Genetic diversity and community composition of arbuscular mycorrhizal fungi associated with root and rhizosphere soil of the pioneer plant Pueraria phaseoloides

The pioneering plant Pueraria phaseoloides had a strong modulation effect on arbuscular mycorrhizal fungi (AMF) communities. Irrespective of geographical location, community composition of AMF in rhizosphere soil differed from that of the root. Co‐occurrence network analysis revealed two AMF keystone species in rhizosphere soil (Acaulospora) and roots (Rhizophagus) of P. phaseoloides.


INTRODUCTION
Arbuscular mycorrhizal fungi (AMF) ensure the survival of plants by facilitating access to limited resources, particularly in degraded ecosystems [1], thereby playing a crucial role in sustaining ecosystem processes and functions [2][3][4].The composition, diversity, and abundance of AMF communities depend on several factors.For instance, meta-analysis revealed that AMF exhibit biogeographic patterns at a global scale [5].At a local scale, both soil and biogeographical factors were shown to determine AMF communities [6].Numerous studies have proven that soil conditions also determine AMF diversity and composition (Supporting Information: Table S1).However, AMF have an idiosyncratic response to soil conditions, and there is no consensus on their relative importance [7].Several studies showed that soil pH is an important factor to determine AMF communities [8,9].It was also reported that a higher level of phosphorus (P) limited the diversity of AMF [10]; but another study showed that soil texture, rather than pH or P, affects AMF composition in an agriculture soil [11].This divergence may be the result of having targeted different ecosystems, host plants and sample types (Supporting Information: Table S1).More importantly, plants exert strong effects on the diversity and composition of their asscociated AMF [12,13], but the specificity of the association between plant host and AMF taxa is generally low [14].In addition, the same plant species may reveal differences in AMF communities between plant compartments, that is, root and rhizosphere soil [15][16][17][18].
AMF have been classified into different functional groups based on biomass allocation, that is, rhizophilic guild and edaphophilic guild [19].The rhizophilic guild is thought to allocate more arbuscular mycorrhizal (AM) biomass to roots than soil, such as Rhizophagus; while the edaphophilic guild is thought to allocate more AM biomass to soil than roots, like Acaulospora.However, caution must be taken when classifying AMF families into guilds, due to the different technologies used [20].In addition, plants could affect AMF richness by delivering more carbon to beneficial symbionts which could facilitate the competition with others [21,22].Especially in degraded ecosystems, "founder AMF" species might benefit from plant-derived carbon to colonize the plant roots, thus would outcompete "AMF latecomers", thereby leading to differences between root and rhizosphere soil [23].Thus, understanding the extent to which various factors modulate abundance, diversity, and selective root-colonization of AMF (i.e., niche differentiation) is essential not only for maintenance of ecological processes in agroecosystems [24,25], but also for facilitating the restoration of degraded ecosystems [26][27][28].AMF has been verified as pioneering microorganisms in sand dunes [29], river floodplains [30], and volcanic areas [31].Although the role of AMF has been studied intensively in different ecosystems with different plant species (Supporting Information: Table S1), AMF diversity and communities in association with pioneer plants in heavily degraded ecosystems remain largely unexplored.These include post-mining sites [32], which are frequently found in West Africa.This is especially true for Ghana, which is the largest producer of gold in Africa, and the sixth largest producer in the world [33].With a rate of 2600 hectares per year, natural vegetation in Ghana has been intensively converted into gold mining sites [34].After surface mining, the land is usually left abandoned and free of vegetation, offering space for colonization by pioneer plants.In Ghana, Pueraria phaseoloides (Roxb.)Benth.(tropical kudzu), a perennial, N 2 -fixing legume, has been found as a pioneer plant species that colonizes vigorously post-mining sites (Y.Guo, personal communication, 2019).This was similarly observed in Indonesia [35].It could be speculated that such successful colonization and potential adaptation to various site conditions is reinforced by the symbiosis with AMF.This assumption is rationalized by an earlier study, revealing that P. phaseoloides failed to establish without the presence of symbiotic AMF [36].
In view of the ecological advantage of P. phaseoloides to colonize efficiently post-mining sites, it is imperative to disentangle the factors shaping the mycorrhizal communities associated with P. phaseoloides, considering benefits for degraded land restoration.Hence, we investigated the genetic diversity and composition of AMF communities in the rhizosphere and roots of P. phaseolides growing under the prevailing soil conditions at abandoned, highly disturbed post-mining sites in Ghana.High throughput DNA sequencing was employed to analyze AMF communities in degraded mining soils.Different from morphological methods, which rely mainly on spore identification, DNA-based approaches capture genetic information from hyphae, mycorrhizal roots and spores [37].We hypothesized that, due to the strong adaptation potential of P. phaseolides in different environments, host identity is a major driving factor shaping AMF communities, assumably stronger than local factors related to geography and soil conditions.Secondly, we hypothesized that the high adaptability of P. phaseolides may be reflected in the selection of AMF specific species from the rhizosphere soil.By testing this, our main ambition is to fill a gap in the understanding of the ecological status of AMF in degraded mining soils and to provide a scientific basis for developing AMF-based strategies for restoring degraded lands.

