Plant Growth-Promoting Rhizobacteria Inoculation to Enhance Vegetative Growth, Nitrogen Fixation and Nitrogen Remobilisation of Maize under Greenhouse Conditions

Plant growth-promoting rhizobacteria (PGPR) may provide a biological alternative to fix atmospheric N2 and delay N remobilisation in maize plant to increase crop yield, based on an understanding that plant-N remobilisation is directly correlated to its plant senescence. Thus, four PGPR strains were selected from a series of bacterial strains isolated from maize roots at two locations in Malaysia. The PGPR strains were screened in vitro for their biochemical plant growth-promoting (PGP) abilities and plant growth promotion assays. These strains were identified as Klebsiella sp. Br1, Klebsiella pneumoniae Fr1, Bacillus pumilus S1r1 and Acinetobacter sp. S3r2 and a reference strain used was Bacillus subtilis UPMB10. All the PGPR strains were tested positive for N2 fixation, phosphate solubilisation and auxin production by in vitro tests. In a greenhouse experiment with reduced fertiliser-N input (a third of recommended fertiliser-N rate), the N2 fixation abilities of PGPR in association with maize were determined by 15N isotope dilution technique at two harvests, namely, prior to anthesis (D50) and ear harvest (D65). The results indicated that dry biomass of top, root and ear, total N content and bacterial colonisations in non-rhizosphere, rhizosphere and endosphere of maize roots were influenced by PGPR inoculation. In particular, the plants inoculated with B. pumilus S1r1 generally outperformed those with the other treatments. They produced the highest N2 fixing capacity of 30.5% (262 mg N2 fixed plant−1) and 25.5% (304 mg N2 fixed plant−1) of the total N requirement of maize top at D50 and D65, respectively. N remobilisation and plant senescence in maize were delayed by PGPR inoculation, which is an indicative of greater grain production. This is indicated by significant interactions between PGPR strains and time of harvests for parameters on N uptake and at. % 15Ne of tassel. The phenomenon is also supported by the lower N content in tassels of maize treated with PGPR, namely, B. pumilus S1r1, K. pneumoniae Fr1, B. subtilis UPMB10 and Acinetobacter sp. S3r2 at D65 harvest. This study provides evidence that PGPR inoculation, namely, B. pumilus S1r1 can biologically fix atmospheric N2 and provide an alternative technique, besides plant breeding, to delay N remobilisation in maize plant for higher ear yield (up to 30.9%) with reduced fertiliser-N input.


Introduction
In Malaysia, both field and sweet corn varieties are highly in demand as animal feed and for human consumption. However, only the latter demand is being widely addressed through maize cultivation as cash crop due to its higher return on investment [1]. In 2012, the total maize production in Malaysia was valued at RM 334.4 million, triple the value of 2008 [2] due to the increasing local and foreign (Brunei and Singapore) demands [1]. Current maize varieties require high rates of fertiliser inputs, particularly fertiliser-N for maximum crop yield and profitability. Fertiliser-N is also required to replenish the N harvested in stover and ear of the previous season, which can be as high as 115 kg N ha −1 [3]. Thus, many farmers practise 'insurance' application of fertiliser-N to ensure adequate N supply for crop growth. This practice demands high amounts of fertilisers to grow maize, although only 30-50% of the fertiliser-N applied is absorbed by plants [4], the rest are either rendered unavailable as adsorbed soil organic-N or leached into the environment.
Plant growth-promoting rhizobacteria (PGPR) isolated as free-living soil bacteria from plant rhizosphere can decrease chemical fertiliser-N use and increase plant growth and yield when associated with plant roots and other plant parts [5]. Several bacteria such as Azospirillum [6], Klebsiella [7], Burkholderia [8], Bacillus [9] and Pseudomonas [10] have been identified as PGPR to maize plant through biological nitrogen fixation (BNF), phosphate solubilisation, phytohormone production (e.g., auxin, gibberellin and cytokinin) and biological control of soil pathogens. BNF by PGPR has been reported to contribute up to 12-70% of total N uptake in field crops or 26.7 kg N ha −1 (70% of total N uptake) in maize [6], sugarcane [11] and oil palm [9]. Generally, it has been estimated that up to 65% of N used in agriculture is contributed by BNF, and that it will be an increasingly important component in future plant-N management [12]. In addition, N remobilisation in plant plays a crucial role to reutilise the N from vegetative plant parts for developing organs, especially seeds/grains. N remobilisation occurs naturally throughout plant growth and whenever the plant requires it, although the onset of leaf senescence is identified as the main stimulus [13]. It was reported that 50-90% of N in wheat and maize grains is remobilised from the leaf-N [14].
However, present information on indigenous PGPR associations with maize plant towards BNF and their influence on N remobilisation is still limited. Thus, a concerted effort is needed for an effective plant-N management. Therefore, the aim of this study was to select effective PGPR strains from a series of indigenous bacterial strains by biochemical characterisations and plant growth promotion assays. These selected strains and a reference strain, UPMB10 were identified using 16S rDNA gene analysis and further inoculated to maize plants grown under greenhouse conditions to estimate the amount of N 2 fixed and their influence on plant-N remobilisation prior to anthesis and ear harvest, using the 15 N isotope dilution technique.

