Endolichenic Fungal Community Analysis by Pure Culture Isolation and Metabarcoding: A Case Study of Parmotrema tinctorum.

Lichen is a symbiotic mutualism of mycobiont and photobiont that harbors diverse organisms including endolichenic fungi (ELF). Despite the taxonomic and ecological significance of ELF, no comparative investigation of an ELF community involving isolation of a pure culture and high-throughput sequencing has been conducted. Thus, we analyzed the ELF community in Parmotrema tinctorum by culture and metabarcoding. Alpha diversity of the ELF community was notably greater in metabarcoding than in culture-based analysis. Taxonomic proportions of the ELF community estimated by metabarcoding and by culture analyses showed remarkable differences: Sordariomycetes was the most dominant fungal class in culture-based analysis, while Dothideomycetes was the most abundant in metabarcoding analysis. Thirty-seven operational taxonomic units (OTUs) were commonly observed by culture- and metabarcoding-based analyses but relative abundances differed: most of common OTUs were underrepresented in metabarcoding. The ELF community differed in lichen segments and thalli in metabarcoding analysis. Dissimilarity of ELF community intra lichen thallus increased with thallus segment distance; inter-thallus ELF community dissimilarity was significantly greater than intra-thallus ELF community dissimilarity. Finally, we tested how many fungal sequence reads would be needed to ELF diversity with relationship assays between numbers of lichen segments and saturation patterns of OTU richness and sample coverage. At least 6000 sequence reads per lichen thallus were sufficient for prediction of overall ELF community diversity and 50,000 reads per thallus were enough to observe rare taxa of ELF.

Introduction

Lichen is a symbiotic mutualism of lichen-forming fungi that live as obligate biotrophs in symbiosis with green algae or cyanobacteria as photosynthetic partners [1]. The mutualism of lichen symbiosis benefits both organisms [2]. Lichen thalli can protect photosynthetic partners from harmful environmental factors such as UV radiation [3,4], and these photobionts do not appear to survive in nature by themselves [5,6]. Mycobionts in lichens depend on photosynthates such as ribitol, produced by photobionts for carbon source [1,6]. From this perspective, lichens were called fungal farmers of algae [6] or fungi that have discovered agriculture [7].

Lichen thalli are microbial consortiums that consist of diverse organisms including bacteria [8], filamentous fungi [9], and basidiomycete yeasts [10]. Endolichenic fungi (ELF) are distinct from mycobionts [11] and considered similar to endophytic fungi because they reside in lichen thalli without visible symptoms like endophytes [12,13]. ELF are considered not accidental colonizers of lichen thalli but rather considered a distinct ecological group [14] closely connected with photobionts as heterotrophs [12]. A Lower Devonian fossil lichen specimen indicated that the history of ELF associations inside lichen thalli is quite old-established [15]. However, the evolution and ecological roles of ELF inside lichen thalli remain unclear.

Recently, research on ELF has gained greater attention as a fascinating bioresource because of their relatively untapped secondary metabolites [16,17], which are distinct from the natural substances produced by lichens [17]. The diverse metabolites isolated from ELF include alkaloids [18,19], quinones [20,21], terpenes [22], and peptides [23]. Various substances produced by ELF have been reported to exhibit remarkable biological activities [16] including anti-cancer [24], anti-bacterial [25,26], anti-fungal [27,28], anti-inflammatory [29], and anti-oxidant [30] activities. Because of the wide range of secondary metabolites, ELF may play ecologically important roles in resistance to abiotic and biotic stresses and detoxification of substances that endolichenic bacteria secrete [12,31].

