Uncovering multi-faceted taxonomic and functional diversity of soil bacteriomes in tropical Southeast Asian countries

The soil bacterial diversity

The soil bacteriome dataset in this study included 558 soil samples collected from Thailand, the Philippines, Malaysia, and Indonesia (Fig. 1).

Mapping to the global gridded soil information system: SoilGrids²¹, the soil samples of each selected country encompassed different soil classes (Supplementary Figure S2). The soil from Thailand samples were mostly Acrisols, which comprise clay-rich subsoil with low fertility and high aluminium content. The soil from the Philippines samples were mostly Gleysols, iron-rich wetland soil saturated with groundwater or underwater or in tidal areas. The soil from Malaysia samples were mostly Ferralsols. The soils from Indonesia samples were of mixed soil classes; nearly half (45%) of them belonged to Nitisols, well-drained soil with a moderate-to-high clay content and limited phosphorus availability. Ferralsols took up about 20% of the Indonesia soil samples while another 18% were Histosols (moist soils with thick organic layers). The soil pH levels were significantly different among the soil samples of 4 selected countries (ANOVA, P value < 0.01) (Fig. 2). The location of soil samples included in this study were shown in Fig. 3A.

Alpha diversity of bacterial species in the soil samples were assessed using Chao1 species richness estimator (Fig. 3B) and Shannon’s diversity index (Fig. 3C). Thailand soil bacteriomes harbored significantly higher species richness and Shannon’s diversity index than the others (P-value < 0.001). The Shannon’s diversity indices of Indonesia soil samples had wider range than those of other countries’ samples. In total the soil bacteriomes of the soil samples from the selected four countries contained 11,893 bacterial species, which predominantly belonged to Proteobacteria, Actinobacteria, Firmicutes, Bacteroidetes, and Cyanobacteria (Supplementary Figure S1).

Differential taxonomic richness in the soil bacteriomes

The taxonomic richness comparison showed that, when compared the taxonomic richness based on the number of genera, there were 350 differential families among the 4 countries (Supplementary Table S1). On the other comparisons which based on the number of species, significant species richness shifts were detected in 18 differential families and 117 differential genera (Supplementary Table S1-S2). The differential taxon was a family or a genus that demonstrated a significant differential shift in richness (number of member genera or species within that family or genus) across 4 countries.

Based on genus richness (number of member genera within each family), Flavobacteriaceae was found to be one of the most diverse bacterial families (i.e., families with high numbers of member genera) in the soil bacteriomes and differentially exhibited the highest genus richness in Thailand with 24 genera (Supplementary Table S1). Moraxellaceae and Chitinophagaceae were the differential bacterial families that harbored a high genus richness specifically to Malaysia. Several families including Peptococcaceae, Paenibacillaceae, Spirochaetaceae, Desulfobulbaceae, Syntrophobacteraceae, and Alteromonadaceae, exhibited significantly higher genus richness in the Philippines soil bacteriome than in the other countries. On the other hand, Bradyrhizobiaceae exhibited differentially high richness in Indonesia.

But when comparing species richness (number of member species within each family), Micromonosporaceae was the differential family with the highest species richness. The soil bacteriome of each country contained more than 30 species of Micromonosporaceae (Philippines = 59, Indonesia = 49, and Thailand = 36), except in Malaysia’s where only 9 species of this family were found (Supplementary Table S1). Several families including Anaerolineaceae, Nostocaceae, Sporomusaceae, Desulfobulbaceae, Syntrophobacteraceae, and Syntrophobacteraceae, exhibited differentially higher species richness in the Philippines soil bacteriomes than in the other countries.

At the genus level, it was noted that differential genera were small genera (i.e., genera with small numbers of member species). The most diverse differential genus was Legionella, which differentially harbored higher species richness in Thailand than in the other countries (Supplementary Table S2).

Species abundance patterns in the soil bacteriomes

An enriched species was defined as a species with an observed abundance in a particular sample group significantly higher than expected by random chance²². In this study, a total of 282 enriched soil bacterial species was found among the 4 countries. The enriched bacterial species taxonomically belonged to 25 phyla, 40 classes, 72 orders, and 125 families (Supplementary Table S3). Overall, Proteobacteria and Firmitcutes was the largest taxon groups of the enriched bacterial species in all 4 countries (Fig. 4), constituting 41.84% (118 out of 282 species) and 16.67% (47 out of 282 species), respectively.