Bacterial isolation
Bacterial strains were isolated from roots of healthy maize plants grown at University Agricultural Park, Universiti Putra Malaysia (UPM), Selangor (2°58'52.17" N, 101°42'44.94" E) and Lentang Village in Sik, Kedah (6°2'57.84" N, 100°50'24.60" E) using the modified method of Hoben and Somasegaran [15]. Three fresh root tips (3 cm) with sufficient adhered rhizosphere soils were collected in McCartney bottles which contained 10 mL of sterilised distilled water. The bottle was shaken for 30 s with a vortex mixer at 1000 rpm and serially diluted with tenfold dilutions prior to spreading the root suspension on tryptic soy agar medium (TSA; Merck KGaA, Germany) to isolate the rhizospheric bacteria. The same roots were surface-sterilised with 70% ethanol for 5 min, followed by 1% of sodium hypochlorite for 30 s and washed five times with sterilised distilled water. The roots were streaked on TSA plates to check the sterilisation efficiency and aseptically smashed with mortar and pestle to isolate the endophytic bacteria on TSA medium. The TSA plates were invertedly incubated for 24 h at 30±2°C. Colonies with visual morphological differences were selected and sub-cultured to obtain pure colonies. A total of 57 bacterial strains were isolated and screened for N 2 -fixing activity, phosphate solubilisation, indole-3-acetic acid (IAA) production and plant growth promotion assay. A parallel experiment on plant growth promotion assays indicated that PGPR inoculations with Fr1, S1r1 and S3r2 produced the highest maize plant top biomass and N uptake (S1 and S2 Figs). Thus, these three PGPR strains and a negative and reference strain, Br1 and UPMB10, respectively, were subsequently selected for the pot experiment. The UPMB10 strain was isolated from oil palm root [16] and is used in the commercial product Bacto-10™. Maize and oil palm being monocotyledonous plants form a monophyletic group that shares similar arrangements of vascular bundles in the stem, parallel major leaf veins and adventitious root system.

Biochemical characterisation
Initial screening for N 2 -fixing activity of the pure bacterial cultures was determined on N-free semi-solid malate medium (Nfb) [17]. The plates were incubated for 24 h at 30±2°C, whereby a colour change from pale green to blue would qualitatively indicate the positive effect of N 2 -fixing activity. Phosphate solubilisation test was conducted according to the method of Pikovskaya [18], where the presence of clear halo zones around the colonies would indicate positive phosphate solubilisation. IAA production was determined according to the modified method of Glickmann and Dessaux [19]. PGPR cultures were grown in tryptic soy broth (TSB) which contained 2 mg mL −1 of L-tryptophan and incubated on a rotary shaker at 200 rpm under room temperature, 28±2°C for 24 h. The cultures were centrifuged at 7000 rpm for 7 min and the supernatants were quantified spectrophotometrically at 535 nm with 2 mL of Salkowski's reagent (2% of 0.5 M FeCl 3 in 35% perchloric acid) after 25 min. The IAA concentrations were estimated from a standard IAA curve.