Lichenicolous fungi represent a successfully developed group on lichen thalli visibly [32]. Being found that asymptomatic lichens harbored lichenicolus fungi [33], it is hard to distinguish lichenicolous fungi and ELF scientifically. ELF are comprised of diverse taxa that mainly belong to major lineages of Ascomycota [13]. The diversity and community structure of ELF are known to be affected by host lichen lineage, photobiont type, geographic location, and climate [13,14,34]. In recent years, studies on ELF diversity have been conducted mainly by culture-based analysis [14,34-36] and metabarcoding [33,37,38], but the structures of ELF communities determined by these two approaches were different from each other. Xylariaceous fungi (Sordariomycetes), which might play the role of an endophyte or late decomposer [39], were reported repetitively that they possessed a large amount of ELF as determined by culture-based studies [14,34,40-42]. However, ELF belonging to Sordariomycetes were not frequently detected in metabarcoding-based studies [33,38]. As Chagnon et al. [43] and Suryanarayan and Thirunavukkarasu [12] discussed, both culture-dependent and independent assays should be used for detailed ELF diversity assessments. Notably, U’Ren et al. firstly compared the endolichenic fungal diversity between culture-based approach and pyrosequencing [44]. They found that ELF species richness was significantly different in culturing and pyrosequencing at the same sampling effort. However, further detailed information like commonly observed species in two methods is needed to be revealed.

Here, we investigated and compared the diversity and community structure of the ELF in the foliose lichen Parmotrema tinctorum by metabarcoding and culture-based analysis. Also, we used metabarcoding to analyze ELF community similarities between lichen segments and thalli. Finally, relationship assays between numbers of lichen segments, saturation patterns of operational taxonomic unit (OTU) [45] richness and sample coverage were performed to test how many fungal sequence reads would be needed to assess ELF diversity.

Materials and methods

Lichen sampling

We collected three thalli (thallus A, B, and C) of P. tinctorum on pine trees (Pinus thunbergii) from Songho beach located in Haenam-gun, South Korea (34°31'52" N, 126°51'88" E). Surface organic residues of thalli were removed using a syringe tip and running tap water, and thalli surfaces were sterilized with 70% ethanol followed by 0.4% sodium hypochlorite (both for 90 sec) as described by Yang et al. [40]. After rinsing with sterilized distilled water (SDW), lichen thalli were divided into 10 equally sized segments (1 cm2), which were subjected to culture isolation and metabarcoding of their ELF (Figure 1).

Figure 1.

Schematic diagram of the study design. Three thalli of Parmotrema tinctorum were divided into 30 segments (1 cm2) and each of the 15 segments was used for culture and metabarcoding analysis.

Culture-based analysis

Lichen segments were suspended in SDW (up to 2.0 mL) in sterilized tubes (2 mL) and inoculated into four types of media: potato dextrose agar (PDA, Difco, Spark, MD, USA), Bold’s basal medium (BBM) [46], Water agar (WA, 2% agarose), and host lichen medium (HM) each in two 24-well plates. We made the HM by adding 2% crushed lichen thallus to WA before sterilization. Eight 24-well plates were inoculated with 1920 μL (10 μL per well) of the lichen mixture: in total, 3 lichen thalli were inoculated into 2880 wells (3 thalliㆍ5 segmentsㆍ4 mediaㆍ48 wells). The plates were incubated at room temperature (ca. 21.5 °C) for more than 2 months. Newly growing mycelia were transferred to PDA, and pure isolates were incubated at the RT until sufficient biomass for DNA extraction was obtained.

The isolates were grouped by morphotypes based on their colony morphology [47]. The genomic DNA (gDNA) of each morphotype was extracted using the PowerSoil DNA isolation kit (Qiagen, Hilden, Germany). The nuclear ribosomal internal transcribed spacer (ITS) region was amplified using ITS4 and ITS5 primers [48]. PCR was performed using AccuPower PCR PreMix (Bioneer, Daejeon, Korea) in a final volume of 20 μL containing 1 μL of 10 pmol of bidirectional primer, 2 μL of gDNA, and 16 μL of SDW. The PCR conditions were as follows: 95 °C for 5 min, and 30 cycles of 94 °C for 30 s, 56 °C for 30 s, 72 °C for 60 s, and a final extension at 72 °C for 5 min. The PCR product was checked by 1% agarose gel electrophoresis and sequenced by Macrogen (Seoul, Korea). Bidirectional sequences were edited and quality-checked using BioEdit v.7.0.5.3 [49] and assembled using ATGC ver. 1.03 (GENETYX Co., Tokyo, Japan). ITS sequence-based identification was conducted using BLAST against NCBI GenBank. Phylogenetic analysis was conducted using MEGA v.7.0 software [50] using the maximum-likelihood (ML) at the species level. Whole representative sequences determined in this study were deposited in GenBank (MZ855362 – MZ833462).

