One ring to rule them all: an ecosystem engineer fungus fosters plant and microbial diversity in a Mediterranean grassland
Summary
- Species coexistence in grasslands is regulated by several environmental factors and interactions with the soil microbial community.
- Here, the development of the Basidiomycetes fungus Agaricus arvensis, forming fairy rings, in a species-rich Mediterranean grassland, is described. Effects of the mycelial front on plants, fungi and bacteria were assessed by vegetation survey and next generation sequencing approaches.
- Our results showed a fungal-dependent shift in the community structure operated by a wave-like spread of fairy rings that decreased plant, fungal and bacterial diversity, indicating a detrimental effect of fairy rings on most species. The fairy rings induced successional processes in plants that enhanced the replacement of a community dominated by perennial plants with short-living and fast-growing plant species. In parallel, fungal and bacterial communities showed evident differences in species composition with several taxa associated within distinct sampling zone across the fairy rings. Notably, bacteria belonging to the Burkholderia genus and fungi of the genus Trichoderma increased in response to the advancing mycelium of A. arvensis.
- The profound changes in community composition and the overall increase in taxa diversity at ecosystemic scale suggest that fairy ring-forming fungi may act as ecosystem engineer species in Mediterranean grasslands.
Introduction
Among global ecosystems, temperate grasslands can be considered biodiversity hotspots, with a species richness challenging that of tropical forests (Mittermeier et al., 2011; Wilson et al., 2012; Dengler et al., 2013; Habel et al., 2013). Such high level of species richness is maintained by the combined and complex interaction of both environmental and biological factors: grazing, fire, hail, frost, snow cover, water relations, plant–plant competition, facilitation, phytotoxicity and autotoxicity (Tilman & Downing, 1994; Grace, 1999; Callaway & Aschehoug, 2000; Bonanomi et al., 2005; Mazzoleni et al., 2007; Mazzoleni et al., 2010; Rice, 2013). Moreover, the relationship between plant species distribution and soil microbial community has been highlighted in recent years (Bezemer et al., 2006; Van Der Heijden et al., 2008). In particular, the balance between fungal symbionts, pathogens and saprotrophs (Klironomos, 2002; Teste et al., 2017; Semchenko et al., 2018), and the extent to which decomposition of plant materials affects the microbial community is crucial to understanding plant community patterns (Bartelt-Ryser et al., 2005; Bonanomi et al., 2019). Recently, Mazzoleni et al. (2015a) reported an inhibitory effect of the extracellular DNA of plant species released by litter decomposition on the individuals of the same species. The same effect was also observed in fungi and bacteria (Mazzoleni et al., 2015b). Then, this phenomenon has been suggested to be a driving factor in the development of vegetation patterns, including clonal plant rings (Cartenì et al., 2012), and species coexistence (Cartenì et al., 2016; Vincenot et al., 2017).
A specific case of a plant–soil interaction that mediates species richness in grassland ecosystems is that generated by fungal fairy rings (Edwards, 1984; Bonanomi et al., 2012; Caesar et al., 2013). Fairy rings are concentric vegetation bands caused by the expansion of basidiomycetous fungi in the soil that, in most cases, strongly alter the standing plant community (Shantz & Piemeisel, 1917; Halisky & Peterson, 1970; Edwards, 1988). Fairy rings have been described as exerting either negative, positive or neutral effects on plant communities. Shantz & Piemeisel (1917) classified grassland fairy rings into three types: ‘Type-1’ formed by a bare area of dead vegetation in proximity to the underground-expanding fungal front and followed by a belt of luxuriant vegetation, as in the cases of Marasmius oreades (Cosby, 1960) and of fungi belonging to Agaricus genus (Halisky & Peterson, 1970). ‘Type-2’, clearly recognisable by the presence of luxuriant vegetation but without an external area of dead/stunting plants, for example rings formed by Calvatia cyathiformis (Shantz & Piemeisel, 1917). ‘Type-3’, ephemeral rings, periodically revealed by the presence of fungal carpophores, with no detectable changes in vegetation patterns, for example those formed by Macrolepiota procera and Infundibulicybe geotropa (Halisky & Peterson, 1970).
Historically, greater attention has been directed towards Type-1 and Type-2 fairy rings. Several attempts have been made to describe both the former vegetation decline and the later plant vegetative regrowth following fungal development. In some cases related to M. oreades, fairy rings were considered agents of phytopathological problems associated with turf management (Elliott et al., 2002; Fidanza et al., 2007; Miller et al., 2012). Other studies have reported that basidiomycetes growth in the soil affected the plant community by modifying nutritional and physical soil properties, thus also changing the species competition outcome (Shantz & Piemeisel, 1917; Elliott, 1926; Cosby, 1960; Wang et al., 2005; Xu et al., 2011).
