Skip to content
Frontiers
Search

Original Research

Frontiers in Soil Science ·

Volume 6 - 2026 | https://doi.org/10.3389/fsoil.2026.1911455

Edaphic heterogeneity shapes bacterial microbiome composition and rhizobial diversity in smallholder cowpea farms of semiarid eastern Kenya

  • Ahmed Kassem Khater1,2*corresponding author
  • Ezekiel Mugendi Njeru3
  • Stephen Mwangi Githiri4
  1. 1Department of Molecular Biology and Biotechnology, Pan African University Institute for Basic Sciences, Technology and Innovation (PAUSTI), Jomo Kenyatta University of Agriculture and Technology (JKUAT), Nairobi, Kenya
  2. 2Department of Agricultural Biochemistry, Faculty of Agriculture, Al-Azhar University, Cairo, Egypt
  3. 3Department of Biochemistry, Microbiology and Biotechnology, Kenyatta University, Nairobi, Kenya
  4. 4Department of Horticulture and Food Security, Jomo Kenyatta University of Agriculture and Technology (JKUAT), Nairobi, Kenya
Article metrics0 viewsView details
PDFCiteMetrics

Abstract

Cowpea (Vigna unguiculata [L.] Walp.) is an important grain legume in semiarid Africa, where it contributes to food security and livelihood resilience in smallholder farming systems. However, information on the soil bacterial microbiome associated with cowpea production in semiarid eastern Kenya remains limited. This study investigated heterogeneity of soil physicochemical properties and bacterial community structure in cowpea farms from Machakos and Kitui Counties via 16S rRNA amplicon next-generation sequencing (NGS), with the aim of improving the understanding of belowground microbial patterns in these production systems. Most measured soil physicochemical properties varied significantly among farms, indicating substantial field-level heterogeneity, whereas county-level differences were less pronounced. Across all the samples, the bacterial communities were dominated by Actinomycetota, followed by Pseudomonadota, Bacillota, Chloroflexota, and Acidobacteriota. Rhizobial genera of agronomic relevance, including Bradyrhizobium, Rhizobium, Ensifer, and Mesorhizobium, were detected across farms, with Bradyrhizobium showing the highest relative abundance. The alpha-diversity indices varied among the samples, while the beta-diversity analysis revealed significant differences in the bacterial community composition between the counties. Preliminary correlation analysis revealed associations between selected soil variables and the distributions of dominant bacterial taxa and rhizobial genera, however none of these pairwise correlations remained significant after Benjamini-Hochberg FDR correction. Cowpea farm soils in semiarid eastern Kenya harbor compositionally distinct bacterial communities that vary with farm-level edaphic heterogeneity, with native rhizobial genera detected across all sites indicating ecologically relevant symbiotic bacterial resources; a direct test (Mantel test, constrained ordination) did not establish the measured soil variables as significant drivers of this compositional variation at the present sample size. These findings establish a baseline for soil bacterial ecology in Kenyan cowpea systems and demonstrate that site-specific, microbiome-informed strategies are needed for sustainable soil fertility management and effective deployment of native rhizobial inoculants in semiarid smallholder agriculture.

Keywords

1 Introduction

Cowpea (Vigna unguiculata [L.] Walp.) is one of the most important grain legumes in sub-Saharan Africa, where it contributes substantially to household nutrition, income generation, and the resilience of smallholder farming systems. Its value is particularly evident in semiarid environments because cowpea is cultivated both as a food and fodder crop and is relatively well adapted to heat, intermittent drought, and low-input conditions (1, 2). In addition to its agronomic and nutritional importance, cowpea can improve soil fertility through symbiotic nitrogen fixation, making it especially relevant in production systems where access to mineral fertilizers is limited (2, 3). In eastern Kenya, where rainfall variability and declining soil fertility constrain crop productivity, cowpea remains an important component of smallholder agriculture and a valuable crop for dryland food security (4). Understanding the belowground microbial context of these farming systems is therefore directly relevant to improving cowpea yields, as soil bacterial communities, particularly rhizobia, are integral to biological nitrogen fixation and plant productivity in low-input systems (3). Kitui and Machakos are among the leading cowpea-producing counties in Kenya and, together with neighboring semiarid counties, account for approximately 90% of national cowpea production, underscoring the crop’s economic and food-security relevance to smallholder livelihoods in this region (5, 6).

Despite its adaptability, cowpea productivity in semiarid regions remains inconsistent because crop performance is influenced not only by water limitation but also by soil fertility constraints and variation in belowground ecological conditions (7). Many smallholder fields in eastern Kenya are characterized by nutrient depletion, heterogeneous soil properties, and management differences that may influence plant growth directly and indirectly through their effects on soil microbial communities (8). Consequently, variation in the soil bacterial microbiome may represent an important but insufficiently characterized component of cowpea production systems in semiarid environments (9–11).

Bacterial community assembly in semiarid soils is governed primarily by environmental filtering, the selective pressure exerted by abiotic conditions such as pH, water availability, and nutrient gradients that determine which taxa can persist and proliferate, rather than by stochastic dispersal alone (12, 13). Lauber et al. (14) demonstrated that soil pH was a strong predictor of bacterial community structure across soils, whereas Zhou et al. (15) further reported at the global scale that more than 75% of bacterial genera are predicted by soil pH. In agricultural soils, these edaphic effects interact with management-related factors, and Mo et al. (16) reported that the fertility source can be a pronounced factor influencing soil microbiome assembly. These controls often operate at relatively fine spatial scales, meaning that neighboring farms may harbor distinct microbial communities even within the same broader agroecological zone (17–19). Understanding how such variability occurs in cowpea-growing soils is important for interpreting the ecological context within which crop performance is expressed.

For legumes such as cowpea, the importance of soil bacterial communities extends beyond general soil processes to include taxa associated with nodulation, biological nitrogen fixation, and rhizosphere functioning. Rhizosphere microorganisms modulate host plant immunity, stress tolerance, and symbiotic nitrogen fixation through direct root-associated interactions (20). In Kenya, previous studies have demonstrated substantial diversity among indigenous cowpea-nodulating bacteria and have shown that native rhizobial populations can vary across locations and differ in symbiotic effectiveness (21). Ondieki et al. (22) characterized the morphological and genetic diversity of cowpea rhizobia from lower eastern Kenya, including Machakos, Makueni, and Kitui, whereas Muindi et al. (23) reported that native rhizobia and associated endophytes from semiarid lower eastern Kenya displayed considerable diversity and variable symbiotic efficiency. Earlier work by Kimiti and Odee (3) also revealed that soil fertility management influences the population and effectiveness of indigenous cowpea rhizobia in semiarid eastern Kenya. These findings highlight the agricultural relevance of native microbial populations and suggest that soil biological variation may contribute to differences in cowpea performance across farms.

Although earlier work in Kenya characterized rhizobial diversity and symbiotic potential in cowpea systems, microbiome-scale information on whole-soil bacterial communities remains limited, particularly under semiarid smallholder conditions (24). In particular, little is known about how farm-level variation in soil physicochemical properties is associated with bacterial microbiome structure in cowpea farms in eastern Kenya. Addressing this gap is important for improving the ecological understanding of these production systems and for informing future microbiome-based approaches to sustainable soil fertility management and the potential use of beneficial native microbial resources.

Therefore, this study aimed to characterize soil physicochemical heterogeneity and its associations with bacterial microbiome composition and diversity in smallholder cowpea farms across semiarid eastern Kenya using 16S rRNA amplicon sequencing. We hypothesized that edaphic variation among farms would be a primary driver of bacterial community structure, and that rhizobial genera relevant to cowpea nodulation would be detectable in native soil communities across farms in the absence of commercial inoculation.

