Introduction

The phyllosphere refers to the shoots of plants, including leaves, stems, fruits and flowers1,2. As with almost every other environmental niche of plants, the phyllosphere is colonized by microorganisms, with bacteria being the most dominant1,2,3,4. Microorganisms coexisting with multicellular organisms confer a wide range of benefits1, including protection from pathogens and stress, nutrient modulation, synthesis of secondary metabolites and growth promotion. For many host organisms, the microbiome in each niche is shaped by antimicrobial principles produced by the host. While this was shown for the rhizosphere5,6, less is known about the phyllosphere.

A great proportion of research on medicinal plants has focused on their antimicrobial properties, leading to the discovery of compounds with antibacterial and other biological activities, but the microbiome is almost totally neglected in such bioprospecting initiatives7,8,9,10. Medicinal plants should be seen as meta-organisms consisting of the plants themselves and the microbes they harbour and efforts are thus being made to investigate the microbiology of medicinal plants’ phyllospheres. This interest is aroused partly because some bacteria coexisting with medicinal plants contribute to the production of secondary metabolites and active compounds5,11,12, and play roles in the metabolism and ecology of the plant. Indeed, some studies have confirmed that some plant-associated microbes themselves directly produce antimicrobial compounds, while others are involved in the production of secondary metabolites by such plants13,14,15,16,17,18. It is also possible that the microbiomes themselves are shaped by bioactive agents produced by the plants6. Phyllosphere studies are increasingly featured in the literature, but the few from Africa focus predominantly on plant pathogens of agricultural crops, and few studies worldwide examine the culturable phyllosphere19,20,21.

Euphorbia lateriflora Schumach. & Thonn. and Ficus thonningii Blume are medicinal plants of the Euphorbiaceae and Moraceae families, respectively. E. lateriflora is a shrub growing to a height of approximately 1.7 m with nearly vertical waxy branches19. Abundant whitish latex is produced when the leaves shed or when the plant is otherwise injured. The latex is toxic to the eyes and may result in blindness in some cases. Scientific investigation into the toxicity of E. lateriflora discovered varying levels of dose-dependent toxicity in experimental animals22,23. E. lateriflora is used in ethnomedicine for treating prostatic diseases, skin and soft tissue infections, sexually transmitted infections and a variety of other syndromes 19,24, and various solvent extracts produce in vitro antimicrobial activity. F. thonningii is a large, multistemmed, flowering tree with a dense spreading crown. The tree bark on old branches is grey and smooth, while younger branches have hairy bark. F. thonningii has green leaves, red to yellow flowers and yellow or pink fruits25,26. F. thonningii is generally considered safe, with minimal or no toxicity recorded27,28,29. Several parts of F. thonningii have been reportedly used in ethnomedicine, but the leaves are more commonly used in treating diarrhea, gonorrhea, wound infections, urinary tract infections and skin and soft tissue infections30,31,32,33,34. In addition to surveys, various extracts of F. thonningii have been shown to exhibit in vitro antimicrobial activity against microbial pathogens35,36,37,38,39,40,41,42.

This study investigated the phyllosphere microbiome of E. lateriflora and F. thonningii. We hypothesized that due to the different secondary metabolites produced, E. lateriflora and F. thonningii are colonized by diverse bacteria that are nonsusceptible to the antimicrobial principles of the plants.

Methods

Plant collection

Euphorbia lateriflora for this study was sampled from the botanical garden of the University of Ibadan, Ibadan, Nigeria (GPS coordinates 7.377536, 3.94704), while Ficus thonningii was sampled from a mature tree near Ojoo, Ibadan, Nigeria (GPS coordinates 7.49529, 3.93415). Up to 10 leaves of E. lateriflora were sampled each time, in contrast to three leaves of F. thonningii due to differences in leaf surface area. Healthy, fresh leaves of the plants were aseptically plucked from the petiole with the aid of presterilized forceps, placed directly into sterile laboratory collection bags and transported to the laboratory for culture and DNA extraction within 1 h. For solvent extraction, the plants were collected and dried to constant weight at ambient temperature (28 °C), away from direct sunlight, before further processing. E. lateriflora and F. thonningii were authenticated by Mr Adeyemo Adejola at the Forestry Research Institute of Nigeria (FRIN) and herbarium specimens were deposited with the respective voucher numbers FHI 110801 and FHI 1106898.

DNA extraction

DNA was extracted from two biological replicate leaf samples of both plants with the PowerSoil DNA isolation kit (MOBIO Laboratories Int.) according to the manufacturer’s instructions.

