Spatial turnover of fungi and partner choice shape mycorrhizal networks in epiphytic orchids

Most vascular plants harbour mutualistic fungi in their roots, forming associations called mycorrhizas (Brundrett & Tedersoo, 2018; Van Der Heijden et al., 2015). Mycorrhizas are chimeric organs, composed of both plant tissues and fungal hyphae, where nutrient exchanges can occur directions (Smith Read, 2008). While provide photosynthetic acquired carbon to partners, the latter improve hydro-mineral nutrition host enhance defence against biotic abiotic stresses. As mycorrhizal associate with same (Beiler 2010) or different species (Richard 2005; Selosse 2002), they known form complex networks interactions roots soils (e.g. Jacquemyn 2015; Rog 2020; Toju 2014, 2016). Plant–fungus be classified into two categories: ecological networks, which operational taxonomic units (OTUs) interact that do not obligately co-occur, meaning necessarily link a given site, physical single mycelia simultaneously several individuals. Regarding mycorrhizas, have often been referred as common (CMNs; Simard Durall, 2004; Horton, 2009). Because involve contacts between mediated by such lead fungal-mediated plant–plant interactions. First, connected fungus may asymmetrically contribute its need, resulting benefit for less providing one (Walder 2012). Second, nutrients information (Klein 2016; Roy, 2009; 1997, 2015) even though consequences always easy experiment observe (Booth, Liang 2020). interaction mainly studied soils, especially temperate ecosystems. By contrast, tropical ecosystems largely overlooked, when considering epiphytic habitat grow on tree bark twigs. Indeed, epiphytes represent 9% world-wide (Zotz, 2016), reaching up 25% (Nieder 2001). Notably, these undergo transient water limitation rainfall events (Yoder 2010; Zotz, mediating and/or transfers co-occurring plants, formed could thus highly advantageous (Leake Cameron, 2012) also previously suggested terrestrial mutualisms (Bingham Simard, 2011). Epiphytic were long considered lack 2018) but many works now demonstrating types compared counterparts, frequent Kottke Martos 2012; Petrolli 2021; Rowe Pringle, 2005). Particularly, 70% members Orchidaceae family This comprises no than 28,000 (Christenhusz Byng, >80% (Givnish Orchids major 2018), orchid endomycorrhiza, is constantly found throughout (Selosse 2022; Smith 2008; Wang 2021) well orchids (Dearnaley 2012, The orchid–fungus symbiosis atypical because it takes intracellular hyphal structures pelotons, partners required at each step life cycle from symbiotic germination dust seeds colonization adult (Rasmussen, 1995; Orchid (OMF) polyphyletic group mostly Basidiomycota and, lesser extent, Ascomycota 2021). In green orchids, main belong historically ‘rhizoctonias’ 2011) include Cantharellales (namely Ceratobasidiaceae Tulasnellaceae; Veldre 2013) Sebacinales Serendipitaceae; Weiß On bark, seem limited range including rhizoctonias (Herrera 2019; Izuddin 2021), other families Mycena (Zhang 2012), Fusarium (Jiang 2019) some Pucciniomycota belonging Atractiellales 2010). Since require germination, sometimes ‘take pay later’, first receive during subsequently reward stage. Thus, CMNs particularly relevant adults germinating co-occur (Jacquemyn Orchid-fungal investigated (reviewed Li, Shimao, is, large scale instead individuals, ignoring direct, Herrera 2008, 2013; Xing limiting individuals growing trees hardly share hyphae. Yet, indirect interactions, transfers, only shared among plants. recent investigation communities Brazilian we communities, OMF, extend away forage surrounding rhizosphere (Petrolli More importantly, OMF species, suggesting existence tree. However, results supported few number trees. Additionally, previous focused (merging trees; see references above), factors shaping plant–fungal still poorly understood. whether shaped identity (i.e. fungi), spatial distribution spatially close similar regardless species) stochastic determinants remains question. Using explicit sampling design, six trunks rainforest La Réunion (French oversea island Indian Ocean) colonized 10 species. Our working hypotheses that, close, (i) (ii) symbionts, should prevail determining plant–fungus stressing habitat. (iii) ‘seedlings facilitation’ habitat, growth stages an sustain adjacent seedlings. study was conducted February 2020 well-preserved fragment lowland (Mare-longue, 21°21?14?S 55°44?29? E, elevation 205–560 m; Strasberg, 1994) under permission DIR-I-2020-013. Six trees, three Labourdonnaisia calophylloides Bojer (Sapotaceae) Nuxia verticillata Lam. (Stilbaceae; Table 1), selected according diameter (approx. 