2 Materials and methods

2.1 Study design and setting

This cross-sectional farm-based study was conducted to assess the associations between soil physicochemical properties and bacterial microbiome structure in cowpea farms located in Machakos and Kitui Counties in semiarid eastern Kenya (Figure 1). Kitui and Machakos were selected as study sites because they are among the leading cowpea-producing counties in Kenya (5), while also representing two agroecologically distinct semiarid landscapes within the same broader climatic zone, allowing farm-level heterogeneity to be assessed within a nationally important cowpea production region. The farms were purposively selected on the basis of active cowpea cultivation and the absence of commercial rhizobial inoculation, ensuring that detected rhizobial communities reflected native soil populations. Sampled fields were not under cowpea monocropping but were part of mixed smallholder cropping systems, with cowpea cultivated alongside maize (Zea mays) and green gram (Vigna radiata), consistent with common intercropping and rotation practices in the region; this is considered a further contributor to the farm-level heterogeneity discussed below. Kitui County soils are predominantly classified as Vertisols, Cambisols, Luvisols, Ferralsols, Arenosols, Gleysols, and Acrisols, generally low in organic matter, nitrogen, and phosphorus (26); Machakos County lies within the same semiarid Eastern Kenya (Ukambani) agroecological belt and is characterized by broadly similar ferralitic and cambic soil associations. This exploratory cross-sectional design, involving twelve smallholder farms (six per county), was appropriate for generating initial microbiome baseline data in an understudied system. A sample size of twelve farms is consistent with published amplicon-based microbiome surveys in smallholder agricultural systems across Sub-Saharan Africa and other semiarid regions (27, 28), and was considered adequate for robust baseline characterization of microbiome patterns and edaphic associations across contrasting smallholder farms, while larger randomized surveys in future work will further extend spatial coverage. Farm selection criteria prioritized geographic representation of each county’s dominant smallholder agro-ecological conditions.

Sampling locations of cowpea farms in semi-arid Eastern Kenya. Maps show Kenya within Africa and the study counties of Machakos and Kitui. Enlarged panels indicate sampling sites in Matuu (Machakos County) and Matinyani (Kitui County). Dots represent individual sampled farms. Source: Reproduced from Khater et al. (25), © 2026 Khater, Kirui, Njeru and Githiri, used under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/deed.en) license.
Figure 1. Sampling locations of cowpea farms in semi-arid Eastern Kenya. Maps show Kenya within Africa and the study counties of Machakos and Kitui. Enlarged panels indicate sampling sites in Matuu (Machakos County) and Matinyani (Kitui County). Dots represent individual sampled farms. Source: Reproduced from Khater et al. (25), © 2026 Khater, Kirui, Njeru and Githiri, used under CC BY 4.0 license.

2.2 Soil sampling and physicochemical analysis

Soil sampling was conducted in April 2025, when the cowpea crop was at the flowering stage (40–45 days after sowing). At each farm, soil samples were collected from the 0–20 cm depth via sterile hand shovels. Samples were collected as bulk soil, away from root zones. Visible roots and plant debris were manually removed from each subsample prior to compositing, and the composite samples were sieved through a 2 mm mesh to remove remaining root fragments and coarse material before storage. Twenty subsamples were collected across each field and combined into one composite sample to represent the farm. The composite soil samples were transported to the Tissue Culture Laboratory at Kenyatta University for processing and storage until analysis. The soil physicochemical properties were determined via standard laboratory procedures as described by Carter and Gregorich (29). The measured parameters included the soil pH, electrical conductivity, total organic carbon, total nitrogen, available phosphorus, potassium, calcium and magnesium.

2.3 DNA extraction, amplification, and sequencing

Total genomic DNA was extracted from the soil samples via the DNeasy PowerSoil Pro Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. The DNA concentration was quantified via a VICTOR Nivo™ system (PerkinElmer) with PicoGreen reagents.

Bacterial community profiling targeted the V3–V4 region of the 16S rRNA gene. The sequencing libraries were prepared according to the Illumina 16S Metagenomic Sequencing Library Preparation protocol using the 341F/805R primer pair, which has been widely applied and validated for amplicon-based bacterial diversity studies (30). For library preparation, the following locus-specific primers were used with Illumina adapter overhang sequences: forward primer 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG-3′ and reverse primer 5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC-3′.

For the first PCR, 5 ng of genomic DNA was amplified in a reaction containing 5× reaction buffer, 1 mM dNTP mixture, 500 nM of each primer, and Herculase II Fusion DNA Polymerase (Agilent Technologies, Santa Clara, CA, USA). The thermal cycling conditions consisted of initial denaturation at 95 °C for 3 min, followed by 25 cycles of denaturation at 95 °C for 30 s, annealing at 55 °C for 30 s, and extension at 72 °C for 30 s, with a final extension at 72 °C for 5 min.

The first PCR products were purified via AMPure beads (Agencourt Bioscience, Beverly, MA, USA). Following purification, 10 µL of the first PCR product was used as a template in a second PCR for index attachment via Nextera XT Indexed Primers. The second PCR was performed under the same thermal conditions as the first PCR, except that amplification was carried out for 10 cycles. Indexed PCR products were purified again using AMPure beads.

The final purified libraries were quantified via the VICTOR Nivo™ system with PicoGreen reagents and assessed for fragment quality via the TapeStation D1000 ScreenTape system (Agilent Technologies, Waldbronn, Germany). Libraries were then normalized, pooled, and quantified by qPCR using the KAPA Library Quantification Kit for Illumina Sequencing Platforms. Sequencing was performed on the Illumina MiSeq i100 platform (Illumina, San Diego, CA, USA).

2.4 Sequence processing and taxonomic assignment

Raw paired-end reads were processed in R via the DADA2 pipeline. Low-quality reads were filtered and trimmed, sequencing error rates were determined from the data, identical reads were dereplicated, amplicon sequence variants (ASVs) were inferred, paired reads were merged, and chimeric sequences were removed. The DADA2 pipeline enables exact sequence inference from Illumina amplicon data and resolves biological variation at single-nucleotide resolution (31). Because ASVs provide greater taxonomic resolution and reproducibility than traditional operational taxonomic units (OTUs), they are now widely recommended for marker–gene studies (32).

Taxonomic assignment was performed via the SILVA ribosomal RNA gene reference database (version 138.1) (33). After taxonomic classification, sequences assigned to chloroplasts, mitochondria, and other nonbacterial taxa were removed prior to downstream analyses. A phyloseq object was then constructed to integrate the ASV table, taxonomy table, and sample metadata for community-level analysis (34).

2.5 Microbial diversity and community composition analysis

Alpha diversity was quantified via four complementary indices: observed richness, which counts the total number of detected ASVs; Chao1, which estimates true species richness including rare and undetected taxa; the Shannon index, which integrates both richness and evenness of the community; and the Simpson index, which emphasizes the probability that two randomly drawn individuals belong to different taxa and is sensitive to dominant taxa. Together, these indices capture different dimensions of bacterial diversity that no single metric addresses alone. Prior to diversity analyses, reads were rarefied to the minimum sample sequencing depth (32,620 reads; sample KF4) to normalize for unequal sequencing effort across samples, ensuring comparability of diversity estimates. Beta diversity was evaluated via Bray–Curtis dissimilarity, a widely used ecological distance metric for community composition data (35). Differences in bacterial community structure were visualized via principal coordinate analysis (PCoA), and variation in community composition between counties was tested via permutational multivariate analysis of variance (PERMANOVA), a robust nonparametric approach for multivariate ecological datasets (36).

The bacterial community composition was summarized at the phylum and genus levels, and the dominant taxa were visualized via relative abundance plots and clustered heatmaps. Microbiome data integration, management, and visualization were performed via the phyloseq framework (34).

