Three sampling locations were selected within a 50 km radius in the Nama Karoo, South Africa, during August 2023: Location 1 [L1; 29°21’53.7”S, 18°28’01.8”E], Location 2 [L2; 28°27’49.7”S, 20°17’02.3”E], and Location 3 [L3; 28°38’51.4”S, 20°26’48.5”E]. These regions have a mean annual precipitation between 127 mm and 254 mm with temperatures ranging from 2 °C to 40 °C (Figure 1). A total of 96 bulk soil samples were collected. Briefly, at each location, random plots of 1 m × 1 m were identified and three soil subsamples of approximately 150 g each were taken randomly within the first 10 cm of the soil surface. All samples were collected away from plants to avoid roots. Where present, the top layer of organic material was removed before soil collection (Keet et al., 2019). The three soil subsamples were combined and homogenized to form a composite sample.

Soil moisture, electrical conductivity (EC), and pH were measured as outlined by Anderson and Ingram (1994). Soil pH was determined in a soil slurry in ddH₂O (1:5) solution using an Ohaus Starter 2100 bench pH meter (Ohaus, Switzerland), and EC was measured using an Aqualytic AL15 electrochemistry meter (Lab Unlimited, Ireland). Soil moisture was calculated by weighing 10 g of soil into a sterile falcon tube, desiccating at 37 °C for 7 days, and recording post-incubation weight loss per gram of dry soil. All absorbance readings of downstream tests were read on an iMark Microplate Absorbance Reader (Bio-Rad, USA) and 0.1 g of soil was used unless stated otherwise.

To determine nitrogen mineralization ammonium concentrations in soil were measured at day 0 and day 7. All required reagents were prepared following the protocol outlined by Anderson and Ingram (1994). Absorbance at 655 nm was recorded for day 0 samples, and net nitrogen mineralization was calculated based on the change in ammonium concentration after 7 days of incubation at 37 °C. Nitrate levels were measured using a salicylic acid colorimetric method. Soil extracts were mixed with 5% salicylic acid in sulfuric acid and left for 30 minutes. Sodium hydroxide (4 M) was then added, and the samples were incubated for 1 hour to allow full-color development and absorbance was measured at 410 nm. Available phosphorus was determined with Bray II solution and organic carbon with Walkley-Black method (Bray and Kurtz, 1945; Walkley, 1947; Anderson and Ingram, 1994).

Active carbon, a labile fraction of soil organic carbon, was measured using the potassium permanganate (KMnO₄) oxidation method (Weil et al., 2003). Soil samples were treated with 0.2 M KMnO₄ in 1 M calcium chloride (CaCl₂) solution, and the reaction was conducted in the dark for 10 minutes to oxidize easily oxidizable organic matter. After settling, 20 µL of the supernatant was transferred to a new tube with 1.98 mL ddH₂O water. The absorbance of the diluted solution was measured at 550 nm.

Soil enzyme activities measured included: β-glucosidase, urease, and alkaline-acid phosphatase (Dick et al., 1997). Assays were performed with a total of 0.1 g of soil, in triplicate. For β-glucosidase activity, p-nitrophenyl-β-D-glucopyranoside (PNG) was used as the substrate. Samples were incubated with the substrate and buffer solution (THAM 12.1 g/L, Maleic acid 11.6 g/L, Boric acid 6.3 g/L, Citric acid 14.0 g/L, NaOH 19.52 g) at pH 6.0 at 37 °C for 1 hour. After incubation, 0.5 M CaCl₂ and THAM buffer (pH 12) were added, and absorbance was measured at 410 nm. Urease activity was determined by incubating soil samples in a urea solution (720 mM) with borate buffer (pH 10) at 37 °C for 2 hours. After incubation, a KCl–HCl solution was added, and the mixture was incubated for an additional 30 minutes at room temperature and absorbance measured (660 nm). Both acid and alkaline phosphatase activities were measured using p-nitrophenyl phosphate as the substrate at pH 6.5 and 11, respectively. Samples were incubated at 37 °C for 1 hour, and p-nitrophenyl was extracted using 0.5 M CaCl₂ and 0.5 M NaOH, with absorbance measured at 410 nm.