Increase in soil hydrophobicity from hyphal expansion (Gramss et al., 2005), immobilisation of nutrients (Fisher, 1977), pathogenic behaviour (Terashima et al., 2004; Fidanza et al., 2007), release of phytotoxic compounds like cyanide (Blenis et al., 2004; Caspar & Spiteller, 2015), and soil microbiota disequilibrium (Bonanomi et al., 2012), have all been proposed as triggering mechanisms of the detrimental behaviour of fairy rings on plant communities in grasslands. In parallel, causative mechanisms of the rearrangement of the plant community after the passage of fairy rings follow the same hypothetical lines such as disruption of existing pathogenic guilds, release of nutrients from dead microbial biomass, formation of vacant niches, or biosynthesis of plant hormones known as ‘fairy chemicals’ (Choi et al., 2010; Caspar & Spiteller, 2015; Suzuki et al., 2016).
While the impact of fairy rings on soil chemistry is well known (Gramss et al., 2005; Fidanza et al., 2007), little information is known about the modification exerted on the fungal and bacterial communities. Moreover, the persistence of the effects following fungal progression is still not clear. Several studies have described community changes of fungi and bacteria in soil subjected to fairy rings (Ohara & Hamada, 1967; Kataoka et al., 2012; Caesar et al., 2013). Unfortunately, early studies were limited by culture-based methods (Nesme et al., 2016). Currently, molecular techniques boost the inferential power of studies aimed to describe microbial composition in soils (Leff et al., 2018).
- to assess the differences of soil microbial communities before and after the passage of the fairy ring fungus A. arvensis;
- to describe the plant community response to the fairy ring dynamics;
- to define the whole impact of the fairy ring fungus A. arvensis on the above- and below-ground ecosystem structure.
Materials and Methods
Study site
The study was carried out in a species-rich calcareous grassland in the Puro-Rogedano-Valleremita mountain in central Italy (43.29°N, 12.85°E) at 869 m above sea level (asl). In the site, the presence of fairy rings is clearly recognisable by the changing in vegetation (Fig. 1). The same study area was previously used for other researches on fairy rings (Bonanomi et al., 2012) and plant species richness (Bonanomi et al., 2009). The study site is subject to periodic mowing for fodder production, and grazing is restricted in the fall to a few wild populations of ungulates. As a species-rich hotspot, the habitat is subject to biodiversity conservation policy (EU Directive 6210 ‘seminatural dry grassland and scrubland facies on calcareous substrates’ (Festuco-Brometalia – *important orchid sites)).

The underlying soil is shallow and sandy (sand 75%, silt 12%, clay 13%), with an average depth profile of c. 20 cm, and rich in organic matter (9.4%) and total N (8.7 g kg−1). The soil is characterised by pH 5.8, P2O5 9.4 mg kg−1, K2O 118 mg kg−1, Ca 2.9 mg kg−1, C/N 10.7 and total cation exchange capacity (CEC) of 28.9 mEq 100 g−1. The soil has a low pH because it is a fersialitic paleosol that has been decarbonated. The mean annual rainfall is 945 mm, with a moderate dry season during the summer months that can lead to dry conditions in some years. The mean monthly temperatures are between 21.9°C (July) and 3.8°C (January) (means over 46 yr of observations; Fabriano meteorological station, 357 m asl, 7 km from the study site).
Survey design
Fairy rings produced by the basidiomycete Agaricus arvensis were recognised by identification of the basidiomes produced. The diameter of fairy rings in the Puro-Rogedano grassland area has been monitored since 2008 and has an average annual expansion rate of 60 cm yr−1 (Bonanomi et al., 2012). Fairy rings in the range 10–15 m in diameter were selected for inclusion in the present study.
Vegetation analysis was done along transects across four randomly selected fairy rings in mid May 2017. For each ring, 720 cm long transects were established. Each transect consisted of contiguous 100 × 100 cm plots with the exception of the zones containing the fungal front that, given their narrow extension, were sampled in 50 × 20 cm plots. Six different zones were identified across transects for vegetation analysis and soil sampling, proceeding from the outer to the inner areas of the rings. These are referred to as: OUT, external grassland; FF, fungal front of A. arvensis; Belt, zone with flourishing vegetation inside the fairy rings; IN1, zone adjacent to the inner Belt border (1 m from FF); IN2, zone at a distance of 2 m from FF; IN3, zone at 5 m from the FF (Fig. 1c). Thus, 24 plots were surveyed for vegetation analysis.
In late May 2017, soil samples for metagenomics analysis were collected from all plots and in two additional transects located in two other randomly selected fairy rings nearby the original survey area. Thus, six fairy rings were sampled for metagenomics analyses. Six plots in each transect, corresponding to the OUT, FF, Belt, IN1, IN2 and IN3 sampling zones, were sampled for metagenomics analyses for 36 samples (Fig. 1). However, three samples were lost during laboratory processing, corresponding to the innermost part of the rings (IN3), thus resulting in 33 soil samples.