2.6 Environmental associations between soil properties and bacterial taxa

Relationships between soil physicochemical properties and bacterial taxa were assessed via Spearman correlation analysis. Correlation matrices were constructed to evaluate associations between selected soil variables and the relative abundances of the top 10 bacterial phyla, as well as between soil variables and selected rhizobial genera. These relationships were visualized as clustered heatmaps to highlight patterns of positive and negative associations between soil physicochemical variables and bacterial groups across the sampled farms. To account for the large number of pairwise comparisons in these correlation matrices (~70 tests spanning Figures 2, 3), P values were adjusted for multiple testing via the Benjamini-Hochberg false discovery rate (FDR) procedure (37). To formally test whether farm-level edaphic variation was directly associated with bacterial community structure, a Mantel test (Spearman correlation, 999 permutations) was performed between the Bray-Curtis community dissimilarity matrix and a Euclidean dissimilarity matrix of standardized soil variables (38). A distance-based redundancy analysis (RDA) constrained by the eight soil variables was additionally performed on Bray-Curtis dissimilarities via capscale, with overall and per-variable significance assessed by permutation ANOVA (999 permutations) (36).

Correlations between soil variables and dominant bacterial phyla. Red and orange colors indicate positive correlations, whereas blue colors indicate negative correlations. Rows represent soil variables and columns represent bacterial phyla. None of these pairwise correlations remained significant after Benjamini-Hochberg FDR correction (q < 0.05); patterns shown are exploratory.
Figure 2. Correlations between soil variables and dominant bacterial phyla. Red and orange colors indicate positive correlations, whereas blue colors indicate negative correlations. Rows represent soil variables and columns represent bacterial phyla. None of these pairwise correlations remained significant after Benjamini-Hochberg FDR correction (q < 0.05); patterns shown are exploratory.
Correlations between soil variables and rhizobial genera. Red and orange colors indicate positive correlations, whereas blue colors indicate negative correlations. Rows represent soil variables and columns represent rhizobial genera. None of these pairwise correlations remained significant after Benjamini-Hochberg FDR correction (q < 0.05); patterns shown are exploratory.
Figure 3. Correlations between soil variables and rhizobial genera. Red and orange colors indicate positive correlations, whereas blue colors indicate negative correlations. Rows represent soil variables and columns represent rhizobial genera. None of these pairwise correlations remained significant after Benjamini-Hochberg FDR correction (q < 0.05); patterns shown are exploratory.

2.7 Statistical analysis and visualization

All analyses were performed in R (version 4.5.2). Differences in soil physicochemical properties among farms were evaluated by one-way ANOVA followed by Tukey’s HSD test for pairwise comparisons. County-level differences in soil variables were assessed by independent samples t-test or Mann-Whitney U test based on data normality. Differences in alpha diversity indices between counties were tested using Mann-Whitney U tests. Statistical significance was set at P < 0.05 throughout.

3 Results

3.1 Soil physicochemical properties

The soil physicochemical properties differed among the sampled cowpea farms in Kitui and Machakos Counties (Table 1). The soil pH ranged from 5.82 ± 0.10 in KF1 to 6.67 ± 0.07 in MF6. Significant farm-level differences were observed for pH, electrical conductivity, total nitrogen, phosphorus, potassium, magnesium, and calcium, whereas total organic carbon did not differ significantly among the farms. At the county level, most soil variables, including pH, electrical conductivity, total organic carbon, total nitrogen, phosphorus, potassium, and calcium, did not differ significantly between Kitui and Machakos (Table 2). In contrast, magnesium was significantly greater in Machakos (0.33 ± 0.03 mg kg−1) than in Kitui (0.19 ± 0.03 mg kg−1; P = 0.0036). Principal component analysis revealed variation in the soil physicochemical properties among the sampled farms in Kitui and Machakos Counties (Figure 4).

Table 1. Soil physicochemical properties of the sampled cowpea farms.

FarmpHEC(dS/m)TOC(%)N(%)P(mg kg−1)K(mg kg−1)Mg(mg kg−1)Ca(mg kg−1)
KF15.82 ± 0.10 d0.08 ± 0.00 de0.68 ± 0.09 a0.03 ± 0.00 j8.35 ± 0.12 f26.93 ± 0.23 i0.20 ± 0.00 g3.43 ± 0.06 f
KF25.90 ± 0.08 cd0.04 ± 0.00 h0.96 ± 0.10 a0.09 ± 0.00 g13.53 ± 0.06 d38.97 ± 0.32 g0.26 ± 0.00 ef3.64 ± 0.02 ef
KF36.16 ± 0.04 bcd0.39 ± 0.01 a0.66 ± 0.15 a0.07 ± 0.00 h6.73 ± 0.06 g23.23 ± 0.34 j0.11 ± 0.00 h3.82 ± 0.06 de
KF46.63 ± 0.12 a0.08 ± 0.00 e0.94 ± 0.08 a0.12 ± 0.00 e16.50 ± 0.22 c64.67 ± 0.91 c0.24 ± 0.00 f4.96 ± 0.04 b
KF56.50 ± 0.10 ab0.06 ± 0.00 g0.66 ± 0.11 a0.12 ± 0.00 e16.18 ± 0.17 c59.04 ± 0.49 d0.21 ± 0.00 g4.77 ± 0.10 b
KF65.87 ± 0.09 d0.07 ± 0.00 f0.91 ± 0.15 a0.14 ± 0.00 d12.84 ± 0.15 d54.27 ± 0.51 e0.10 ± 0.00 h4.03 ± 0.06 d
MF16.33 ± 0.08 ab0.10 ± 0.00 c0.85 ± 0.10 a0.17 ± 0.00 b6.57 ± 0.08 g74.67 ± 0.82 b0.37 ± 0.01 b2.50 ± 0.03 g
MF26.29 ± 0.07 abc0.04 ± 0.00 h0.67 ± 0.13 a0.15 ± 0.00 c4.85 ± 0.07 h52.64 ± 0.53 e0.26 ± 0.00 de2.12 ± 0.05 h
MF36.32 ± 0.08 ab0.04 ± 0.00 h0.76 ± 0.11 a0.30 ± 0.00 a17.47 ± 0.37 b63.27 ± 0.40 c0.41 ± 0.00 a4.43 ± 0.04 c
MF46.43 ± 0.06 ab0.09 ± 0.00 d0.66 ± 0.11 a0.10 ± 0.00 f24.68 ± 0.26 a78.28 ± 1.17 a0.41 ± 0.01 a4.03 ± 0.03 d
MF56.60 ± 0.09 a0.12 ± 0.00 b0.72 ± 0.08 a0.16 ± 0.00 c6.31 ± 0.07 g42.55 ± 0.42 f0.28 ± 0.01 c6.05 ± 0.06 a
MF66.67 ± 0.07 a0.03 ± 0.00 i0.60 ± 0.12 a0.05 ± 0.00 i9.42 ± 0.09 e34.19 ± 0.34 h0.28 ± 0.00 cd4.46 ± 0.06 c
P-value<0.0001<0.00010.3614<0.0001<0.0001<0.0001<0.0001<0.0001

The values represent mean ± standard error (n = 5). Different lowercase letters within a column indicate significant differences among farms according to Tukey’s honestly significant difference (HSD) test at P < 0.05. The p-values shown at the bottom of each column represent the overall one-way ANOVA for farm effects.

Bold values in the P-value row indicate statistically significant differences among farms (P < 0.05) based on one-way ANOVA.

Table 2. County-level comparison of soil physicochemical properties.