Metabarcoding analysis

Surface-sterilized thallus segment (1 cm2) were homogenized and DNA was extracted using a PowerSoil DNA isolation kit. PCR amplification of the fungal ITS1 subregion [51] was conducted using ITS1F [52] and ITS2 [48] primers ligated to Illumina sequencing adaptors. PCR was conducted three times for each sample using the AccuPower PCR PreMix kit using the following conditions: 94 °C for 5 min, 25 cycles of 94 °C for 30 s, 55 °C for 30 s, and 72 °C for 40 s with a final extension at 72 °C for 10 min. PCR products were checked on 1% agarose gel and purified using the Expin™ PCR SV kit (GeneAll Biotechnology CO., Seoul, South Korea). Second PCR for barcoding was conducted to attach multiple index delimiters as required by the Nextera XT Index kit protocol (Illumina, San Diego, CA, USA). After purification, concentrations of amplicon libraries were measured using a NanoDrop2000 (Thermo Fisher Scientific, Waltham, MA, USA) and pooled in equimolar quantities. Amplicon libraries were sequenced using an Illumina Miseq platform by Macrogen (Seoul, South Korea).

Raw sequences were processed using QIIME2 v.2021.4 [53] by demultiplexing and denoising reads following the DADA2 pipeline [54]. Taxonomic assignment was performed according to the VSEARCH guideline [55] using identity of 97% on the UNITE fungi 99% OTU database [56]. The OTU table of ELF was constructed after filtering sequences of the host lichen P. tinctorum. A phylogenetic tree was constructed using the q2-alignment plugin (https://docs.qiime2.org/2021.4/tutorials/phylogeny) based on maximum likelihood. To identify OTUs commonly detected by culture-based analysis and metabarcoding, OTU clustering of sequence reads was conducted using VSEARCH pipeline using an identity of 97% on the sequence database obtained by culture-based analysis. Sequencing data were deposited in NCBI SRA under accession number PRJNA755844.

Statistics and visualizations

The following analyses were performed using R v.3.5.3 [57] and RStudio v.1.2.5 [58]. The OTU table constructed by metabarcoding was imported from QIIME2 to R using qiime2R (https://forum.qiime2.org/t/tutorial-integrating-qiime2-and-r-for-data-visualization-and-analysis-using-qiime2r/4121) and rarefied to a minimum number of sequences using phyloseq package [59] before conducting further community analyses. The OTU table from culture-derived results was rarefied using vegan package [60]. Alpha diversity indices were calculated using phyloseq and vegan. Community dissimilarity was calculated with Bray-Curtis distance [61] using phyloseq and vegan. We calculated alpha and beta diversity indices from lichen segments rather than one lichen thallus to consider intra-sample variability. For the statistical analysis, we initially conducted the Shapiro test [62] to check data normality. Since the normality assumption was violated, we compared alpha diversity indices and community dissimilarities using the Mann–Whitney U test [63].

All of the graphs produced except a heatmap, rarefaction curves, and a non-metric multidimensional scaling (NMDS) were visualized using ggplot2 package [64]. The heatmap representing commonly observed OTUs was constructed using pheatmap package [65]. Relative abundance in the heatmap was calculated as by dividing the number OTUs by the total number of commonly observed 37 OTUs because the relative abundances of these OTUs according to metabarcoding-derived result was extremely low and not well visualized. Constructing OTU richness rarefaction curve and sample coverage were performed using iNEXT package [66]. NMDS ordination based on Bray-Curtis dissimilarity was plotted using phyloseq.

Results

Overall classification output and alpha diversity of the ELF community

One thousand twenty-nine ELF isolates were obtained by culture-based analysis and classified into 101 OTUs representing 2 phyla, 6 classes, 23 orders, and 58 genera (Table 1). ITS amplicon sequencing yielded a total of 17,844,170 reads in 15 lichen thallus segments, which was reduced to 308,065 by denoising and filtering out sequences of the host lichen. After rarefaction, the number of sequence reads was reduced to 36,960, and taxonomic classification showed these sequences represented 4 phyla, 19 classes, 65 orders, 206 genera, and 473 OTUs.

Table 1.

Isolated endolichenic fungal OTUs in culture-based analysis.