20 cm) breast height coexistence least trunk. L. calophylloides, dominant this forest, has straight trunk cracked whereas N. (also common) twisted, crevassed leafy exfoliating (Figure S1). Both endemic Mascarene islands shown 15 14 respectively survey 1-ha plot forest (Lancaster, 2004). (labelled A–F), harboured total epidendroid genera morphologically identified (Table 1; S1), except protocorms F later DNA analysis (see below). All listed ‘Least Concern’ ‘Data Deficient’ IUCN, Beclardia macrostachya (Thouars) A. Rich ‘Near Threatened’ (UICN France, CBNM, FCBN MNHN, 2013). trunk, position sample (roots, bark) precisely determined coordinate system comprising c. 1 3 m horizontal distance vertical reference line, using measuring tape Yellow–orange portions (c. 2 contact (Martos sampled sterile razor blades leaving alive Their surface sterilized ethanol, then quickly dried silica gel. Each root assigned stage: protocorm, leaf; stage 2, 3, young seedling (not age flowering); 4, S2). Plant counted half sympodial allowing vegetative multiplication. addition seedlings, preserved 50% ethanol 4°C distribute around collected 50 mg outer samples regular intervals blade. These immediately individual bags gel, preventing any cross-contamination samples. end, 341 sections, 7 protocorms, 105 pieces 453 samples, used molecular analyses. powdered TissueLyser II (Qiagen) tungsten beads per tube, genomic extracted DNeasy® Mini Kit NucleoSpin Soil (Macherey-Nagel, GmbH Co KG.) respectively. Polyvinylpolypyrrolidone (PVPP) added cell lysis precipitate tannins. Due small size, ground manually buffer pellet pestles, roots. amplified PCR sets primers targeting internal transcribed spacer (ITS2) nuclear ribosomal provides more suitable ITS1 (Yang namely widely ITS86-F/ITS4 (White 1990) amplifying Asco- Tulasnellaceae (Rammitsu 5.8S-OF/ITS4-Tul (Vogt-Schilb 2020) specifically amplify Tulasnellaceae. amplification forward reverse tagged experimental design included negative (ultrapure water) mock community controls plate, 36 positive total. We tagging assess trials, library preparation sequencing (Hornung Zinger 2019). consisted defined mixture strains, organization plates randomized. performed 25 ?l containing 0.2 mM dNTP, 0.4 ?M primer, unit AmpliTaq Gold® 360 Polymerase (Thermo Fisher Scientific), 1X Buffer supplied MgCL2 1.5 template DNA, following program: initial denaturation min 95°C; 35 cycles 30s 95°C, annealing 56.5°C (ITS86-F/ITS4) 52°C (5.8S-OF/ITS4-tul), elongation 72°C; final 12 72°C. Whenever did universal primer set (47 samples), nested protocol ITS1/TW13 follows—initial 55°C 72°C—then above. trial replicated twice, pooled loaded agarose gel visualization amplicons. amplicon purified NucleoMag® NGS Clean-up Size Select (Macherey-Nagel), quantified fluorescence Qubit™ dsDNA High-Sensitivity (Invitrogen™), mixed equimolarly amplicons four pools. pools sequenced × 250 bp paired-end runs Illumina MiSeq platform Fasteris (Geneva, Switzerland). For identification seminested PCRs ITS1P/ITS4 (Ridgway 2003) ITS1/ITS4 1990), Sanger method Eurofins Genomics (Germany). obtained sequences aligned retrieved GenBank MAFTT (Katoh phylogenetic inference ML IQTREE 2.1.3 (Minh comparison, default parameters. Paired reads processed customed pipeline based VSEARCH (Rognes 2016) Perez-Lamarque, Krehenwinkel, al. (2022). Briefly, assembled fastq_mergepairs function parameters errors alignment discarded. tag+primer CUTADAPT (Martin, zero discrepancies. Tag jumps (Zinger estimated represented 0.04% reads. Operational clustered 97% similarity threshold algorithm. Chimeras detected uchime_denovo option removed dataset. OTUs taxonomically usearch_global UNITE V8.3 (2021-05-10) eukaryotes database (Nilsson Singletons short (< 200 bp) before OTU table building. Nonfungal abnormally discarded dataset, contaminants decontam r-package frequency (Davis 2018). Finally, tag-jumps rate, <0.5% (hereafter referenced bioinformatics threshold) 0, hence minimizing impact artefactual tag jumps. Samples <1000 without subsequent pairs varying specificity, read counts comparable therefore, presence–absence (1/0) colonizing 368 remaining statistical analyses (except Figure Figures S3, S15 all analysed). low abundances over-represented data, repeated excluding