The study was conducted in the Scientific City of Brazzaville (4˚16’42, 1439’’S, 15˚14’24, 6538’’E). The climate of Brazzaville is characterized by two seasons. The rainfall occurs between September and May whereas the dry season takes place between June and August. Annual rainfall averages are 1200 to 1500 mm. Relative humidity is always high around 75%. Temperatures are often around 25˚C [6] .

Soil sampling was taken at 1m from the tree (Cajanus cajan and Milletia laurentii) in the 0 - 10 cm horizon, using an auger. Four soil subsamples were taken near each tree (three tree per specie). The subsamples of each tree were mixed to form a composite sample. Each composite sample was separated in two and packaged in sterile glass jars and transported to the laboratory using a cooler. At the laboratory, the stones and roots were removed and the soils were kept at 4˚C until the use. One part of soil samples was used for the analysis of the bacterial communities and the other part was used for physicochemical characterization and isolate cultivable bacteria.

Soil texture was assessed using the pipette method, pH was measured with pHmeter. Total carbon of soil was determined by the Walkley-Black method [12] . determined The Kjeldahl method [13] was used to assess total nitrogen. Ammoniacal nitrogen was determined using Nessler’s reagent [13] . Phosphorus was determined by Olsen’s method [14] . The DEB method [15] was used to determine total iron, the Magnesium was determined by spectrophotometry.

The soil physicochemical characteristics used in Table 1 was determined by [6] .

DNA extraction, Illumina-Hiseq sequencing and bioinformatics analyses were carried out at Mr DNA laboratories (USA). Genomic DNA was extracted from 0.5 g of dry soil sample using the PowerSoil kit (MOBIO Laboratory, Carlsbad, CA, USA) following the manufacturer’s instructions. The concentration of the extracted DNA was estimated using the Nanodrop 2000C spectrophotometer (Thermo Scientific, Wilmington, DE, USA). Finally, DNA extracts from soil samples were stored at 80˚C until use. The bacterial 16S rRNA gene was amplified with primers 515F (5’-GTGCCAGCMGCCG CGGTAA-3’) and 806R (5’-GGACTACHVGGTWTCTA AT-3’). The PCR reaction was conducted as follows: a denaturation was carried out at 94˚C for 3 minutes, 30 - 35 denaturation cycles at 94˚C for 30 s for amplification, hybridization at 53˚C for 40 s, then elongation at 72˚C for 1 minute and a final extension at 72˚C for 5 minutes. After amplification, PCR products were visualized by electrophoresis on 2% agarose gel. Then, the two samples were grouped and purified together in equal proportions based on their DNA concentrations. The samples were purified using the ampure XP calibrated ball method. Then, pooled and purified PCR products are used to prepare the Illumina DNA bank. Sequencing was performed at MR DNA (www.mrdnalab.com, Shalowater, TX, USA). Sequences were assembled and the barcodes eliminated. Then 150 bp sequences and chimeras were removed. The OTUs were defined by grouping the sequences at 3% divergence. The final OTUs were taxonomically classified using the BLAST program against the organized database derived from RDPII and NCBI ( https://www.ncbi.nlm.nih.gov/ , https://rdp.cme.msu.edu ).

Soils were rehydrated at 50% of field capacity. They were then pre-incubated for 7 days at 30˚C under aerobic conditions to allow microbial activity to stabilize [16] . Twenty grams of moist soil were extracted with 50 mL of 0.5 M K₂SO₄ after stirring for 30 min. Another 20 g of moist soil was put into 100mL Erlenmeyer flasks. This soil was incubated under vacuum in a desiccator in the presence of 25mL of Chloroform without ethanol for 24 hours. The experiment was carried out in triplicate. After the incubation, chloroform was removed by ventilation then the soil was extracted with K₂SO₄ as described above. Organic carbon in the soil extracts was measured by the HACH TM COD method after heating at 150˚C for 2 hours [17] . The dosage was carried out by colorimetry with the DR890. Extractable nitrogen was measured by the spectrophotometric method after mineralization of N to ${\text{NH}}_4^{+}$ using indophenol blue [18] . Microbial Biomass carbon (MBC) and Microbial Biomass nitrogen (MBN) were calculated as differences of C and N concentrations between fumigated and non-fumigated extractions. Factors of extraction efficiency of 0.45 and 0.54 were used for MBC and MBN respectively [19] : MBC = Ec/KE_C with KE_C = 0.45 and Ec = (Organic C extracted from fumigated soil) − (Organic C extracted from non-fumigated soil). The microbial carbon biomass is expressed as µg C/g soil; MBN = E_N/KE_N with KE_N = 0.45 and E_N = (organic N extracted from fumigated soils) − (organic N extracted from non-fumigated soils). The microbial nitrogen biomass is expressed as µg N/g soil.