VariableKituiMachakosP_value
pH6.15 ± 0.146.44 ± 0.070.1034
EC (dS/m)0.12 ± 0.050.07 ± 0.020.4035
TOC (%)0.81 ± 0.060.71 ± 0.040.2348
N (%)0.10 ± 0.020.16 ± 0.030.1534
P (mg kg−1)12.35 ± 1.6411.55 ± 3.210.829
K (mg kg−1)44.52 ± 7.0857.60 ± 7.190.224
Mg (mg kg−1)0.19 ± 0.030.33 ± 0.030.0036
Ca (mg kg−1)4.11 ± 0.253.93 ± 0.590.7909

The values represent means ± standard errors of the farms means (n = 6 farms per county). P values were obtained from t tests comparing county means.

Principal component analysis of soil physicochemical properties. PCA ordination shows variation in soil physicochemical properties among sampled farms. Each point represents one farm, and colors indicate county of origin.
Figure 4. Principal component analysis of soil physicochemical properties. PCA ordination shows variation in soil physicochemical properties among sampled farms. Each point represents one farm, and colors indicate county of origin.

3.2 Sequencing output and ASV recovery

Amplicon sequencing generated substantial read numbers across all the soil samples (Supplementary Table 1). The number of input reads ranged from 83,144 in KF4 to 157,037 in KF1. After filtering, denoising, merging, and chimera removal, the number of nonchimeric reads ranged from 32,650 to 147,791 per sample. The final number of retained reads ranged from 32,620 in KF4 to 147,742 in KF1 (Supplementary Table 1; Supplementary Figure S1). All the samples retained sufficient sequencing depth to support downstream diversity and community composition analyses.

3.3 Bacterial community composition

The bacterial communities across all the cowpea farm soils were dominated by five major phyla (Figure 5). Actinomycetota was the most abundant phylum (46.23%), followed by Pseudomonadota (15.87%), Bacillota (12.11%), Chloroflexota (10.16%), and Acidobacteriota (6.97%). Additional phyla, including Myxococcota, Bacteroidota, Gemmatimonadota, Planctomycetota, and Verrucomicrobiota, were consistently detected but occurred at relatively low relative abundances. Relative abundances varied among farms, although the dominant phyla were shared across both counties.

Relative abundance of dominant bacterial phyla. Stacked bar plot showing the relative abundance of the major bacterial phyla across sampled farms. Actinomycetota, Pseudomonadota, Bacillota, Chloroflexota, and Acidobacteriota were the dominant phyla. Farms are grouped by county.
Figure 5. Relative abundance of dominant bacterial phyla. Stacked bar plot showing the relative abundance of the major bacterial phyla across sampled farms. Actinomycetota, Pseudomonadota, Bacillota, Chloroflexota, and Acidobacteriota were the dominant phyla. Farms are grouped by county.

Among the detected rhizobial genera, Bradyrhizobium was the most abundant and occurred across all the sampled farms (Figure 6). Rhizobium, Ensifer, and Mesorhizobium were also detected but at lower relative abundances and with greater variation among farms. No consistent county-level pattern was observed for these genera.

Relative abundance of selected rhizobial genera. Bar plot showing the relative abundance of selected rhizobial genera, including *Bradyrhizobium*, *Rhizobium*, *Ensifer*, and *Mesorhizobium*, across sampled farms. Farms are grouped by county.
Figure 6. Relative abundance of selected rhizobial genera. Bar plot showing the relative abundance of selected rhizobial genera, including Bradyrhizobium, Rhizobium, Ensifer, and Mesorhizobium, across sampled farms. Farms are grouped by county.

The genus-level community structure was further resolved by hierarchical clustering of the dominant taxa (Figure 7). The heatmap showed marked variation in the abundance of dominant genera among farms, with clustering patterns reflecting both within-county and between-county differences in bacterial composition.

Heatmap of dominant bacterial genera. Clustered heatmap showing the relative abundance patterns of the dominant bacterial genera across soil samples collected from cowpea farms in Kitui and Machakos counties. Rows represent bacterial genera and columns represent individual farm samples. Color intensity indicates standardized relative abundance, with warmer colors representing higher abundance and cooler colors representing lower abundance. The top annotation bar indicates county of origin for each sample.
Figure 7. Heatmap of dominant bacterial genera. Clustered heatmap showing the relative abundance patterns of the dominant bacterial genera across soil samples collected from cowpea farms in Kitui and Machakos counties. Rows represent bacterial genera and columns represent individual farm samples. Color intensity indicates standardized relative abundance, with warmer colors representing higher abundance and cooler colors representing lower abundance. The top annotation bar indicates county of origin for each sample.

3.4 Bacterial diversity and richness

3.4.1 Alpha diversity

The alpha diversity varied among the samples and between the counties (Table 3). The Chao1 richness ranged from 1381.27 in KF1 to 2232.07 in MF4, the Shannon diversity ranged from 6.29 in MF5 to 6.92 in KF5, and the Simpson diversity ranged from 0.9942 in MF5 to 0.9982 in KF5. The mean Chao1 richness was greater in Machakos (1974.4 ± 83.0) than in Kitui (1812.1 ± 134.1). In contrast, the Shannon (6.83 ± 0.03) and Simpson (0.9978 ± 0.0002) values were greater in Kitui than in Machakos (6.53 ± 0.07 and 0.9945 ± 0.0006, respectively). Mann-Whitney U tests revealed no significant difference between counties for Chao1 richness (W = 15, P = 0.699). However, Shannon diversity was significantly greater in Kitui than in Machakos (W = 36, P = 0.005), and Simpson diversity was also significantly greater in Kitui (W = 36, P = 0.004), indicating that bacterial community evenness differed significantly between counties despite the absence of a significant difference in overall species richness.

Table 3. Alpha diversity indices of the soil bacterial communities.

Sample_IDChao1ShannonSimpson
KF11381.276.750.998
KF21453.086.870.998
KF31981.686.750.997
KF41840.816.830.997
KF52209.276.920.997
KF62006.276.850.997
MF11788.856.380.993
MF22216.256.650.994
MF31967.056.510.994
MF42232.076.630.994
MF51815.606.290.992
MF61826.356.700.996

3.4.2 Beta diversity and community structure

Beta diversity analysis revealed significant differences in bacterial community composition between Kitui County and Machakos County (Figure 8; Supplementary Table 2). PERMANOVA based on Bray–Curtis dissimilarity revealed a significant effect of county type on community structure (F = 2.076, R² = 0.1719, P = 0.004), indicating that county type explained 17.2% of the total variation in bacterial community composition.

Principal coordinates analysis of bacterial community composition. The ordination is based on Bray–Curtis dissimilarity, and each point represents one farm sample.
Figure 8. Principal coordinates analysis of bacterial community composition. The ordination is based on Bray–Curtis dissimilarity, and each point represents one farm sample.

3.5 Associations between soil physicochemical properties and bacterial taxa

Spearman correlation analysis revealed variable associations between the soil physicochemical properties and the dominant bacterial phyla (Figure 2). Positive correlations were observed between several nutrient-related variables and dominant phyla, whereas negative correlations were observed for other variables. In particular, Bacillota was positively associated with pH and magnesium, whereas Chloroflexota was negatively associated with several nutrient-related variables. Hierarchical clustering grouped the soil variables and bacterial phyla according to similarity in their correlation patterns.

Correlation analysis between the soil variables and rhizobial genera also revealed distinct association patterns (Figure 3). Bradyrhizobium and Rhizobium were positively associated with nitrogen, phosphorus, and potassium, whereas Ensifer and Mesorhizobium were negatively associated with several of these variables. Contrasting relationships were also observed with magnesium and pH among the rhizobial genera. After applying a Benjamini-Hochberg FDR correction across the ~70 pairwise tests represented in Figure 2 and 3, none of the individual soil-taxon correlations remained significant at q < 0.05; the patterns described above should therefore be interpreted as descriptive, exploratory associations rather than statistically robust relationships.