Vegetation analysis
The cover of each plant species along the transects was visually estimated during the growing season. Plant species determination followed the bibliographic repository of Italian flora. Plant cover was recorded with the Braun-Blanquet abundance–dominance scale transformed into percentages as follows: 5 = 85%; 4 = 60%; 3 = 35%; 2 = 15%; 1 = 5%; + = 1%.
Soil sampling, DNA extraction and amplification
Soil samples were collected by a 2 cm diameter soil corer, at a depth of 10 cm, in four randomly selected points within each sampling plot. Subsequently, soil was pooled and sieved (2 mm mesh) on site resulting in a single sample for each sampling plot. The samples were stored in sterile plastic bags and labelled. Before every sampling operation, the soil corer was thoroughly cleaned and sterilised to avoid between samples contamination. After collection, samples were rapidly frozen and stored at −80°C until the analyses; 200 mg of homogenised soil for each sampling zone were used for total microbial DNA extraction, carried out by using the DNeasy Power Soil kit (Qiagen) according to the manufacturer’s instructions.
Bacterial and fungal diversity was assessed by high-throughput sequencing of the amplified V3–V4 regions of the 16S rRNA gene (c. 460 bp) and ITS1-2 (c. 300 bp), respectively. PCR was carried out with the primers S-D-Bact-0341-b- S-17/S-D-Bact-0785-a-A-21 (Berni Canani et al., 2017) and BITS1fw/ B58S3-ITS2rev (Bokulich & Mills, 2013) using conditions reported in the original studies.
For bacterial primers S-D-Bact-0341-b-S-17 (5′-CCTACGGGNGGCWGCAG-3′) and S-D-Bact-0785-a-A-21 (5′-GACTACHVGGGTATCTAATCC-3′), PCR conditions were: 25 cycles of 95°C for 3 min, 95°C for 30 s, 55°C for 30 s, 72°C for 30 s, 72°C for 5 min and held at 4°C. For fungal primers BITS1fw (5′-ACCTGCGGARGGATCA-3′) and B58S3-ITS2rev (5′-GAGATCCRTTGYTRAAAGTT-3′) PCR conditions were: 35 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 60 s, and a final extension of 72°C for 5 min. Subsequently, PCR products were purified with Agencourt AMPure beads (Beckman Coulter, Milan, Italy) and quantified using an AF2200 Plate Reader (Eppendorf, Milan, Italy). Equimolar pools were sequenced on an Illumina MiSeq platform, yielding 2× 250 bp, paired-end reads.
Sequence data analysis
Raw reads were filtered and analysed using the Qiime 1.9.0 software (Caporaso et al., 2010). Sequences were filtered for reads shorter than 300 or 150 bp (for bacteria and fungi, respectively), with more than one primer mismatch, or with an average quality score (Phred) below 25. operational taxonomic units (OTUs) were assembled through a de novo approach and the uclust method with a similarity threshold of 97%, and taxonomic assignment was obtained by using the RDP classifier and the Greengenes (McDonald et al., 2012) or the UNITE v.8 database (Nilsson et al., 2018). Chloroplast and Streptophyta sequences, as well as singleton OTUs, were removed and the relative abundance of other taxa was recalculated. In order to avoid biases due to the different sequencing depth, OTU tables were rarefied to the lowest number of sequences per sample. Raw sequences are available at the Sequence Read Archive (SRA) of the National Center for Biotechnology Information (NCBI), under accession number PRJNA573680.
Data analysis
Statistical analyses and plotting were carried out using the Primer 7 software (PRIMER-E Ltd, Plymouth; UK). Alpha diversity metrics were calculated. Heatmaps were generated to assess the variation in community composition at lower taxonomic levels. In heatplots, variables were reordered according to the results of an index of association similarity matrix. The 50 most abundant taxa of the plant, fungi and bacterial communities are shown in the heatplots.
A resemblance matrix calculated on Bray–Curtis dissimilarity was used to perform nonmetric multidimensional scaling (nMDS) to assess variation in species composition across sampling points for plant, bacterial and fungal communities. In association with nMDS, the significance of changes in composition and alpha diversity of the three communities analysed were tested through Permanova (999 permutations), using the sampling zones across fairy rings as fixed factor and fairy rings identity as random factor.
We further analysed functional group variation across the fairy rings in each community in question. We used different ordination methods: (1) for the plant community, the dataset was reordered according to family taxonomic level and life form, namely perennials and annuals; (2) for the fungal community, a rarefied OTU table was submitted to FUNGuilds tools to evaluate trophic strategies of each fungal taxon available in the unite database (Nguyen et al., 2016); (3) for the bacterial community, we performed PICRUSt analysis to predict KEGG pathway variation during fairy rings development (Langille et al., 2013). For each community, datasets were percentage transformed according to total values of each variable in the transects and normalised in Primer 7 software. In order to assess association between the three recorded or predicted datasets and sampling zones we performed principal component analysis (PCA).
