues of pearl millet.

Table 2 16S rRNA gene partial sequence-based identification of representative isolates of osmotolerant endophytic bacteria from ARDRA clusters
Full size table

Endophytes belonging to the genus Bacillus could be clustered into three distinct groups (Fig. 5a). In the first group were B. cereus, B. thuringiensis, B. licheniformis, and Bacillus sp. KPRL9; the second group had only one isolate Bacillus sp. KPRR22 while B. anthracis, B. endophyticus, and B. axarquiensis belonged to the third group. On the other hand, all the Stenotrophomonas sp. appeared to be closely related and fell into a single cluster which also included S. maltophila (Fig. 5b). Both the Pseudomonas sp. were not very closely related and were also very distinct from those available in the GenBank (Fig. 5c).

Fig. 5
figure 5

Phylogenetic relationships showing relatedness of 16S rDNA between the endophytic isolates of species of Bacillus (a), the isolates of Stenotrophomonas and the selected reference isolates derived from the GenBank database (b), the isolates of Pseudomonas sp. and the selected reference isolates derived from the GenBank database (c). Values at the node represent the bootstrap confidence estimates. GenBank accession numbers have been indicated for the reference isolates and the isolates obtained in the present study. The scale bar indicates 0.1 nucleotide substitution per nucleotide position

Full size image

Highest diversity of osmotolerant endophytic bacteria was observed in the roots, followed by the stem tissues of the drought-tolerant genotype of pearl millet. All the tissues of the drought-susceptible genotype had a very low diversity of bacterial endophytes. B. axarquiensis was the most abundant osmotolerant species and represented 21 isolates which were all grouped into a single cluster (Suppl. Table 1). This species was ubiquitous and different isolates belonging to B. axarquiensis were recovered from all the tissues, namely the root, the stem, and the leaves of both the genotypes of pearl millet (Table 3). B. cereus was the second most abundant osmotolerant species. However, B. cereus isolates were obtained only from the roots and the leaves of the drought-tolerant genotype of pearl millet. Many of the clusters were either genotype specific or tissue specific. Isolates belonging to seven clusters were unique to the roots of the drought-tolerant genotype. Species belonging to the genus Stenotrophomonas was found only in the different tissues of the drought-tolerant genotype while M. caseolyticus was isolated only from the drought-susceptible genotype and was found in all the tissues of this genotype. Surprisingly, B. thuringiensis was found only in the stem tissues of both the genotypes, and B. cereus was found only in the roots and the leaves of the drought-tolerant genotype.

Table 3 Distribution of the bacterial species in the different tissues of drought-tolerant and drought-susceptible genotypes of pearl millet
Full size table

PGP and Biocontrol Activities of the Representative Isolates

Most of the representative isolates possessed multiple plant growth-promoting traits. All the representative isolates were IAA producers (Table 4). IAA production ranged from 0.60 to 50.89 (μg IAA/mg protein). Highest IAA production was observed in B. endophyticus strain KPRR16 isolated from the roots of the drought-tolerant genotype. Seven out of the 16 representative isolates had the ability to solubilize phosphate. P solubilization ranged from 0.09 to 4.66 (μg P solubilized/ml. Only one B. cereus strain MPSL7 was observed to possess nitrogen fixing ability (282.0 nmoles of ethylene produced/mg protein/h). Four bacterial isolates out of the 16 representative isolates were observed to possess ACC deaminase activity. Out of the 16 representative isolates, 13 showed siderophore production ability, and only one of the isolate possessed HCN production ability.