3.6 Direct test of the edaphic-driver hypothesis

To directly test the hypothesis that farm-level edaphic variation drives bacterial community structure, a Mantel test was performed between the Bray-Curtis community dissimilarity matrix and the Euclidean soil-variable dissimilarity matrix. This test did not detect a significant association (Mantel r = -0.105, P = 0.691). A distance-based redundancy analysis constrained by the eight soil variables similarly did not explain a significant proportion of community variation (pseudo-F test, P = 0.634; adjusted R2 = -0.050), and none of the eight soil variables was individually significant as a marginal term (permutation P range = 0.118-0.834). Given the modest sample size relative to the number of predictors (n = 12 farms, 8 variables, 3 residual degrees of freedom), this constrained ordination is likely underpowered, and these results should be interpreted as an absence of detectable evidence for a direct edaphic-community link at the present sample size rather than definitive evidence against it.

4 Discussion

This study characterizes farm-scale associations between soil physicochemical heterogeneity and bacterial microbiome composition in smallholder cowpea farms across semiarid eastern Kenya. The significant variation observed in most soil properties among farms, in contrast with the relatively limited county-level differences, indicates that local edaphic conditions are likely stronger determinants of soil heterogeneity than county identity alone. This interpretation is consistent with broader soil microbiome literature showing that bacterial community structure is often shaped more strongly by local environmental filters than by geographic distance per se. Lauber et al. (14) demonstrated that soil pH is a major predictor of bacterial community structure across broad spatial scales, whereas Fierer and Jackson (39) similarly reported that edaphic variation, especially pH, strongly influences bacterial biogeography. At the global scale, Bahram et al. (40) reported that microbial gene composition varies more strongly with environmental variables than with geographic distance, reinforcing the idea that local soil conditions can outweigh coarse spatial grouping in structuring microbiomes. It is important, however, to reconcile this county-level interpretation with the direct hypothesis test reported in Section 3.6: the Mantel test and constrained ordination did not detect a significant relationship between the eight measured soil variables and community dissimilarity. This apparent contrast is not contradictory. The PERMANOVA result establishes that Kitui and Machakos harbor compositionally distinct bacterial communities as a grouping factor, whereas the Mantel/RDA tests specifically ask whether that compositional difference is explained by the eight chemical variables measured here; a non-significant result for the latter indicates that this particular set of variables, at this sample size, does not provide statistical evidence of a direct link, not that no edaphic driver exists. Given only three residual degrees of freedom in the constrained model, this analysis was likely underpowered to detect a real but modest effect. We therefore revise our interpretation accordingly: farm- and county-level bacterial community differentiation is a robust finding of this study, but the specific claim that the measured edaphic variables are a demonstrated primary driver of that differentiation is not supported by the present data and should be treated as an open question for future, adequately powered studies.

Although the county-level contrasts in the measured soil properties were modest, the Bray–Curtis PERMANOVA results indicated significant separation in the bacterial community composition between Machakos and Kitui. This suggests that county-level microbiome differentiation may reflect the combined influence of subtle edaphic gradients, environmental variables not captured in the present dataset, and differences in field history or management. This pattern is ecologically plausible because bacterial beta diversity often responds to the interaction of multiple drivers rather than to a single soil variable. Fierer (41) emphasized that soil microbiomes are shaped by a complex combination of abiotic and biotic filters rather than by simple geographic proximity. Similarly, Delgado-Baquerizo et al. (42) reported that relatively few bacterial taxa dominate globally, but their ecological distributions are strongly associated with environmental conditions, whereas Mo et al. (16) recently demonstrated that agricultural practices, particularly fertility sources, are pronounced drivers of microbiome assembly in managed soils.

At the phylum level, the dominance of Actinomycetota, followed by Pseudomonadota, Bacillota, Chloroflexota, and Acidobacteriota, is consistent with the environmental context of semiarid agricultural soils. Actinomycetota are frequently associated with drought-prone and nutrient-variable soils, where their broad metabolic capacity and stress tolerance may confer an ecological advantage. Barnard et al. (43) reported that bacterial communities, including abundant Actinobacteria, respond strongly to drying–rewetting cycles, whereas Ebrahimi-Zarandi et al. (44) highlighted the ecological importance of Actinobacteria under drought stress and their persistence in water-limited environments. Similar phylum-level patterns have also been reported in arid and semiarid soils, including in recent work from Botswana, where Actinomycetota, Pseudomonadota, and Acidobacteriota were among the dominant bacterial groups (27). These patterns are also consistent with the copiotroph–oligotroph framework of soil bacterial ecology (45): the dominance of Actinomycetota and Chloroflexota reflects an oligotrophic, stress-tolerant life strategy suited to the nutrient-variable, moisture-limited conditions of dryland soils, whereas the comparatively lower abundance of typically copiotrophic Pseudomonadota and Bacillota fits the low organic-matter, low-input character of these smallholder soils. These observations support the view that the present microbiome composition reflects a dryland edaphic setting rather than an anomalous community structure.

The genus-level heatmap further revealed considerable farm-level heterogeneity, suggesting that bacterial assembly in these cowpea soils may be influenced by local environmental filtering. This is expected in smallholder systems, where neighboring farms can differ in nutrient inputs, residue return, tillage intensity, crop sequence, and organic matter management. Hartmann et al. (46) demonstrated that long-term farming system differences reshape soil microbial diversity and community structure, whereas Bier et al. (10) reported that the fertility source and soil depth are important determinants of the soil microbiome composition in agricultural systems. Peralta et al. (47) likewise reported that crop rotation history influences the soil microbiome and disease-suppressive potential. Taken together, these studies support the interpretation that the microbiome patterns observed here likely reflect cumulative field-level ecological histories rather than simple spatial proximity.

One of the most agriculturally relevant findings of this study was the detection of Bradyrhizobium, Rhizobium, Ensifer, and Mesorhizobium across the sampled farms. These genera include important legume-associated taxa involved in nodulation and biological nitrogen fixation, and their occurrence indicates that the studied soils harbor native bacterial groups with potential relevance to cowpea nutrition. The relatively greater abundance of Bradyrhizobium is biologically credible because this genus is widely associated with cowpea nodulation in African soils and frequently dominates symbiotic associations under stressful field conditions. Zahran (48) reviewed the importance of rhizobial adaptation in harsh edaphic environments, whereas Broughton et al. (49) emphasized the central role of rhizobial compatibility in legume symbiosis. In eastern Kenya specifically, Kimiti and Odee (3) demonstrated that integrated soil fertility management affects the population and effectiveness of indigenous cowpea rhizobia. Ondieki et al. (22) documented substantial morphological and genetic diversity among cowpea rhizobia from Machakos, Makueni, and Kitui, and Muindi et al. (23) reported that the symbiotic efficiency of native rhizobia and associated endophytes from semiarid Kenya varies. The present study extends those isolate-based findings by showing that rhizobial genera are also detectable in whole-soil community profiles across farms. However, this taxonomic signal should still be interpreted cautiously because 16S rRNA amplicon sequencing does not confirm the effectiveness of nodulation or nitrogen fixation under field conditions.

The alpha diversity results provide additional ecological insight that complements the beta diversity findings. While species richness as estimated by Chao1 did not differ significantly between counties (W = 15, P = 0.699), both Shannon diversity (W = 36, P = 0.005) and Simpson diversity (W = 36, P = 0.004) were significantly greater in Kitui than in Machakos. This pattern indicates that the two counties differed not in the total number of bacterial taxa detected, but in community evenness — the degree to which reads were equitably distributed across taxa. Greater Shannon and Simpson values in Kitui suggest that bacterial communities in Kitui farms were more evenly structured, with less dominance by individual taxa, relative to Machakos farms. This finding is consistent with the significant county-level difference in beta diversity (PERMANOVA: F = 2.076, R² = 0.172, P = 0.004), together indicating that Kitui and Machakos harbored compositionally and structurally distinct bacterial communities. Bahram et al. (40) and Delgado-Baquerizo et al. (42) both reported that large-scale microbiome differentiation often reflects ecological turnover among taxa more than simple richness changes, whereas Hartmann et al. (46) reported that management can shift evenness and dispersion in ways not fully captured by richness alone, patterns consistent with the present findings.