Quantitative polymerase chain reaction (qPCR)

Quantitative polymerase chain reaction (qPCR) was performed with the aim of estimating the abundance of archaea, bacteria and fungi in the extracted community DNA. Archaeal and bacterial qPCRs targeted the 16S rRNA gene with the respective domain-specific primer pairs 344aF and 517uR43 and Unibac-II-515f. and Unibac-II-806r44, while fungal qPCR targeted the internal transcribed spacer (ITS) gene with ITS1f. and ITS2r primers45. All reactions consisted of the following components: sample DNA (1 µl) KAPA SYBR Green (5 µl) (Bio-Rad, Hercules, CA, USA), 0.5 µl each of forward and reverse primers (10 µM) and PCR-grade water (3 µl) making up a total volume of 10 µl. Tenfold dilutions of appropriate archaeal, bacterial and fungal standards that contained known abundance estimates were prepared according to previously described protocols44 and were used to calculate gene copy numbers. All reactions were performed in duplicate and run on a Rotor-Gene™ 6000 series thermal cycler (Corbett Research, Sydney, Australia). The qPCR protocol for archaea was as follows: initial denaturation (95 °C/5 min) followed by 39 cycles of denaturation (95 °C/15 s), annealing (60 °C/30 s) and extension (72 °C/30 s) with a final step from 72 to 96 °C. The protocol for bacteria and fungi was as follows: initial denaturation (95 °C/5 min) followed by 35 cycles of denaturation (95 °C/10 s), annealing (54 °C/15 s) and extension (72 °C/10 s) with a final step from 72 to 96 °C. Statistical analyses of gene abundance estimates were computed with the Kruskal‒Wallis test49.

Amplicon sequencing

Amplicon libraries of the bacterial 16S rRNA and fungal ITS genes were prepared for technical replicates as described below. PCR targeting the 16S rRNA gene was conducted with the 515f. and 806r primer pair46. Both forward and reverse primers contained barcodes at the five prime ends, and these were assigned to samples to yield unique combinations for sample multiplexing and sequencing. To minimize the amplification of plant mitochondrial 16S rRNA and plastid genes, peptide nucleic acid (PNA) clamps were incorporated in the PCRs47,48. Each PCR contained 6 µl of 5XTaq & GO premix for PCR (MP Biomedicals), 0.6 µl each of forward and reverse primers, 0.45 µl each of mPNA and pPNA, 1 µl of template DNA and 20.9 µl of molecular biology grade water to yield a final volume of 30 µl. The PCR conditions consisted of an initial denaturation at 60 °C for 5 min and subsequent 30 cycles of 94 °C for 60 s, 78 °C for 5 s, 54 °C for 1 min and 72 °C for 60 s with a final extension at 72 °C for 10 min49. PCRs were run in triplicate. The respective replicates were pooled and stored at − 20 °C for onwards processing.

The ITS primer pair ITS1f. and ITS2r was used to amplify the ITS 1 region of the fungal ITS gene45,50. The PCR consisted of 1 µl of template, 6 µl of 5XTaq &Go pre-PCR mix (MP Biomedicals), 0.6 µl each of forward and reverse primers and 21.8 µl of molecular biology grade water to a final volume of 30 µl. The PCR conditions included an initial denaturation at 98 °C for 5 min and then 30 cycles of denaturation at 98 °C for 60 s, primer annealing at 58 °C for 60 s, extension at 74 °C for 60 s with a final extension at 72 °C for 10 min49. PCRs were run in three replicates, and the replicates were combined and kept at − 20 °C prior to further processing.

Pooled amplicons were purified with a Wizard SV Gel and PCR clean-up system (Promega, Ref: A9282), after which DNA concentrations were estimated with a Qubit dsDNA broad range kit. Samples were normalized to equimolar concentrations and pooled before being sent for Illumina sequencing by the commercial company Novogene Corporation Incorporated.

Sequence analyses of the amplicon libraries

Raw sequence reads obtained were demultiplexed. Reads with identical barcodes were combined, and primers/barcodes were removed with cutadapt51. The processed reads were then imported into QIIME252, where the input was summarized, quality filtered, denoised (with the DADA algorithm)53 and classified. Operational taxonomic units were apportioned with the VSEARCH algorithm54 implemented within QIIME2 using curated reference databases, SILVA132 for bacterial 16S rRNA55,56 and UNITE for fungal ITS57.

Statistical analyses and data visualization were performed in Rstudio in addition to the online microbiome analyst platform58. The generated OTU table and taxonomic classification were analysed with phyloseq59 and vegan packages within RStudio. The Shannon index was used to estimate alpha diversity after rarefication of data to the minimum library size. Upon realization, after initial analyses that the bacterial sequences were overdominated by OTUs mapping to the Sphingomonas genus, the genus was omitted for subanalyses, creating two sets of 16S rRNA data. The relative abundance of taxa was computed across three levels of taxonomy: class, order and family.

