The soil microbiomics of intact, degraded and partially-restored semi-arid succulent thicket (Albany Subtropical Thicket)

DNA Extraction and Sequencing

Metagenomic DNA (i.e., DNA found in an environmental sample) was extracted from 0.5 g soil samples using the QIAGEN DNeasy PowerSoil kit (Qiagen, Venlo, Netherlands). The protocol was modified to include an additional step of soil agitation by two 40 s cycles of 2,500 rpm in a Powerlyzer 24 (Qiagen, Venlo, Netherlands). We evaluated the quantity and quality of the DNA with a Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and by amplifying the bacterial 16S rRNA V3–V4 region (E9F and U1510R primers) using OneTaq^® Hot Start DNA Polymerase (New England Biolabs, Ipswich, MA, USA). Metagenomic DNA was submitted to Omega Bioservices (Norcross, Georgia, United States) for sequencing of the V3–V4 region of the 16S rRNA gene (Forward primer: 5′-CCTACGGGNGGCWGCAG-3′, reverse primer: 5′-GACTACHVGGGTATCTAATCC-3′ (Klindworth et al., 2013)) on an Illumina MiSeq v3 with paired-end 300 bp reads. Each sample was sequenced twice.

Biodiversity and microbial composition differences between the intact and degraded zones

Alpha-diversity analysis of the 26,759 observed ASVs revealed no significant differences in biodiversity between intact and degraded zones (Fig. S5). This was unexpected and contradicts a recent study that found higher levels of diversity in arid soils with plant cover compared to those without (Kushwaha et al., 2021). However, previous studies have disagreed whether herbivory increases (Eldridge et al., 2017) or decreases (Cheng et al., 2016) bacterial diversity and the link between browsing and soil microbial diversity is probably very complex and may depend on both browsing intensity and the plant community species diversity and composition (Qu et al., 2016). Despite the similarities in biodiversity summary statistics, microbial communities from the different zones formed distinct clusters, as indicated by the beta-diversity distances between the samples (Fig. 2B and Fig. S6). Clustering of the microbial populations for the three zones (intact, degraded and restored) captured 15.13% of the variation in the samples, with restored samples located in an intermediate position between the intact and degraded samples. Together these results suggest that vegetation loss has an impact on the community structure of the succulent thicket soil microbiome, but not on its overall biodiversity.

The core microbial community (defined as ASVs present in at least 95% (i.e., ≥33/35) of samples: Table S3) only accounted for 103 genus-level ASVs (9.87%) but comprised 70.29% of the sequence reads. If the threshold were raised to 100% (i.e., 35/35 sites), then 60 genus-level ASVs, comprising 5.75% of the total genus-level ASVs and 54.81% of the reads, would be detected but if it were lowered to 89% (i.e., 31/35 sites) then 142 genus-level ASVs comprising 13.60% of the total genus-level ASVs and 77.84% of the reads, would be detected. Each of the three sampled zones also had their own unique core communities which showed a level of similar dominance across the reads (Tables S4–S6); the core community accounted for 118 ASVs (13.58% of genus-level ASVs, 74.95% of reads) in the intact zone, 166 ASVs (22.10% of genus-level ASVs, 88.43% of reads) in the degraded zone and 207 ASVs (30.49% of genus-level ASVs, 86.40% of reads) in the restored zone. If the threshold were raised to 100% (i.e., 15/15 or 5/5 sites) then the core community would consist of 78 ASVs (8.98% of genus-level ASVs and 62.75% of reads) in the intact zone and 121 ASVs (16.11% of genus-level ASVs and 80.85% of reads) in the degraded zone while the restored zone would be the same as at the standard 95% threshold. If the threshold were lowered to 89% (i.e., 13/15 or 4/5 sites) then the core community would consist of 155 ASVs (17.84% of genus-level ASVs and 81.68% of reads) in the intact zone, 189 ASVs (25.16% of genus-level ASVs and 90.77% of reads) in the degraded zone and 295 ASVs (43.45% of genus-level ASVs and 93.74% of reads) in the restored zone. The dominance of a relatively small number of taxa is a well-known phenomenon in soils (Delgado-Baquerizo et al., 2018). The composite microbial community in all experimental samples comprised 36 prokaryotic phyla, the most abundant of which (based on ASV assignments of 16S rRNA gene amplicon reads) were Actinobacteriota (28.76%), Proteobacteria (21.39%), Acidobacteriota (11.40%), Plantomycetes (9.84%) and Bacteroidota (7.71%) (Fig. 3A).

Open in a separate window

Figure 3

Relative abundance of bacterial taxa.

Relative abundance of (A) bacterial phyla and (B) genera. Taxa which have a relative abundance less than 1% are combined in the NA category in (A) and are completely removed in (B) for ease of viewing. The taxa displayed in (B) account for approximately 25% of the total abundance in the samples.

