Screening Digitaria eriantha cv. Suvernola Endophytic Bacteria for Maize Growth Promotion

The search for sustainable agriculture has increased interest in using endophytic bacteria to reduce fertilizer use and increase stress resilience. Stress-adapted plants are a potential source of these bacteria. Some species of these plants have not yet been evaluated for this, such as pangolão grass, from which we considered endophytic bacteria as potential plant growth promoters. Bacteria from the root, colm, leaves, and rhizospheric soil were isolated, and 132 strains were evaluated for their in vitro biological nitrogen fixation, IAA and siderophores production, and phosphate solubilization. Each mechanism was also assessed under low N availability, water stress, and low-solubility Fe and P sources in maize greenhouse experiments. All strains synthesized IAA; 63 grew on N-free media, 114 synthesized siderophores, and 46 solubilized P, while 19 presented all four mechanisms. Overall, these strains had better performance than commercial inoculant in all experiments. Still, in vitro responses were not good predictors of in vivo effects, which indicates that the former should not be used for strain selection, since this could lead to not testing strains with good plant growth promotion potential. Their heterologous growth promotion in maize reinforces the potential of stress-adapted plant species as potential sources of strains for inoculants.


Introduction
Population growth demands increased food production, while current and forecasted environmental conditions require a significant decrease in agriculture's ecological footprint. This has led to the increasing use of plant-growth-promoting bacteria (PGPB), since these can reduce fertilizer use and increase crops' resilience to environmental stresses [1]. These are usually endophytic or rhizospheric bacteria that do not provoke any harm to the plant and may, under some conditions, promote their growth or reduce the effects of environmental stresses [2].
Endophytic bacteria may increase plant growth and adaptation to environmental stress [3]. This plant growth promotion is variously credited to several mechanisms, such as biological nitrogen fixation, synthesis and release of phytohormones such as IAA, phytopathogen control through synthesis and release of siderophores, or phosphate solubilization, several of which are frequently evaluated under in vitro conditions [4][5][6].
On the other hand, several plant species well adapted to stressful environments, such as pangolão grass (Digitaria eriantha cv. Survenola), have not been well studied until now, even though we have found it to harbor a very diverse bacterial endophyte population [13], since the literature contains several examples of plants from these environments being effective sources of plant-growth-promoting bacteria [14][15][16]. At the same time, it is also known that bacteria isolated from one species may be highly effective on another; for example, as happens with the commercial inoculants based on Azospirillum brasilense in Brazil [17]. Other example is the use of bacteria isolated from Stipa tenacissima L. and used to reduce salt stress on tomato [16], from Opuntia ficus-indica increasing wheat growth under drought [18], from garlic on common beans growth [19], or from Suaeda nudiflora on Amaranthus viridis under salt stress [3].
We aim, thus, to evaluate this diversity as a source of plant-growth-promoter bacteria both in vitro and for maize plants.

Binding of In Vitro Growth-Promoting Mechanisms
Grouping these characteristics led to eight groups with 100% similarity (Figure 2). Group I (GI) included two strains that grew without N and produced IAA, while GII also had two strains that, besides the above, also solubilized phosphate. GIII with four strains was like GII but did not grow in N-free media. In comparison, the 10 strains of GIV could only produce IAA, the 19 strains of GV presented all mechanisms, and the 40 strains of GVI had all but P solubilization. GVII is formed by 34 IAA-and siderophore-producing strains, and GVIII by 21 strains with all mechanisms bar growth in N-free media.

Binding of In Vitro Growth-Promoting Mechanisms
Grouping these characteristics led to eight groups with 100% similarity (Figure 2). Group I (GI) included two strains that grew without N and produced IAA, while GII also had two strains that, besides the above, also solubilized phosphate. GIII with four strains was like GII but did not grow in N-free media. In comparison, the 10 strains of GIV could only produce IAA, the 19 strains of GV presented all mechanisms, and the 40 strains of GVI had all but P solubilization. GVII is formed by 34 IAA-and siderophore-producing strains, and GVIII by 21 strains with all mechanisms bar growth in N-free media.