Accession no.	Isolation no.	Operational taxonomic unit	Class	Meta*
MZ855362	KoLRI_053464	Acremonium sp.	Sordariomycetes
MZ855363	KoLRI_053361	Acremonium polychromum	Sordariomycetes
MZ855364	KoLRI_053921	Alternaria cf. alternata	Dothideomycetes	Observed
MZ855365	KoLRI_053241	Amphisphaeria sp.	Sordariomycetes
MZ855366	KoLRI_053516	Annulohypoxylon atroroseum	Sordariomycetes
MZ855367	KoLRI_053245	Annulohypoxylon annulatum	Sordariomycetes	Observed
MZ855368	KoLRI_053394	Annulohypoxylon areolatum	Sordariomycetes
MZ855369	KoLRI_053392	Annulohypoxylon cohaerens	Sordariomycetes
MZ855370	KoLRI_053322	Annulohypoxylon sp.1	Sordariomycetes	Observed
MZ855371	KoLRI_053387	Annulohypoxylon sp.2	Sordariomycetes
MZ855372	KoLRI_053922	Antennariella placitae	Dothideomycetes
MZ855373	KoLRI_053461	Arcopilus aureus	Sordariomycetes
MZ855374	KoLRI_053924	Aspergillus cf. amstelodami	Eurotiomycetes	Observed
MZ855375	KoLRI_053925	Aspergillus cf. flavus	Eurotiomycetes	Observed
MZ855376	KoLRI_053373	Biscogniauxia atropunctata	Sordariomycetes	Observed
MZ855377	KoLRI_053242	Biscogniauxia petrensis	Sordariomycetes
MZ855378	KoLRI_053927	Bjerkandera adusta	Agaricomycetes
MZ855379	KoLRI_053346	Camaropella pugillus	Sordariomycetes	Observed
MZ855380	KoLRI_053458	Chaetomium cf. pachypodioides	Sordariomycetes
MZ855381	KoLRI_053355	Chaetomium cf. pseudoglobosum	Sordariomycetes	Observed
MZ855382	KoLRI_053418	Chaetomium cf. critinum	Sordariomycetes
MZ855383	KoLRI_053260	Chaetomium cf. globosum	Sordariomycetes
MZ855384	KoLRI_053465	Chaetomium trigonosporum	Sordariomycetes
MZ855385	KoLRI_053928	Cladophialophora sp.	Eurotiomycetes	Observed
MZ855386	KoLRI_053929	Cladosporium cf. halotolerans	Dothideomycetes
MZ855387	KoLRI_053504	Coniochaeta prunicola	Sordariomycetes
MZ855388	KoLRI_053364	Coniochaeta sp.1	Sordariomycetes
MZ855389	KoLRI_053283	Coniochaeta sp.2	Sordariomycetes
MZ855390	KoLRI_053243	Coniochaeta velutina	Sordariomycetes
MZ855391	KoLRI_053930	Coprinellus radians	Agaricomycetes
MZ855392	KoLRI_053931	Coprinopsis atramentaria	Agaricomycetes	Observed
MZ855393	KoLRI_053367	Cosmospora sp.	Sordariomycetes
MZ855394	KoLRI_053267	Creosphaeria sassafras	Sordariomycetes	Observed
MZ855395	KoLRI_053272	Daldinia childiae	Sordariomycetes	Observed
MZ855396	KoLRI_053478	Dichotomopilus cf. funicola	Sordariomycetes
MZ855397	KoLRI_053408	Dichotomopilus cf. ramosissimus	Sordariomycetes	Observed
MZ855398	KoLRI_053932	Capnodiales sp.	Dothideomycetes	Observed
MZ855399	KoLRI_053933	Dothideomycetes sp.2	Dothideomycetes
MZ855400	KoLRI_053934	Dothideomycetes sp.3	Dothideomycetes	Observed
MZ855401	KoLRI_053935	Dothideomycetes sp.4	Dothideomycetes	Observed
MZ855402	KoLRI_053936	Dothideomycetes sp.5	Dothideomycetes
MZ855403	KoLRI_053937	Dothideomycetes sp.6	Dothideomycetes
MZ855404	KoLRI_053939	Exophiala bergeri	Eurotiomycetes
MZ855405	KoLRI_053940	Exophiala sp.1	Eurotiomycetes
MZ855406	KoLRI_053941	Exophiala sp.2	Eurotiomycetes
MZ855407	KoLRI_053942	Herpotrichiellaceae sp.3	Eurotiomycetes	Observed
MZ855408	KoLRI_053292	Fusarium cf. acaciae-mearnsii	Sordariomycetes	Observed
MZ855409	KoLRI_053248	Fusarium cf. babinda	Sordariomycetes
MZ855410	KoLRI_053944	Gymnopilus luteofolius	Agaricomycetes
MZ855411	KoLRI_053946	Herpotrichiellaceae sp.2	Eurotiomycetes
