Associative bacterial diversity of pangolão, a stress-resilient tropical grass

: Some forage species, such as pangolão grass ( Digitaria eriantha Steud. cv. Survenola), are resilient in tropical semiarid regions. A possible reason for this is the presence of endophytic and rhizospheric microorganisms. Thus, this study evaluated the diversity of associative bacteria in pangolão grass. Bacteria associated with the roots, culm, leaves, and rhizospheric soil were isolated and characterized in three municipalities of Pernambuco, Brazil. An initial phenotypic characterization was followed by a genotypic assessment by based repetitive extragenic palindromic-polymerase chain reaction (BOX-PCR) and partial sequencing of the 16S rRNA gene. We obtained 325 phenotypically-characterized isolates grouped into 243 strains with 100% similarity by BOX-PCR. The most diverse sampling environment was Araripina, and all factors affected bacterial diversity. There were 135 groups with 90% similarity, that were represented by a single strain each for sequencing. Among the sequenced strains, 118 showed 96.84–99.9% similarity with previously described strains, whereas 17 could not be identified. The following 18 genera were identified from three phyla, five classes, seven orders, and 13 families: Achromobacter , Agrobacterium , Bacillus , Burkholderia , Curtobacterium , Enterobacter , Herbaspirillum , Kosakonia , Ochrobactrum , Paenibacillus , Pantoea , Priestia , Pseudomonas , Rhizobium , Serratia , Shinella , Stenotrophomonas , and Variovorax . The diversity of endophytic and rhizospheric bacteria may contribute to the resilience of pangolão , as various strains of these genera have been described as plant growth promoters. This is the first evaluation of pangolão bacterial diversity under tropical semiarid conditions. Since several of the genera include strains known to promote plant growth, we propose further research to evaluate this on crops.


INTRODUCTION
Brazilian livestock production is mainly based on pastures, either native or cultivated. Certain species located in these pastures, such as Digitaria eriantha Steud. cv. Survenola, known as pangolão grass, are tolerant to water deficit and low soil fertility (Navarro et al. 2005).
A possible coping mechanism for these conditions is the presence of endophytic microorganisms, that assist in plant growth and development and induce resistance to biotic and abiotic stresses, such as through phytostimulation, biofertilization, or biocontrol (Afzal et al. 2019).
Endophytic bacteria are found in the majority of plant species (Afzal et al. 2019), and many endophytes have been detected in association with a wide range of grasses with variable diversity (Lu et al. 2021). Unfortunately, no data were found in literature on endophytic bacteria from Digitaria species, and little information is available on these bacteria in tropical semiarid conditions. https: //doi.org/10.1590/1678-4499.20220071

MATERIAL AND METHODS
Sampling was conducted in three municipalities of Pernambuco, Brazil, covering climates ranging from hot dry semiarid (Araripina) to hot tropical with winter rains (Gravatá and Nazaré da Mata, with lower average rainfall in the former), classified as BSh and As according to Köppen-Geiger's classification (CONDEPE/FIDEM, 2016) (Fig. 1). Gravatá and Nazaré da Mata pangolão pastures were on private properties and approximately 6 and 11 years old, with unknown fertilization practices. In Araripina, the samples were collected at the Experimental Station of Pernambuco Agronomic Institute and were at least 30 years old, with unknown fertilization during this time. After the first sampling in this field, part of the area was limed by the research station, allowing observation of the effects of liming on endophytic bacterial diversity.