The numbers and taxonomic distribution of the enriched species differed from country to country. Some bacterial taxa were specifically prevalent in the enriched species set of particular countries. For instance, there were 7 enriched species of Chitinophagaceae, 5 enriched species of Paenibacillaceae, and 5 enriched species of Isosphaeraceae in Malaysia while the members of these families were not present in the enriched species set of the other 3 countries (Supplementary Table S3). The uneven taxonomic proportion of the enriched species set was also observed at the phylum level as shown in Fig. 4.

Interspecific association of bacteria in the soil bacteriomes

For each of the selected country, the communities in the association networks of the top abundant bacterial species were identified as shown in Fig. 5.

The total 56 soil bacterial species in the association networks of all 4 countries belonged to diverse taxa (12 phyla, 24 classes, 35 orders, 42 families) (Supplementary Table S4-S7). Out of these 56 species, Proteobacteria was the most prevalent taxon group with 19 species, followed by Firmicutes (7 species). At the family level, in Thailand, Malaysia, and the Philippines, some of the bacterial families to which the species in the association networks belong were the differential families (based on species richness): 6 out of 17 families in Thailand network, 4 out of 14 families in Malaysia, and 10 out of 22 families in the Philippines (Supplementary Table S1, S4-S6). However, none of 14 bacterial families in the association network of Indonesia were differential taxa (Supplementary Table S1, S7). These differentiated taxonomic distribution may result from the heterogeneity of soil class in Indonesia soil samples while the soil samples in other three countries were more homogeneous, i.e., dominated by a single soil class for each country.

Based on the interaction degrees of the species as described in the method section, the association networks included hub and/or semi-hub species depending on the community structure.

However, the interaction degree of species did not relate to the taxonomic group to which that species belongs. Neither significant correlations between interaction degree and local abundance of the species could be found. Even when considering the same species, the interaction degree in the association network seemed to vary from country to country. For example, Acidibacter ferrireducens (Gammaproteobacteria), which was the only species commonly found in all 4 countries, was the hub species of the community I3 in Indonesia (degree = 8, betweenness = 93) and the semi-hub of the community T3 in Thailand (degree = 2, betweenness = 18), but had low interaction degrees in Malaysia’s community M4 and Philippines’ community P2 (degree = 1, betweenness = 0) (Fig. 5). This pattern indicates that the interaction of bacteria in the association network along their distribution range can be adaptively altered by external factors, such as surrounding environment and co-occurring microbial species.

The soil bacteriomes and biogeochemical cycles

Using the genome data of 164 enriched soil bacterial species available in the NCBI Genome Database, the functional profiles related to biogeochemical cycles of N, S, C, and P were analysed as displayed in Fig. 6.

Nitrogen biogeochemical pathway

The variation was observed in the association networks of the top abundant soil species (Fig. 5 and Supplementary Table S8). Most of the nitrogen-fixing bacterial species in the association networks of top abundant species belonged to Firmicutes (Supplementary Table S8). Exceptionally, in the Malaysia’s association network, there were no Firmicutes species present (Fig. 5). The only nitrogen-fixing species in the Malaysia’s association network was Roseiarcus fermentans, which belongs to the phylum Proteobacteria.

According to Fig. 6, nitrogen metabolic profiles also varied depending on the enriched species in the soil bacteriomes of the countries. There were 27 enriched species of nitrogen-fixing bacteria in this study, all of which encompassed the molybdenum-dependent nitrogenase genes (nifD, nifK, nifH, and anfG) (Supplementary Table S8). Thus, it is clear that the molybdenum is crucial in bacterial nitrogen fixation²³. However, one nitrogen-fixing species, Rhodopila globiformis (Proteobacteria), also contained the genes that encoded vanadium-dependent nitrogenase (vnfD, vnfK, vnfG, vnfH). The presence of the alternative nitrogenase might result from the limited amount of molybdenum in the environments^24,25.

In the assimilatory nitrate reduction process, the primary nitrate reductase gene, nasAB was found in 36 enriched soil bacterial species, most of which belonged to Proteobacteria (Supplementary Table S8), while the alternative nitrate reductase, narB were found in 4 non-Proteobacteria species. Likewise, the nitrite reductase gene, nirA were found in 20 enriched bacterial species (Supplementary Table S8), while another nitrite reductase gene, NIT-6, was not found in any enriched bacterial species in this study (Fig. 6). In contrast, every community of the top-abundant soil networks encompassed species containing genetic capabilities to produce essential enzymes for dissimilatory nitrate and nitrite reduction (Fig. 5 and Supplementary Table S8). Even in the cases where a single species had genetic machinery to carry out only dissimilatory nitrate or nitrite reduction, the missing genes would be compensated by another co-occurring species. For example, Gaiella occulta, which had the dissimilatory nitrate reductase genes (narGHI) but lacked the dissimilatory nitrite reductase genes (nirBD and nrfAH), co-occurred in the same communities with other soil bacterial species with nitrite reductase genes, namely Limisphaera ngatamarikiensis (Verrucomicrobia) in the community M4 in Malaysia and Geminisphaera colitermitum (Verrucomicrobia) in the community P5 in the Philippines.

