Two Widespread Green Neottia Species (Orchidaceae) Show Mycorrhizal Preference for Sebacinales in Various Habitats and Ontogenetic Stages

Molecular Ecology (2015) 24, 1122–1134 doi: 10.1111/mec.13088

Two widespread green Neottia species (Orchidaceae) show mycorrhizal preference for Sebacinales in various habitats and ontogenetic stages

1 TAMARA TESITELOVA,* MILAN KOTILINEK,* JANA JERSAKOVA,* FRANCß OIS-XAVIER JOLY,* JIRIKOSNAR,* IRINA TATARENKO†‡ and MARC-ANDRE SELOSSE§ *Faculty of Science, University of South Bohemia, Branisovska 31, 37005 Ceske Budejovice, Czech Republic, †Moscow Pedagogic State University, 1/1 M. Pirogovskaya Str., Moscow 119991, Russia, ‡Department of Environment, Earth and Ecosystems, Open University, Walton Hall, Milton Keynes MK7 6AA, UK, §Departement Systematique et Evolution (UMR 7205 ISYEB), Museum national d’Histoire naturelle, CP 50, 45 rue Buffon, 75005 Paris, France

Abstract Plant dependence on fungal carbon (mycoheterotrophy) evolved repeatedly. In orchids, it is connected with a mycorrhizal shift from rhizoctonia to ectomycorrhizal fungi and a high natural 13C and 15N abundance. Some green relatives of mycoheterotrophic species show identical trends, but most of these remain unstudied, blurring our understanding of evolution to mycoheterotrophy. We analysed mycorrhizal associations and 13C and 15N biomass content in two green species, Neottia ovata and N. cordata (tribe Neottieae), from a genus comprising green and nongreen (mycoheterotrophic) species. Our study covered 41 European sites, including different meadow and forest habitats and orchid developmental stages. Fungal ITS barcoding and electron microscopy showed that both Neottia species associated mainly with nonectomycorrhizal Sebacinales Clade B, a group of rhizoctonia symbionts of green orchids, regardless of the habitat or growth stage. Few additional rhizoctonias from Ceratobasidiaceae and Tulasnellaceae, and ectomycorrhizal fungi were detected. Isotope abundances did not detect carbon gain from the ectomycorrhizal fungi, suggesting a usual nutrition of rhizoctonia-associated green orchids. Con- sidering associations of related partially or fully mycoheterotrophic species such as Neottia camtschatea or N. nidus-avis with ectomycorrhizal Sebacinales Clade A, we pro- pose that the genus Neottia displays a mycorrhizal preference for Sebacinales and that the association with nonectomycorrhizal Sebacinales Clade B is likely ancestral. Such a change in preference for mycorrhizal associates differing in ecology within the same fungal taxon is rare among orchids. Moreover, the existence of rhizoctonia-associated Neot- tia spp. challenges the shift to ectomycorrhizal fungi as an ancestral pre-adaptation to mycoheterotrophy in the whole Neottieae.

Keywords: germination, mixotrophy, Neottieae, orchid mycorrhiza, Orchidaceae, Sebacinales Received 21 May 2014; revision received 16 January 2015; accepted 19 January 2015

fungi (Leake 1994): such mycoheterotrophic (MH) spe- Introduction cies evolved repeatedly, with ca. 45 independent origins Some achlorophyllous heterotrophic plants acquire all in both monocots and eudicots, and involve all mycor- nutrients, including carbon, from their mycorrhizal rhiza-forming fungal taxa (Glomeromycota, Basidiomy- cota and Ascomycota), as well as some saprobic fungi (Merckx 2013). The level of MH nutrition may vary, and Correspondence: Tamara Tesitelova, Fax: (+420)385 310 366; E-mail: [email protected] species that combine MH with autotrophic nutrition in 1Present address: Centre d’Ecologie Fonctionnelle et Evolutive, an intermediate strategy called partial mycoheterotrophy CNRS, 1919 route de Mende, 34293 Montpellier Cedex 5, France (PMH, a kind of mixotrophy) have been discovered