The correlation analyses provided preliminary, exploratory insight into the soil variables potentially associated with the observed taxonomic patterns, though as noted in Section 3.6, none of these pairwise correlations remained significant after FDR correction and should be read as hypothesis-generating rather than confirmed relationships. At the phylum level, the positive associations of Bacillota with pH and magnesium and the more negative associations of Chloroflexota with several nutrient-related variables are consistent with broader evidence that edaphic gradients influence major bacterial groups differently. Lauber et al. (14) reported that bacterial community composition closely tracks soil pH across landscapes, and Rousk et al. (50) reported that in an arable soil system, bacterial communities are strongly influenced by pH across a liming gradient. More recently, Xiong et al. (51) reported that soil pH amendment can alter microbial abundance, diversity, and composition, further reinforcing the importance of pH as a structuring variable. The present genus-level correlations, particularly the positive associations of Bradyrhizobium and Rhizobium with nitrogen, phosphorus, and potassium, are ecologically plausible, but they should not be interpreted causally because correlation analysis reflects covariation rather than mechanistic control. As reported in Section 3.6, a constrained ordination directly testing these relationships did not reach significance at the present sample size; larger, adequately powered surveys would be valuable for disentangling the relative contributions of individual soil variables more rigorously.

From an applied perspective, the results suggest that cowpea farms in semiarid eastern Kenya should not be regarded as microbiologically uniform. The coexistence of considerable farm-level soil heterogeneity, recurring dominant phyla, and variable rhizobial distributions suggests that local soil conditions shape the ecological background within which plant–microbe interactions operate. This information is relevant for future efforts aimed at improving soil fertility and developing microbial inoculants. Van der Heijden et al. (52) reported that soil microbes are major drivers of plant productivity, whereas Wagg et al. (53) reported that soil biodiversity and community composition influence ecosystem multifunctionality. Mendes et al. (54) and Luo et al. (20) further reported that specific rhizosphere microbiome configurations can be functionally important for plant health. In that context, the present findings support the idea that site specific, microbiome-informed approaches may be more appropriate than uniform recommendations when designing future interventions for semiarid cowpea systems.

The present findings should be interpreted within the context of the study design. The observational nature of the study means that associations between soil variables and microbiome composition reflect statistical covariation rather than causal relationships; future controlled soil amendment experiments with concurrent microbiome profiling would establish directionality. The cross-sectional design, involving twelve purposively selected farms sampled at a single time point, was appropriate for generating baseline ecological data, and the composite sampling approach effectively captured farm-level averages across twenty subsampling points per field. Longitudinal surveys spanning multiple seasons and expanded randomized farm networks would build on this baseline by resolving temporal dynamics and strengthening spatial generalizability. The 16S rRNA amplicon approach provided high-resolution taxonomic profiling, though species-level identification and functional capacity, including confirmed nitrogen fixation activity, would require complementary metagenomics and symbiotic effectiveness assays. Two further limitations should be noted. First, farm-level management history, including fertilizer or manure inputs, tillage intensity, and crop rotation or intercropping, was not systematically recorded at each site in this baseline survey, and unrecorded differences in management likely contribute to the observed farm-level heterogeneity in both soil chemistry and community composition, consistent with evidence that management is a strong driver of soil microbiome assembly (10, 16). Second, key soil physical properties known to shape dryland microbiomes, including soil moisture status, texture, bulk density, aggregate stability, and water-holding capacity, were not measured at the single sampling time point used here (55), and their absence represents the principal scope limitation of this chemical-property baseline relative to a complete edaphic characterization. Systematic recording of field management history and measurement of these physical variables are identified as priorities for the planned longitudinal follow-up survey. Taken together, these considerations define the scope of this baseline study and outline a productive research agenda for microbiome-informed soil fertility management in semiarid cowpea systems.

5 Conclusion

This study provides the first amplicon sequencing-based characterization of soil bacterial microbiome structure in smallholder cowpea farms across semiarid eastern Kenya, revealing substantial farm-level edaphic heterogeneity and associated variation in bacterial community composition and diversity. Actinomycetota dominance and the detection of agriculturally relevant rhizobial genera, particularly Bradyrhizobium, across all sampled farms reflect the ecological characteristics of semiarid dryland soils and highlight the presence of native symbiotic bacterial resources with potential agronomic value. The significant compositional differentiation between Machakos and Kitui counties, accompanied by significantly greater community evenness in Kitui, indicates that both local edaphic conditions and broader spatial factors structure these communities, although a direct statistical test (Mantel test, constrained ordination) did not confirm the eight measured soil variables as significant drivers of this compositional variation at the present sample size, underscoring the need for larger, adequately powered studies to formally establish edaphic drivers. These baseline findings demonstrate that cowpea-growing soils in this region are microbiologically heterogeneous and should not be managed uniformly. Future research should integrate functional metagenomics, seasonal sampling, and controlled inoculation trials using native rhizobial isolates to evaluate the nitrogen-fixing potential of the microbial communities characterized here, advancing the development of site-specific, microbiome-informed approaches for sustainable cowpea production in semiarid Sub-Saharan Africa.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fsoil.2026.1911455/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Viviane Cordovez, Netherlands Institute of Ecology (NIOO-KNAW), Netherlands

Reviewed by: Aura Liz García Serquén, National Institute of Agricultural Innovation (INIA), Peru

Andrej-Nikolai Spiess, Freelance, Germany

Statements

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

The author(s) declared that financial support was received for this work and/or its publication. This work was supported by a Pan African University (PAU) scholarship awarded to Ahmed Kassem Khater. Article publication fees were supported by Google Ireland Limited. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data availability

The original contributions presented in the study are publicly available. This data can be found here: NCBI Sequence Read Archive, accession PRJNA1446837.

Author contributions

AK: Methodology, Investigation, Writing – original draft, Software, Conceptualization, Writing – review & editing, Formal analysis, Visualization, Validation. EN: Conceptualization, Supervision, Project administration, Investigation, Validation, Writing – review & editing, Methodology, Writing – original draft, Resources. SG: Supervision, Methodology, Investigation, Conceptualization, Writing – review & editing, Resources, Validation, Project administration, Writing – original draft.

Acknowledgements

We express our sincere gratitude to Pan African University, Institute of Science, Technology, and Innovation (PAUSTI) for supporting this study. The authors also thank the Department of Agricultural Biochemistry, Faculty of Agriculture, Al-Azhar University, Cairo, Egypt, for its support. Special thanks to laboratory staff from Microbiology and Molecular Biology, Kenyatta University, for their participation, cooperation, time, and efforts during the collection of microbiological and molecular data and analysis. Additionally, the authors wish to thank the farmers from Kitui and Machakos Counties who participated in the fieldwork.