Significant differences were observed at the genus level between the intact and degraded zones (Fig. 3B and Fig. S7). The largest differences were observed for Rubrobacter (1.72% intact vs 6.16% degraded), Conexibacter (3.57% intact vs 0.14% degraded), RB41 (2.22% intact vs 5.02% degraded), Bryobacter (4.45% intact vs 1.82% degraded) and Mycobacterium (2.69% intact vs 0.53% degraded). ASVs with a relative abundance of less than 1% accounted for between 66.05% and 69.94% of all reads in the various zones. Soil samples from the restored zone showed a larger number of taxa classified as ‘uncultured’ than either the intact or degraded zone samples (4.06% restored vs 1.50% intact and 0.29% degraded).

In addition, LEfSe analysis identified four biomarker taxa for intact zone soils and five for degraded zone soils (Fig. 4). The intact biomarker taxa were derived from the families Acidobacteriaceae (Subgroup 1) and Myxococcaceae, the order Frankiales and the class Verrucomicrobiae. The order Frankiales is one of the most abundant in the dataset and includes many root-nodule associated taxa (Pawlowski & Demchenko, 2012; Battenberg et al., 2017); its over-representation in intact sites suggests that the changes in the soil microbiome may be due to the reduction or disappearance of plant-associated taxa with the loss of vegetation. By comparison, biomarker taxa for degraded zone soils were the genera Ensifer and Exiguobacterium, members of which are found in diverse environments (Kasana & Pandey, 2018), the cyanobacterial family Coleofasiculaceae, the order Puniceispiralles and the Chloroflexi class Anaerolineae, commonly found in anaerobic digesters (Xia et al., 2016).

Open in a separate window

Figure 4

Taxa identified as potential biomarkers.

Taxa identified as potential biomarkers according to LEfSe analysis. Only the lowest taxa in the hierarchy is displayed for a particular condition. Taxa were identified as biomarkers for (A–E) degraded vegetation, (F–I) intact vegetation or (J–P) restored vegetation. Colours represent the different zones; orange = degraded, green = intact, blue = restored.

Abiotic drivers of microbial community structure in both intact and degraded zones

In order to determine which soil physicochemical properties were potentially important for microbial community structure, we used RDA (Fig. 5). At the phylum level, the ratio of Ca to Mg and the soil density explained 47.78% of the variation in the microbial community (Fig. 5A) and, at the genus level, the ratio of Ca to Mg, the soil density and the soil pH were able to explain 50.10% of the variation in the community structure (Fig. 5B, a third axis is not plotted). pH has frequently been identified as a major driver of bacterial community composition in soils (Rousk et al., 2010; Qu et al., 2016). Contrary to one of the initial expectations of this study, differences in soil relative humidity and temperature resulting from the loss of vegetation did not appear to significant affect microbial community structure in the intact and degraded zones. However, the interplay of relative humidity and temperature may affect the water balance of the soil, which could potentially be responsible for the shift in pH (Slessarev et al., 2016).

Open in a separate window

Figure 5

Redundancy analysis and associated soil physicochemistry.

On the top row are the RDA results when assessing the communities at the (A) phylum and (B) genus level. Only a limited number of taxa names are displayed to prevent label overlap. The bottom row shows box-and-whisker plots of the variables which were determined to play a significant role in the RDA; (C) pH, (D) ratio of Ca to Mg and (E) density. Significance was determined using the Wilcox test.

Unique to near-unique members of the core microbial community in each zone

To identify unique members of the common core microbial community (Risely, 2020) within each zone, genus-level ASVs were filtered using the following two criteria: the ASVs were present in >95% of the sites within a zone and in fewer than 10% of sites within the other zones (Table S7). The number of near-unique taxa in each zone were also tested at different thresholds to assess how the threshold affected the results (Table S8). The threshold for the number of sites in which an ASV had to present for a specific zone had the greatest effect on the results, while the threshold for the number of non-zone sites in which an ASV could be present only had an effect when the threshold was raised to 20% or 3/15 sites. While there were several unique core members for the restored zone, there were fewer sites for that zone and all the unique core members were at or below the mean relative abundance of detected taxa; for these reasons the unique core taxa of only the intact and degraded zone samples are discussed.