β-Glucosidase was measured following the method proposed by [20] . Briefly, 1 g of soil was mixed with 4 mL of modified universal buffer (pH 6), 1 mL of a 0.025M toluene solution and 1mL of a p-nitrophenol-β-D-glucoside solution. The mixture is then incubated at 37˚C for 1 hour. The released p-nitrophenol is quantified using a UV-visible spectrophotometer at 410 nm. β-Glucosaminidase was quantified according to the method described by [21] . Briefly, to 1 g of soil are added: 4 mL of 0.1M acetate buffer (pH 5.5) and 1mL of a 10mM p-nitrophenyl-N-acetyl-β-D-Glucosaminide solution. The mixture is incubated at 37˚C for 1 hour. Then, 1mL of a 0.5 M CaCl₂ solution and 4 mL of 0.5 M NaOH are added to stop the reaction. The mixture is stirred then filtered on Watman No. 2 filter paper. The intensity of the filtrate coloring is measured using a spectrophotometer at 405 nm.

Acid phosphatase was determined by adding to 1 g of dry soil: 4mL of modified universal buffer (pH 6.5), 1mL of a 0.025M toluene solution and 1mL of p-nitophenyl phosphate. The mixture is incubated at 37˚C for 1 hour. The released p-nitrophenol (PNP) was quantified using a spectrophotometer at 410 nm [22] [23] .

Dehydrogenase activity was determined according to the method of [24] . This method is based on the estimation of the concentration of 2,3,5-triphenyl formazan (TPF) released by dehydrogenases when the soil is incubated with a 2,3,5-tryphenyl tetrazolium chloride (TTC) buffer solution at 30˚C for 24 hours. In short, in a beaker we mix 20 g of soil with 0.2 g of calcium carbonate and 4mL of reagent. The optical density of the formed TPF was determined at 485 nm using U-V spectrophotometer.

To assess family and genus level of culturable bacteria distribution, a heatmap was done with the relative abundance of each family or genus. Alpha diversity indices: Shannon index, Simpson index, Evenness, Equitability and Chao-1 was calculated using the Past software 4.03. To compare the properties under the rhizospheres of Cajanus cajan and Milletia laurentii on a multivariate scale, principal components analysis (PCA) was performed with Past software 4.03. The non metric multidimensional scaling (NMDS) plot, cosine method, was used to short a graphical ordination (two synthetic axes) of the impact of soil characteristics, microbial biomass and enzymes activities on distribution of uncultured bacteria under the rhizospheres.

From Illumina sequencing, 45,412 and 32,781 raw sequences were obtained respectively for the rhizosphere soil of Cajanus cajan and Milletia laurentii. These sequences mainly belong to culturable and uncultured soil bacteria. After statistical processing, 1434 OTUs (Operational Taxonomic Unit) were obtained from the valid sequences. In this work, only families and genera of culturable soil bacteria as well as the genera of uncultured soil bacteria were taken into account.

Figure 1 shows that the nine families most representative culturable soil bacteria under Cajanus cajan were: Hyphomicrobiaceae (35.05%), Bacillaceae (18.98%), Bradyrhizobiaceae (11.70%), Streptomycetaceae (3.66%), Rhizobiaceae (3.65%), Methylobacteriaceae (3.44%), Hyphomonadaceae (3.00%), Pseudomonadaceae (2.40%), Sphingomonadaceae (1.91%). The most dominant genera of culturable soil bacteria under Milletia laurentii were: Hyphomicrobiaceae (24.22%), Bradyrhizobiaceae (17.32%), Bacillaceae (6.49%), Sphingobacteriaceae (5.80%), Chitinophagaceae (4.02%), Flavobacteriaceae (3.70%), Nitrospiraceae (2.42%), Xanthobacteriaceae (2.30), Sinobacteriaceae (2.11%).