16S rDNA gene analysis of selected bacterial strains
Total genomic DNA was extracted using Genomic DNA Mini Kit (Yeastern Biotech Co. Ltd.) according to the supplier's instructions and used as DNA template in polymerase chain reaction (PCR) for amplification of the 16S rDNA gene. The DNA purity was quantified at 260 nm and 280 nm using NanoDrop Spectrophotometer (ND1000, Thermo Fisher Scientific Inc.), 1.6 to 2.2, to detect protein contamination in the DNA [20]. PCR amplification was performed in a reaction mixture containing 50 ng genomic DNA template, 1X reaction buffer, 200 μM dNTPs mixture, 1.5 mM MgCl 2 , 2.5 U Taq DNA polymerase (Thermo Fisher Scientific), 0.16 μM of each primer 27F (5'-AGAGTTTGATCTTGGCTCAG-3') and 1492R (5'-TACGGTTACC TTGTTACGACTT-3') [21], and ultra-pure sterilised water to a final volume of 50 μL. These primers allowed an approximate 1500 bp of DNA fragments to be amplified. The DNA amplification was performed in a thermal cycler (MJ Mini Personal Thermal Cycler PTC-1148, Bio-Rad) by an initial denaturation at 94°C for 5 min, followed by 30 amplification cycles of denaturation at 94°C for 30 s, annealing at 55°C for 30 s, extension at 72°C for 1 min and a final extension step at 72°C for 5 min. The suitability of DNA amplification was visualised by electrophoresis of PCR products with 6X loading dye (Thermo Scientific™) at 5:1 (PCR product: dye) ratio and a marker (1kb DNA ladder, Fermentas GeneRuler™) in 1% (w/v) agarose gel in 1X Tris-acetate EDTA (TAE) buffer for 1 h at 80 V. The agarose gel was stained with GelRed™ (Biotium Inc.) for 40 min and examined under UV light in a UV transilluminator (Bio-Rad Molecular Imager 1 Gel Doc™ XR+ System). Gel image was captured with Image Lab software (Version 4.1, Bio-Rad Laboratories) (S3 Fig). DNA fragments were extracted from the gel by slicing the target bands under UV light and purified using QIAquick gel extraction kit (Qiagen 1 ). The purified DNA products were sent for sequencing by First BASE Laboratories Sdn. Bhd. (www.baseasia.com) using BigDye 1 Terminator 3.1 cycle sequencing kit (Applied BioSystems). The quality of 16S rDNA sequences was checked manually by using Applied Biosystems Sequence Scanner Software (version 2.0). The sequences were analysed using the BLAST Sequence Similarity Search to identify the most closely related members in the NCBI GenBank DNA database (www.ncbi. nlm.nih.gov/geo). The partial 16S rDNA sequences of the PGPR strains were submitted to the NCBI database under their respective accession number as follows: Bacillus subtilis UPMB10 (KP641618), Klebsiella sp. Br1 (KP257586), Klebsiella pneumoniae Fr1 (KP641617), Bacillus pumilus S1r1 (KP295962) and Acinetobacter sp. S3r2 (KP295963).

Phylogenetic analysis
The partial 16S rDNA sequences of UPMB10, Br1, Fr1, S1r1 and S3r2 were aligned with the most closely related bacteria sequences obtained from the NCBI database using MUSCLE [22]. The tree was constructed with Mega version 5 software package [23] by using the maximum likelihood method from distance calculated by the method of Kimura two-parameter model with a discrete Gamma distribution [24]. Gaps were treated by partial deletion and bootstrap analysis was done by using 2000 replications. Streptomyces griseus FGQ9 (HQ202539) was used as an outgroup sequence.

Pot experiment
The experimental soil (Serdang series soil, Typic Paleudult) was collected from 0-15 cm soil depth at the University Agriculture Park, UPM (2°59'12.5" N, 101°38'52.9" E). The soil was airdried, ground, sieved (2 mm mesh) and analysed for its physico-chemical properties. The soil had pH 5.1 (1:2.5, soil:water ratio; MeterLab 1 PHM210); total N (semi-micro Kjeldahl method) [25], 0.46%; total C (Leco CR-12 carbon analyser) [ [30], respectively, were applied to each polybag. The 15 N-labelled urea (46% N, 4.72 at. % 15 N e ) at 40 kg N ha −1 equivalent (a third of recommended fertiliser-N rate) [30] was dissolved in 100 mL distilled water and applied to each polybag as a tracer. The experiment was conducted in a randomised complete block design, with two harvests at 50 and 65 days after planting (DAP; D 50 and D 65 ) with four blocks and 0.75 m × 0.25 m planting distance between the polybags. The six treatments imposed were: (i) killed B. subtilis UPMB10 inoculum, autoclaved at 121°C for 20 min (Uninoculated control), (ii) B. subtilis UPMB10 inoculation (Reference control), (iii) Klebsiella sp. Br1 inoculation (Negative control), (iv) K. pneumoniae Fr1 inoculation, (v) B. pumilus S1r1 inoculation and (vi) Acinetobacter sp. S3r2 inoculation. Six maize seeds of similar size and shape, var. Hibrimas, were sowed and thinned to two per pot at 7 th DAP (D 7 ). The PGPR inocula were grown in TSB medium for 24 h (200 rpm, 26±2°C) and cell suspensions were adjusted to OD 600 between 1.4 and 2.0, which corresponded to the total plate counts of ca. 10 9 cfu mL −1 , as determined on the TSA medium. Each polybag was inoculated with 20 mL of respective inocula (live or killed) on D 0 , D 7 and D 35 . The polybags were watered and weighted daily to maintain at field capacity throughout the experiment to avoid any possible loss of applied 15 N labelled fertiliser through denitrification, if the soil contained excess moisture. The total bacterial colonisations in soil, rhizosphere and root-endosphere were determined according to the method of Hoben and Somasegaran [15] from the collected soil and fresh root samples at D 65 harvest.