whose per-sample relative abundance ?10% (instead 0.5% mentioned confirm robustness our potential mid-to-low-abundance errors. Amplicon sequence variant (ASV) construction methods developed delineation microbial taxa reproducibility metabarcoding although relevance comparison questioned ITS regions Dikarya (Tedersoo 2022). therefore analyses, network below), most abundant unique (up five) inside OTU. finer 100%-similarity OTUs, aimed haplotype beta diversity vegan (Oksanen 2013), Jaccard index implemented vegdist function. To differences composition substrates vs. bark), overall nonmetric multidimensional scaling (NMDS) approach metaMDS tested PERMANOVA, 999 permutations, adonis grew area, permutations constrained strata adonis. test effect within PERMANOVA (A–F) explanatory variable strata. Bipartite describe analyse patterns groups organisms, example, 2011; usually nonrandom, rather modular structure (Chagnon network, generalist specialist little reciprocal specificity (Bascompte 2003). Nestedness feature supposed confer stability (Thébault Fontaine, 2010), result neutral processes (Krishna however, modules (Olesen 2007). Modularity indicate presence specialization parasitic co-evolution), environmental constraints individual- versus species-level influence network's Olesen order mycelia, matrices independently, corresponding studied. Binary (rows) (columns; i.e. bipartite networks), after low-count occurrences filtering described above, inputs From matrix, evaluated nestedness modularity observed significance 1000 null models reconstructed permatswap r-package. calculated (Dormann 2008), ‘NODF2’ method. ‘Q’ computeModule package (Newman) steps. module detection slightly vary calculation due randomized procedure, 100 times kept reliability module, searched how additional calculation, p-values accordingly them. distances coordinates taken field, smallest them cylindric trunks. separated branches located passing through fork. orchid–OMF interaction, autocorrelations analysed mantel correlograms vegan, ‘spearman’ method, permutations. corroborate Atlantic Forest decay above reference. full component proxy illustrations), principal neighbour (PCNM) approach. PCNM allows reduction vectors eigenvectors. eigenvectors computed pcnm package. Eigenvectors axes) associated eigenvalues submitted selection (Blanchet 2008) code provided (Bauman Variance partitioning varpart r-package, second variable. Associated redundancy (RDAs) cited variables, followed ‘anova’ either variables communities. if module. so, 10,000 names matrix mean (mD) (sC) original matrices. proportion mDrandomized < mDoriginal sCrandomized = sCoriginal. A 25,806,753 24,989,711 raw 32,593,936 selective process 4564 expected controls. After OTUs' representing sample, belonged community. Applying stringent curation data led 794 27,595,918 reads). pair 57 ITS86-F/ITS4. Despite fine height, rarefaction curves reach plateau, neither nor S3). Ninety-one (11.5%) (57 OTUs) (3 Serendipitaceae (31 Sebacinales, average 12% root-associated 1C). Other putative Atractiellomycetes (two OTUs), (one OTU), too occasional key here guilds, 703 non-OMF saprotrophs (13.6%), pathogens (7.7%) lichenized (4.8%; S4A). Remaining (62.5%) unambiguously trophic guild insufficiently resolution. almost S4B). equally distributed 2). Conversely, notably macrostachya. significantly differed (nested PERMANOVA; 4.53, R2 0.05, p 0.001; S5). varied across (F 47.27, 0.11, 0.001) 35.87, 0.32, 0.001, Considering only, 38.18, 0.33, S6) 52.45, 0.12, 0.001). They controlling 7.99, 0.25, while 25.98, e.g. Polystachya concreta S6). sharing high level OTU's clearly visible TUL1 SER21 (six each) CER4 (Figures 6). very consistent showing significant antinestedness S8). change S9). Modules OMFs involved (29.6%), (37.0%), (25.9%) >4 (7.4%) S10). Similarly, (41.7%), (50.0%) (8.3%) families. individual, autocorrelation ranging 90 cm 4; S11). suggests structured, over (~20 cm). addition, visualizations 3; approaches quantify tendencies. scale, variance revealed tree, (as assessed PCNMs) had 5; contribution networks' comparing sized Results (synthetized S3) showed alone, explained 22.2% 11.1% modules. interplay components 37.0% Accordingly, closer OTU, S12). tended S13) intensively chance colonize finally greater distances, S14). reveals effectively expansion rhizosphere. seedlings successfully cucullatum. fungi, protocorm S15), cucullatum additionally When looking haplotypes, 6; S16)