The fluorescein diacetate (FDA) hydrolysis assay, was used as a proxy of total microbial activity, following the method described by Green et al. (2006). A total of 0.5 g of soil was combined with 12.5 mL of 1× PBS buffer (pH 7.4) and 0.25 mL of 4.9 mM FDA dissolved in acetone and incubated at 43 °C for 2 hours under constant agitation. After incubation, FDA hydrolysis was halted by adding 40 μL of acetone to 1 mL of soil slurry. Samples were then centrifuged at 8800 g for 5 minutes, and fluorescence was measured (490 nm).

DNA of soil samples were extracted using the ZR Quick-DNA Fecal/Soil Microbe Kits (Zymo Research, USA) according to the manufacturer’s instructions with the following modifications to increase final DNA concentration; 0.1% (W/V) of Polyvinylpyrrolidone (PVP) was added to genomic lysis buffer and DNA Elution Buffer was heated at 65 °C for 5 minutes on a heating block (Omega Scientific, USA) (Gray and Herwig, 1996). PCR amplification of successfully extracted genomic DNA was performed with primers specific to the V1-V9 region of the 16S rRNA gene (Dos Santos et al., 2019). The total reaction volume (25 μL) contained 5 µL of 1x KAPA Taq Hotstart Buffer, 2 µL of 1.5 mM MgCl₂, 0.5 μL dNTP mix, 075 μM forward [27F (5’-ACTCCTACGGGAGGCAGCAG-3’)] and reverse primers [1492R (5’-GGTTACCTTGTTACGAGTT-3’)], 0.2 µL of 0.5 U KAPA Taq HotStart DNA Polymerase, and 2 μL of template DNA (Dos Santos et al., 2019). Products were amplified using the 2720 Thermal Cycler (Applied Biosystems, USA) with following settings: Initial denaturing at 95 °C for 5 minutes, followed by 35 cycles of 95 °C for 30 seconds, 58 °C for 30 seconds and 72 °C for 1 minute. A final extension was completed at 72 °C for 1 minute and the PCR samples were held at 4 °C. Successful amplification was visualized on a 1% agarose gel stained with ethidium bromide and final DNA concentration determined with BioDrop µLite (BioDrop, UK).

Following PCR amplification, DNA barcoding and library preparation were performed using the Native Barcoding Kit 96 V14 (SQK-NBD114.96, Oxford Nanopore Technologies (ONT), UK) according to the manufacturer’s protocol. For end-repair and dA-tailing, the NEBNext Ultra II module (ONT, UK) was employed with 11.5 µL of amplified DNA at a final concentration of 200 fmol. The reaction was incubated at 20 °C for 5 minutes, followed by 65 °C for 5 minutes, using a 2720 Thermal Cycler (Applied Biosystems, USA). Each sample was assigned a distinct barcode and subsequently pooled. Adapter ligation and clean-up were performed using the NEBNext Quick Ligation Module (ONT, UK) as per the manufacturer’s guidelines. DNA fragments within the 1500–1900 bp range were selected using the long fragment buffer. The final library concentration was quantified using a BioDrop µLite spectrophotometer (BioDrop, UK) and adjusted to 12 µL (50 fmol) for sequencing. The prepared sequencing library, along with the Flow Cell Priming Kit (ONT, UK), was loaded onto an R10.4.1 flow cell (ONT, UK). Sequencing was conducted on the MinION Mk1B platform using MinKNOW software (Version 24.02.6). Basecalling was performed with Dorado (Version 7.2.0) using the super-accurate basecalling model, with barcode trimming enabled at both ends and in the middle.

The quality of the sequencing reads was assessed and visualized using Nanoplot (Version 1.42.0; de Coster et al., 2018). Following quality assessment, the data was quality filtered using MOTHUR (Version 1.48.0; Schloss et al., 2009; Winand et al., 2019; Akaçin et al., 2023). Initially, quality filtering was applied to remove primers, sequences with ambiguous bases, homopolymers longer than 8 bp, and those outside the correct size range (1000 bp to 1900 bp) or with a quality score below 25. Chimeric sequences were identified and removed using VSEARCH (Version 2.16.0; Rognes et al., 2016). Following filtering and trimming, the remaining sequences were clustered into OTUs and classified against two reference databases: SILVA Version 132 (http://www.arb-silva.de/) and EzBioCloud (https://www.ezbiocloud.net/) with a cut-off value of 80 (Quast et al., 2013; Yoon et al., 2017; Chalita et al., 2024). Sequences classified as unknown, eukaryotes, mitochondria, or chloroplasts were removed before further analysis. The raw sequence data was submitted to GENBANK as FASTQ files under accession number PRJNA1207931.