MZ855412	KoLRI_053255	Hypoxylon investiens	Sordariomycetes
MZ855413	KoLRI_053459	Hypoxylon monticulosum	Sordariomycetes
MZ855414	KoLRI_053426	Hypoxylon perforatum	Sordariomycetes
MZ855415	KoLRI_053334	Hypoxylon pilgerianum	Sordariomycetes
MZ855416	KoLRI_053947	Kirschsteiniothelia sp.	Dothideomycetes
MZ855417	KoLRI_053335	Microascales sp.1	Sordariomycetes
MZ855418	KoLRI_053263	Microascales sp.2	Sordariomycetes
MZ855419	KoLRI_053338	Microascales sp.3	Sordariomycetes
MZ855420	KoLRI_053460	Microascales sp.4	Sordariomycetes
MZ855421	KoLRI_053383	Phomatospora sp.	Sordariomycetes
MZ855422	KoLRI_053254	Microascales sp.6	Sordariomycetes
MZ855423	KoLRI_053251	Microcera sp.	Sordariomycetes	Observed
MZ855424	KoLRI_053247	Nemania diffusa	Sordariomycetes	Observed
MZ855425	KoLRI_053352	Nemania sp.	Sordariomycetes
MZ855426	KoLRI_053384	Nigrospora oryzae	Sordariomycetes	Observed
MZ855427	KoLRI_053948	Paracamarosporium hawaiiense	Dothideomycetes	Observed
MZ855428	KoLRI_053949	Paraconiothyrium brasiliense	Dothideomycetes	Observed
MZ855429	KoLRI_053950	Penicillium cf. citreonigrum	Eurotiomycetes	Observed
MZ855430	KoLRI_053951	Penicillium cf. sajarovii	Eurotiomycetes
MZ855431	KoLRI_053952	Penicillium cf. cairnsense	Eurotiomycetes	Observed
MZ855432	KoLRI_053953	Penicillium cf. cosmopolitanum	Eurotiomycetes
MZ855433	KoLRI_053954	Penicillium cf. sumatraense	Eurotiomycetes
MZ855434	KoLRI_053955	Penicillium citrinum	Eurotiomycetes
MZ855435	KoLRI_053428	Pestalotiopsis cf. oryzae	Sordariomycetes
MZ855436	KoLRI_053336	Pestalotiopsis neglecta	Sordariomycetes
MZ855437	KoLRI_053261	Pestalotiopsis cf. lespedezae	Sordariomycetes	Observed
MZ855438	KoLRI_053956	Phanerochaete concrescens	Agaricomycetes
MZ855439	KoLRI_053483	Diaporthe elonis	Sordariomycetes	Observed
MZ855440	KoLRI_053957	Pseudeurotium cf. hygrophilum	Leotiomycetes
MZ855441	KoLRI_053958	Pseudochaetosphaeronema sp.	Dothideomycetes	Observed
MZ855442	KoLRI_053381	Pseudovalsaria sp.	Sordariomycetes
MZ855443	KoLRI_053291	Purpureocillium lilacinum	Sordariomycetes
MZ855444	KoLRI_053320	Rosellinia aquila	Sordariomycetes
MZ855445	KoLRI_053959	Sarea difformis	Sareomycetes	Observed
MZ855446	KoLRI_053960	Sarea resinae	Sareomycetes	Observed
MZ855447	KoLRI_053259	Microascales sp.7	Sordariomycetes
MZ855448	KoLRI_053497	Sordariomycetes sp.1	Sordariomycetes
MZ855449	KoLRI_053409	Sordariomycetes sp.2	Sordariomycetes
MZ855450	KoLRI_053962	Teichospora sp.	Dothideomycetes
MZ855451	KoLRI_053489	Thielavia arenaria	Sordariomycetes	Observed
MZ855452	KoLRI_053268	Tolypocladium sp.1	Sordariomycetes	Observed
MZ855453	KoLRI_053339	Tolypocladium sp.2	Sordariomycetes	Observed
MZ855454	KoLRI_053264	Tolypocladium tropicale	Sordariomycetes
MZ855455	KoLRI_053288	Trichoderma cf. lixii	Sordariomycetes	Observed
MZ855456	KoLRI_053246	Trichoderma cf. koningiopsis	Sordariomycetes	Observed
MZ855457	KoLRI_053963	Tricholomataceae sp.	Agaricomycetes	Observed
MZ855458	KoLRI_053262	Whalleya microplaca	Sordariomycetes
MZ855459	KoLRI_053429	Xylaria arbuscular	Sordariomycetes
MZ855460	KoLRI_053424	Xylaria longipes	Sordariomycetes
MZ855461	KoLRI_053337	Xylariaceae sp.	Sordariomycetes	Observed
MZ855462	KoLRI_053448	Xylomelasma sp.	Sordariomycetes