Results
Plant community composition
By analysing the plant communities at family level, we observed a clear-cut turnover in plant community structure (Fig. 2a; Supporting Information Table S1). Before the passage of A. arvensis (OUT plots), the community was mainly characterised by Poaceae (55.90%), followed by Asteraceae (12.9%), Fabaceae (10.6%), Lamiaceae (5.4%) and Rosaceae (3.0%). At the passage of the fungal front (FF zone), the community shifted towards a structure where Caprifoliaceae became heavily dominant (47.2%). Other families associated with the FF zones but with lower levels of abundance were Poaceae, Lamiaceae, Caryophyllaceae and Rosaceae (15.0%, 10.8%, 6.8% and 5.5%, respectively). In the inner zones of the fairy rings, a generalised legacy effect of the fungal passage is evident with differences relative to the original community structure observed in each zone. A high level of contribution of Poaceae (68.5%), Caryophyllaceae (13.1%), Rubiaceae (6.4%), and Fabaceae (3.5%) was observed in the Belt zone. In IN1, the community mainly consisted of Poaceae (49.3%) Fabaceae (18.0%), Caryophyllaceae (12.1%), Valerianaceae (5.0%) and Rubiaceae (4.6%). In the IN2 zone, community structure was almost similar to that of IN1, being composed of Poaceae (41.5%), Fabaceae (17.9%), Caryophyllaceae (7.7%) and Scrophulariaceae (4.9%). In the IN3 zone, plant families occurred at similar proportions as in the OUT zone, with Poaceae, Asteraceae and Fabaceae at 48.6%, 14.8% and 10.8% of the total, respectively.

Differentiations in the plant community among the zones across the fairy rings were well represented by the nMDS plot (Fig. S1a), according to a stress indicator value of 0.11 (Clarke, 1993). Significant species level changes in plant community composition across the fairy rings was observed between the sampling zones (Fig. 2b; P = 0.001; Table S4). The plant community outside the fairy rings (OUT) was characterised by dominant populations of Bromopsis erecta (53.8%) and Briza media (12.5%), furthermore Festuca circummediterranea and Festuca stricta subsp. sulcata (10.0% and 5.0%, respectively). At the highest level of abundance of A. arvensis (FF) (Fig. 1b), the plant community was clearly different compared with the sampling zones outside and inside the fairy rings. A lower level of similarity is observed when comparing FF zone to the other sampling zones within the transects, with similarity percentage values below 25.0% and a similarity of 21.5% with respect to the previous zone (OUT) (P = 0.002). Plant community in the FF zone was characterised by a generalised decline in the whole plant community. The sole exception was the increase in Knautia calycina comprising 36.3% of the total plant community and Betonica officinalis increasing from 2.8% in OUT to 4% in FF zones. In the Belt zone, community composition showed a high level of differentiation compared to the previous zone with low levels of similarity (17.5%; P = 0.003). The plant community in the Belt zone differed in the presence of Cynosurus echinatus (35.0%), Bromopsis erecta (15.0%), Cerastium glutinosum (17.5%), Anthoxanthum odoratum (15.0%) and Festuca stricta subsp. sulcata (5.0%). Within the rings, the IN1 zone showed similarity values of 36.6% with Belt zones (P = 0.01), indicating a less abrupt change in the plant community, which was dominated by Lolium perenne (30.0%) and Trifolium incarnatum (18.3%), and with the persistence of C. echinatus (17.5%) and C. glutinosum (15.0%). A high level of similarity was observed between zones IN1 and IN2 (54.0%; P = 0.281) with nonsignificant changes in the plant community structure. Lower level of similarity in plant community was again observed when shifting from the IN2 to IN3 (21.6%; P = 0.004) whereas a high level of similarity was observed between IN3 and OUT (61.3%; P = 0.699), indicating that the plant community structure is restored inside the rings with a community consisting of B. erecta (48.8%), C. ambigua (10.0%), F. circummediterranea (7.5%), B. media (8.0%) and Trifolium ochroleucon (6.5%).
Fungal community composition
Analysed at phylum level, the differentiation in fungal community was clear only in the case of the FF zone while other zones have a similar level in taxa composition (Fig. 3a; Table S2). In details, the OUT zone is dominated by the Basidiomycota phylum (44.1%), Ascomycota (28.0%) and a large fraction of unidentified fungi (23.9%), with smaller contributions from Zygomycota (3.5%) and Glomeromycota (0.5%). With fungal front development (FF zone), a sharp increase in Basidiomycota is recorded at the expense of all other phyla. Within the fairy rings (Belt, IN1, IN2 and IN3) the organisation of phyla is restored to similar levels of those within the OUT zone.