References

  1. 1 Boukar O, Togola A, Chamarthi S, Belko N, Ishikawa H, Suzuki K. Cowpea [Vigna unguiculata (L.) Walp.] breeding. In: Advances in Plant Breeding Strategies: Legumes: Volume 7. Cham: Springer (2019). p. 201–43. doi: doi:10.1007/978-3-030-23400-3_6
  2. 2 Kim DK, Ochar K, Iwar K, Ha BK, Kim SH. Cowpea (Vigna unguiculata L.) production, genetic resources and strategic breeding priorities for sustainable food security: a review. Front Plant Sci. (2025) 16: 1562142. doi: doi:10.3389/fpls.2025.1562142
  3. 3 Kimiti JM, Odee DW. Integrated soil fertility management enhances population and effectiveness of indigenous cowpea rhizobia in semi-arid eastern Kenya. Appl Soil Ecol. (2010) 45: 304–9. doi: doi:10.1016/j.apsoil.2010.05.008
  4. 4 Ngetich FK, Mairura FS, Musafiri CM, Kiboi MN, Shisanya CA. Smallholders' coping strategies in response to climate variability in semi-arid agro-ecozones of Upper Eastern Kenya. Soc Sci Humanities Open. (2022) 6: 100319. doi: doi:10.1016/j.ssaho.2022.100319
  5. 5 Chumo SK, Korir H, Opiyo AM, Mungai NW. “ Land suitability, production and utilization of cowpea in selected counties of Kenya ”, in: Egerton University International Conference. Njoro, Kenya: Egerton University (2022). Available online at: https://conferences.egerton.ac.ke/index.php/euc/article/view/135 (Accessed June 5, 2026).
  6. 6 Kenya Agricultural and Livestock Research Organization (KALRO). Cowpeas. In: Unified Technology Transfer Platform. Nairobi, Kenya: Kenya Agricultural and Livestock Research Organization (KALRO). Available online at: https://agrificsapp.kalro.org/valuechain/12 (Accessed June 5, 2026).
  7. 7 Nkomo GV, Sedibe MM, Mofokeng MA. Production constraints and improvement strategies of cowpea (Vigna unguiculata L. Walp.) genotypes for drought tolerance. Int J Agron. (2021) 2021: 5536417. doi: doi:10.1155/2021/5536417
  8. 8 Pérez-Aguilar LY, López-Osorio RF, Zambrano-Medina YG, Avila-Aceves E, Monjardin-Armenta SA, Plata-Rocha W. Soil degradation: causes, impacts, and mitigation strategies in the context of climate change. In: Climate Change, Land Degradation, and Sustainability: Insight Towards Innovative Solutions. Cham: Springer (2026). p. 3 – 36. doi: doi:10.1007/978-3-032-00704-9_1
  9. 9 Pandey K, Saharan BS. Soil microbiomes: a promising strategy for boosting crop yield and advancing sustainable agriculture. Discover Agric. (2025) 3: 54. doi: doi:10.1007/s44279-025-00208-5
  10. 10 Bier RL, Daniels M, Oviedo-Vargas D, Peipoch M, Price JR, Omondi E. Agricultural soil microbiomes differentiate in soil profiles with fertility source, tillage, and cover crops. Agric Ecosyst Environ. (2024) 368: 109002. doi: doi:10.1016/j.agee.2024.109002
  11. 11 Gao C, de Vries FT, Bezemer TM, Moinet GYK, Sur M, de Kroon H. Dual‐scale drivers of soil biodiversity in agroecosystems: Field management outweighs landscape effects, but both matter. J Appl Ecol. (2026) 63: e70295. doi: doi:10.1111/1365-2664.70295
  12. 12 Zhang C, Nie S, Liang J, Zeng G, Wu H, Hua S. Effects of heavy metals and soil physicochemical properties on wetland soil microbial biomass and bacterial community structure. Sci Total Environ. (2016) 557: 785–90. doi: doi:10.1016/j.scitotenv.2016.01.170
  13. 13 Nasreen S, Arshad R, Parveen K, Mansab S. Soil chemical Properties, nutrient dynamics and fertility management. In: Soils and Sustainable Agriculture: Interplay of Soil, Plant, Water and Environmental Systems for Sustainable Agriculture. Cham: Springer (2025). p. 57 – 77. doi: doi:10.1007/978-3-031-91114-9_4
  14. 14 Lauber CL, Hamady M, Knight R, Fierer N. Pyrosequencing-based assessment of soil pH as a predictor of soil bacterial community structure at the continental scale. Appl Environ Microbiol. (2009) 75: 5111–20. doi: doi:10.1128/aem.00335-09
  15. 15 Zhou X, Tahvanainen T, Malard L, Chen L, Pérez-Pérez J, Berninger F. Global analysis of soil bacterial genera and diversity in response to pH. Soil Biol Biochem. (2024) 198: 109552. doi: doi:10.1016/j.soilbio.2024.109552
  16. 16 Mo Y, Bier R, Li X, Daniels M, Smith A, Yu L. Agricultural practices influence soil microbiome assembly and interactions at different depths identified by machine learning. Commun Biol. (2024) 7: 1349. doi: doi:10.1038/s42003-024-07059-8
  17. 17 Zhang H, Jiang N, Zhang S, Zhu X, Wang H, Xiu W. Soil bacterial community composition is altered more by soil nutrient availability than pH following long-term nutrient addition in a temperate steppe. Front Microbiol. (2024) 15: 1455891. doi: doi:10.3389/fmicb.2024.1455891
  18. 18 Zhao W, Hou R, Liu M, Shen H, Deng X, Wang M. Analysis of soil microbial community structure changes in the drainage field of the Shengli coalfield based on high-throughput sequencing. BMC Microbiol. (2025) 25: 132. doi: doi:10.1186/s12866-025-03761-7
  19. 19 Wei Y, Chen L, Feng Q, Xi H, Zhang C, Gan K. Structure and assembly mechanism of soil bacterial community under different soil salt intensities in arid and semiarid regions. Ecol Indic. (2024) 158: 111631. doi: doi:10.1016/j.ecolind.2024.111631
  20. 20 Luo C, He Y, Chen Y. Rhizosphere microbiome regulation: Unlocking the potential for plant growth. Curr Res Microb Sci. (2025) 8: 100322. doi: doi:10.1016/j.crmicr.2024.100322
  21. 21 Nyaga JW, Njeru EM. Potential of native rhizobia to improve cowpea growth and production in semiarid regions of Kenya. Front Agron. (2020) 2: 606293. doi: doi:10.3389/fagro.2020.606293
  22. 22 Ondieki DK, Nyaboga EN, Wagacha JM, Mwaura FB. Morphological and genetic diversity of Rhizobia nodulating cowpea (Vigna unguiculata L.) from agricultural soils of lower eastern Kenya. Int J Microbiol. (2017) 2017: 8684921. doi: doi:10.1155/2017/8684921
  23. 23 Muindi MM, Muthini M, Njeru EM, Maingi J. Symbiotic efficiency and genetic characterization of rhizobia and non rhizobial endophytes associated with cowpea grown in semi-arid tropics of Kenya. Heliyon. (2021) 7: e06867. doi: doi:10.1016/j.heliyon.2021.e06867
  24. 24 Pinto Y, Bhatt AS. Sequencing-based analysis of microbiomes. Nat Rev Genet. (2024) 25: 829–45. doi: doi:10.1038/s41576-024-00746-6
  25. 25 Khater AK, Kirui CK, Njeru EM, Githiri SM. (2026). 16S rRNA-based genetic diversity and symbiotic efficiency of indigenous cowpea-nodulating rhizobia from semiarid Eastern Kenya. Front. Microbiol. 17: 1875429. doi: doi:10.3389/fmicb.2026.1875429
  26. 26 Yageta Y, Osbahr H, Morimoto Y, Clark J. Comparing farmers' qualitative evaluation of soil fertility with quantitative soil fertility indicators in Kitui County, Kenya. Geoderma. (2019) 344: 153–63. doi: doi:10.1016/j.geoderma.2019.01.019
  27. 27 Tidimalo C, Maximiliano O, Karen J, Lebre PH, Bernard O, Michelle G. Microbial diversity in the arid and semi‐arid soils of Botswana. Environ Microbiol Rep. (2024) 16: e70044. doi: doi:10.1111/1758-2229.70044
  28. 28 Muzami EM, Kitundu G, Muriithi OM, Kavoo AM, Gichuru VG, Mbinda WM. Metagenomic insights to bacterial communities, functional traits, and soil health in banana smallholder agroecosystems of Kenya. Front Microbiol. (2025) 16: 1582271. doi: doi:10.3389/fmicb.2025.1582271
  29. 29 Carter MR, Gregorich EG. Soil Sampling and Methods of Analysis. Boca Raton, FL: CRC Press (2007). doi: doi:10.1201/9781420005271
  30. 30 Klindworth A, Pruesse E, Schweer T, Peplies J, Quast C, Horn M. Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next-generation sequencing-based diversity studies. Nucleic Acids Res. (2013) 41: e1–1. doi: doi:10.1093/nar/gks808
  31. 31 Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. DADA2: High-resolution sample inference from Illumina amplicon data. Nat Methods. (2016) 13: 581–3. doi: doi:10.1038/nmeth.3869
  32. 32 Callahan BJ, McMurdie PJ, Holmes SP. Exact sequence variants should replace operational taxonomic units in marker-gene data analysis. ISME J. (2017) 11: 2639–43. doi: doi:10.1038/ismej.2017.119
  33. 33 Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. (2012) 41: D590–6. doi: doi:10.1093/nar/gks1219
  34. 34 McMurdie PJ, Holmes S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PloS One. (2013) 8: e61217. doi: doi:10.1371/journal.pone.0061217
  35. 35 Bray JR, Curtis JT. An ordination of the upland forest communities of southern Wisconsin. Ecol Monogr. (1957) 27: 326–49. doi: doi:10.2307/1942268
  36. 36 Anderson MJ. A new method for non‐parametric multivariate analysis of variance. Austral Ecol. (2001) 26: 32 – 46. doi: doi:10.1111/j.1442-9993.2001.01070.pp.x
  37. 37 Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Society: Ser B. (1995) 57: 289 – 300. doi: doi:10.1111/j.2517-6161.1995.tb02031.x
  38. 38 Mantel N. The detection of disease clustering and a generalized regression approach. Cancer Res. (1967) 27: 209–20.
  39. 39 Fierer N, Jackson RB. The diversity and biogeography of soil bacterial communities. Proc Natl Acad Sci. (2006) 103: 626–31. doi: doi:10.1073/pnas.0507535103
  40. 40 Bahram M, Hildebrand F, Forslund SK, Anderson JL, Soudzilovskaia NA, Bodegom PM. Structure and function of the global topsoil microbiome. Nature. (2018) 560: 233–7. doi: doi:10.1038/s41586-018-0386-6
  41. 41 Fierer N. Embracing the unknown: disentangling the complexities of the soil microbiome. Nat Rev Microbiol. (2017) 15: 579–90. doi: doi:10.1038/nrmicro.2017.87
  42. 42 Delgado-Baquerizo M, Oliverio AM, Brewer TE, Benavent-González A, Eldridge DJ, Bardgett RD. A global atlas of the dominant bacteria found in soil. Science. (2018) 359: 320–5. doi: doi:10.1126/science.aap9516
  43. 43 Barnard RL, Osborne CA, Firestone MK. Responses of soil bacterial and fungal communities to extreme desiccation and rewetting. ISME J. (2013) 7: 2229–41. doi: doi:10.1038/ismej.2013.104
  44. 44 Ebrahimi-Zarandi M, Etesami H, Glick BR. Fostering plant resilience to drought with Actinobacteria: Unveiling perennial allies in drought stress tolerance. Plant Stress. (2023) 10: 100242. doi: doi:10.1016/j.stress.2023.100242
  45. 45 Fierer N, Bradford MA, Jackson RB. Toward an ecological classification of soil bacteria. Ecology. (2007) 88: 1354–64. doi: doi:10.1890/05-1839
  46. 46 Hartmann M, Frey B, Mayer J, Mäder P, Widmer F. Distinct soil microbial diversity under long-term organic and conventional farming. ISME J. (2015) 9: 1177–94. doi: doi:10.1038/ismej.2014.210
  47. 47 Peralta AL, Sun Y, McDaniel MD, Lennon JT. Crop rotational diversity increases disease suppressive capacity of soil microbiomes. Ecosphere. (2018) 9: e02235. doi: doi:10.1002/ecs2.2235
  48. 48 Zahran HH. Rhizobium-legume symbiosis and nitrogen fixation under severe conditions and in an arid climate. Microbiol Mol Biol Rev. (1999) 63: 968–89. doi: doi:10.1128/mmbr.63.4.968-989.1999
  49. 49 Broughton WJ, Jabbouri SD, Perret X. Keys to symbiotic harmony. J Bacteriol. (2000) 182: 5641–52. doi: doi:10.1128/jb.182.20.5641-5652.2000
  50. 50 Rousk J, Bååth E, Brookes PC, Lauber CL, Lozupone C, Caporaso JG. Soil bacterial and fungal communities across a pH gradient in an arable soil. ISME J. (2010) 4: 1340–51. doi: doi:10.1038/ismej.2010.58
  51. 51 Xiong R, He X, Gao N, Li Q, Qiu Z, Hou Y. Soil pH amendment alters the abundance, diversity, and composition of microbial communities in two contrasting agricultural soils. Microbiol Spectr. (2024) 12: e04165-23. doi: doi:10.1128/spectrum.04165-23
  52. 52 Van Der Heijden MG, Bardgett RD, Van Straalen NM. The unseen majority: soil microbes as drivers of plant diversity and productivity in terrestrial ecosystems. Ecol Lett. (2008) 11: 296 – 310. doi: doi:10.1111/j.1461-0248.2008.01199.x
  53. 53 Wagg C, Bender SF, Widmer F, van der Heijden MGA. Soil biodiversity and soil community composition determine ecosystem multifunctionality. Proc Natl Acad Sci. (2014) 111: 5266–70. doi: doi:10.1073/pnas.1320054111
  54. 54 Mendes R, Kruijt M, de Bruijn I, Dekkers E, van der Voort M, Schneider JHM. Deciphering the rhizosphere microbiome for disease-suppressive bacteria. Science. (2011) 332: 1097–100. doi: doi:10.1126/science.1203980
  55. 55 Lüneberg K, Schneider D, Siebe C, Daniel R. Drylands soil bacterial community is affected by land use change and different irrigation practices in the Mezquital Valley, Mexico. Sci Rep. (2018) 8: 1413. doi: doi:10.1038/s41598-018-19743-x