Isolation of bacteria

Fresh leaves (2–4) were aseptically collected into sterile bags, transported to the laboratory within an hour and processed following previously established procedures60. In summary, the leaves were washed off in sterile water containing Tween 20 with 5 min of sonication at ambient temperature. The sonicate was serially diluted, and aliquots were inoculated into 0.1 × tryptone soya agar (TSA) containing 100 µg/mL cycloheximide. Plates were then incubated at room temperature for 5–7 days. Plate counts were computed after incubation, after which approximately 150 colonies were selected for each plant. The colonies were selected from all plates with distinctive colonies (counts between 30 and 300). All the different colony morphologies were picked in relation to their abundance as much as possible. The selected colonies were then subcultured to purity, gram-stained and cryo-preserved at − 80 °C.

Isolate identification by 16S rRNA gene sequencing

Isolates were identified by analysis of the DNA sequences of their 16S rRNA genes. Five to six colonies of each isolate were harvested into sterile, nuclease-free microcentrifuge tubes containing 100 µl of sterile molecular biology grade water. This was then boiled at 95 °C for 10 min60 and used as a template for PCR. PCR amplification of 16S rRNA genes was conducted for all isolates using (illustra TM) puRe Taq Ready-T-Go PCR beads in a 25 µl reaction volume. The amplification protocol for 10F and 1507R began with initial denaturation at 94 °C for 2 min, followed by 36 cycles of denaturation at 94 °C for 30 s, annealing at 59 °C for 30 s and extension at 72 °C for 1 min and 30 s, with a final step of 72 °C for 5 min. The 27F and 1492R primer set was as follows: initial denaturing temperature 94 °C for 2 min, denaturing temperature 94 °C for 30 s, annealing temperature 53 °C for 30 s, extension temperature 72 °C for 1 min and 30 s, and final step 72 °C for 5 min for a total of 36 cycles61. The primers employed are listed in the Supplementary Material.

Following amplification, 5 µl of each PCR product was electrophoresed on a 1.5% (w/v) agarose gel containing gel red at 110 V for 45 min. The gels were thereafter visualized on a transilluminator, after which PCR amplicons were Sanger sequenced by Inqaba Biotec (Ibadan). Raw reads with acceptable peak quality on traces were analysed using the basic local alignment search tool (BLAST)62 version 2.6.0 + tool of the National Center for Biotechnology Information (NCBI) 16S rRNA database for bacteria and archaea. Strains with sequences 97% identical to the database match were presumed to belong to the same species as the matching organism in the database, and a 95% cut-off was used to define genera63. Sequence data with low quality were exempted from downstream analyses. The sequences were deposited in GenBank under Bioproject SUB12689536.

Phylogenetic analyses

Multiple sequence alignments and phylogenetic tree construction were performed using the MEGA tool (http://www.megasoftware.net/) following directions and recommendations by Hall64. All sequences were trimmed to approximately 840 bp, equivalent to positions 20–855 of the E. coli 16S rRNA gene. The alignment was then executed, and the data were exported as a mega file, after which they were used to build a phylogenetic tree (construct/test maximum likelihood tree). Phylogenetic trees were visualized and annotated with the interactive Tree of Life tool (iTOL)65.

Isolate characterization by phenotypic coliform test

All identified E. coli and E. fergusonnii isolates were tested for lactose fermentation at 44 °C to determine if they are type 1 E. coli typically associated with the gastrointestinal tract of homothermic animals66,67. Briefly, isolates were subcultured on 0.1 × tryptone soy agar (TSA) for 24 h and subsequently incubated in 5 mL of MacConkey broth (Oxoid) at 44 °C for 24 h. A colour change from purple to yellow indicated a positive test and was used to differentiate type 1 from other E. coli.

Isolate characterization by antimicrobial screening using plant extracts

Ethyl-acetate extracts of dried pulverized leaves of E. lateriflora and F. thonningii were obtained with the Soxhlet extraction method68. The extracts were then concentrated to constant weight with the aid of a rotary evaporator before storage at − 20 °C for future use.

To determine whether the plants had antimicrobial activity against phyllosphere isolates, the agar diffusion method was employed. Briefly, the ethyl acetate extracts of both plants were dissolved in methanol to a concentration of 100 mg/mL. Phyllosphere isolates and standard control strains were grown on 0.1× TSA, after which they were suspended in 0.85% physiological saline to a turbidity level equivalent to 0.5% McFarland’s standard. Sterile cotton swabs were used to evenly spread the inoculum on pre-prepared sterile plates of 0.1× TSA. Equidistant wells were thereafter bored with the aid of an 8 mm cork-borer. One hundred microliters of prepared extracts was then dispensed in the respective wells and allowed sufficient time to diffuse. Ciprofloxacin (10 µg/mL) and chlorocresol (500 µg/mL) were also dispensed in 100 µL volumes as active compound controls. The plates were incubated at room temperature with daily observations. At the end of 5 days of incubation, zone diameters were recorded.