N abundance plant analysis
At harvest (D 50 and D 65 ), each maize plant was separated into roots, young, ear and old leaves, stalk, tassel and ear (D 65 only). The ear bearing leaf and its immediate lower leaf were labelled as ear leaves. The upward and downward leaves from the ear leaves were respectively labelled as young and old leaves. Every component was weighed, oven-dried (65±2°C, 72 h) and ground (<1 mm) to determine the plant tissue N (except root) using the semi-micro Kjeldahl method [25] and 15 N abundance using an emission spectrometer (NOI7, Fisher Germany) at Malaysian Nuclear Agency in Bangi, Selangor. The 15 N abundance in the sample was corrected for the natural 15 N abundance present in the environment (0.3663 at. % 15 N e ). Meanwhile, the N 2 fixation in the whole maize plant (plant top basis) was calculated from the mean weighted atom excess (WAE) using the following formula [31]: Where AE = at. % 15 N e , TN = total N of a, b, c, d, e and f = tassel, young, ear and old leaves, stalk and ear, respectively, ear was only available in D 65 harvest. The proportions of N derived from the atmosphere (% Ndfa) was calculated, as follows: Ndfa ð%Þ ¼ 1 À at: % 15 N e fixing plant at: % 15 N e non À fixing plant Â 100

Statistical analysis
Data were analysed using one-way analysis of variance procedure (ANOVA) followed by Duncan's Multiple Range Test (DMRT) at p<0.05 using Statistical Analysis System software (SAS version 9.0) [32]. Pearson correlation of coefficient test was performed to estimate the relationships between related variables.

PGPR strain characterisation
The plant growth-promoting (PGP) abilities of the four selected strains and a reference strain, UPMB10 are shown in Table 1. These five strains gave positive reactions on Nfb media. In particular, four strains (UPMB10, Br1, S1r1 and S3r2) showed the clearest halo zones around their bacterial colonies grown on the Pikovskaya media, indicating higher phosphate solubilisation abilities compared to Fr1 strain. Treatment with Fr1 strain, however, produced the significantly highest IAA at 13 μg mL −1 compared to the other strains (5-11 μg mL −1 ).

Molecular identification of PGPR strains
Analysis of 16S rDNA sequences from the NCBI database suggested that UPMB10, Br1, Fr1, S1r1 and S3r2 were most closely related to Bacillus subtilis (98% similarity), Klebsiella sp. (99% similarity), Klebsiella pneumoniae (99% similarity), Bacillus pumilus (99% similarity) and Acinetobacter sp. (98% similarity), respectively. A phenogram showing the genetic relationship among the PGPR strains and their most closely related bacteria obtained from NCBI database is presented in Fig 1. The PGPR strains were indicated to belong to two subdivisions and three different genera of bacteria: (i) Gamma-proteobacteria: Klebsiella spp. (Br1 and Fr1) and Acinetobacter spp. (S3r2); (ii) Firmicutes: Bacillus spp. (UPMB10 and S1r1). Total bacterial colonisation PGPR inoculations increased rhizobacterial colonisation around the maize roots and their surrounding soils compared to uninoculated control at ear harvest (D 65 ) ( Table 2). Rhizosphere had the highest total bacterial population, followed by soil and root-endosphere. Inoculation with Klebsiella sp. Br1 and Acinetobacter sp. S3r2 gave significantly higher bacterial populations in their respective soils, with around 250-300% over the uninoculated control (1.98 x 10 7 cfu g −1 ). In the root-endosphere, the PGPR inoculations particularly with Acinetobacter sp. S3r2 significantly stimulated up to 757 folds increase in total bacterial population compared to the uninoculated control (96 cfu cm −1 fresh root).

Total nitrogen uptake
Inoculation with PGPR significantly increased the N uptake in all plant parts and plant tops of maize at D 50 and D 65 harvests, except in the tassel (D 65 ) (Table 4). Meanwhile, inoculation with B. pumilus S1r1 significantly increased the total N uptake in plant top by 55.1% and 50.1%, followed by K. pneumoniae Fr1 with similar increments of 61.4% and 48.4% at D 50 and D 65 , respectively, in comparison to uninoculated control. B. pumilus S1r1 and K. pneumoniae Fr1 also significantly increased the N uptake in tassel (D 50 only), young, ear and old leaves, stalk and ear (D 65 ) of maize. Generally, the significant differences in the N uptake of plant parts between D 50 and D 65 harvests were shown as a decrease in total N in the tassel and increased total N in the young leaves (S1r1 only), stalk and plant top.
In the tassel, the mean N uptake at D 65 (119 mg plant −1 ) was significantly lower compared to D 50 (160 mg plant −1 ), as shown among the inoculated treatments (except for Klebsiella sp. Br1). In young leaves, the mean N uptake was significantly higher at D 65 (184 mg plant −1 ) compared to D 50 (142 mg plant −1 ). However, this effect was evident only in inoculated treatment with B. pumilus S1r1. Similarly, the mean N uptake in the stalk and plant top significantly increased at D 65 (347; 1108 mg plant −1 ) compared to D 50 (253; 829 mg plant −1 ). The highest accumulation of N was found in plant stalk, containing 30.5% and 31.3% of the total plant top N at D 50 and D 65 , respectively. The lowest accumulation of N was found in old leaves (D 50 ) and plant tassel (D 65 ) at 15.1% and 10.7% of the total plant top N, respectively. In general, majority of total N (70-81%) in the plant top of maize was accumulated before anthesis (D 50 ).