Figure 2 shows that the most dominants genera under the rhizosphere soil of Cajanus cajan were: Acidobacterium (26.55%), Skermanella (17.52%), Rhodoplanes (7.50%), Bacillus (6.29%), Chloroflexus (4.74%), Steroidobacter (2.34%), Sphingomonas (2.32%), Bradyrhizobium (1.96%), Dongia (1.80), Sphingobacterim (1.74%), Nitrospira (1.37%), Microvirga (1.30%), Cupriavidus (1.12%), Gemmatimonas (1.12%), Chelatococcus (1.03%) and Streptomyces (1,02%). In the rhizosphere soil of Milletia laurentii the principal dominants genera were : Rhodoplanes (21.63%), Bradyrhizobium (17.27%), Bacillus (6.43%), Sphingobacterium (5.68%), Acidobacterium (4.58%), Mesorhizobium (4.06%), Nitrospira (2.42%), Steroidobacter (2.05%), Hyphomicrobium (1.53%), Pedomicrobium (1.53%), Streptomyces (1.45%), Burkholderia (1.45%), Nitratireductor (1.33%), Niastella (1.33%), Sphingomonas (1.22%), Gemmatimonas (1.22%), Mycobacterium (1.08%) and Holophaga (1.08%).

Table 2 shows that the nine predominants of uncultured soil bacteria genera (relative abundance ≥1%) in the rhizosphere soil of Cajanus cajan, were mainly: Acidobacterium (34.02%), Skermanella (15.83%), Nitrosovibrio (8.29%), Chloroflexus (7.99%), Sinorhizobium (4.79%), Polyangium (4.31%), Blastopirellula (2.98%), Opitutus (2.47%), Sphingobacterium (2.45). The relative abundance of other genera was between 1.01 and 1.74%. Nine genera (Blastopirellula, Candidatus lariskella, Chitinophaga, Ktedonobacter, Nitrosovibrio, Opitutus) were exclusively founded among genera of uncultured bacteria in the rhizosphere soil. On

the other hand, in the rhizosphere soil of Milletia laurentii ( Table 2 ), uncultured soil bacteria were predominant in the genera Rhodoplanes (12.59%), Brayrhizobium (8.73%), Chloroflexus (7.99%), Acidobacterium (4.55%), Blastopirellula (3.39%), Gemmata (3.03%), Flavisolibacter (2.25%), Candidatus (2.12%), Flavobacterium (1.95%) while the other genera of uncultured bacteria had relative abundances between 1.02 and 1.93%. Table 2 also reveals that relative abundance of uncultured soil bacteria genus Acidobacterium, Skermanella, Chloroflexus, Nitrosovibrio and Sinorhizobia were higher under rhizosphere of Cajanus cajan than under Milletia laurentii. However, the results indicated that a high uncultured soil bacteria genera of Rhodoplanes, bacillus, Blastopirellula, bradyrhizobium, Gemmata were most observed under rhizosphere soil of Milletia laurentii than under Cajanus cajan. Distribution of the other uncultured soil bacteria genera was linked with the stand.

Table 3(a) shows that under rhizosphere soil of Cajanus cajan all uncultured soil bacteria genera had high relative abundance than culturable soil bacteria outside of genus Skemanella. The results indicated that six culturable soil bacteria (Sphingomonas, bradyrhizobium, Microvirga, Cupriavidus, Chelatococcus and streptomyces) were not retrieved among the uncultured soil bacteria genera. On the other hand, several uncultured soil bacteria were not in the list of genera of culturable soil bacteria for example: Nitrosovibrio, Sinorhizobium and Polyangium. Ten genera of culturable soil bacteria were not found among the uncultured soil bacteria genera. This was the case for Sphingomonas, Bradyrhizobium ( Table 3(b) ). On the other hand, thirteen uncultured soil bacterial genera have not been listed among the culturable soil bacteria, for example: Blastopirellula, flavobacterium, rhizobium, Skermanella.