Minimum inhibitory concentrations (MICs) were determined using the agar dilution method. TSA (0.1×) plates containing doubling dilutions of extract in concentrations ranging between 0.39 mg/mL and 50 mg/mL were prepared in distilled methanol. Methanol was also used as a solvent control for the assay. Plates containing twofold dilutions of chlorocresol from 7.8 µg/mL to 1000 mg/mL were similarly prepared. An overnight broth culture of each isolate on 0.1× tryptone soy broth (TSB) was diluted, and 200 µl of dilutions equivalent to a 0.5 McFarland standard was transferred into a sterile 96-well microtiter plate, which was used to surface-inoculate duplicate plates of each extract or chlorocresol concentration using a sterile 48-point multipoint inoculator. Sterile plates of 0.1× TSA and MacConkey Agar with neither extract nor antibacterial were similarly inoculated. Plates were incubated at ambient temperatures for up to 5 days, with daily inspection. The lowest concentration of antimicrobials that inhibited the growth of each isolate over the entire incubation period was recorded as its MIC.

Plant authentication

The plant leaves were collected and processed according to laid down guidelines of the Department of Pharmaceutical Microbiology and the University of Ibadan with approval for collection given by the board of the department. Plant samples were authenticated by Mr Adeyemo Adejola at the Forest Herbarium Ibadan, located within Forestry Research Institute of Nigeria (FRIN), and herbarium specimens were deposited with voucher numbers FRIN 110801 and FRIN 1106898 for E. lateriflora and F. thonningii respectively.

Results

Marker gene copy estimation for archaea, bacteria and fungi by qPCR revealed higher copy numbers for fungi and bacteria than for archaea for both E. lateriflora and F. thonningii (Fig. 1). Kruskal‒Wallis statistics indicated a significant difference in gene abundance between archaea, bacteria and fungi (p = 0.01832) but no significant difference comparing the estimated abundance of these domains across the two plants (p = 0.52).

Figure 1
figure 1

qPCR estimation of the abundance of archaeal 16S rRNA, bacterial 16S rRNA and fungal ITS genes in the community DNA of E. lateriflora and F. thonningii. Gene copies were normalized to the weight of DNA extraction starting plant material.

Full size image

The microbiome of both plants was dominated by Pseudomonadota (alpha- and gamma-Protobacteria), with over 65% of OTUs assigned to the Sphingomonas genus (Fig. 2). Excluding Sphingomonas, Gammaproteobacteria, Bacilli, Alphaproteobacteria and Actinomycetia were the most commonly detected taxonomic classes with visible but statistically insignificant variation between replicates. Bacillales, Burkholderiales, Enterobacteriales, Rhizobiales and Xanthomonadales were the dominant orders across replicates of both plants. A high proportion of Pseudomonadales was observed for F. thonningii compared to E. lateriflora. At the family level, Bacillaceae, Beijerinckiaceae, Enterobacteriaceae and Xanthomonadaceae were abundant on both plants. While Beijerinckiaceae and Vagococcaceae were relatively more abundant in E. lateriflora, Moraxellaceae was more abundant in F. thonningii, and Oxalobacteriaceae was exclusively detected in F. thonningii. Overall, both plants were similar in their bacterial composition at higher taxonomic levels, but the similarity was reduced at lower taxonomic levels.

Figure 2
figure 2

16S rRNA- and ITS-based bacterial and fungal diversity, composition and abundance in the phyllosphere of E. lateriflora and F. thonningii. Sample labels EL1 and EL2 represent technical replicates of E. lateriflora, while FT1 and FT2 represent technical replicates of F. thonningii.

Full size image

The two most frequently detected fungal clases were Dothideomycetes and Sordariomycetes (Fig. 2). The orders Capnodiales, Mycosphaerellales and Pleosporales were commonly identified from both plants. Higher, almost exclusive abundances of Glomerallales and Hypocreales were observed in Ficus, while Trichosphaeriales and Xylariales were observed in Euphorbia. At the family level, Cladosporiaceae, Corynesporascaceae and Mycosphaerellaceae were identified at high frequencies on both plants. Didymellaceae, Massarinaceae, Nectriaceae and Neodevriesiaceae were almost exclusive to F. thonningii, and Dissoconiaceae, Schizothyriaceae and Amphisphaeriaceae were more commonly identified in E. lateriflora. Similar to the observation for 16S rRNA sequences, both plants were composed of similar taxa, but the proportions decreased at lower taxonomic levels.