Citation

Khater, A. K., et al. (2026). Edaphic heterogeneity shapes bacterial microbiome composition and rhizobial diversity in smallholder cowpea farms of semiarid eastern Kenya. Frontiers in Soil Science, Volume 6. https://doi.org/
Received
Accepted
Published
Volume
6 - 2026

Copyright

Copyright © 2026 Khater, Njeru and Githiri.

License: https://creativecommons.org/licenses/by/4.0/

*Correspondence: Ahmed Kassem Khater

Article metrics

Views
0
Downloads
0
Citations
0
Shares
0

Related articles

Original ResearchSiderophores and siderophore-associated metabolites from Antarctic Pseudomonas sp. ANT_H12B contribute to iron and phosphorus mobilization in soils with contrasting propertiesXiaomeng Chen, Marcin Musiałowski, Klaudia Dębiec-Andrzejewska, Robert Stasiuk, Luisella Celi, Michela Schiavon, Maria Martin, Silvia CellettiPerspectiveSoil contaminants at the environment-health interface require function-centered risk assessmentSylvester Chibueze Izah, Ebinyo Rebecca Aseibai, Vivian Ibienebakabobo Promise, Matthew Chidozie OgwuOriginal ResearchSoil organic carbon storage across Mediterranean mountain forests: a comparative assessment of forest ecosystem diversity and geological substrate in Ifrane and Talassemtane National Parks, MoroccoMohamed El Mderssa, Meysara ELMALKI, Fouad Malki, Taha Abdelhakim, Younes Abbas, Jamal Ibijbijen