Meta*: commonly observed OTUs in metabarcoding analysis.

Alpha diversity was calculated using four indices: the number of observed OTUs, Chao1 richness, Shannon’s diversity, and Shannon’s equitability (evenness). Alpha diversity of the ELF community was greater in metabarcoding analysis than in culture-based analysis (Figure 2), and the number of observed OTUs in metabarcoding analysis (78.80 ± 18.89) was also significantly greater than in culture-based analysis (15.73 ± 3.13) (Mann–Whitney, p < 0.001). Chao1 richness in metabarcoding (86.35 ± 20.96) was also remarkably higher than in culture analysis (29.77 ± 9.94) (Mann–Whitney, p < 0.001). However, Shannon’s diversity was significantly higher in culture analysis (3.68 ± 0.42) than in metabarcoding (2.79 ± 0.75) (Mann–Whitney, p < 0.005). Evenness indices determined in metabarcoding (0.64 ± 0.16) and culture analyses (0.65 ± 0.02) were not significantly different (Mann–Whitney, p > 0.05).

Figure 2.

Alpha diversity of the ELF community. Alpha diversity of ELF communities was calculated using four indices: the number of observed OTUs (a), Chao1 richness (b), Shannon’s diversity (c), and Shannon’s equitability (evenness, d). Meta, metabarcoding; ****p < 0.001; nsp > 0.05.

Taxonomic composition of ELF community

Relative abundances of ELF communities observed in metabarcoding and in culture-based analysis were remarkably different (Figure 3). In metabarcoding, Dothideomycetes (31%) was the most abundant class followed by Sordariomycetes (18%), and Agaricomycetes (9%). At the order level, Capnodilaes (18%) was the most dominant followed by Hypocreales (17%), and Pleosporales (10%). However, culture-based analysis showed that Sordariomycetes (62%) was the dominant class followed by Dothideomycetes (12%) and Eurotiomycetes (7%). At the order level, Xylariales (31%) was the most abundant followed by Sareales (17%) and Hypocreales (13%).

Figure 3.

Taxonomic composition ELF community. Taxonomic proportion was calculated in relative abundance of taxa at class level (a) and at order level (b).

Commonly observed OTUs by culture-based and metabarcoding analyses

After taxonomic classification of amplicon sequences based on the culture-derived database using 97% identity, 37 OTUs were detected by culture-based analysis and metabarcoding (Figure 4(a); Table 1). Culture-based and metabarcoding analyses differed in terms of the relative abundances of commonly observed OTUs. At the class level, Dothideomycetes was underrepresented in culture-based analysis as compared with metabarcoding analysis (Figure 4(b)). On the other hand, Sareomycetes was abundant by culture-based analysis but of low abundance in metabarcoding analysis. Overall, the relative abundances of the 37 OTUs in metabarcoding analysis, were extremely low compared to in culture-based analysis, with the exception of Capnodiales sp. (Figure 4(c)).

Figure 4.