Figure 2.
Pangolão grass endophytic bacteria grouped by in vitro growth promotion mechanisms. N-growth in N-free media; I-IAA production; S-siderophore production; P-calcium phosphate solubilization; +-bacteria presents the trait; −-bacteria does not present the trait.

Efficiency of Growth-Promoting Bacteria in the Early Stage of Maize
Generally, the pangolão isolated strains yielded higher results than CI for at least two variables in the N-restricted experiment (Table 1). Although no strain led to significantly higher SDM than CI, several were significantly higher in RECI, indicating an overall Pangolão grass endophytic bacteria grouped by in vitro growth promotion mechanisms. N-growth in N-free media; I-IAA production; S-siderophore production; P-calcium phosphate solubilization; +-bacteria presents the trait; −-bacteria does not present the trait.

Efficiency of Growth-Promoting Bacteria in the Early Stage of Maize
Generally, the pangolão isolated strains yielded higher results than CI for at least two variables in the N-restricted experiment (Table 1). Although no strain led to significantly higher SDM than CI, several were significantly higher in RECI, indicating an overall higher growth promotion activity. However, all were significantly lower in RE + C than the +C treatment, indicating that the growth promotion could not fully overcome N deficiency effects.  Although no strain from Pangolão led to higher LA than CI, most other variables had at least two strains with higher results ( Table 2). Most importantly, two strains had higher SDM and RECI, while several increased RDM, and four achieved higher RE + C than fully irrigated plants. 0.03 0.04 0.08 0.10 0.11 0.10 0.09 0.08 0.08 0.08 +C = 100% pot-water holding capacity; NI = no inoculation; CI = commercial inoculant; PH = plant height; CD = colm diameter; LA = leaf area; SDM = shoot dry mass; RDM = dry mass of the root system; SAN = accumulation of N in the aerial part; SAP = accumulation of P in the aerial part; RECI = efficiency relative to the commercial inoculant; REN = non-inoculated control relative efficiency; RE + C = relative efficiency positive control. CV = coefficient of variation. SDE = residual standard deviation. Height, LA, SDM, RDM, SAN, and SAP data were transformed in to log10. Means followed by an asterisk differ significantly from inoculation with the recommended strain at 0.05 probability by Dunnett's test.
Although no strain significantly differed from CI for any of the variables on the low-Fe solubility experiment (Table S2), plants with strain 252 were significantly higher, and strain 5227 had a higher SDM. While no strain achieved higher SDM, SAN, or relative efficiencies than CI, several led to significantly higher PH, CD, LA, RDM, and SAP (Table 3). Some had non-significant higher RE + C than +C plants.  14 0.14 0.14 0.14 +C = recommended dose of P 2 O 5 provided by triple superphosphate; NI = no inoculation; CI = commercial inoculant; PH = plant height; CD = colm diameter; LA = leaf area; SDM = shoot dry mass; RDM = dry mass of the root system; SAN = accumulation of N in the aerial part; SAP = accumulation of P in the aerial part; RECI = efficiency relative to the commercial inoculant; REN = non-inoculated control relative efficiency; RE + C = relative efficiency positive control. CV = coefficient of variation. SDE = residual standard deviation. All data were transformed by log10. Means followed by an asterisk differ significantly from the commercial inoculant control at 0.05 probability, by Dunnett's test.

Binding of In Vitro and In Vivo Growth-Promoting Mechanisms
Most in vitro plant-growth-promoting characteristics were not significantly correlated to plant growth effect in maize (Table 4), and even those which were significant, such as SI and PH for the siderophore experiment (r = 0.25, p < 0.01), or between SI and SDM in the P solubilization experiment (r = 0.30, p < 0.01), had relatively low predictive values.