At each sampling (Table 1), 20 plants were randomly collected from the field, grouped, and separated into leaves, culms, roots, and rhizospheric soil, followed by bagging and refrigeration until further analysis. All samplings were conducted in the rainy seasons for Gravatá and Nazaré da Mata, while Araripina was sampled during both the rainy and dry seasons. A single field was used for each location, except for Araripina, in which samples were collected in both limed and non-limed fields during the rainy season. In each site in Araripina, soil was collected, dried at 60 °C, sieved through a 2 mm mesh, and subjected to textural and chemical characterization based on Brazilian standard protocols (Teixeira et al. 2017). Leaves and colm were disinfected by washing in water, drying, swabbing in 70% ethanol, and rinsing in distilled autoclaved water. The roots were washed in water, cut into 10 cm pieces, immersed in 70% ethanol for 30 s, followed by 2.5% sodium hypochlorite, and washed five times with distilled autoclaved water. All plant samples were subsequently blended in autoclaved saline solution at a 10 -1 dilution, followed by serial dilution from 10 -3 to 10 -7 in triplicate, also using autoclaved saline solution (Döbereiner et al. 1995).
At each dilution, samples were inoculated into penicillin flasks with semi-selective N-free semi-solid NFB culture media (Döbereiner et al. 1995), while in Araripina JNFB (Baldani et al. 1986) and JMV (Baldani et al. 1996) were also used. In all cases, the flasks were incubated at 28 °C for at least 48 h, after which samples from any dilution with visible growth were plated onto Petri dishes with yeast malt agar (YMA) media and bromothymol blue (Vincent 1970) and characterized by pH change (acidic, neutral, or alkaline), mucous production (absent or present), colony color (pink, yellow, cream or cream and yellow), opacity (opaque or translucent), form (circular or irregular), perimeter (full or irregular), and surface (smooth or irregular) (Silva et al. 2012).  All isolates were inoculated in 5 mL of trypticase soy broth (TSB) medium and shaken at 180 rpm for 72 h at 28 °C for DNA extraction (Lyra et al. 2019). Then, 2 mL of the bacterial suspension was centrifuged at 7,500 rpm for 3 min, the supernatant was discarded, and the pellet formed was used for DNA extraction using the MiniPrep Kit (Axygen), according to the manufacturer's recommendations. DNA integrity was analyzed by electrophoresis on a 0.8% agarose gel for 30 min in 0.5 × tris/borato/EDTA (TBE) buffer at 100 V, after being stained with SybrGold (Thermo Fisher Scientific). The 100 bp Plus DNA Ladder (Thermo Fisher Scientific) was used as a molecular standard, followed by DNA quantification using the NanoDrop 2000c (Thermo Fisher Scientific), standardization of the concentration to 20-30 ng•μL -1 , and storage at -20 °C.
Dendrograms of the sample areas and plant parts, individually and jointly, were constructed using Geljv2 (Heras et al. 2015) with the Jaccard coefficient and UPGMA algorithm. Groups with 100% similarity were considered as distinct strains and evaluated using the Shannon-Weaver's diversity (Shannon and Weaver 1949), Pielou's uniformity (Pielou 1959), Simpson's diversity and dominance (Simpson 1949) and Margalef 's richness (Margalef 1956) indexes.
The sequences were compared to the type strains in the National Center for Biotechnology Information (NCBI) database. To determine the molecular identity, each strain was individually subjected to a similarity analysis using the MEGABLAST algorithm (highly similar sequences) or BLASTn algorithm (similar sequences). The same sequences were then analyzed for the percentage of molecular identity using the CLUSTAL W multiple progression method (Thompson et al. 1994) with the MEGA7.1 program (Kumar et al. 2018). The Juke-Cantor neighbor-join method was used to determine the similarity value and distance matrix and to build gene trees of the concatenated sequences for each isolate (Kumar et al. 2018). The significance of branching within trees was assessed by bootstrap analysis of 1,000 computer generated replicates. Sequences that did not show high similarity were not included in the phylogenetic tree.

RESULTS AND DISCUSSION
Using the semi-specific culture media, 325 isolates were obtained, which restricted the endophyte population that could be isolated (Hernández-Pacheco et al. 2021).