Overall sequencing and taxonomic assignments
In total, 2,312,972 raw reads were obtained with 301 bp average read length from Illumina MiSeq® sequencing.The quality control (quality scores > 35) reduced the reads to 2,039,447 with 271 bp sequences on average (i.e., an average of 11.8% of the reads was discarded).Rare amplicon sequence variants (ASV) with a frequency of less than 0.1% of the mean sample depth were removed (see explanation in "Bioinformatic analysis" section).After removal of rare ASV (16.8%), a total of 1,746,146 reads remained (Supporting Information: Table S2).

Alpha diversity of AMF
Rarefaction curves display the number of sequences, which have reached adequate coverage (saturation) of AMF diversity, as a quality requirement for downstream analysis (Supporting Information: Figure S3).The α-diversity indices were displayed and statistical differences were annotated for both locations (Konongo, KN; Bosome-Freho, BF) (Figure 1C,D and Supporting Information: Figure S4).At both locations, more ASV were observed in rhizosphere soil than in roots, although this difference was not significant (p > 0.05) (Supporting Information: Figure S4A).At both locations, ASV evenness and diversity of rhizosphere soil were, however, higher than in roots (p < 0.05) (Figure 1C,D).ASV richness of rhizosphere soil in BF was higher than in KG (p < 0.05), while evenness and diversity of ASV did not reveal any difference between the two locations (data not shown).

Beta diversity (community composition) of AMF
AMF composition between the two compartments was clearly separated, but not between the locations (Figure 1B).The stress value of NMDS analysis was 0.09, reflecting the significant variation of AMF communities among factors [38].To verify the significance of NMDS analysis, factors (plant compartment, location) were further examined using permutational multivariate analysis of variance (MANOVA) test (Figure 1B).Results indicate that the plant compartment was a crucial determinant to modulate AMF composition (p < 0.05), explaining more than 20% of variation.However, the effect of location on AMF composition was not significant, explaining only 4% of variation (p > 0.05).

Relationship of AMF with soil characteristics
The relationship between AMF richness (observed ASV) and diversity (Shannon diversity) and soil characteristics was analyzed (Supporting Information: Table S3).The contents of soil calcium (Ca) and soil pH had negative correlations with both AMF richness and diversity (p < 0.05).The content of soil zinc (Zn) only had a negative correlation with AMF richness (p < 0.05).Distance-based redundancy analysis (db-RDA) was carried out to examine the influence of soil characteristics on AMF composition.Results indicate that soil characteristics had no effect on AMF composition (p > 0.05) (Supporting Information: Table S4).