Table 4 Plant growth-promoting and biocontrol activities of the representative isolates
Full size table

Discussion

The aim of our investigation was to study and compare the colonization pattern and diversity of the endophytic bacteria in the different tissues of the pearl millet genotypes differing in their drought tolerance and to functionally characterize them. In both the drought-tolerant and the drought-susceptible genotypes, there was a higher population of Gram positive bacteria and a lower population of Gram negative bacteria in the root tissues. However, throughout all the plant tissues, population densities of Gram positive bacteria were higher in the tissues of the drought-tolerant genotype as compared to the same tissues of the drought-susceptible genotype. In contrast, higher population densities of Gram negative bacteria were noted in the tissues of the drought-susceptible genotype as compared to the same tissues of the drought-tolerant genotype. The leaf tissues of the drought-tolerant genotype were the only exception, where a higher population of Gram negative bacteria was present, as compared to the drought-susceptible genotype. Predominance of Gram positive bacteria in the root and Gram negative bacteria in the stem and the leaf tissues have been noted by earlier workers in rice [13, 30]. Zinniel et al. [52] reported isolation of equal percentage of Gram positive and Gram negative bacteria from the aerial parts of agronomic crops such as soybean, sorghum, wheat, and corn. However, they obtained a higher percentage of Gram negative bacteria from the aerial tissues of prairie plants.

Certain similarities in the endophytic colonization pattern of both the genotypes of pearl millet were observed, like absence of fluorescent pseudomonads as an endophyte. Surprisingly, putative diazotrophs formed the most predominant group of endophytes in all the tissues of both the genotypes. However, in spite of maintaining high populations as endophytes, except for the stem tissues of the drought-tolerant genotype and the leaves of the drought-sensitive genotype, a very low diversity of this group was observed in the different plant tissues. Grasses and other monocots including cereals and millets are known to harbor a high population of diazotrophic endophytes [22]. However, based on morphological and physiological characteristics, these could be grouped into four groups in cultivated and wild rice. Similarly, although in pearl millet the endophytic diazotrophic population ranged between 103 and 104 cfu/g fr wt, it was colonized by only five different morphotypes [10]. Diazotrophs are known to supplement part of the nitrogen requirement in sugarcane, rice, and maize [19, 21]. Similar mechanism may be helping pearl millet obtain its requirement of nitrogen through these diazotrophs leading to the noted lack of response of different genotypes of pearl millet to nitrogen fertilization [35].

During the screening for osmotolerance, isolates obtained from the drought-tolerant pearl millet genotype showed a higher incidence of osmotolerance as compared to the isolates from the drought-susceptible genotype. Along with the role of genotype in imparting drought tolerance, microbes are also known to enhance environmental fitness of a plant. Higher incidence of osmotolerant bacteria colonizing the interiors of this genotype may be one of the factors contributing towards improving their drought tolerance. Higher incidence of fermentative siderophore producing bacteria (SPB) in the roots under oxic conditions and their displacement by oxidative SPB under anoxic conditions in rice, indicated that physiological state of the plant influenced the composition of SPB community [25]. Similarly, increase in population of the endophytic P solubilizing bacteria was noted in plants treated with insoluble phosphate, and no such changes were observed in the control plants [13].

All the osmotolerant isolates from both the genotypes of pearl millet could be grouped into 16 clusters based on ARDRA pattern. However, only two of the clusters were represented by a number of isolates, while most of the clusters were represented by a single isolate only. This indicated the uniqueness of these endophytes. Identification of the representative isolate from each cluster indicated that Bacillus was the predominant genera. Out of the 16 clusters, ten were identified as different species of Bacillus. Osmotolerance, spore-forming ability, and motility presumably played an important role in favoring endophytic colonization of Bacillus [36]. The other important genera observed were Stenotrophomonas and Pseudomonas. Bacillus, Pseudomonas, and Stenotrophomonas have been reported to be the most common endophytes [9, 15, 20]. Further, 16S rRNA gene sequence analysis indicated that there was not much diversity within Bacillus isolates as these could be grouped into only three clusters. Many of the species were found to be closely related to each other. Similarly, not much diversity was observed among isolates of Stenotrophomonas sp. and S. maltophila, while isolates belonging to Pseudomonas sp. were not observed to be closely related.