Phenotypic characterization revealed that 80.3% of the isolates acidified, 3.7% alkalized, and 16% did not change the pH of the medium (Fig. 2 and Suppl. Table 1), which may be associated with the assimilation or production of organic acids and potentially aid in the solubilization of phosphates . Mucus production occurred with 72.3% of the isolates, which may indicate resistance to environmental stress (Khan and Singh 2021). Cream and yellow colony colors predominated with 45%, followed by cream (32%), yellow (21.2%), and pink (1.3%). As for the edge, shape, opacity, and surface, 88.9% were solid, 92.9% circular, 75.4% opaque, and 98.5% smooth, respectively ( Fig. 2 and Suppl. Table 1).
Out of the 325 isolates, 315 amplified the BOX elements, forming bands from 228 to 1,736 bp, which identified 243 strains ( Table 2), 78% of which were isolated at a single time, which indicates high diversity. This high diversity was also observed when plant parts or sampling areas were compared through the diversity indexes (Table 2), while dominance, equability, and high Margalef 's richness indexes confirmed that the strains were equally abundant, which usually indicates greater stability and resistance to environmental stresses (Zhao et al. 2022). Considering the overall environments, Araripina had the most isolates and higher Shannon's (5.241), Simpson's (0.99) and Margalef 's (7.502) indexes, followed by Nazaré da Mata and Gravatá. The lower diversity and higher dominance indexes found at Gravatá (Table 2) are likely due to this stand being the youngest (6 years compared to 11 in Nazaré da Mata and > 30 in Araripina), which is comparable to that observed in Lolium perenne plants, which showed higher endophytic bacterial diversity in older plants than in younger ones (Tannenbaum et al. 2020). Another possible reason for the greater diversity found in Araripina is the drier environment, which could be more inducive to bacterial diversity (Wei et al. 2020   Considering only Araripina, higher diversity was found in the dry season than that in the rainy season, with and without liming. While liming effects in the rainy season are reasonable, the higher diversity during the dry season is uncommon (Xu et al. 2018, Ou et al. 2019, Firrincieli et al. 2020), but it may relate to the use of multiple culture media, allowing for a broader range of cultivable bacteria (Döbereiner et al. 1995, Jia et al. 2022). Comparing the different plant parts and rhizospheric soil, diversity decreased on the culm, rhizospheric soil, leaves, and root, in that order (Table 2), although rhizospheric soil presented the highest uniformity. Since the difference between these sources is relatively small, and higher uniformity was found in the rhizospheric soil, the results likely do not strongly deviate from the usual pattern for bacterial diversity found in these sources (Liu et al. 2017, Huang 2018, Zhang et al. 2020, Wang et al. 2022. Bacterial diversity varies between plant parts in a species dependent manner, with the most diverse part of maize being the culm, millet the roots, and both the culm and roots being more diverse for rice (Patel and Archana 2017). This also occurred on a cultivar basis, with drought resistant and tolerant millet genotypes presenting different bacterial diversity patterns (Manjunatha et al. 2019). Therefore, it is important to evaluate bacterial diversity in different plant parts, despite repeated isolation of the same strain.
Among the 135 groups with 90% similarity, 118 were identified at the genus level (96.84-99.9% similarity), while 13 (68-92.4% similarity) had insufficient similarity for taxonomic classification, and four strains showed no similarity with any genetic sequence in the GenBank database (Suppl. Table 2).
Proteobacteria domination found in pangolão grass has been previously described for widely divergent species, such as wheat (Robinson et al. 2016) and bamboo , which indicates a possible general pattern among endophytes. The genera found included several species described as plant growth promoters, such as Stenotrophomonas (Ramos et al. 2011), Pseudomonas (Josephine CM and Thomas 2021), Enterobacter, Pantoea (Lu et al. 2021), Burkholderia, Herbaspirillum (Van Deynze et al. 2018), Rhizobium (Hahn et al. 2016, Silva et al. 2020, Bacillus, and Paenibacillus (Govindasamy et al. 2010, Kavamura et al. 2013. A commercial product based on Bacillus strains has previously been licensed for use in maize to reduce drought effects, which was originally isolated from the native Brazilian cactus, Cereus jamacaru (Kavamura et al. 2013).