All soil abiotic properties and enzyme activity results with corresponding significant p-values are summarized in Table 1. The table also highlights where significant differences were observed between the sampling locations. A significant difference was observed for EC, total nitrate, and available phosphate. Enzyme activities that displayed significance included acid phosphatases and urease. The significant difference for EC occurred between L 1 (65.38 ± 10.93) and both L 2 (47.49 ± 5.28) and L 3 (48.18 ± 23.27). The same pattern was identified for available phosphate (Table 1). The significant difference observed for acid phosphatase, urease and total nitrate occurred between L1 and L2 and L2 and L3 (Table 1). No significant correlation was observed for pH. Increased acid and alkaline phosphatases occurred at all three locations, with low available phosphate. Microbial activity was the highest at L1, followed by L3 and L2, respectively.

In the Nama Karoo, as well as in secondary data analyses, SDs 4, 3, and 6 collectively dominated the Acidobacteriota community. Subdivision 4 was the most abundant and consistently stable, reflecting its ability to thrive in arid environments characterized by low moisture, slightly alkaline pH, and nutrient limitation (Wüst et al., 2016; Ivanova et al., 2020). Cultivated members of SD 4, such as Aridibacter, Blastocatella, and Pyrinomonas, demonstrate metabolic strategies that support survival under these conditions, including oligotrophic growth, utilization of diverse carbon substrates, and resistance to desiccation, collectively enabling carbon cycling, polysaccharide degradation, and soil stabilization (Huber et al., 2014; Crowe et al., 2014; Ivanova et al., 2020; Huber et al., 2017).

Similarly, SD 3 and SD 6 showed clear associations with soil pH and nutrient availability. SD 3 members are adapted to mildly acidic soils and appear capable of exploiting available nitrate and organic carbon, suggesting metabolic versatility that supports persistence in oligotrophic, arid soils (Huber and Overmann, 2018). Subdivision 6, including representatives such as Vicinamibacter silvestris, tolerates a broad pH range and utilizes diverse carbon sources, likely explaining its occurrence in arid soils with variable pH and low nutrient content (Mukherjee et al., 2014; Kielak et al., 2016; Kalam et al., 2020). The consistent clustering of these SDs further indicates shared ecological preferences, metabolic traits, and niche adaptations that underpin their dominance in arid landscapes.

Rare SDs exhibited narrower ecological niches and strong location-specific dependencies. In this study, SD 9 was exclusively present at Nama Karoo L3, highlighting its restricted habitat preference. Notably, SD 17 (Class Vicinamibacteria) and SD 25 consistently clustered together across all three locations, suggesting shared ecological adaptability. Consistent with these findings, Vicinamibacteria and SD 25 were among the most abundant classes in a previous study by Dedysh et al., 2022. Their research showed that fens were dominated by Vicinamibacteria, particularly members of the order Vicinamibacteriales, which comprised 23.1% to 58.9% of all Acidobacteriota sequences, alongside a significant presence of uncultured SD 17. This consistent clustering pattern further suggests functional similarities and ecological overlap between SD 17 and SD 25, reinforcing their adaptability across different environments. Additionally, SD 2 was exclusively found at L3, further emphasizing its potential niche specialization and highlighting the role of location-specific factors in shaping the distribution of rare SDs.

Subdivision 3 members are commonly found in mildly acidic soils, consistent with their physiological tolerance for a broader pH range (Jones et al., 2009; Ivanova et al., 2020). The observed negative correlation with acid phosphatase activity suggests that SD 3 may not rely on phosphatase-mediated phosphorus acquisition. Instead, these organisms may obtain phosphorus indirectly through microbial interactions, such as cross-feeding, where other microbes release phosphate as a metabolic by-product that SD 3 can utilize (Deng et al., 2022; Jara-Servin et al., 2023). Additionally, the mildly acidic soils preferred by SD 3 may naturally contain more bioavailable phosphorus, further reducing the necessity for acid phosphatases (Kielak et al., 2016; Kalam et al., 2020). In contrast, soils with high acid phosphatase activity may indicate phosphorus limitation, favoring microbes capable of producing these enzymes, whereas SD 3 thrives in less phosphorus-limited conditions.