Leaves of E. lateriflora and F. thonningii were abundantly colonized: the colony forming units per gram of leaf were 1.72 × 107 and 7.63 × 106, respectively, with significantly more colonies from E. lateriflora in comparison to F. thonningii. Overall, 127 and 98 isolates were identified from Euphorbia lateriflora and Ficus thonningii, respectively. Bacterial species of Actinomycetota, Bacillota and Pseudomonadota were identified from both plants (Table 1). A total of 11 and 16 different genera were identified on E. lateriflora and F. thonningii, respectively. Only seven genera, Bacillus, Lysinibacillus, Staphylococcus, Methylobacterium, Sphingomonas, Escherichia and Pseudomonas, were shared across both plants. Four genera were unique to E. lateriflora, while nine were exclusive to F. thonningii. Some isolates could not be assigned to the genus level and were thus classified as ‘unassigned’ and placed in the family of the closest BLAST hit. There were 15 such isolates from E. lateriflora and 14 from F. thonningii.

Table 1 Taxonomic summary of identified isolates from E. lateriflora and F. thonningii.
Full size table

A total of nine families were identified in E. lateriflora and 13 in F. thonningii. While Enterobacteriaceae, Pseudomonadaceae, Bacillaceae, Staphylococcaceae, Methylobacteriaceae, Sphingomonadaceae, Comamonadaceae, and Burkholderales were common to both plants, Brevibacteriaceae (five isolates) was exclusive to E. lateriflora, while Microbacteriaceae (eight isolates), Micrococcaceae (one isolate), Xanthomonadaceae (two isolates), Dermabacteriaceae (one isolate) and Vibronaceae (one isolate) were unique to F. thonningii (See Fig. 1 in the Supplementary Material).

Figure 3 shows 16S rRNA sequence-based phylogeny of the identified isolates with associated metadata- source plant and bacterial classification at the phyla, order and family levels. Isolates clustered according to established phylogeny. There was no amplification of the 16S rRNA gene for seven isolates from E. lateriflora and 14 isolates from F. thonningii, after trying both universal primers, and as a result, their 16S rRNA genes were not sequenced.

Figure 3
figure 3

16S rRNA gene sequence-based phylogeny of cultured isolates from E. lateriflora and F. thonningiiI. Metadata bars represent host plant and phyla, order and family classification.

Full size image

A total of 65 isolates from E. lateriflora and and thirty-three from F. thonningii were identified as Escherichia coli., Pair-wise analyses detected nucleotide differences within their 16S rRNA genes, suggesting the possibility of multiple clones and this was tested by performing the simple coliform assay. Representatives of identical isolates were selected and the coliform testing was performed for 41 isolates of E. lateriflora and 11 isolates of F. thonningii. Figure 4 shows the result of the coliform test on Escherichia isolates from both plants. Thirty-eight out of 41 isolates of E. lateriflora and one out of eleven isolates of F. thonningii fermented lactose at 44 °C.

Figure 4
figure 4

Distribution of type 1 and type 2 Escherichia spp. on E. lateriflora and F. thonningii.

Full size image

Figure 5 shows the phylogeny of representative strains from both plants and four control strains that were subjected to antimicrobial susceptibility testing of ethyl acetate extracts of the plants and chlorocresol. Zones of inhibition from the agar diffusion method ranged from 13 to 34 mm, depending on the plant and isolates. Overall, the extract of E. lateriflora produced wider zones of inhibition when compared to F. thonningii. Both extracts consistently produced wider zones of inhibition than ciprofloxacin (10 µg/mL) and chlorocresol (500 µg/mL). The MICs of the plant extracts ranged from 3.125 to 12.5 mg/mL, while those of chlorocresol ranged from 62 to > 1000 µg/mL Overall, the MICs produced by E. lateriflora extracts were lower than those produced by F. thonningii extracts. However, Xanthomonas perforans, Micrococcus luteus and two strains of Microbacterium testaceum from F. thonningii had relatively higher MICs to the Euphorbia lateriflora extract.

Figure 5
figure 5

16S rRNA gene sequence-based phylogeny of selected cultured strains representing the different OTUs showing susceptibility to ethyl acetate extracts of E. lateriflora and F. thonningii. The letters preceding organism names in the tree represents the plant origin of isolate; EL and FT for E. lateriflora and F. thonningii respectively, while the numerals represent isolate numbers. Associated metadata are source plant, phyla and order taxonomy, minimum inhibitory concentrations of E. lateriflora and F. thonningii extracts and chlorocresol.