PGPR biological nitrogen fixation and plant nitrogen remobilisation
Maize inoculated with PGPR generally had significantly lower weighted 15 (Table 6). In the maize plant top, inoculation with B. pumilus S1r1 gave 30.5% Ndfa (262 mg N fixed plant −1 ) and 25.5% Ndfa (304 mg N fixed plant −1 ) at D 50 and D 65 , respectively. Generally, the mean % Ndfa in the inoculated maize plant top at D 65 (17.5%) was significantly reduced in comparison to D 50 (21.1%), despite the significant increase in the mean amount of fixed N by 12.9%. Based on the fixed N content of the inoculated maize plants, the estimated mean N 2 fixation rate was higher prior to anthesis (D 50 ) at 3.5 mg N fixed plant −1 day −1 compared to after anthesis (D 65 ) at 3.1 mg N fixed plant −1 day −1 .
Inoculation with B. pumilus S1r1 recorded significantly highest % Ndfa and fixed N content in tassel (D 50

Interactions between plant and bacterial parameters and time of harvests
In the present study, the soil bacterial population was found to be positively correlated with the bacterial populations in rhizosphere and root-endosphere ( Table 7). The IAA production of PGPR strains was positively correlated with all the plant parameters, except in the top and root biomass at D 65 . There were significant interactions between PGPR strains and time of harvests for the parameters on total N uptake (tassel and plant top) and at. % 15 N e (tassel, young leaves and plant top) (S1 and S2 Tables).