The positive correlation of SD 3 with nitrate suggests that some members may exploit nitrate as a nitrogen source, consistent with known capabilities of taxa such as Bryobacter aggregatus (Kulichevskaya et al., 2010). Taken together, these patterns indicate that SD 3’s distribution is shaped by a combination of metabolic traits, nutrient acquisition strategies, and interactions with co-occurring microbes, providing a mechanistic explanation for its ecological niche in mildly acidic, moderately nutrient-rich soils.

Subdivision 5 is one of the lesser-known Acidobacteriota subdivisions, with limited information available due to the absence of described cultivated members. To our knowledge, this study represents the first report of SD 5 in arid regions of South Africa. Microbial communities in arid environments commonly employ survival strategies such as dormancy, slow growth, or biofilm formation to persist during prolonged dry periods (Huber et al., 2022a). The observed negative correlation between moisture and SD 5 suggests an adaptation to drier conditions, where reduced water availability may limit competition for resources and favor organisms capable of maintaining metabolic activity under stress (Huber et al., 2022b). These organisms may therefore persist or increase in relative abundance during dry periods when moisture-dependent taxa are less active, potentially relying on low-energy, oligotrophic metabolic strategies to maintain viability under nutrient- and water-limited conditions (Flieder et al., 2021; Kristensen et al., 2021; Cao et al., 2022; Huber et al., 2022a). The negative correlation between SD 5 and both nitrate concentrations and urease activity further supports a mechanistic interpretation based on metabolic efficiency. Genomic and physiological studies indicate that not all Acidobacteriota subdivisions utilize nitrate as an electron acceptor, instead relying on low-energy, oligotrophic metabolic strategies that minimize energetic costs under nutrient-limited conditions (Flieder et al., 2021; Kristensen et al., 2021; Cao et al., 2022; Kato et al., 2022). Such energy-efficient metabolic configurations are advantageous in arid soils, where nitrogen availability is limited, spatially heterogeneous, and temporally variable. The association of SD 5 with low-nitrogen environments is further supported by its negative correlation with high urease activity, which often reflects intensified nitrogen mineralization in nitrogen-limited soils (Roscoe et al., 2000).

Subdivision 1 displayed a significant negative correlation with EC, indicating a preference for soils with lower salinity. Although studies directly linking SD 1 abundance to soil EC are limited, physiological evidence from cultivated representatives provides mechanistic insight into this pattern. Several described members of SD 1, including Silvibacterium dinghuense, Acidobacterium ailaaui, Acidicapsa ligni, and Silvibacterium bohemicum, exhibit limited tolerance to NaCl concentrations above 0–2.5%, reflecting sensitivity to osmotic and ionic stress (Kulichevskaya et al., 2012; Lladó et al., 2016; Myers and King, 2016; Zhang et al., 2022). This physiological constraint suggests that SD 1 lacks specialized salt-adaptation mechanisms, such as efficient osmolyte accumulation or ion extrusion systems, thereby favoring its persistence in low-salinity arid soils rather than saline environments.

Subdivision 2 exhibited a negative correlation with high alkaline phosphatase activity, suggesting that SD 2 may not directly contribute to phosphatase production. This pattern implies that other organisms are primarily responsible for phosphorus mineralization in these soils. Alternatively, SD 2 could participate indirectly in phosphorus cycling through mutualistic or commensal interactions with microbes that produce or utilize alkaline phosphatase, allowing it to access released phosphate without expending energy on enzyme production (Bakki et al., 2024). Its lower abundance in soils with high alkaline phosphatase activity may indicate competitive exclusion, where SD 2 is less adapted to environments dominated by fast-growing, phosphatase-producing taxa. Overall, these patterns suggest that SD 2 occupies a niche strategy focused on indirect nutrient acquisition, highlighting the importance of microbial interactions in shaping its ecological distribution.

The only described member from SD 22 has been isolated from marine sediments where its role in sulphur cycling was reported (Flieder et al., 2021). Therefore, the detection of this group in these soil environments was surprising. However other members of this phylum have been identified in both aquatic and terrestrial environments (Kishimoto et al., 1991; Barns et al., 1999).