nMDS based on Bray–Curtis similarity of zones across the fairy rings of A. arvensis gave a good representation of the differentiation in the fungal community (Stress: 0.14). A marked separation of the fungal community was observed in the case of FF zone, while a level of differentiation was detected in the Belt zone. Fungal communities in the IN1, IN2 and IN3 were similarly composed (Fig. S1b). Significant changes in fungal community composition were observed across the fairy rings (Fig. 3b; P = 0.001; Table S4). The OUT zone had a community dominated by unidentified fungi (23.9%), and OTU within the Hygrocybe genus (20.7%), unidentified Ascomycetous fungi (15.3%), an OTU within the Clavariaceae family (5.0%) and an OTU within the genus Morteriella (2.9%). The development of A. arvensis imposed significant changes in the fungal community (P = 0.001) where the fairy ring-forming fungus reaches considerable abundance (74.8%) and was associated with increased abundance of Trichoderma (7.2%). The overall changes between OUT and FF zones was supported by a low level of similarity of mycobiota (9.8%). In the Belt zone, significant changes in the fungal community were observed compared with the FF zone (P = 0.007), although higher similarity (24.5%) was recorded compared with the OUT to FF transition. In the Belt zone, the community was again dominated by unidentified fungi (14.1%), unidentified Ascomycetous fungi (13.0%), a residual presence of A. arvensis (12.4%), and, at lower abundances, OTU within Myxotrichaceae (6.2%), Mycena (5.8%), Clitocybe (4.7%) and Chetomium (2.8%). Higher similarity (43.5%; P = 0.047) was observed between the Belt and IN1 zones. The fungal community in IN3 remained dominated by unidentified fungi (19.7%), unidentified Ascomycetous fungi (12.1%) and an OTU within the Agaricales (14.0%). Other OTUs with a lower contribution belonged to Morteriella (5.5%). In the inner zones of the fairy rings no significant changes were observed between IN1 and IN2 (similarity 48.9%, P = 0.566), or between IN2 and IN3 (similarity 48.6%; P = 0.269). Finally, the innermost community IN3 showed no evident changes compared with the external grassland (OUT) (similarity 47.7%; P = 0.780).
Bacterial community composition
At phylum level, a considerable change was recorded in the bacterial community after passage of the fairy ring fungus (Fig. 4a; Table S3). Grassland soils outside the fairy rings (OUT) hosted mainly Proteobacteria (24.9%), followed by Acidobacteria (21.1%), Verrucomicrobia (17.5%), Firmicutes (9.7%), Planctomycetes (8.4%), Actinobacteria (6.9%) and Bacterioidetes (4.0%). In the FF zone, Proteobacteria (54.9%), Actinobacteria (17.9%) and Bacterioidetes (10.5%) were more abundant compared with soil outside the rings, while Acidobacteria (6.2%), Verrucomicrobia (3.4%) and Plantomyces (3.1%) declined. Throughout the inner zones (Belt, IN1, IN2 and IN3), the phyla compositions observed in the FF zone gradually change back, to become similar to that of the external grassland. The gradual community development within the fairy rings, was characterised by a decline in Proteobacteria, from a 40.8% in the Belt to 22.7% in IN3, and Bacterioidetes, which decrease from 15.5% in the Belt to 2.5% in IN3.

nMDS succeeded well in describing the changes in taxa composition of the bacterial community (Stress: 0.008; Fig. S1c), which were significant across different sampling zones of the fairy rings (P = 0.001; Table S4). Starting from outside towards inside the fairy rings (Fig. 4b), the community had a higher portion of DA101 (14.9%) followed by an OTU within Koribacteriaceae (3.3%), Candidatus solibacter (3.2%) and WD2101 (3.1%). Concurrently with the development of A. arvensis, OTUs within Burkholderia (26.6%), Sphingobacteriaceae (8.0%), Sphingomonas (6.4%), Actinoallomurus (4.4%) and Streptomyces (3.8%) were found in FF zone. Modification in the organisation of the community resulted in a low level of Bray–Curtis similarity (26.8%) between the external grassland (OUT) and the zone with higher influence of A. arvensis (FF) (P = 0.001). Further significant changes were recorded when passing from the FF to the Belt zone (P = 0.001), but with higher similarity levels (44.4%) compared with the transition from OUT to FF. Composition of the community but with different relative abundances. OTUs within Sphingobacteriaceae, Chitinophagaceae and Actinoallomurus were the most abundant taxa (4.5%, 4.5% and 4.3%, respectively).
The IN1 zone was more similar relative to the previous zone (56.9%; P = 0.01) within the main taxa contributing to community organisation being OTUs in Actinoallomurus (5.7%), Sphaerisporangium (5.0%) and Micromonosporaceae (3.5%). Increasing similarity was also observed between IN1 and IN2 (65.5%; P = 0.213) for which the recurring DA101 was noted as the main contributing taxon (7.6%) and the Actinoallomurus (5.2%) and Micromonosporaceae (4.0%) OTUs persisted. The bacterial community was maintained in the more internal area of the fairy rings (IN3) (similarity 54. 5%; P = 0.208) and became similar to the external grassland (OUT) (72.8%; P = 0.533). The IN3 bacterial community comprised DA101 as the main taxon (16.7%), followed by OTUs in Koribacteriaceae (4.2%) and WD2101 (3.8%).