Discussion
In the present study, four PGPR strains (Klebsiella sp. Br1, K. pneumoniae Fr1, B. pumilus S1r1 and Acinetobacter sp. S3r2) isolated from maize roots and a reference strain (B. subtilis UPMB10) from oil palm roots were used as inoculants for maize plants grown under greenhouse conditions. According to the 16S rDNA sequence analysis, these PGPR strains belonged  [6] reported that most of their isolated bacterial genera from maize belonged to gamma-proteobacteria subdivision and indicated a selective maize plant association with some of these bacterial genera such as Acinetobacter and Klebsiella. Other researchers have also observed that bacteria from Klebsiella genus were commonly sighted near the maize root system and soil environment [8,33]. Bacterial genera such as Acetobacter, Pseudomonas, Azospirillum, Azotobacter, Burkholderia, Herbaspirillum and Rhizobium have also been reported as effective maize PGPR. Inoculation with Azotobacter and Azospirillum on field-grown maize significantly increased the plant biomass by 30.7% [10]. Similarly, the co-inoculation of Bacillus megaterium, Azotobacter chroococcum and Bacillus mucilaginous significantly increased maize biomass and height equivalent to half of the chemical fertiliser inputs [34]. Many similar effective N 2 -fixing PGPR inoculation results have been reported on maize plant under low fertiliser-N (ca. 48 kg N ha −1 ) condition, with strains such as Bacillus spp., Klebsiella spp., Azospirillum spp., Azotobacter spp. and Pantoea spp. [6,34]. These researchers attributed the increase in plant-N uptake and dry biomass of inoculated plants to PGP abilities such as BNF, phosphate solubilisation and root promoting phytohormone production namely IAA, cytokinin and gibberellin.
The five selected PGPR strains showed positive reactions on Nfb media, a differentiating media for screening PGPR with potential BNF ability [34]. The colour change phenomenon from green (pH 7.0) to blue (above pH 7.6) is due to the bromothymol blue content which changes colour with increase in pH above neutral. This increase in pH is due to the formation of fixed-ammonia from atmospheric N 2 through natural N 2 fixation phenomenon. In addition, this N 2 fixation screening method allows selection of strains with higher survival traits in Ndeprived soil condition [10]. Besides N 2 fixation, these five strains have other PGP abilities, namely phosphate solubilisation and IAA production. The clear halo zones exhibited around the bacterial colonies in Pikovskaya media were in response to the solubilisation of insoluble inorganic phosphates by organic acids. Song et al. [35] attributed the efficiency to solubilise insoluble phosphates in Burkholderia cepacia, a PGPR isolated from cultivated soils in Korea, to its high level of organic acid productions namely gluconic acid. Other organisms such as Pseudomonas aeruginosa and B. megaterium have also been reported to have high phosphate solubilisation efficiency [36].
The PGPR strains in this study produced considerable amounts of IAA (5-13 μg mL −1 ) which were comparable to those presented in other reports. Among other, Egamberdiyeva et al. [37] reported 0.3 μg mL −1 and 0.5 μg mL −1 for Pseudomonas alcaligenes and Mycobacterium phlei, whereas Sachdev et al. [38] reported the highest level of 27.5 μg mL −1 from their collections of K. pneumoniae strains. In nature, IAA is synthesised by plants and PGPR from amino acid tryptophan, a common precursor in root exudates through transamination and decarboxylation biochemical reactions [19]. It is estimated that 80% of soil rhizosphere bacteria can produce IAA, whereas almost all (98%) the PGPR strains isolated by Arruda et al. [7] from the plant rhizospheres were able to produce IAA. In this study, the positive correlations between IAA productions by PGPR and most maize plant growth parameters (plant top and root biomass) at D 50 strongly suggested the influence of phytohormone IAA on plant growth. However, these positive plant growth effects from IAA on plant top and root biomass were not apparent at ear harvest (D 65 ), possibly due to the pot environment. IAA enhances plant growth mainly through extension of root system to reach larger soil volume for increased water and nutrient uptake. The common soil volume in all treatments could gradually limit this beneficial effects of IAA on root development, especially at the later stage of plant growth, ear harvest (D 65 ) [39].
Rhizosphere usually has higher bacterial biomass and activity than the bulk soil due to the photosynthetically assimilated carbon sources such as carbohydrates, amino acids, amides, vitamins and organic acids in the root exudates [5]. This phenomenon is known as the "rhizosphere effect", where essential nutrients for soil bacterial growth are abundantly available [40]. In this study, although the PGPR inoculated rhizospheres had higher bacterial colonisation than the uninoculated control, their differences were not statistically significant, which was possibly due to the nature of unsterilised soil condition that allowed competition between the introduced PGPR and indigenous soil bacteria [5]. In general, the total bacterial populations (10 8 cfu g −1 ) in the PGPR inoculated rhizospheres were deemed as sufficient to promote nitrogenase activity [34,41] in their free living state [40]. The significant correlation between bacterial populations in rhizosphere and soil suggests the phenomenon of bacterial migration from the bulk soil to the rhizosphere due to the bacterial ability to respond chemotactically [42]. Furthermore, it was hypothesised that bacterial mobility could be accelerated by soil water movement [43] in which the soil was maintained at field capacity throughout the study. Similarly, the root-endosphere of maize inoculated with PGPR showed higher bacterial populations, particularly with Acinetobacter sp. S3r2 treatment.
Generally, PGPR inoculations (B. subtilis UPMB10, Klebsiella sp. Br1, K. pneumoniae Fr1, B. pumilus S1r1 and Acinetobacter sp. S3r2) significantly increased the total N content and dry biomass of maize throughout the study (D 50 and D 65 ). Among the PGPR strains, inoculation with B. pumilus S1r1 and K. pneumoniae Fr1 recorded higher amounts of total N content and dry biomass in the respective whole plants, plant tops and different plant parts (tassel, young, ear and old leaves, stalk, ear and root) of maize prior to anthesis (D 50 ) and ear harvest (D 65 ). These increments were strongly attributed to the inherent BNF abilities, as indicated by the significantly lower at. % 15 N e in the plant tops and different plant parts of the inoculated maize compared to the uninoculated control. Lower at. % 15 N e value indicates a marked increase in percentage of N within inoculated maize plant is derived from atmospheric-N [44]. Notably, maize plants in this study were grown under unsterilised condition to simulate the actual field environment where the agricultural soil is generally unsterilised. This practice is prone to underestimate the N 2 fixation rate of PGPR since the rate is calculated based on at. % 15 N e of reference plant, which will be indiscriminately influenced by the BNF of indigenous soil bacteria. Therefore, a suitable reference plant with similar plant type, root system and plant growth is crucial in a 15 N isotope dilution study to minimise plant variations due to the influence of indigenous soil bacteria [45]. Factoring the conditions into account, the 15 N isotope dilution technique is widely regarded as the most accurate and the only direct method available to quantify N status in plant and soil studies for short and long term experiments without any isotope effect or health risk under growth chamber, greenhouse and field conditions [9,46,47].
In the present study, maize inoculated with B. pumilus S1r1 accorded the lowest at. % 15 N e or the highest % Ndfa (as well as relative N 2 fixed amount) in the plant top, followed by K. pneumoniae Fr1. Inoculation with B. pumilus S1r1 contributed substantial amounts of fixed N to maize plant top of 262 mg N 2 fixed plant −1 (30.5% Ndfa) and 304 mg N 2 fixed plant −1 (25.5% Ndfa) at D 50 and D 65 , respectively, which were equivalent to 14.0 kg N ha −1 and 16.2 kg N ha −1 , respectively, based on an equivalent planting density of 53,333 plants ha −1 . According to Schröder et al. [48], the initial three weeks of maize plant growth requires approximately 0.50 kg N ha −1 day −1 ; thus, the amount of fixed N (0.25-0.28 kg N ha −1 day −1 ) from B. pumilus S1r1 inoculation could potentially complement ca. 50% of the total plant-N requirement. Other researchers have reported contributions of fixed N up to 26.7 kg N ha −1 or 70% of total plant-N in sugarcane inoculated with Azospirillum spp. [11]. The main advantage of N derived from BNF is due to the complete uptake of readily fixed ammonia within the plant with no losses to the environment. N losses between 50-70% of the inorganic fertiliser-N in soils through natural processes such as volatilisation, denitrification and nutrient leaching have been reported [4]. Moreover, plants cultivated in the tropical soils of low pH favour the uptake of N from ammonium or amino acid sources [49]. These N sources can influence the plant-N content since plants grown under ammonium-N condition will have twice the amount of N in their vegetative parts compared to those grown under nitrate-N condition [50].