Full size image

Discussion

Our research on two common Nigerian medicinal plants, Euphorbia lateriflora and Ficus thonningii, showed that they are densely colonized at similar abundance but by different phyllosphere microorganisms. In detail, fungal and bacterial marker genes were abundantly detected through qPCR in relation to archaea. Bacteria and fungi were frequently shown the most abundant groups inhabiting the phyllosphere of diverse plant species1,69,70. It has been argued that the abundance of archaea is frequently underestimated, and the full potential of archaeal communities is poorly understood due to limitations in current assays71,72,73. However, archaea have been sometimes reported in high abundances in the rhizosphere, especially in perennial plants70,73,74.

The high abundance determined by estimation of colony forming units of bacteria per gram (CFU/g) of leaves suggests that the leaves of E. lateriflora and F. thonningii in Ibadan are heavily colonized. Phyllosphere bacterial populations typically range from 101 to 108 CFU/g depending on the plant species and on its environmental, climatic and geographical habitat5,60,75,76,77,78,79. The numbers recorded in this study are in the upper range, similar to those obtained from sampling of roots and rhizosphere environment5,80. This may be attributable to the plants’ intrinsic properties and differences in environmental conditions from other phyllosphere studies, as the vast majority of studies originated outside Nigeria and West Africa. There are relatively less fluctuations in temperature and humidity in tropical Nigeria, which may favour the proliferation of bacteria in the phyllosphere, resulting in large populations. There was a contrast in the abundance of taxa by culture and culture-independent methods, which is usually the case70 but both methods indicate a high colonisation density in the phyllosphere.

Overall, Pseudomonadota was the most abundant phylum in all replicates of the two plants. Pseudomonadota is frequently the most dominant phylum in several phyllosphere studies1,81. We observed that the dominance of Pseudomonadota was highly influenced by Alphaproteobacteria (75 to 90% of total ASVs depending on the sample), which was due to an overabundance of the Sphingomonas genus. Sphingomonas is a major genus in the phyllosphere of many plants and colonizes a wider variety of plants when compared to the more commonly detected Pseudomonas genus82,83,84,85,86. Sphingomonads are well adapted to withstand harsh conditions and competitive life in the phyllosphere82,87,88, where they are associated with antagonism towards pathogens, plant protection from disease, and functions in geochemical cycling83,89,90.

To visualize the composition of other taxa in the microbiome more clearly, we removed all ASVs mapping to Sphingomonas spp. and repeated the analyses. The most abundant class shifted to Gammaproteobacteria, followed by Bacilli, Alphaproteobacteria and Actinomycetota. There were visible differences in the abundance between the two plants. For instance, Actinomycetia, Alphaproteobacteria and Bacilli were more abundant in E. lateriflora, while Gammaproteobacteria were more abundant in F. thonningii. However, the observed differences were not statistically significant. The differences in composition and abundance between the plants were more evident at lower taxonomic levels, where Oxalobacteraceae was exclusively detected in F. thonningii. Overall, the compositions of the microbiome of both plants were similar, which may be partly ascribed to similar environmental and climatic conditions, as the plants were collected in close locations. It has been established that environmental conditions and geographical location have a profound influence on phyllosphere microbiome composition91. The host plant is also considered a significant influential factor of microbiome composition and abundance and is argued to be the most significant single factor from studies that have observed similar composition from identical plants in different locations92,93. However, several studies have also reported similarities in different plants within the same location, and differences between the same plant species in different locations91,94. It is evident that an interplay of factors influences the composition of the phyllosphere microbiome, and none of these can be singly ascribed to a single factor95. However, the combination of location and host plant seems to be significantly involved in microbiome composition4. Nevertheless, there are remarkable overlaps between the phyllosphere microbiome composition of diverse plants irrespective of geographical location, plant genetics, and other factors81,91.

The most dominant fungal classes, Dothideomycetes and Sordariomycetes, are commonly detected in high numbers in the phyllosphere81,91,95. Differences in fungal composition between the plants were more visible compared to bacteria, and this was statistically significant. Similar to bacterial composition, the plants shared some ASVs, notably Cladosporiaceae, Corynesporascaceae and Mycosphaerellaceae. The commonly detected fungal families, Cladosporiaceae, Didymellaceae and Nectriaceae, were also found in our study. We observed less fungal diversity than bacterial diversity, which is consistent with other reports of phyllosphere epiphytes81,95.