B. axarquiensis isolates probably comprised the core endophytic microflora of these genotypes since different isolates belonging to this species were recovered from all the tissues of both the genotypes of pearl millet. The results suggested that most of the endophytic bacteria displayed genotype/tissue specific distribution. Most of the species of Bacillus showed specificity to the drought-tolerant genotype and were not recovered from the drought-susceptible genotype. The only exception was Bacillus thuringiensis, which showed tissue specificity. M. caseolyticus showed genotype specificity towards the drought-susceptible genotype. Few species of Bacillus such as B. endophyticus, B. anthracis, and B. licheniformis were found colonizing only the roots of the drought-tolerant genotype. While Stenotrophomonas sp. and S. maltophila were localized to only the root and the stem tissues of the drought-tolerant genotype. Maize genotypes significantly influenced the relative abundance of the main bacterial genera indicating specific relationship between the endophytic community and the host genome [5]. In an investigation conducted to study the relative species abundance among the plant tissues, it was noted that some species were widely distributed in all the plant tissues while others could be isolated only from specific tissues [4, 49,50,51].

Bacterial endophytes are known to possess multiple PGP activities [44] which were also observed in the case of the osmotolerant endophytes obtained from both the pearl millet genotypes. The core microflora comprising of isolates of B. axarquiensis showed nearly similar incidence of different PGP activities in both the genotypes of pearl millet. These isolates had a high incidence of IAA production ability and moderate incidence of P solubilization and ACC deaminase activities (Suppl. Fig. 1). Higher incidence of IAA production and P solubilization abilities were observed in other endophytes isolated from the roots of the drought-tolerant genotype in comparison to the drought-susceptible genotype indicating preferential selection/recruitment of endophytes with these PGP abilities. Drought stress is known to affect hormonal homeostasis in plants leading to stunting of growth. IAA-producing endophytes and rhizobacteria have been reported to mitigate drought stress in plants possibly by supplementing the endogenous pool of IAA and modifying root system architecture [7, 47]. Based on their physiological needs, the drought-tolerant genotype may favor colonization by such beneficial endophytes which may contribute to the higher tolerance of the plant to stresses [33, 42].

None of the osmotolerant isolates from the drought-tolerant genotype were observed to possess nitrogen fixing ability while a moderate incidence of nitrogen fixing ability was observed in the osmotolerant isolates from the drought-susceptible genotype. Incidence of ACC deaminase activity was higher in the osmotolerant endophytes isolated from the different tissues of the drought-susceptible genotype of pearl millet as compared to the drought-tolerant genotype, indicating their preferential recruitment by the drought-susceptible genotype. Genotypic variation in accumulation of ethylene under water deficit stress has been reported in wheat [48]. Significantly higher accumulation of ethylene was observed in the drought-susceptible genotype, which is known to adversely affect plant growth. Endophytic bacteria with ACC deaminase activity can reduce stress by reducing ethylene levels in the plant, thereby promoting plant growth [46], which may have been the probable reason for their preferential recruitment by the drought-susceptible genotype of pearl millet. It was earlier reported that plant species-specific selection of endophytes was driven by functional traits based rather than phylogeny in three agriculturally important grasses species, viz., Dactylis glomerata L., Festuca rubra L., and Lolium perenne L. [49].

In conclusion, in the present investigation, we found that B. axarquiensis was uniformly presented in all the tissues of both the genotypes of pearl millet and may comprise the core microflora. Isolates belonging to this species showed a high incidence of IAA production and a moderate incidence of P solubilization and ACC deaminase activity. However, other endophytes showed genotype or tissue specificity. Other species of Bacillus and Stenotrophomonas sp. were specific to the drought-tolerant genotype while M. caseolyticus was specific to the drought-susceptible genotype. B. thuringiensis showed tissue specificity only. Functional characterization of the endophytes indicated that there was higher incidence of IAA producers and P solubilizers in endophytes from the roots of the drought-tolerant genotype as compared to the drought-susceptible genotype. In contrast, a higher incidence of ACC deaminase activity was found in endophytes from the drought-susceptible genotype as compared to the drought-tolerant genotype. Hence, host played a role in the selection of not just the microbial community but their recruitment was also driven based on microbial functional traits.