Plant and microbial diversity changes following fairy ring passage
Across the sampling zones, a general decrease in the diversity metrics was observed from the outside to the inside of the fairy rings, although in some cases the decrease was not significant (Table S5). Overall, the lowest level of diversity was observed in the FF zones where the active mycelium of A. arvensis reaches the highest abundance. In the zone just behind the area of influence of A. arvensis, a gradual restoration of diversity was observed, with complete re-establishment of the grassland communities in the more internal zones of the fairy rings (Fig. 5). Significant variations across the fairy ring zones were detected in plant species richness (S, P = 0.001) and Shannon index (H′, P = 0.003). But not in species evenness (J′), although the community in IN1 was significantly less even in IN2 (P = 0.005) (Table S6).

For fungal diversity significant variation was detected in all the indices (P = at 0.004, 0.001 and 0.001 for OTU number, H′ and J′, respectively). All the metrics dramatically decreased from outside the rings following fungal development in FF zones, with subsequent recovery of fungal diversity in the inner zones (Table S7). Diversity indices for the bacterial community showed evident changes in species richness (H′) and species equitability (J′) (P = 0.002 for both indexes) across fairy rings, while no significant variation was found for the number of OTUs (P = 0.510). In both the Shannon index and community evenness, the OUT zone showed high equitability in species distribution, but evenness decreased in FF zones. Within the fairy rings, diversity (evenness) quickly recovered, reaching maximum values in the Belt and IN1 zones and decreased in IN2 and IN3 (Table S8).
Plant and microbial functional aspects
PCA of the plant community dataset according to life form showed a clear pattern associated to fairy rings development. Overall, the first two components explained 60.0% of the total variance in the dataset (PCI: 42.9% and PCII: 17.1%) (Fig. 6a). In the OUT zone, the plant community was dominated by perennial plants. Proliferation of the fungus altered community to dominance by annual plants, which persisted in the internal zones of Belt, IN1 and IN2, but in the inner zone of the rings (IN3), the perennial-dominated pattern community was re-established.

The PCA based on the fungal community composition explained 53.9% of the total variance (PCI:28.0% and PCII: 25.9%). Among the six guilds distinguished, arbuscular mycorrhizal taxa (AM) were more abundant in the inner zones in the of the fairy rings (Fig. 6b).
PICRUSt analysis provided an overview of the genomic potential of the bacterial community and how it changed to cope with the different conditions occurring across fairy rings (Fig. 6c). PCA of the predicted KEGG pathways explains an intrinsic variance in the data of 92.8% (PCI: 87.1% and PCII: 5.7%). Among the 32 variables employed in our data analysis, only xenobiotic biodegradation and metabolism showed a clear pattern of association with the FF and Belt zones. By contrast, IN1, OUT, IN2 and IN3 were associated to an increase in bacterial cellular processes.
Discussion
Plant community response to fairy ring
Our results showed a fungal-dependent shift in the plant community structure operated by a wave-like spread of the fairy rings in the grassland soil. On the one hand, we detected a general decrease in plant diversity, indicating a detrimental effect of fairy rings on most plant species rooted in the path of the fungus. On the other hand, the legacy of the fairy ring fungus led to a rapid increase in plant species richness with a major shift in plant species composition, whereas a restoration process of the previous vegetation extended over years. This finding is in accordance with previous works focused on Type-1 fairy rings, especially those of Agaricus spp. (Shantz & Piemeisel, 1917; Edwards, 1984; Bonanomi et al., 2012).
The development of the fairy rings has a detrimental effect on the abundance of perennial species and especially on the dominant perennial Poaceae, as also described in American grasslands for fairy rings of M. oreades in pastures (Cosby, 1960). The mechanisms causing the death of most of the plants is still a matter for debate. Induction of soil hydrophobicity from accumulation of dense mats of mycelium (Gramss et al., 2005), direct pathogenic activity (Terashima et al., 2004), and release of toxins-like cyanide (Blenis et al., 2004; Caspar & Spiteller, 2015) are among the most cited hypotheses. In this regard, it is remarkable that, among plants, only K. calycina, and to a less extent B. officinalis, were able to withstand the passage of the fairy rings. Some specific ability to tackle the passage of the fairy rings could explain the persistence of K. calycina and B. officinalis in the FF zone. For example, among the species of the studied plant community, these species have tap-root systems instead of fibrous roots.
After the passage of the fairy rings, plant composition underwent a total turnover with a shift towards short-lived, annual species. This phenomenon was already described in Type-1 fairy rings of M. oreades, A. campestris and Agaricus tabularis (Shantz & Piemeisel, 1917; Cosby, 1960; Bonanomi et al., 2012). The structural reorganisation of the plant community and subsequent gradual restoration reflect typical responses to intense abiotic disturbance, where short-lived plants rapidly recolonise the empty niche (Pärtel et al., 2005; De Luis et al., 2006). Taken together, our results indicate that the sudden disappearance of a stable grassland community causes a competitive release that enable faster-growing and short-lived species to colonise the gaps created by the passage of the fungus. The possibility that short-lived species take advantage of the soil chemo-physical modification induced by fairy rings cannot be ruled out (Bonanomi et al., 2012). In the inner ring zones, however, progressive restoration of the original plant community was observed.