In the intact (vegetated) zone core community, two genera were unique: Acidipila-silvibacterium and Burkholderia-caballeronia-paraburkholderia. Acidipila-silvibacterium is a member of the Acidobacteriota, commonly found in soils and capable of tolerating tolerate low pHs (Kielak et al., 2016; Kalam et al., 2020). The family Acidobacteriaceae (Subgroup 1), which contains Acidipila-silvibacterium, was also identified by LEfSe analysis as a biomarker of intact succulent thicket soil samples (Fig. 4), consistent with the lower mean pH values of these samples (Intact: pH 5.48, degraded: pH 6.84). It is likely that the presence of these taxa, almost exclusively in the vegetated soil samples, is due to the abundance of plant biomass, given that Acidipila-silvibacterium has been identified as a core operational taxonomic unit (OTU) of decaying wood (Tláskal et al., 2017) and Burkholderia-caballeronia-paraburkholderia contains many plant-associated species (Compant et al., 2008) which degrade cellulose (Štursová et al., 2012) and are associated with wood-decaying fungi (Christofides et al., 2020).

The two unique genera identified in the degraded zone core community were an uncultured member of the order Azospirillales and the genus Arthrobacter. Both of these taxa are potentially nitrogen-fixing (Steenhoudt & Vanderleyden, 2000; Fernández-González et al., 2017) and Arthrobacter has been implicated in the recovery of polluted soils and soils where vegetation has been lost (Wang, Xie & Hu, 2013; Kim & Song, 2014; Fernández-González et al., 2017).

Correlations of taxa in the intact and degraded sites

In order to understand how the structure of the microbial communities changed between intact and degraded zones, co-occurrence networks of the 50 genus-level ASVs with the highest variation in abundance were filtered from the main dataset and compared between the intact and degraded zones. The network analysis showed a marked decrease in community complexity in degraded compared with intact zone microbiomes (Fig. 6) as measured by clusters of co-occurring taxa. In the intact zone samples, ASVs were grouped in six clusters, whereas in the degraded zone samples, the same ASVs formed only two clusters. The cluster assignments were consistent, even when the number of genus-level ASVs were increased, indicating that the differences in complexity were robust for the ASVs included in the network.

Open in a separate window

Figure 6

Co-occurrence network.

Co-occurrence network showing the 50 ASVs with the highest variance. Bold text indicates hub nodes and nodes of the same colour were clustered together. Blue edges indicate positive correlations and red edges indicate negative correlations between the connected nodes. The thicker the edge, the more similar the two nodes are to one another.

In the intact zone network, only positive correlations are observed within groups; surprisingly negative correlations were completely absent. This suggests that the intact zone microbial community exists in a stable state, where each cluster of taxa may occupy a distinct niche and where inter-taxon competition is minimal. In stark contrast, the degraded zone network showed very large numbers of both positive and negative correlations between the two groups (Fig. 6). A potential ecological implication of this observation is that niches in the degraded zone are largely homogenised, resulting in high levels of inter-taxon competition.

A closer examination of the 50 genus-level ASVs showed that 30 had significantly different abundances between the intact and degraded zones (Fig. S8); 15 in cluster 1 and 15 in cluster 2. Thirteen of the 15 nodes belonging to cluster 1 had higher abundance in the intact zone and 13/15 nodes belonging to cluster 2 had higher abundance in the degraded zone. The genus-level ASVs found to be higher in the intact zone generally belonged to taxa which have been reported to be plant-associated, such as Connexibacter (Dong et al., 2018; Dobrovolskaya et al., 2020), Mycobacterium (Bouam et al., 2018; Pan et al., 2020), Pseudonocardia (Chen et al., 2009; Zhao et al., 2012; Li et al., 2012) and Microlunatus (Tuo et al., 2016). We note that these plant-associated taxa were also present in the degraded zone where vegetation is largely absent. It is unclear whether their presence is due to associations with the sparse vegetation, the presence of species in the plant-associated taxon category which do not undergo obligate interactions with plants or bacteria from the intact zone being carried downhill into the degraded zone by rain water (Abu-Ashour & Lee, 2000; Caillon & Schelker, 2020).

Although there are known pitfalls in interpreting microbial co-occurrence networks (Armitage & Jones, 2019; Carr et al., 2019), we suggest that the presence of abundant vegetation (in the intact zone), and the existence of plant-and plant root-associated microbiome niches, likely underlie the observed differences in the two networks (Fig. 6). Plant-root associated niches such as the rhizoplane and rhizospheric zones provide spatial and physicochemical separation for their intrinsic microbial communities (Ofek-Lalzar et al., 2014; Battenberg et al., 2017; Morella et al., 2020); consistent with the well-discriminated clustering structure of the intact zone network, the limited number of inter-cluster correlations and the absence of negative correlations. Conversely, the loss of these defined niche structures in the largely unvegetated degraded zone appears to spatially homogenise the microbial community, leading to a weak clustering structure and a high level of inter-taxon competition.

The research presented here was part of a BSc (Hons) thesis (MS). We wish to thank Timm Hoffman for providing repeat photographs of the study site (used in Fig. S1).