Differences in AMF communities between plant compartments
With the analysis of DeSeq2, 12 ASV were found to be more abundant in the rhizosphere soil than in root samples, whereas 7 ASV were more abundant in the root compartment than in the rhizosphere soil (Figure 2A).In the root compartment, ASV were affiliated to two genera: Rhizophagus (4) and Paraglomus (3).In rhizosphere soils, ASV were assigned to Acaulospora (5), Paraglomus (4), Dominikia (2) and Claroideoglomus (1).Network of the rhizosphere soil had more nodes and edges (123 nodes, 948 edges) than the root compartment network (86 nodes, 468 edges).The modularity index (rhizosphere soil: 0.78; root: 0.79) was above 0.4, indicating a modular network structure [39].The average degrees of rhizosphere soil and root were 17.6 and 10.9, respectively, and the average clustering coefficient was 0.84 for rhizosphere soil and 0.95 for root.The detailed results of co-occurrence networks are listed in Supporting Information: Tables S5 and S6.On basis of the network analysis, ASV230 (Acaulospora) and ASV238 (Rhizophagus) were determined as keystone taxa for the rhizosphere soil and root compartment, respectively (Figure 2B and Supporting Information: Tables S5  and S6).

DISCUSSION
Our results showed that the root of the pioneer plant P. phaseoloides revealed a lower ASV richness and diversity of AMF than the associated rhizosphere soil, a finding in line with other plant species [15,16].This confirms the general view that the rhizosphere soil provides an important reservoir of AMF for plants, from which plants only recruit a proportion at a certain time [40].More importantly, the plant compartment of P. phaseoloides exerts, independent of geographical location or microbiome provenance, a strong effect on microbial consortia shift, indicating a selective preference for associated AMF.It was shown that geographic distance had a small effect on AMF communities, because a single plant species in agricultural land may homogenize the AMF communities over a certain distance [41].Furthermore, the host compartmentalization of microbial communities facilitates the decoupling from effects of habitat fragmentation [42].In our study, P. phaseoloides plays an important role in shaping AMF communities, thereby overriding geographic factors.However, the symbiosis between plants and AMF is generally considered as nonspecific [43], which ascribes the small number of characterized AMF species (~300) compared to that of plant species (~300,000) [44].Nevertheless, evidence exists that a preference of AMF-plant associations was present in different ecosystems [13,14,44].
Apart from the release of carbonaceous root exudates triggering the most preferred symbiont [45], soil conditions have been acknowledged to modify the ability of plants to attract selected AMF [46].Our results showed that soil conditions had no significant effect on AMF composition.This divergence seems likely due to the strong regulation of the plant host on the rhizosphere species pool via root exudates, thereby masking soil conditions in structuring the soil AMF communities.In line with other studies, it was demonstrated that the host plant exerts a much stronger selectivity for AMF colonization of roots than prevailing soil conditions [13,47,48].This confirms the acknowledged community assembly concept of AMF, whereby the host filter was decisive for AMF assemblage within the plant [7].On the other hand, the effect of soil conditions on AMF richness and diversity was detected.First, AMF richness and diversity correlated negatively with soil pH, as it was observed in other situations [8,16,49].Soil pH controls AMF richness and diversity through influencing nutrient and ion availability [50].Consequently, we found that the content of soil zinc (Zn) correlated with AMF richness but not diversity, while soil calcium (Ca) correlated with both AMF richness and diversity.Both nutrients facilitate plant metabolic processes, and their uptake is supported by AMF [51].It is known that high Zn soil levels can negatively affect the abundance and composition of AMF in polluted soils [52,53], while other studies have shown that Zn could influence AMF diversity and richness in non-polluted ecosystems [8,16,54].These inconsistent results were most likely explained by differences in ecosystem type and structure.On the other hand, only limited information is available about the influence of Ca on AMF diversity and composition.Ca is not only a nutrient, but also considered as a messenger which initializes the communication between plants and AMF [55].Further research would be needed to explore the influence of Ca on AMF diversity and composition under the given environmental conditions of our study.Although pH, Zn and Ca had a significant effect on richness and/or diversity of AMF in our study, they did not trigger a significant difference in terms of community assembly.
Our data further supported the concept of host preference, revealing for the first time two different and highly abundant AMF keystone species in two distinct plant compartments: Rhizophagus in roots and Acaulospora in rhizosphere soils, associated with a single plant species (i.e., P. phaseoloides).It was reported that keystone taxa with high abundance have vital contributions for maintaining ecosystem functioning [56].This may also apply to Rhizophagus, which exists in high abundance in the root compartment of P. phaseoloides.Rhizophagus is a generally fast growing species (r-strategy) and, based on its phenotypic traits, was classified as "competitor" in the life history classification system [57].Thus, Rhizophagus may have a competitive advantage to occupy efficiently the root niche with immediate access to plant-derived resources.Furthermore, plants prefer to deliver more carbon to beneficial symbionts [21,22], like Rhizophagus, which may help plants to be successful in the harsh environmental conditions of abandoned mining sites.With regard to the colonization of degraded ecosystems (e.g., abandoned mining sites), it was proposed that socalled "founder AMF" species might benefit from this plant-derived carbon to colonize the plant roots in the early stage of ecological succession [58].Accordingly, this ecological advantage would outcompete so-called "AMF latecomers", benefiting the proliferation of Rhizophagus through the soil via colonization of newly formed roots.More importantly, Rhizophagus has been recognized as a dominant AMF species that supports plants in the early stages of growth development in agricultural system [41,59].This might also be true for degraded ecosystems.In fact, Rhizophagus was considered as a prominently abundant taxon, since it has been found in diverse host species and environments (Supporting Information: Table S1).However, a phylogenetic meta-analysis of most abundant AMF taxa across different ecosystems, such as Rhizophagus, indicated that they do not necessarily have the same phylogenetic structure [60].On the other hand, Acaulospora is a slow growing species (k-strategy) and the trait-based framework presented Acaulospora as "stresstolerant" AMF [57].However, stress-tolerant AMF were believed to provide a delayed benefit to their host, which is accompanied by an excessive carbon demand from their host [57].Therefore, our study suggested that the abundant Rhizophagus in the roots of P. phaseoloides was the result of a good functional match between both partners in this degraded ecosystem.