Fairy ring affects the soil mycobiota
The advancement of mycelial fronts of A. arvensis causes a clear destabilisation of the fungal community assemblage, with only a member of Trichoderma genus increasing its relative abundance in proximity of A. arvensis. Trichoderma is a well known fungal genus, members of which are heavily employed as beneficial organisms in biological crop protection because of their ability to parasitise other fungi (Harman et al., 2004; Vinale et al., 2008). The increase in abundance of Trichoderma OTUs in correspondence with the active FF of A. arvensis, suggests that A. arvensis is followed by a weak parasite. Alternatively, the increasing abundance of Trichoderma could be related to the increase in senescent mycelium of A. arvensis. The occurrence of Trichoderma in proximity of the mycelium of fairy ring-forming fungi was also described for ectomycorrhizal fairy rings of T. matsutake on P. densiflora (Park et al., 2014; Oh et al., 2018). In the Belt zone of the rings, a Mortierella species and the fungi in the Myxotrichaceae family became more abundant. This result is consistent with similar work reporting fairy rings development and effects on the fungal community (Ohara & Hamada, 1967). Both taxa are known for their chitinolytic activity (Edgington et al., 2014; Johnson et al., 2019), suggesting that the zones that are enriched by a residual mycelium of A. arvensis can support fungal taxa specialised in the degradation of chitin. As observed for bacteria and higher plants, the mycobiota recovered quickly after the fairy rings passage.
Noteworthy, the waxcap Hygrocybe is the most abundant genus in grassland before FF passage. Together with Clavaria and Entoloma, Hygrocybe indicates a nutrient poor and undisturbed grassland (Griffith et al., 2002). Disappearing in the Belt, IN1 and IN2 portions of the fairy rings and reappearing in the more internal zones, these fungi suffer a decline in abundance with the passage of the mycelial front of A. arvensis and recover only after the legacy of the fairy rings has disappeared.
The arbuscular mycorrhizal fungal guilds (AM) increased in the inner zone of the rings. The stimulation of these taxa is coincident with the establishment of annual plants in Belt zones, and the increase in AM fungi could relate to the enhanced growth and lush vegetation characterising the plant community in Belt zones. Further investigation is required to understand the mechanisms promoting AM increase but it could be argued that the formation of symbiosis can be favoured in phosphorous limitation (Nouri et al., 2014), as it has been demonstrated that fairy ring fungi are able to immobilise this nutrient during their proliferation (Edwards, 1988).
Fairy ring alters grassland prokaryotic communities
As for the plant community, fairy rings trigger a complete compositional change in grassland prokaryotic assemblages. Expansion of A. arvensis results in a temporal decrease in bacterial community evenness, which however quickly recovered. This effect is consistent with metagenomics studies describing a dramatic decline in bacteria in the proximity of the vegetative mycelia of T. matsutake ectomycorrhizal fairy rings (Kim et al., 2014; Yang et al., 2018).
The decrease in bacterial diversity caused by fairy rings was characterised by the proliferation of a few taxonomic groups. Members of the genus Burkholderia became most abundant at the fairy rings front, followed by Streptomyces, Actinoallomurus and the Sphingobacteriaceae family with the related Sphingomonas genus. In studies of the effect of ectomycorrhizal fairy rings development on the bacterial community, these taxa were described as potential mycorrhizal-helper bacteria (Kim et al., 2014; Oh et al., 2016). Despite this, upon direct application of Burkholderia strains to pure cultures of T. matsutake the bacteria acted as fungal suppressors (Oh & Lim, 2018). In addition, some of these bacteria are commonly applied as suppressors of fungal plant pathogens in agroecosystems (Holmes et al., 1998). Indeed, studies on Streptomyces and Burkholderia clearly demonstrated the ability of members of these genera to form a mycoparasitic relationship (Trejo-Estrada et al., 1998; Govan, 2000). Additional studies are required to reveal the specific interactions between some of these bacteria and the fairy ring fungi. As also suggested for the changes in the mycobiota, it cannot be excluded that specific bacterial taxa may take advantage of the large amount of mycelium deposed in the soil by the fairy ring fungus. Indeed, the specialised bacterial community associated with the FF of A. arvensis is replaced by species able to hydrolyse chitin, such as members of the Chitinophagaceae and Actinoallomurus (Whitman et al., 2015; Brabcová et al., 2016). Plausibly, this indicates that the conditioning effect of the fungus and the degradation of mycelium of A. arvensis affect the microbial community in the inner soils of fairy rings for years after ring passage.