Commonly observed OTUs by culture-based and metabarcoding analyses. Thirty seven OTUs were commonly observed by metabarcoding and culture (a). These OTUs showed a difference in relative abundance at class level (b) and at OTU level (c) from two approaches. Full scientific names of these OTUs are listed in Table 1.

Unexpectedly, ITS amplicon data did not cover the whole range of culture-derived species richness even though it showed higher OTU richness. Notably, the OTU richness in culture-based analysis was not sufficiently high: the rarefaction curve showed that our sampling effort (isolation depth) did not saturate isolated OTU richness (Figure 5). OTU richness was not saturated when the number of isolates per thallus exceeded approximately 1000, which was computed by extrapolation excluding thallus C (Figure 5(a)). Generally, OTU richness per thallus segment showed that OTU richness was not saturated at the isolation depth (Figure 5(b)).

Figure 5.

Rarefaction curves of ELF isolates. Saturation pattern of OTU richness was computed based on the number of isolates per thallus (a) and per segment (b). The 95% confidence intervals of OTU richness calculated on the bootstrap method had been shaded.

ELF community similarity in metabarcoding analysis

To investigate how the diversity of ELF communities changed as the distance between lichen segments increased, we conducted linear regression analysis (Figure 6(a)). Generally, ELF community dissimilarity, based on Bray-Curtis distance, increased with segment distance (F1,28 = 1.05, r2 = 0.01, p = 0.31). We also compared ELF community dissimilarities between thalli (inter-thallus) and within thalli (intra-thallus) (Figure 6(b)). Intra-thallus ELF community dissimilarity calculated by Bray-Curtis distance (0.34 ± 0.18) was significantly smaller than inter-thallus dissimilarity (0.42 ± 0.15) (Mann–Whitney, p < 0.005). In addition, NMDS ordination based on Bray-Curtis dissimilarity revealed that ELF communities were clustered by host lichen thalli (Figure 6(c)).

Figure 6.

The ELF community similarity in metabarcoding analysis. (a) Linear regression analysis showed the relationship between the distance of lichen segment and the ELF community dissimilarity. (b) Differences of community dissimilarity between intra- and inter-thallus were visualized on the violin plot. (c) NMDS ordination showed the clustering pattern of ELF communities with ellipses representing the 90% confidence interval of multivariate distribution. All the dissimilarities were calculated in Bray–Curtis distance. ***p < 0.005.

Saturation degree of OTU richness and sample coverage in the ELF community in metabarcoding analysis

Saturation degree of OTU richness and sample coverage was investigated to find out how many lichen segments were required for covering sufficiently diverse taxa of ELF and analyzing ELF community in a single lichen thallus. The rarefaction curve of OTU richness showed that the richness was saturated when the number of sequence reads reached approximately 100,000 in thallus A and 50,000 in thalli B and C (Figure 7(a)). Coverage-based rarefaction sampling curves compute diversity estimates for samples with sample completeness as measured by sample coverage [66]. In our case, it showed a saturation pattern when the number of reads was > ∼6000 (Figure 7(b)), which indicated that although OTU richness was not saturated when the number of reads was ∼6000, those uncaptured OTUs accounted for a small portion of overall ELF communities. Therefore, we regarded at least 50,000 reads per lichen thallus sufficient to observe highly diverse taxa of ELF. Consequently, 6000 reads per thallus was considered sufficient to observe general diversity of ELF community, and for community analysis.

Figure 7.

Saturation pattern of OTU richness and sample coverage in the ELF community in metabarcoding analysis. Rarefaction curves were computed using two diversity indices: OTU richness (a) and sample coverage (b).

Discussion

We compared ELF communities utilizing culture-based analysis and metabarcoding. ELF diversities obtained were remarkably different using these two approaches, especially in terms of alpha diversity and taxonomic proportions. Metabarcoding showed that ELF communities were quite different in lichen segment pairs and the presence of intra- and inter-thallus differences. In addition, we identified the number of sequence reads and the number of lichen segments that would saturate OTU richness and sample coverage.