The taxonomic composition of bacterial isolates via culture was specific and unique for each plant species. We note that 71 (56%) and 38 (40%) of the isolates characterized from E. laterifora and F. thoningii, respectively, were Pseudomonadota belonging to the species Escherichia coli. Phenotypic examination demonstrated that 38 out of 41 of these isolates from E. lateriflora produced both acid and gas from lactose at 44 °C, strongly suggesting that their origin is recent feacal contamination from warm-blooded animals. All but one of the E. coli isolates from F. thonningii did not show these properties, suggesting that they are likely of environmental origin, preferably called type 2 Escherichia sp., and may indeed be stable members of the phyllosphere. F. thonningii trees are large with dense spreading crown and the effect of faecal contamination on its leaves would be less pronounced when compared to E. lateriflora which is usually smaller and grows closer to the ground. This unique ecological succession on the leaves presumably indicates that height is a factor to consider in the type of organisms that colonize the phyllosphere microbiome, as F. thonningii is a tree, while E. lateriflora is a shrub. The fact that only tryptone soya agar was used for bacteria could favour selection of organisms that grow well in the medium, such as Enterobacteriaceae and Pseudomonadaceae, although we tried to reduce this effect by diluting the medium in 10-folds, as done by Yashiro et al.60 to mimic the nutrient deficient nature of the phyllosphere. Furthermore, our selection of cultured isolates for identification based on visual colony morphologies may have created bias in sampling. At least some mammalian- and avian-associated E. coli lineages are known to be associated with diminished phyllosphere abundance96, which may account for the relatively smaller number of non-E. coli isolates in this study compared with other culture-based phyllosphere studies. Dominance of Enterobacteriaceae other than Pantoea sp. in the phyllosphere, although relatively unpopular, has been several times reported1,97,98. Enterobacteriaceae, including antimicrobial resistant Shigella and Enterobacter, were found to dominate the phyllosphere or arugula (Eruca sativa), a commonly consumed vegetable in Graz, Austria, with resultant implications for the heath of those who consume it raw97. A study of banana phyllospheres in Uganda similarly reported an overdominance of enteric bacteria in the banana pseudostems98. The overdominance of Enterobacteriaceae may be linked to the ubiquitous use of organic manure in agriculture, and as the average annual temperature is approximately 26 °C and 27 °C in Uganda and Nigeria, respectively, enteric bacteria will be able to thrive for extended periods. Additionally, birds perching on trees may produce droppings that may introduce enteric organisms into phyllospheres.

All non-E. coli isolates belonged to three phyla, Pseudomonadota, Actinomycetota and Bacillota, which are common on plants74,99,100. The majority of isolates from E. lateriflora belonged to the families Comamonadaceae (Pseudomonadota), Enterobacteriaceae (Pseudomonadota) and Pseudomonadaceae (Pseudomonadota), and those from F. thonningii belonged to the families Enterobacteriaceae (Pseudomonadota) and Bacillaceae (Bacillota). Other less dominant families include Brevibacteriaceae (Actinomycetota), Micrococcaceae (Actinomycetota), Xanthomonadaceae (Pseudomonadota), Microbacteriaceae (Actinomycetota), Sphingomonadaceae (Pseudomonadota) Staphylococcaceae (Bacillota) and Methylobacteriaceae (Pseudomonadota). Thirteen and nineteen genera were represented on E. lateriflora and F. thonningii, respectively, while 29 isolates, designated ‘unknown’, could not be classified into any genera based on the cut offs as set by Kim et al., (2014). Fourteen (48.3%) of these isolates were Burkholderiales, and work to characterize and report these potential novel species is ongoing. Isolates belonging to all three phyla, Pseudomonadota, Actinomycetota and Bacillota, were detected in both E. lateriflora and F. thonningii. However, the composition of the phyllosphere at lower classification levels is unique to each plant. The Bacillota genera Bacillus, Lysinibacillus and Staphylococcus were found in both plants but in higher proportions in F. thonningii. Similarly, Pseudomonas was detected in higher proportions in E. lateriflora. Important plant-associated taxa, including Pantoea and Xanthomonas, were found only in F. thonningii. Actinomycetales, represented only by Brevibacterium, were isolated exclusively in E. lateriflora, while Micrococcales were isolated exclusively in F. thonningii. Overall, E. lateriflora and F. thonningii shared seven bacterial genera and differed by four and nine unique genera, respectively. Several factors, including host plant intrinsic characteristics, plant location, environmental factors and microbiome interactions, influence phyllosphere microbiome composition101,102,103. Of these factors, the host plant is the most significant, as evident from the similarity in the phyllosphere microbiome of identical plants across different locations, and may account for the differences observed in the present study4,104,105,106. Some of the isolates identified are dynamic in their interactions with plants. Pseudomonas oryzihabitans, for instance, could be pathogenic, causing stem and leaf rot in Chinese muskmelon107 and rice panicle blight and grain discolouration108. Despite being a plant pathogen, P. oryzihabitans can also be beneficial and has been shown to promote plant growth and prevent colonization by pathogenic fungi and nematodes109,110,111. Other plant pathogens identified include Pantoea allii, Xanthomonas perforans and Xanthomonas euvesicatoria112,113,114. The phyllosphere of these plants could thus serve as a reservoir for these plant pathogens and this may have impact in agricultural production. Xanthomonas perforans and Xanthomonas euvesicatoria for example, are pathogens that influence tomato and pepper plants which are farmed in Nigeria in large quantities. Transfer of these pathogens may directly influence production and lead to economic losses to both farmers and the entire country.