Earlier research has consistently shown that both SDs 1 and 3 thrive in slightly acidic environments. The negative correlation observed in the secondary data further supports this pattern, reinforcing the association between SD 1 and SD 3 abundance and lower soil pH (Jones et al., 2009; Lauber et al., 2009; Kielak et al., 2016; Dedysh et al., 2017; Dedysh and Sinninghe Damsté, 2018; Ivanova et al., 2020; Kalam et al., 2020; Conradie and Jacobs, 2021). Their positive correlation with total carbon aligns with the metabolic versatility reported in related Acidobacteriota, which possess the genomic potential to degrade a range of organic compounds, including complex carbon substrates (Kielak et al., 2016; Kalam et al., 2020). This suggests that SDs 1 and 3 could contribute to soil carbon cycling by utilizing both readily available and more recalcitrant carbon sources. The consistent clustering of SDs 1 and 3 indicates shared ecological preferences, potential overlap in metabolic traits, and similar niche adaptations. While direct experimental data on SD 1 and SD 3 remain limited, these patterns collectively suggest that their distribution is influenced by pH tolerance, carbon utilization strategies inferred from related taxa, and niche complementarity, highlighting directions for future functional research.

Subdivision 4 is frequently associated with arid lands as previously mentioned, where nutrient levels, including phosphorus, are often low (Huber et al., 2014). Its ability to survive in higher pH could be explained, as currently described members of this SD can tolerate a range from 4 to 10, with optimum at between 7 and 9 (Huber et al., 2014). The negative correlation with both P and C % highlights its ability to compete effectively in oligotrophic environments with limited phosphorus resources.

The first cultured representative of SD 6, Vicinamibacter silvestris, was found to grow within a pH range of 4.7 to 9.0, demonstrating adaptability to diverse soil pH conditions (Mukherjee et al., 2014). This broad pH tolerance likely explains its presence in arid soils, where pH can vary due to low organic matter content and mineral heterogeneity (Maestre et al., 2015; Tidimalo et al., 2024). SD 6 also displays a positive correlation with both active and total organic carbon, suggesting a capacity to utilize a range of carbon substrates, which would further support its survival in nutrient-limited arid environments (Kielak et al., 2016; Kalam et al., 2020). While research on SD 6 and soil phosphorus interactions is limited, its ability to tolerate variable pH and utilize diverse carbon sources likely underpins its ecological success in arid regions. Further studies are required to experimentally confirm these functional traits.

In contrast, SD 8 showed no significant correlations with any abiotic factors. This SD may exhibit broad tolerance or versatility, allowing it to adapt to a wide range of environmental conditions. This adaptability suggests that its abundance and distribution are not strongly influenced by specific abiotic factors such as pH, nutrients, or moisture. Instead, SD 8 might rely more on biotic interactions, including relationships with other microorganisms to shape its ecological role and distribution. Additionally, SD 8 may perform generalist roles in soil ecosystems, enabling it to function effectively across diverse environments without dependence on specific abiotic conditions (von Meijenfeldt et al., 2023). This functional redundancy could explain its even distribution and lack of strong associations with abiotic factors.

A large portion of the Acidobacteriota found in secondary data and the current study remained unknown due to challenges in cultivation, slow growth, and the limited resolution provided by culture-independent techniques. Advances in genomic tools and improved culturing methods may help uncover more information about these elusive bacteria in the future. This observation is not a new phenomenon as other studies have reported the same limitation (Conradie and Jacobs, 2021; Coluccia and Besaury, 2023).

It is important to note that in this study, Oxford Nanopore Technologies was used in this study, offering significant advantages due to its ability to generate longer read lengths and capture greater variability by targeting the entire V1–V9 region of the 16S rRNA gene. This approach enabled the detection of a broader spectrum of bacterial diversity, including elusive groups such as SD 9, which were previously underrepresented. Compared to earlier sequencing methods that focused on shorter regions, ONT sequencing provided superior resolution, allowing differentiation between closely related SDs. The improved resolution is evident in the detection of SDs in this study compared to secondary data, where shorter read lengths likely led to the underrepresentation of certain groups. Numerous studies have also demonstrated ONT’s ability to enhance microbial diversity detection in both mock communities and environmental samples (Szoboszlay et al., 2023; Waechter et al., 2023), reinforcing its value in microbial ecology research.