CONCLUSION
Our results provided fundamental genetic insights into AMF communities associated with the pioneering plant P. phaseoloides colonizing abandoned mining sites in Ghana.Our study showed that geographic and prevailing soil conditions only exerted significant effects on AMF richness and diversity, but not on AMF community composition.Instead, the plant compartment largely explained the differences in AMF composition, with two different functional species in two distinct plant compartments (i.e., Acaulospora in rhizosphere soil; Rhizophagus in root).This implied that P. phaseoloides has a strong selectivity for AMF species, irrespective of soil conditions, emphasizing the ecological plasticity of the host in selecting AMF.The present study was based on a one-time point sampling.Hence, to fully understand the ecological effects of AMF communities in degraded ecosystems, further studies, including a broader range of abandoned mining sites with distinct environmental conditions and considering multiple AMF proxies (i.e., spore density, intraradical and extraradical hyphae) across various seasons, would provide a more profound insight into the plasticity and responsiveness of AMF compartmentation in association with P. phaseoloides, as a suitable ecological basis for restoration of degraded ecosystems.distance of at least 50 m between each other.Both locations had similar climate conditions (Supporting Information: Table S7), but differed in physic-chemical soil characteristics (Supporting Information: Table S8).At each site, three spatially separated plant individuals of P. phaseoloides with a distance of 1.5 m from each other were randomly selected to ensure independence of samples [54].The entire plants with soil (approximately 10 cm width and 20 cm depth) were excavated and put into poly bags and transported in cooling boxes.Intact root systems (root balls) were conserved at 4°C [61], permitting samples to be in a semi-natural state before shipping to Germany (University of Hohenheim, Stuttgart).Upon arrival, samples were conserved at 4°C for further processing.