Discussion
This is the first evaluation of the plant-growth-promoting potential of Pangolão grass endophytic bacteria, which is particularly interesting since endophytic bacteria from plants from stressful environments such as this might help reduce environmental stress effects on crops [5,18]. Strains previously isolated from this plant species [13] included several genera known to harbor plant-growth-promoting strains. Indeed, all 132 strains tested under in vitro conditions here had at least one commonly held plant-growth-promoting trait.
These traits occurred in different proportions when other plant species were evaluated. For example, while 48% of the strains grew in N-free media, strains from other grasses grown in the Brazilian tropical semiarid included 66% diazotrophs [20], and signal grass root and rhizosphere bacteria had 58% diazotrophs [21].
IAA production, on the other hand, seems to be much more widely spread, since 100% of the strains were positive for this trait, as also happened with strains from cold-tolerant rice [22,23], and strains from plants under saline conditions also had 86% occurrence for this trait [24]. A similarly frequent feature seems to be siderophore production since we found 86% of siderophore-producing strains and signal grass had very similar results at 84% occurrence [25].
Most Pangolão-isolated strains had at least one of the commonly evaluated in vitro growth-promoting characteristics [26], while strains with more than one of these tended to have stronger plant-growth-promoting effects under environmental stress [8]. This may indicate that these bacteria are an important part of Pangolão adaptation mechanisms to the environmental stress typical of where it was collected. This is furthered by the presence of Pantoea, Enterobacter, Rhizobium, Pseudomonas, and Stenotrophomonas strains, which are frequently described as promoting plant growth for several species [14,[27][28][29][30][31][32].
The high proportion of in vitro plant growth characteristics found here agree with the frequent consideration that plants adapted to stressful environments tend to have plant-growth-promoting endophytic bacteria [8,26,33]. Although we evaluated only the initial maize growth, several strains increased some of the plant variables found in plants receiving a CI with known effective strains for this crop under low N availability [17]. This is likely due to IAA effects on root growth, and thus on water and nitrogen absorption [2].
Interestingly, the strains with higher N accumulation in the low-water-availability experiment were all identified as Rhizobium, Paenibacillus, and Agrobacterium species, all of which are known as potentially diazotrophic [34,35], which may indicate some level of unmeasured biological nitrogen fixation.
On the other hand, the small effects of both low P and Fe solubility experiments might be due to the early stage on which we harvested the maize, as also observed for wheat [32]. Still, even under those limiting conditions, some strains induced higher growth than the CI.
A common occurrence in our experiments was bacteria with low values for a given in vitro mechanism achieving higher than CI values for a plant-growth promotion in the experiment designed to evaluate that trait. This is coupled with a generally low and non-significant correlation between in vitro and maize effects and might be due to most of the strains presenting more than one of the mechanisms (Figure 2), as also proposed in the literature and in the typical indication of using a mix of strains for inoculant production [36][37][38][39][40].