Only a few communities in the top abundant soil networks encompassed complete gene sets for the denitrification, in which nitrite is successively converted to nitric oxide, nitrous oxide, and ultimately nitrogen gas. The denitrification in each community was observed to proceed by a single species with all essential denitrification enzymes; for example, Thiobacillus thioparus (Proteobacteria) in Thailand’s community T3, Burkholderia pseudomallei (Proteobacteria) and Brucella suis (Proteobacteria) in Indonesia’s communities I2 and I3. In the Philippines’ network, denitrifying bacteria in the community P5 did not possess all key genes in denitrification; some species possessed nitrite reductase (nirS and nirK) genes while another possessed nitrous-oxide reductase (nosZ) gene (Fig. 5 and Supplementary Table S8). In this community, there was Desulfomonile tiedjei (Proteobacteria), which harbored nitric oxide reductase (norB and norC) genes. The co-occurrence of complementary species seemed to indicate that D. tiedjei might function as a bridge in order to complete the denitrification process.

Sulfur biogeochemical pathway

In accordance with the nitrogen reduction pathway, the variation of sulfur metabolic profiles of the soil bacteriomes among the 4 selected Southeast Asian countries also apparently reflected the differences in enriched species and their taxonomy (Fig. 6). There were 154 enriched soil bacterial species which possessed at least one gene in an assimilatory sulfate reduction process (Supplementary Table S9). Assimilatory sulfate reduction is a pathway to convert inorganic sulfate to sulfide, which is a form of sulfur necessary to form cysteine and homo-cysteine²⁶. While several Proteobacteria enriched species were found to be genetically capable of assimilatory sulfate reduction, the Alphaproteobacteria species enriched in Thailand uncharacteristically lacked the key ATP sulfurylase genes (cysNC and cysH).

125 enriched bacterial species possessed genes for dissimilatory sulfate reduction, which is a form of anaerobic respiration that uses sulfate as the terminal electron acceptor (Supplementary Table S9). The taxonomic diversity of these species was similar among the countries, as a majority belonged to Proteobacteria. However, in terms of abundance, the enriched species with those genes had significantly high abundance in Indonesia and Thailand, and significantly low abundance in Malaysia (ANOVA, P-value < 0.001).

Almost every community encompassed the genes necessary to carry out the assimilatory and dissimilatory sulfate reduction, except the community T2 in Thailand that lacked genes involving in sulfate reduction entirely (Fig. 5 and Supplementary Table S9). The only sulfur functional gene found in the available genomes in T2 was soux, which regulates sulfite detoxification. However, there were only 3 communities that encompassed the necessary gene set to complete dissimilatory sulfate reduction due to the low number of species possessing alkaline proteases (AprAB) among the association networks. The aprA and aprB genes were found in the genomes of Thiobacillus thioparus (Proteobacteria) in Thailand’s community T3, Reyranella soli (Proteobacteria) in the Philippines’ community P2, and Acidisphaera rubrifaciens (Proteobacteria) in Indonesia’s community I1 (Fig. 5 and Supplementary Table S9).

Among the enriched bacterial species in this study, 24 species possessed soxA and soxX, which are essential genes in the first step of the sulfur oxidation (SOX) system that converts sulfur to sulfate through sulfide or thiosulfate and SoxYZ-S-SH complex^27,28 (Supplementary Table S9). The enriched species with soxAX was significantly more abundant in Indonesia than in Thailand and Malaysia (ANOVA, P-value = 0.0003). However, no enriched bacterial species in the Philippines harbored SOX system-related genes. Remarkably, no enriched bacterial species possessed soxB, soxC, and soxD, which are genes involving in the electron release process in the SOX system. Moreover, SOX system-related genes were considerably rare in the association networks of the top abundant soil species (Fig. 5 and Supplementary Table S9). Even in Indonesia’s network where every community possessed anaerobic sulfate-respiring species, only the community I1 harbored the bacterial species with SOX system-related genes, Acidisphaera rubrifaciens.

There were also 11 enriched soil bacterial species harboring fccAB and fsr genes in the sulfur oxidation pathway using ferricytochrome c or quinone as an electron acceptor (Supplementary Table S9). Although the abundance of sulfur-oxidizing species was not significantly different across Thailand, Indonesia, and Philippines (ANOVA, P value = 0.12), the taxonomic composition was considerably different across the countries (Fig. 6). In Thailand and the Philippines, most of the enriched sulfate-oxidizing species belonged to Order Eubacteriales (Phylum Firmicutes). In Indonesia, Salmonella enterica (Proteobacteria) was the only enriched sulfate-oxidizing species.