The most abundant genera were Enterobacter, Pantoea, Pseudomonas, Rhizobium, and Stenotrophomonas, although Achromobacter, Bacillus, Burkholderia, Curtobacterium, Herbaspirillum, Kosakonia, Ochrobactrum, Paenibacillus, Serratia, Shinella, and Variovorax were also present, and these genera have been observed in other grass endophytic communities such as Brachiaria and maize (Mutai et al. 2017, Mashiane et al. 2018. Of the five most abundant genera, Rhizobium, Pantoea, Pseudomonas, and Stenotrophomonas were isolated from all locations (Fig. 3), whereas Enterobacter was not found in Nazaré da Mata, which might be a confirmation of the division of the endophytic community into systemic and transitory groups (Wani et al. 2015). These genera were isolated from all plant parts and the rhizospheric soil.  The phylogenetic tree grouped the genera into six large polyphyletic groups (Fig. 4), with most strains (69%) in GI, GIII, and GVI, with no overlap with the outgroup based on Escherichia coli. The first group (GI) contained 32 strains in two subgroups. The first subgroup consisted of those similar to Rhizobium and Agrobacterium (25 strains, 98% similarity), Shinella (one strain, 100% similarity), Rhizobium (two strains, 58% similarity), Ochrobactrum (three strains, 99% similarity), while the second subgroup was not grouped with any genera. Five strains were included in GII, of which one was similar to Curtobacterium, two are similar to Paenibacillus, and the remaining two to Bacillus at 100% similarity.
GIII included 28 strains of Stenotrophomonas, and Pseudomonas split into two subgroups: one consisting of 22 strains with 70% similarity to these genera according to the bootstrap test to these genera, and the second one with the remaining six strains grouped with Stenotrophomonas at 57% similarity.
GIV contained the 11 Burkholderia, Variovorax, Herbaspirillum, and Achromobacter strains. Subgroup I matched three strains to Burkholderia at 100% similarity, while subgroup II matched four strains to Variovorax with 100% similarity, and the last four were grouped to Achromobacter and Herbaspirillum at 72% similarity.
GV related 11 strains to Pseudomonas (100% similarity). However, GVI connected 31 strains to Enterobacter, Kosakonia, Serratia, and Pantoea, with one strain in the subgroup showing 100% similarity to Kosakonia. In subgroup II, 10 strains were 54% similar to Enterobacter, one strain was 71% similar to Serratia, and 19 strains were similar to Pantoea.
This research is the first one to evaluate endophytic bacterial diversity on pangolão grass. These results indicate several factors, such as environmental conditions, plant parts, pasture age, season and liming, differently affect bacterial diversity. Since these bacteria were evaluated on a species well adapted to environmental stresses predicted to increase with global climate change, several of the genera found are known to include species and strains known to promote plant growth and several other papers, indicating bacteria isolated from one plant species may promote plant growth on other species. We suggest further research evaluating these and similar strains as plant growth promoters for crops such as maize.

CONCLUSION
Bacteria associated with Digitaria eriantha cv. Survenola are highly diverse, and this diversity varies according to environmental conditions, including plant compartment, pasture establishment time, season, and liming history. Proteobacteria were the most frequent bacteria associated with pangolão grass under all environmental conditions. Although diversity was slightly higher in the culm, there were no major differences between plant parts and the rhizospheric soil, and diversity was higher in older pastures.
The diversity of endophytic and rhizospheric bacteria in pangolão grass may have promoted resilience, since many of the identified strains belong to genera known to promote plant growth. These strains should be further evaluated for growth promotion in this and other plant species.
Less studied drought tolerant grass genera, such as Digitaria, might be an interesting source of plant growth promoting bacteria for further studies, due to their endophytic bacteria diversity.

DATA AVAILABILITY STATEMENT
All dataset were generated and analyzed in the current study.