Processing of samples
The samples were processed according to the protocol [62], with minor modifications (Figure 3B).Bulk soil was removed from roots by soft shaking and subjected to physic-chemical analysis (Supporting Information: Table S8).Then, 10-12 roots per plant with a length of 5-8 cm were excised.Excised roots were washed with 35 ml autoclaved, phosphate buffer mixed with 200 g L −1 Tween-20 to detach the rhizosphere soil from roots.The roots were transferred to a new Falcon tube (50 ml) following disinfection procedures: (1) root samples were treated with 35 ml of 50% bleach mixed with 0.01% Tween-20 for 1 min; (2) the liquid phase was replaced by 35 ml of 70% ethanol for 1 min; (3) root samples were rinsed 5 times with sterile water; (4) washed roots were dried on sterile filter paper.Then, roots were cut into small pieces using sterile forceps and pruning scissors, and conserved at −20°C for DNA extraction.Tubes containing the rhizosphere soils were processed as follows: (1) samples were filtered (sterile 100 µm mesh cell strainer) into a new 50 ml tube; (2) samples were centrifuged (3000g, 5 min, room temperature [RT]) and supernatant was removed; (3) tubes were chilled on ice and 1.5 ml of sterile phosphate buffer was added, followed by vortexing; (4) the suspended phase was transferred into a clean 2 ml tube and samples were centrifuged (15,871g, 2 min, RT).Finally, the supernatant was removed and rhizosphere soil pellets were conserved at −20°C for DNA extraction.

Amplicon sequencing
For DNA extraction, 0.1 g frozen roots of each sample homogenized in liquid N, and 0.5 g of frozen rhizosphere soil were used.DNA was extracted with the Fast DNA® spin plant kit (MP Biomedicals).To improve the quantity and quality of rhizosphere soil DNA, 30 mg polyvinylpolypyrrolidone (Thermo Fisher Scientific) was added before buffer addition [63].DNA concentration was measured by a NanoDrop spectrophotometer 2000 (Thermo Fisher Scientific).Working solutions of root (10-fold dilution) and rhizosphere soil (5 ng µl −1 ) DNA were prepared with double-distilled water, and then stored at −20°C for subsequent analysis.Amplicons of Glomeromycota were produced with nested PCR.The first PCR used the primers NS31 [64] and AMDGR [65].
Each PCR (20 µl) comprised 1 µl DNA working solution, 0.2 µM of each primer, 0.2 mM of each deoxynucleoside triphosphate (dNTP), 1.5 mM MgCl 2 and 1 U Taq DNA polymerase (Promega GmbH).Reactions were run on a PeQSTAR thermal cycler (VWR International GmbH, Bruchsal, Germany) using the following conditions: 3 min initial denaturation at 95°C, followed by 35 cycles of 95°C for 30 s, 56°for 30 s, and 72°C for 30 s.Reactions were completed with 72°C for 3 min.The second nested PCR was conducted with the AMF-specific primers AMV4.5NF and AMDGR [65] tagged with Illumina adapters, yielding an amplicon of approximately 300 bp.One µl of 1:10 diluted amplicon of the first PCR was used for the second PCR in 30 µl reactions, applying the same PCR run conditions as mentioned above, except that the annealing time was reduced to 20 s.Amplicons were verified by a 1.5% agarose gel electrophoresis.Then, 25 µl of amplicons with Illumina adaptors were submitted to Eurofins Genomics Europe Sequencing GmbH.Library construction and quality check were done by Eurofins Genomics.Illumina MiSeq was used for sequencing with a 2 × 300 sequence mode (Eurofins Genomics).The raw sequences were deposited in the Genome Sequence Archive [66] under BioProject accession number PRJ011089.