Meanwhile, the at. % 15 N e in the plant tops and in the different plant parts of maize generally decreased upon ear harvest (D 65 ). This phenomenon suggests a continuous contribution of unlabelled N from BNF and soil sources towards the dilution of at. % 15 N e in maize (inoculated and uninoculated control) until ear harvest. Nonetheless, the estimated rate of N 2 fixed plant −1 day −1 in inoculated maize had declined after anthesis (from 3.5 to 3.1 mg N 2 fixed plant −1 day −-1 ), and this was possibly due to the occurrence of N feedback, as reported in Arabidopsis [51]. N feedback occurs when a strong N sink such as seed rapidly develops and induces high N remobilisation from senescing plant parts to phloem [46]. This sudden N spike in plant phloem has been reportedly to inhibit nitrogenase activity [51]. According to Thomas and Smart [52], this phenomenon can be mediated by effective post-anthesis plant nitrogen use efficiency (NUE) and N remobilisation to delay plant senescence.
The 15 N isotope dilution technique can directly label the N in different plant parts to denote the two simultaneous N fluxes: (i) N remobilisation from the plant parts to grain and (ii) exogenous N uptake to the plant parts and grain [47,53]. Conventionally, plant-N remobilisation is estimated using the differences in N contents from respective plant parts and stages of plant developments, namely stalk elongation, prior to anthesis, grain filling and maturation [54,55,56]. This approach is known as balance remobilisation technique which can lead to a biased estimation of remobilised N, as exogenous N uptake distribution is neglected and assumed to be completely allocated to the grain [47]. According to Masclaux-Daubresse et al. [53], the plant grain yield is determined by the collaborate efficiency of N uptake, assimilation, translocation and remobilisation. Most researchers stated that around 45-90% of N in grain of maize is derived from existing N stored in the plant prior to anthesis, while the balance is derived from post-anthesis N uptake, depending on plant genotypes and environmental conditions [14]. In this study, the majority (70-81%) of N content in maize plant top was accumulated prior to anthesis (D 50 ). These stored N is continuously remobilised from structural compounds of senescing plant parts through proteolysis to developing plant parts in order to recycle the N and increase plant NUE [46]. Proteolysis is a controlled and coordinated degradation of photosynthetic protein of plastids such as chloroplast into soluble proteins for N remobilisation [53].
N remobilisation in plants could influence C production, as remobilised N is primarily sourced and initiated from senescing photosynthetic plant parts [57]. According to Uhart and Andrade [13], certain soil conditions such as limited rate of N fertilisation and water supply could increase N remobilisation efficiency in plants but would result in reduced C filling and grain yield. Conversely, delayed plant senescence can increase C filling and grain yield, but it will lead to decreased grain protein content [53]. Consequently, PGPR applications could provide a solution to manipulate plant senescence and continuous exogenous fixed N to plants, and concomitantly maintain high grain yield and protein content [52]. This pioneering study has demonstrated the effects of PGPR on N fluxes in maize plant prior to anthesis (D 50 ) and at ear harvest (D 65 ) which could create a new approach to plant-N management.
Prior to anthesis (D 50 ), the majority of N assimilated during the early growth of maize was derived from labelled fertiliser-urea source, instead of unlabelled soil and atmospheric sources, as indicated by the highest at. % 15 N e in the old leaves of maize (D 50 and D 65 ). This phenomenon demonstrated the role of old leaves as a major sink for N during the early maize growth [57]. Concurrently, N remobilisation from old leaves to photosynthetically active plant parts such as ear and young leaves might have occurred before anthesis, as suggested by the lowest N uptake in the old leaves (D 50 ) and the significant increments in N uptake and dry biomass of young leaves at D 65 . The old leaves continued to be the source of remobilised N for developing plant parts (young leaves, stalk and ear) after anthesis, as indicated in the significantly lower at. % 15 N e value at D 65 , while the N uptake and dry biomass parameters remained similarly low at D 50 . As ear leaves were optimal for photosynthate production at D 50 , the N usage for further organ growth has significantly reduced, as exhibited in the unchanged N uptake and dry biomass parameters of ear leaves between D 65 and D 50 harvests.
At D 50 , some tassels have experienced dehiscence, particularly from the uninoculated and Klebsiella sp. Br1 treated maize plants, as exhibited in their unchanged N uptakes at D 65 harvest. At the same time, inoculations with other superior N 2 -fixing PGPR (B. pumilus S1r1, K. pneumoniae Fr1, B. subtilis UPMB10 and Acinetobacter sp. S3r2) have delayed the dehiscence of tassels, thus exhibiting significant reductions in N uptakes at D 65 harvest. In addition, the influence of PGPR on N remobilisation in maize was clearly demonstrated in the significant interactions found between the PGPR strains and the time of harvests in N uptake and at. % 15 N e parameters of tassel.
During grain formation, the ear leaves functioned as supporting organs to continuously channel exogenous and remobilised N to ears [55]. However, the N remobilisation in ear leaves determined using balance remobilisation technique (N uptake and dry biomass) showed no apparent changes between D 50 and D 65 . Favourably, the significantly lower at. % 15 N e in ear leaves of D 65 harvest suggested that the N content was diluted with "later" unlabelled-N uptakes from BNF and soil sources. The significant increments of fixed N (% Ndfa) in ear leaves of inoculated maize at D 65 further emphasised the role of ear leaves as temporary transits to ears, which had the lowest at. % 15 N e at D 65 . Meanwhile, B. pumilus S1r1 displayed its superiority in BNF as indicated by the highest amount of fixed N found in the ear leaves instead of the plant stalk, where most fixed N of other PGPR strains was located at D 65 harvest.
Amidst the fluctuations of N fluxes in the tassel, leaves (young, ear and old) and ear of maize, the at. % 15 N e in plant stalk remained mostly consistent throughout the plant growth, despite having the highest N uptake and dry biomass at D 50 and D 65 harvests. In addition to the significant increments of plant biomass and N uptake of stalk at D 65 , these phenomena demonstrated the role of plant stalk as an important N reservoir for maize [54,55]. According to Ta and Weiland [57], plant stalks could contribute similar amounts of remobilised N as the accumulated N in plant leaves to the grain, ca. 40%. The present plant stalks which contained between 30.5-31.3% of total plant top N could possibly have several times higher nitrate concentration than the N in the leaves [50] and the nitrate-N is readily available for N remobilisation [54]. In general, the present study has demonstrated that leaves (old, ear and young), tassel and stalk of maize plants had served successively as N sinks and N sources towards ear formation.
Nonetheless, plant roots were not investigated in the present study using 15 N isotope dilution technique due to their lesser involvement in N fluxes, as they exhibited a low representation of plant biomass (<8%) compared to above ground plant parts [54]. According to Salon et al., [46], plant roots function mostly as mechanical support and nutrient uptake for growth. Furthermore, the at. % 15 N e in the above and below ground plant parts are normally similar and either of these plant parts will give similar % Ndfa [31]. This study has clearly demonstrated that the isolated PGPR, particularly B. pumilus S1r1 which fixed a significant amount of atmospheric N 2 , promoted vegetative growth and delayed plant senescence of maize, thereby produced a higher N content and yield of maize ear.