In the Belt and FF zones we found an increase in bacterial xenobiotic biodegradative metabolism, while, in the outer undisturbed zone and in the inner zones of the rings, microbiome analysis predicted functions mainly associated with constitutive molecular pathways. Detoxification processes could be associated with the release of toxic compounds from fungal mycelium that, in turn, should be a triggering factor of the perturbation caused by the fairy rings. Most of the bacterial taxa intervening in the fairy ring active zone have the potential ability to degrade xenobiotic compounds (McMullan et al., 2001). In particular, Burkholderia strains have demonstrated the ability to convert harmful mycotoxins into plant hormone precursors (Choi et al., 2016). However, further studies are needed to identify the nature of the chemicals released by fairy rings into soil and explain the changes observed in grassland communities.
Fairy ring and ecosystem diversity
Overall, we demonstrated that fairy rings foster species richness of higher plants as well as of soil microbiota and mycobiota in Mediterranean grassland. This is consistent with previous studies demonstrating that fairy rings increase higher plant gamma diversity in grassland ecosystems (Bonanomi et al., 2012). Moreover, from our results we concluded that the effect is similar in the microbial community. We observed a limited number of plants and microbial taxa gaining advantage from the direct action of the fairy rings. Instead, a large number of plants and microbes indirectly took advantage of the competitive release or the environmental conditions generated after FF passage. Similarly to other disturbances, fairy rings provide ideal conditions for the proliferation of taxa that depend on the changes imposed by the passage of the FF (Dodds et al., 2004; Kozlowski, 2012; Atwater et al., 2018).
From a mechanistic point of view, several studies have collected evidence and formulated hypotheses regarding the functioning of the detrimental effect of Type-1 fairy rings on soil communities. Changes in the soil structure and texture associated with the conversion of soil into a hydro-repellent matrix have been proposed (Shantz & Piemeisel, 1917; Gramss et al., 2005); a nutrient-deficiency hypothesis, in which the fairy ring fungus exploits and immobilises mineral resources available for the coexisting plant and microbial communities has been suggested (Fisher, 1977; Edwards, 1988). Further hypotheses include the release of mycotoxins such as cyanuric compounds or complexed chemical compounds with phytotoxic and antimicrobial abilities (Elliott, 1926; Blenis et al., 2004), and a possible pathogenic behaviour on the part of the Type-1 fairy ring fungi (Terashima et al., 2004). Although all of the above hypotheses may be considered plausible, what is still required is hierarchical ordination of such factors. However, the hypotheses in question are not mutually exclusive, reinforcing the idea that a combination of factors can act simultaneously and result in a Type-1 fairy ring effect in grassland communities.
Regarding the postpassage effect of fairy rings in grassland soils, plant growth is enhanced in the Belt zone, similar to Type-2 fairy rings. This effect could be explained by production of fungal secondary metabolites promoting plant growth (Choi et al., 2010, 2016), or by the instauration of a beneficial microbial community containing for example AM fungi.
The wave-like centrifugal spread of fairy rings in grasslands induces drastic changes in the plant and soil microbiome community. The profound changes in grassland functioning, composition and diversity induced by a single species suggest that fairy ring fungi can be considered ecosystem engineer species (Jones et al., 1997). The Type-1 fairy ring alteration studied here acts at different levels, modulating the number of species and evenness in the community. The phenomenon as described enhances species turnover, with taxa that uniquely appear or draw advantage from environmental conditions in different zones across the fairy rings. This suggest that several plant and microbial species depend on the passage of fairy rings, similar to other ecological factors that promote species coexistence such as drought, grazing and fire.
In conclusion, there is little doubt that a colonisation of a mycelium of A. arvensis creates a strong impact on other fungal species and bacteria in the soil microbiome and on the corresponding plant communities, but a question remains open on why such dominance is transient in time, with the mycelia keeping their centrifugal movement away from the fairy ring area. Further research will investigate the hypothesis that this can be a case of species-specific negative feedback due to the inhibitory effect of self-DNA (Mazzoleni et al., 2015a,b) released into the soil by the turnover and decomposition of the same A. arvensis mycelia, whose die-back would be self-induced after being dominant.
Acknowledgements
MZ dedicates this work to Prof Vincenzo Zuccarello and Giorgio Mancinelli who introduced him in the study of community ecology. The authors thank Nicole Salvatori for technical assistance and critical revision of the manuscript. This work was supported by No-Self SRL and supported by the System dynamics modelling of microbial cell cultures (MOD_CELL_DEV) Project – Programma di finanziamento della ricerca di Ateneo 2015, University of Naples Federico II.
Author contributions
GB, SM and FG designed the research, MZ and GC collected soil samples, MA and GT collected plant community data and recognised plants at species level, MZ performed the DNA extractions, FDF carried out bioinformatics analysis, MZ performed data analysis and wrote the manuscript, DE helped in interpretation of microbial community data and manuscript preparation.