Table 4 shows that alpha diversity was different under the two rhizosphere soils. Specific richness evaluated by chao-1 had the same value in the two rhizospheres (chao-1 = 44). The others indices exhibited weak differences between: (H = 1.84; 1-D = 0.76), (H = 1.76; 1-D = 0.75) under Cajanus and Milletia respectively. The same trend was observed for evenness and equitability: in Cajanus (e^H/S = 0.14; J = 0.49) and in Milletia (e^H/S = 0.13; J = 0.47). However, the number of individuals was higher in Milletia than in Cajanus.

Soil of rhizosphere associated with Milletia has high levels MBC (688 mgC∙kg⁻¹ soil) and MBN (186.83 mgN∙Kg⁻¹ soil) compared to the rhizosphere of Cajanus (MBC 621 mgC∙Kg⁻¹ soil and MBN 138.57 mgN∙Kg⁻¹ soil) ( Figure 3 ). Enzyme activities into the rhizosphere was different with enzyme (Table V). High β-Glucosidase activity (163 mg∙pN∙kg⁻¹∙soil∙h⁻¹) was observed under Milletia compared in the soil of rhizosphere of Cajanus (121 mg∙pN∙kg⁻¹∙soil∙h⁻¹). However high β-Glucosaminidase activity (64 mg∙pN∙Kg⁻¹∙soil∙h⁻¹) was obtained under Cajanus compared to Milletia (57.50 mg∙pN∙kg⁻¹∙soil∙h⁻¹). Table 5 shows that Acid phosphatase activity was lower in soil under Cajanus than in Milletia, whereas dehydrogenase activity was higher under Milletia than under Cajanus (1.03 µg TTC∙g⁻¹∙soil∙h⁻¹).

Figure 4 shows that MBC and β-Glucosaminidase and dehydrogenase enzymes were more founded in soil of rhizosphere of Cajanus cajan, whereas MBN β-Glucosidase and acid phosphatase enzymes have a favorite distribution in soil rhizosphere of Milletia laurentii.

Considering the component 1, PCA ( Figure 5 ) shows that MBC, the genera (Acidobacterium, Skermanella, Chloroflexi), enzyme (β-glucosaminidase) and the sand were positively correlated with the rhizosphere of Cajanus cajan. MBN, the genera (Mesorhizobium, Bacillus Bradyrhizobium, Rhodoplanes, Burkholderia, Nitratireductor, Hyphomicrobium, Sphingobacterium, Nitrospira), enzymes (β-glucosidase, acid phosphatase), slit, clay, and NH₄ were mainly related with the rhizosphere of Milletia laurentii. Taking into account the component 2, we observed that: Acidobacterium, MBC, enzyme β-glucosaminidase and the sand were positively correlated with Cajanus cajan whereas Skermanella, Chloroflexi, Mesorhizobium, Rhodoplanes, Bradyrhizobium and NH₄ were negatively correlated

with the rhizosphere of this plant. On the other hand, Skermanella, Cloroflexi, Mesorhizobium, Rhodoplanes, Bradyrhizobium and NH₄ had a negative relationship with Milletia, in the same time positive correlation take place between this plant and MBN, Nitrospira, Bacillus, Burkholderia, Nitratireductor, Hyphomicrobium, Sphingobacterium, the enzymes acid phosphatase and β-glucosidase.

The NMDS was used to check the links of uncultured soil bacteria genera with the soil physicochemical, microbial biomass and enzymes. The NMDS (based on both cosine similarity) for genus assemblage, relative abundance was used without transformation. The NMDS graphic shows two groups ( Figure 6 ). The first group included genera: Nitrospira, Chitinophage, Opitutus, Steroidobacter, Nitrosovibrio, Spirulina, Acidobacterium, Gemmatomonas, Sphigobacterium, Dongia, Skermanella, Cloroflexus. The group of these uncultured bacteria were linked with rhizosphere of Cajanus cajan. Under this plant the mainly soil properties were sand, clay, C, NH₄, MBC and β-glucosaminidase. The second group was related to rhizosphere of Milletia laurentii. Its composition was: Ktedonobacter, Bacillus, Candidatus, Nistella, Rhizobium, Rhodplanes, Flavisolibacter, Bradyrhizobium, Flavobacterium, Edaphobacter. Outside the nature of the plant, in this rhizosphere, nine soil properties (Slit, P, Mg, N, Fe, MBN, Acid Phosphatase, Dehydrogenase) could impact the distribution of the genera.