Conclusions
This greenhouse study has demonstrated that inoculation with locally isolated PGPR strains, mainly Bacillus pumilus S1r1, Klebsiella pneumoniae Fr1, Bacillus subtilis UPMB10 and Acinetobacter sp. S3r2 could significantly increase plant-N uptake, dry biomass and ear yield of maize. These increments are mainly attributed to the BNF ability of the strains, namely B. pumilus S1r1, which is able to fix up to 304 mg N 2 fixed plant −1 at ear harvest (D 65 ), and possibly other PGP abilities such as IAA production and phosphate solubilisation. The biomass production of plant tops and roots is correlated positively with IAA production by the PGPR strains. The maize plants inoculated with PGPR strains also exhibited delayed plant senescence, particularly in the tassel, which consequently improved the ear yield. This positively demonstrates that the remobilisation of N accumulated in maize top prior to anthesis from the leaves (old, ear and young), tassel and stalk have served successively as N sinks and N sources toward the ear yield. Thus, this study indicates that PGPR inoculation can be considered as an alternative technique to improve grain yield besides the conventional plant breeding method for "delayed senescence" varieties, which is time consuming and tedious. Further studies are necessary to evaluate (i) the suitability and performance of B. pumilus S1r1 on maize under field conditions and (ii) the PGPR mechanisms involved in delaying plant senescence for higher N accumulation.  Table. ANOVA Output of total N uptake in plant top and in different parts of maize inoculated with PGPR at D50 and D65 harvests. (PDF) S2 Table. ANOVA Output of at. % 15Ne in plant top and in different parts of maize inoculated with PGPR at D50 and D65 harvests. (PDF)