Materials and Methods
Plant sampling, bacterial sampling, isolation, DNA extraction, BOX-PCR grouping, and sampling were all described in Alves et al. [13]. In short, endophytic and rhizospheric bacteria from Pangolão grass from three Pernambuco state municipalities were isolated, through serial dilutions in sterile saline solution, in semi-solid, N-free, NFB media [41], while in one of the locations, JNFB and JMV media were also used [42,43]. Where growth was observed in the N-free media, bacteria were transferred to YMA media for phenotypical characterization [44], followed by DNA extraction and BOX-PCR fingerprinting. Representative strains of the BOX-PCR 90% similarity groups had their 16S rRNA sequenced. They were chosen for the in vitro evaluation of biological nitrogen fixation [41], IAA production [45], calcium phosphate solubilization [46], and siderophore production [47].
Each of these mechanisms was later evaluated in a separate greenhouse experiment with maize to evaluate specific environmental stresses, such as low available nitrogen, water deficit, and low-solubility P and Fe sources, respectively.
All experiments were conducted in 5 L plastic vessels filled with sterile sand: vermiculite 1:1 mixture, using maize cultivar BR-5026, also known as São José. This cultivar is recommended for low-resource agriculture in tropical semiarid regions and was developed as a multipurpose cultivar for forage production, immature corn, or for grain production. It has dented yellow cobs, 300 cm tall plants, a 120-day cycle, and a stable yield.
Seeds were surface-disinfested with 70% ethanol for 30 s, immersed in 2.5% sodium hypochlorite for 2 min, washed eight times with sterile distilled water, and inoculated with 1 mL per seed of 10 9 cells.mL −1 bacterial broth of the strain under evaluation, according to standard Brazilian recommendations for the commercial inoculation of corn [17]. All experiments were conducted on a randomized block design with three replicates.
As biological nitrogen fixation is not considered to support all plant growth for nonlegumes [48], all plants in this experiment were supplied with 30% of the recommended N dose at seeding. Twenty strains were evaluated; seven were positive for N-free media growth, and the remainder did not grow in this media ( Table 6). The control treatments for this experiment were non-inoculated plants inoculated with the commercial Azospirillum brasilense inoculant AzzoFix ® (CI). This is a liquid inoculant for corn with strains Ab-V5 and Ab-V6 and 2.0 × 10 8 UFC·mL −1 , as defined and authorized according to Brazilian legislation [17], and 100% N dose supply (equivalent to 20 kg N·ha −1 ) (+C) under a low N input recommendation for the tropical semi-arid condition in Pernambuco state [49]. Table 6. Strains selected for the maize growth-promotion experiments. BNF-reduced N availability, IAA-water deficit stress, P-low P solubility, and Siderophore-low Fe solubility. Grey cells indicate those in common to all experiments. + and -indicate presence or absence of the trait, respectively.
For the P solubilization capacity, tri-calcium phosphate was used as the sole P source, and 21 bacterial strains were selected to represent the 0-75, 75-95, and 95-100 percentiles for in vitro P solubilization, respecting the proportion of non-solubilizing bacteria effectively found similarly to the BNF experiment (Table 6). Control treatments were similar to the previous experiments, with the positive control being triple superphosphate as the phosphorus source (+C).
Siderophore was evaluated by replacing the Fe source in the nutrient solution with a low-solubility one. In this case, 24 strains were selected to represent the 0-5, 5-25, 25-50, 50-75, 75-95, and 95-100 percentiles for in vitro siderophore production (Table 6), and the control treatments were similar to the previous experiments, with the positive control being a high-solubility Fe source (+C).
For all experiments, harvest was 20 days after emergence, plant height (PH), colm diameter (CD), and leaf area (LA) were measured, and root (RDM) and shoot dry matter (SDM) was determined. Traditional wet methods determined N and P contents, and shoot accumulated N (SAN) and P (SAP) were calculated by the product of shoot dry matter and the appropriate content.
We also calculated the relative efficiencies of each treatment in relation to the noninoculated control (REN), commercial inoculant (RECI), and each positive control (RE + C). In all cases, these were calculated according to this equation.

RE(%) =
Treatment Shoot Dry Matter Average Shoot Dry Matter o f the appropriate control × 100 (1) The in vitro data were grouped at 100% similarity using the Jaccard dissimilarity index and UPGMA. Maize data were evaluated for homoscedastic and outliers. When needed, they were transformed by log 10 and elimination of outliers before ANOVA and Dunnett's test at 5% significance using the commercial inoculant as the control treatment. Later, only the inoculant treatments were evaluated, considering their grouping from the in vitro evaluations, as described for each experiment, and considering the variation within each group as a random effect. The Pearson linear correlation was evaluated for each experiment, considering the in vitro values and the plant variables.

Conclusions
Endophytic bacteria strains from pangolão grass increased initial plant growth for at least some variables of those provided by known effective Azospirillum strains under four different environmental stress conditions, likely due to this plant being adapted to stressful environments. Thus, its microbiome is an essential component of this adaptation.
The lack of substantial predictive value for the in vitro plant-growth-promotion characteristics, as seen by the low and generally non-significant correlations and the apparent disconnect between these and plant effects, indicates that the common practice of choosing strains for plant evaluation based on the in vitro evaluation might be flawed and needlessly reduce the genetic diversity in the plant evaluation.
Although the maize experiments were designed to evaluate specific environmental stresses purportedly linked to each in vitro plant-growth-promoting characteristic, the plant promotion is not directly related to individual mechanisms.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/plants12142589/s1, Table S1: In vitro growth promotion evaluation. BNF-growth in N-free media. IAA-IAA synthesis. SI-calcium phosphate solubilization index. Siderophore-siderophore synthesis; Table S2: Evaluation of strains associated with pangolan grass in terms of in vivo siderophore production and corn growth 20 days after germination.