© 2015 John Wiley & Sons Ltd MYCORRHIZAL PREFERENCES IN NEOTTIA 1123 among relatives of MH species (Gebauer & Meyer 2003; is hitherto known: while the physiological reasons Julou et al. 2005; Tedersoo et al. 2007). Phylogenies of remain unclear, ECM fungi may offer a more continu- taxa with MH and PMH species, where the latter usually ous and massive C supply (e.g. Roy et al. 2013). occupy basal positions, support a gradual shift to full In orchids, a sequential evolution to PMH nutrition MH via PMH (Selosse & Roy 2009; Ogura-Tsujita et al. with mostly nonrhizoctonia ECM fungi, followed by evo- 2012), albeit this requires several steps (Roy et al. 2013; lution to full MH nutrition, occurred in the genus Cymbid- Gonneau et al. 2014). ium (where the scenario is phylogenetically supported; With more than 25 independent origins of MH, orchids Motomura et al. 2010; Ogura-Tsujita et al. 2012), possibly encompass nearly 50% of all MH species (Merckx 2013) in the genus Platanthera (Yagame et al. 2012), and in the and provide useful models to investigate MH diversity tribe Neottieae whose five genera include many (P)MH and evolution. Production of vast amounts of dustlike species and attracted numerous recent works (e.g. Selosse seeds that germinate using a transient MH nutrition et al. 2002b; Bidartondo et al. 2004; Julou et al. 2005; Roy (even in species that turn green at adulthood) is typical et al. 2009; Tesitelova et al. 2012). The common Neottieae for the whole family. The ancestral mycorrhizal associa- ancestor is sometimes assumed to have shifted to nonrhi- tion, typically found in both green adults and their zoctonia ECM fungi as a predisposition of the whole tribe underground MH seedlings (protocorms), is formed by to be PMH and repeatedly evolve MH species (Selosse & rhizoctonias, a polyphyletic group of basidiomycetes Roy 2009). However, some species in the genera Neottia from Tulasnellaceae, Ceratobasidiaceae and Sebacinales and Epipactis may associate with rhizoctonias and display Clade B that are saprotrophic and/or endophytic on non- no isotopic enrichment (e.g. Bidartondo et al. 2004). In orchid plants (Dearnaley et al. 2013). A derived condition spite of their relevance to understanding evolution of MH is found in some orchid species that shifted mycorrhizal at fine phylogenetic scale in Neottieae, they were poorly association from rhizoctonias to various other basidiomy- investigated. In the genus Neottia where the European cetes either purely saprotrophic or ectomycorrhizal MH N. nidus–avis associated with ECM fungi from Sebac- (ECM; Dearnaley et al. 2013) and at same time display inales Clade A is well studied (McKendrick et al. 2002; MH or PMH nutrition (Merckx 2013). Selosse et al. 2002b; Liebel et al. 2010), the two sympatric Fungi deriving nutrients from different resources dif- green species N. ovata and N. cordata have unknown fer in stable isotope 13C and 15N content, so that MH mycorrhizal symbionts (except a Tulasnella-like strain in plants acquiring nutrients from different functional N. ovata, Rasmussen et al. 1991). Stable isotope content is groups of fungi can be distinguished isotopically when- unknown in N. cordata.InN. ovata, it ranges from similar ever the isotopic signature differs from autotrophic to autotrophic plants to enriched as in ECM-associated plants (e.g. Gebauer & Meyer 2003; Taylor et al. 2003). PMH depending on the study site (Gebauer & Meyer Indeed, MH species associated with ECM or sapro- 2003; Abadie et al. 2006; Tedersoo et al. 2007; Liebel et al. trophic fungi are enriched in 13C and 15N as compared 2010). with fully photosynthetic plants, while PMH species Here, we first investigated mycorrhizal fungi of N. ov- have 13C and 15N content intermediate between these ata and N. cordata at multiple sites and additionally one two types depending on MH level and identity of individual of greenish but leafless Asiatic N. camtschatea. mycorrhizal fungi (e.g. Bidartondo et al. 2004; Julou As mycorrhizal fungi may vary between sites and during et al. 2005; Selosse & Roy 2009; Stockel€ et al. 2014). Par- ontogenesis, fungal associates were identified (i) both in tially and fully MH species also feature higher N con- germinating (for N. ovata) and adult individuals and (ii) centration (Selosse & Roy 2009; Merckx 2013; Gonneau from diverse habitats across Europe. Neottia ovata dis- et al. 2014). The repeated evolution of (P)MH plants plays a large ecological range; thus, we sampled its two among orchids prompted questions about rhizoctonia- extreme habitats (open meadows and forests). Second, we associated species. On the one hand, they are often investigated 13C and 15N content in N. ovata and N. corda- assumed autotrophic, based on their 13C and 15N con- ta at two forest sites per species to see whether they obtain tent similar to that of autotrophic plants (e.g. Gebauer nutrients from ECM fungi that would increase their isoto- & Meyer 2003; Dearnaley et al. 2013). On the other pic signature. Finally, we analysed phylogenetic relation- hand, their MH seedlings are also hardly isotopically ships among available Neottia species. distinguishable from autotrophs (Stockel€ et al. 2014), which prompted the recent hypothesis that some adult Materials and methods orchids could receive carbon from rhizoctonias without noticeable change in their stable isotope content (Selosse Study species and sample collection & Martos 2014; Stockel€ et al. 2014). Despite uncertainties on the plesiomorphic nutrition in rhizoctonia-associated We focused on two green Neottia species (Neottieae orchids, no fully MH orchid associated with rhizoctonias tribe) formerly placed in the genus Listera that never