Bioinformatic analysis
Sequences were trimmed to exclude primer sequences and quality-filtered with quality scores > 35 in an initial step [18].Rare amplicon sequence variants (ASV), with a frequency of less than 0.1% of the mean sequence depth, were removed.As Illumina reported to be 0.1% of reads most likely due to MiSeq bleed-through between runs (https://github.com/LangilleLab/microbiome_helper/wiki/Amplicon-SOP-v2-(qiime2-2020.2)).Taxonomic identification of each ASV was performed according to the protocol of Stefani with minor modification [18].First, each ASV was identified with the closest sequences against the National Center for Biotechnology Information (NCBI), using a basic local alignment search tool (BLAST).The search was set to Glomeromycotina (taxid:214504), whereby uncultured/ environmental sample sequences were excluded and the maximum number of similar sequences retrieved (i.e., the number of top hits to record) was set to 10. BLAST results were exported as a single file XML2 and uploaded to Geneious Prime® (version 2020.2), with the aim of downloading the taxonomic information and saving only the first hit (the hit with the highest pairwise similarity and query coverage of >95%).Then, BLAST results with taxonomic information were imported to QIIME2.Using qiime feature-classifier (classify-consensus-blast) to assign ASV, sequences belonging to ASV identified as non-Glomeromycotina (unclassified AMF) at the phylum level were removed.The remaining ASV were considered as effective ASV and used for downstream analysis.Taxonomy assignment was inferred with RAxML (v8.2.12) under the GTRGAMMA model and 1000 bootstraps via the CIPRES web-portal using "phylogenetic backbone tree".Phylogenetic backbone tree was calculated with the same producer as above, in addition to specify outgroups.The reference sequences were acquired from database [67] and recently described AMF species in public repositories.The taxonomy of each ASV was delimitated with its position in the phylogenetic tree (Supporting Information: Table S9).

1
The overall structure of arbuscular mycorrhizal fungi (AMF) communities.(A) Genus distribution of amplicon sequence variants (ASV) using relative abundance of AMF associated with rhizosphere soil and root.Eleven rhizosphere soils and 14 root samples are displayed in separate stacked bars.Different genera are displayed in different colors, and the low abundance genera (<10%) are grouped together (others).Relative abundance of each genus across compartments is displayed in Supporting Information: Figure S2.(B) Nonmetric dimensional scaling (NMDS) analyses of weighted metric distance of AMF communities.(C) and (D) α-diversity of field samples.(C) Pielou's evenness.(D) Shannon diversity.The boxes represent the range between 75th and 25th quartiles.The line within the box represents the median.The whiskers represent the lowest and highest values extending 1.5 of the interquartile range.NS, nonsignificance.* Significance (p < 0.05); ** Significance (p < 0.001).
The differences between amplicon sequence variants (ASV) and keystone taxa identification.(A) DeSeq2 plot showing the difference of amplicon sequence variants (ASV) between the two plant compartments.Different colors represent different genera, and different point sizes represent different mean values, after normalization through DeSeq2.(B) Co-occurrence network in rhizosphere soil and root compartment.Each node represents one ASV labeled by genus.A node was verified by a robust (Spearman's correction coefficient R > 0.6) and significant (p FDR < 0.05) correlation.The size of each node is relational to the number of connections, while nodes with the same color display the same module.The thickness of each connection between two nodes is relational to the strength of Spearman's correlation coefficient.
Soil and root samples were collected at five abandoned gold mining sites distributed across two locations (45 km distance) in the Ashanti region (Ghana) in October 2019.Three sites were located in Konongo (KN, 6°37′ N, 1°14′ E) and two sites in Bosome-Freho (BF, 6°25′ N, 1°18′ E) (Figure3A).At each location, all collecting sites had a (A) (B) F G U R E 3 Sampling locations and sampling methods.(A) Sampling sites across two areas within the Ashanti region in Ghana.(B) Sampling methods to collect rhizosphere and root samples.