Comparison of ELF diversities in metabarcoding- and in culture-based analyses

ELF diversities in metabarcoding and in culture analyses were different. The high-throughput amplification provided by metabarcoding enabled us to observe massive OTU richness [67] in ELF mycobiota, and yet this did not cover the range of OTU richness observed by culture. This result is in accord with the findings of previous studies that investigated OTU diversities as determined by metabarcoding and cultivation [68,69]. We believe this was caused by amplification bias during metabarcoding [70], i.e. it seems that amplification of the ITS1 region of several ELF species failed because of relatively low DNA complementarity between primers and DNAs of uncaptured ELF. The use of sole ITS1-derived diversity results would also cause this disparity between metabarcoding and culture-derived data, as amplification of DNA using the ITS1 and ITS2 regions has been previously reported to result in OTU disparities [68,71,72].

To the best of our knowledge, this study had the largest isolation depth for ELF cultivation and reports the highest culture-derived ELF richness per host lichen thallus. As the result of high isolation depth, we isolated two Sarea species belong to Sareomycetes from the lichen thalli at the first time. However, as mentioned above, the unsaturated rarefaction curve of ELF isolates suggested that diverse strains may have been undetected. Several efforts have been made to isolate diverse ELF. We previously reported the use of small lichen segments increased the richness of ELF isolates because it prevented fast-growing fungal hyphae covering slow-growing fungal colonies [40]. Moreover, Muggia et al. revealed that individual ELF had different optimum incubation conditions in terms of nutrients and pH [73]. Further studies on the isolation of highly diverse ELF are needed.

The taxonomic proportions of ELF biota estimated in metabarcoding and culture analyses were notably different. We believe it is likely that many filamentous fungi belonging to Sordariomycetes prefer the isolation conditions used in the present study than ELF belonging to Dothideomycetes, which constituted the largest proportion by metabarcoding. This accorded with previous reports on the cultivation of ELF that reported Sordariomycetes was the most dominant taxon. Notably, we isolated Capnodiales sp. as an abundant strain by metabarcoding though its isolation frequency was quite low. Capnodiales is considered a core taxon in ELF communities [33]. It would seem that its viability in harsh environments [74,75] enabled it to proliferate successfully inside lichen thalli, which contains various antifungal substances [76,77]. However, few reports have been issued on fungi belonging to Capnodiales isolated as ELF. Muggia et al. successfully isolated several ELF species belonging to Capnodiales utilizing different cultivation strategies [73]. The use of such approaches would be expected to result in ELF taxonomic compositions more resemble metabarcoding results.

ELF community and the origin of ELF

Intra- and inter-thallus ELF community similarities were found to differ in P. tinctorum thalli, ELF biota dissimilarity increased with distance between lichen segments, which caused us to speculate on the origin of ELF. It is generally considered that ELF maintain proliferation by vertical and/or horizontal transmission [12], and our results seem to reflect an aspect of horizontal transmission. Because if ELF community is greatly affected by vertical transmission, ELF community dissimilarities would be constant regardless of distances between lichen segments. Furthermore, considerable numbers of ELF have been reported to be endophytes [9,12,13,34,40]. Hence, horizontal transmission would be reasonable that several endophytes invaded lichen thalli after epiphytic lichen had attached to tree bark. However, no scientific evidence is available regarding the transmission type of ELF. To identify the origin of ELF, comparative investigations on fungal diversity in lichen thalli and surrounding environments are required.

Perspective for further ELF community studies using metabarcoding

In statistical or ecological studies, experimental replication is required to reflect the characteristics of populations [78,79]. To cope with sample variations, we suggest multiple lichen colonies be analyzed per collection site because as the present study shows, inter-thallus ELF community dissimilarities were remarkably greater than intra-thallus dissimilarities. Furthermore, for detailed investigations using several segments of lichen thallus, analysis of multiple segments with a distance would be appropriate because the ELF community was different as the distance between the lichen segments increased. Metabarcoding technique is utilized for identifying major or core taxa in hosts or environments and detecting unique taxa by high-throughput sequencing. In this study, a single segment of lichen thallus (1 cm2) was sufficient for prediction of overall ELF community diversity and four lichen segments per thallus were sufficient to observe rare taxa. Lichens exhibit relatively low growth rates [80,81], and the lack of thallus biomass has limited several studies. Because overall diversity of the ELF community was successfully analyzed by saturated sample coverage of only 6000 sequence reads and thus, small segment of lichen thallus is required for metabarcoding study.