Utilisation of carbohydrates and aromatic compounds

There were 113 enriched soil bacterial species that possessed at least one gene encoding carbohydrate-utilising enzymes (alpha-glucosidase, beta-glucosidase, alpha-galactosidase, beta-galactosidase, alpha-mannosidase, cellobiohydrolase, and pectate lyase) (Supplementary Table S10). These carbohydrate-utilising species mostly belonged to Proteobacteria and Firmicutes which were the most common phyla in the soil bacteriomes. However, some taxonomic groups seemingly exerted specificity to particular types of oligosaccharides and geographical environment. For example, the majority of the enriched species with malZ gene in Thailand and Malaysia soil bacteriomes belonged to Proteobacteria, whereas, in the Philippines and Indonesia, this gene was predominantly found in Firmicutes species (Fig. 6).

In terms of degradation of aromatic compounds, there were 13 enriched species harboring at least one gene involving in hydroxylation of aromatic compounds (ethA, nagH, and nagG) (Supplementary Table S10). Most of these aromatic compound-degrading bacteria belonged to Phylum Proteobacteria, especially Burkholderia and Paraburkholderia spp. Nonetheless, 71 enriched species possessed genes in dct family (dctA, dctP, and dctM), which encodes tripartite ATP-independent periplasmic (TRAP) C4 dicarboxylate transporters. TRAP C4 dicarboxylate transporters are the key regulator of C4 dicarboxylate movement across cellular membranes²⁹. The most common dct gene in the enriched soil bacteria was dctA, which was found in 55 enriched species (Fig. 6). Most of the enriched soil species with the dctA gene (43 out of 55) belonged to Phylum Proteobacteria. Some Proteobacteria species even possessed 3–5 copies of the gene in their genomes (Supplementary Table S10).

Phosphorus uptake and scavenging

The main bioavailable form of phosphorous for soil bacteria is the orthophosphate anion (PO₄^3-), commonly known as inorganic phosphate (P_i)³⁰. The major P_i uptake mechanism in the soil bacteriomes was regulated by PhoR/PhoB two-component system which is a signal transduction controlling gene transcription for phosphorus assimilation and phosphorus scavenging²⁹. There were 118 enriched soil species harboring phoR and/or phoB genes (Supplementary Table S11). PhoR/PhoB-containing species did not show taxonomic preponderance specific to the countries. All enriched Actinobacteria species, however, did not possess phoR or phoB in their genomes (Fig. 6). In terms of comparative abundance, the distribution of the species with phoR gene did not significantly differ across the countries (ANOVA, P value = 0.002) whereas the enriched species with phoB had significantly larger abundance in Indonesia soil bacteriomes than in other countries (ANOVA, P-value = 1.4E−12).

Along with the PhoR/PhoB system, Pst transporter is essential for detecting P_i, and thus effectively regulating P_i uptake³⁰. 149 enriched soil bacterial species harbored at least one gene of pstSCAB operon in their genomes (Supplementary Table S11). The prevalence of the pst genes were not significantly different between the countries (Fig. 6). In addition, another major regulator of the PhoR/PhoB is PhoU, which functions as PhoR/PhoB inhibitor protein. Again, 141 enriched species with the phoU gene were distributed evenly across the countries, signifying the universality of Pst and PhoU as essential PhoR/PhoB regulators in the soil bacteriomes³¹.

Phosphatase genes often act downstream of the PhoR/PhoB system for the hydrolysis of phosphoric esters. Among the P_i scavengers, most of the phosphatase genes found in the enriched soil bacteria were the alkaline phosphatases (phoD and phoAB) (Fig. 6). There were 42 and 65 enriched species harboring phoD and phoAB, respectively (Supplementary Table S11). In contrast, the acid phosphatases (phoN) were a less common group of phosphatases in the enriched soil species, as there were only 14 enriched species possessing the phoN gene.

C-P lyase pathway is a system to uptake phosphorus from C-P compound, especially phosphonates, through carbon-phosphorus bond cleavage³². With regards to genes in the C-P lyase core complex, 23 enriched soil species were shown to harbor phnM, phnL, phnJ, phnI, phnH, and phnG (Supplementary Table S11). According to Fig. 6, most C-P lyase-capable species belonged to Proteobacteria, Firmicutes, and Chloroflexi.

Source: Ecology - nature.com