The majority of the genera identified in this study have been reported to be found in other plants75,76,115,116,117. However, at the species level, only a few species have been reported on plants. Pseudomonas taiwanensis, for instance, was first classified after isolating it from soil118, while Pantoea allii was first classified after isolation from onion plants112. Actinomycetota, represented in this study by Microbacterium, Brevibacteria and Micrococcus, has been reported119,120 to produce secondary metabolites with diverse activities, including antimicrobial activity and promotion of plant growth. Similarly, the genera Pseudomonas and Pantoea consist of members exhibiting antagonistic properties17,75 and are commonly known to coexist with plants in a symbiotic relationship. The occurrence of these genera on E. lateriflora and F. thonningii analysed in this study could contribute positively to the antimicrobial activity of the plant. Indeed, Egamberdieva et al. (2017) reported a correlation between the proportion of antagonistic bacteria and the antimicrobial properties of two plants that were discordant for medicinal activity. The medicinal plant employed in the above study (Hypericum perforatum) housed a significant population of antagonistic bacteria compared to its nonmedicinal counterpart, Ziziphora capitata. As has been reported previously38,121, hexane, ethyl-acetate and methanol extracts of E. lateriflora and F. thonningii show activity against type cultures of Pseudomonas aeruginosa, Staphylococcus aureus, Escherichia coli and Citrobacter freundii, as they did in this study. Phyllosphere isolates obtained from both plants showed wider zones of inhibition to the extracts of both plants, whereas only a handful were susceptible to ciprofloxacin and/or chlorocresol. In the same vein, phyllosphere isolates were typically inhibited by the plant extracts, as shown by the MICs, and not by chlorocresol. MIC values were lower for the E. lateriflora extract than for F. thonningii across board. However, one Xanthonomas perforans, one Micrococcus luteus and two strains of Microbacterium testaceum all isolated from F. thonningii had higher MICs to the E. lateriflora extract.

Overall, the data do not support our hypothesis that phyllosphere bacteria on leaves are resistant to the antimicrobial secondary metabolites extracted from plants. While it is possible that the chemical nature of antimicrobial principles is altered during extract preparation, it is more probable that secondary metabolites produced by healthy plants are embedded within the tissues of the plants, and as such, bacteria leaving on the surfaces of the plants may not have direct contact with such metabolites unless the plant is injured. Also, phyllosphere bacteria may attach firmly to the surface by biofilm formation and this may also reduce the impact of the plants’ active principles on the microbiome. Our findings also point to the possibility that antimicrobial resistance to active plant metabolites may be uncommon in nature, pointing to the potential of active principles to yield therapeutics less likely to be compromised by antimicrobial resistance. As only two plants were evaluated in this study, the generalizability of our findings is presently unknown and should be the subject of future research.

Conclusions

This study has identified the bacteria associated with the leaves of E. lateriflora and F. thonningii, which are important components of the plant, among which are organisms that have probably not been described previously. The phyllosphere bacteria of both plants are dominated by Enterobacteriaceae by culture and Sphingomonas by amplicon sequencing. This study has also shown that culturable components of plant microbiomes can be inhibited by plant extracts. It is, however, noteworthy that F. thoninngii isolates are inhibited by higher concentrations of the extract from this plant than are E. lateriflora isolates, pointing to the possibility that antimicrobial principles may shape the microbiome. This hypothesis is supported by the observation that each of the plants we studied has a distinct, narrowly overlapping culturable phyllosphere. As only culture and amplicon sequencing methods were employed in this study, we may have missed some important components of the phyllosphere: those unculturable under the conditions employed and those that may be negatively biased by the primers used for amplicon library preparation. However, the study has provided insight into the composition of the phyllosphere, which includes some plant-associated taxa as well as some probable new species. Although we did not sample the plants to saturation, we provide baseline data for more in-depth studies on the phyllosphere of these plants and other medicinal plants.