© 2015 John Wiley & Sons Ltd 1124 T. TESITELOVA ET AL. occur sympatrically. Neottia ovata (L.) Bluff & Fingerh. 35-mm plastic slide mounts following Rasmussen & is a common allogamous Eurasian species from open Whigham (1993). Twelve to eighteen packets with seeds forests and meadows. Neottia cordata (L.) Rich. is a tiny of each N. ovata seed mixture were buried at each of allogamous species with circumboreal distribution in the 18 sites where seeds were collected (i.e. 24 to 36 wet acidic soils of open bog spruce forests. Four to six packets per site). Only four packets with N. cordata roots per plant from up to four plants per population seeds were placed at a subset of sites because of the were collected in summer from 2009 to 2012 from 13 few available seeds (see Table S2, Supporting informa- meadow and 15 forest sites for N. ovata, and 13 forest tion for sowing design). In September 2008, seed pack- sites of N. cordata distributed across Europe (see Table ets were placed 5 cm deep in soil close to adults within S1, Supporting information). In addition, roots of one four to six microsites at each site. In 2009, the same individual of greenish but leaﬂess, and likely PMH, N. design was repeated with seeds collected in 2009, with camtschatea (L.) Rchb.f., a central Asiatic Siberian forest the exception of (i) an increase in the number of seeds species, were collected in 2008 in Altai (Russia). Roots per packet to ca. 135 for N. cordata and 185 for N. ovata were washed and processed immediately or stored in and (ii) numbers of packets per site (Table S2, Support- 60% ethanol before inspection by light microscopy. Four ing information). In total, 1246 packets buried in 2008 mycorrhizal root sections (from as many roots as possi- and 2009 were retrieved in both 2010 and 2011, and ble) per plant were pooled for molecular analyses (one kept moist at 4 °C until processing within 4 days. Size pool per sampled individual); 3–6 fresh mycorrhizal of the largest protocorm and the number of protocorms root pieces from one or two plants from two N. ovata per packet were used in statistical analyses. A total of and four N. cordata sites (in total 30 root pieces) were 38 vital protocorms of N. ovata from 28 different packets also harvested for transmission electron microscopy from seven sites were stored at 20 °C for molecular (TEM, Table S1, Supporting information). analysis.

TEM analysis Fungal barcoding To investigate fungal colonization, the root pieces were Total DNA from roots and protocorms was extracted fixed in cacodylate-buffered 2.5% glutaraldehyde and using the Invisorb Spin Plant Mini Kit (Invitek, Berlin, processed in the Laboratory of Electron Microscopy at Germany). Internal transcribed spacer (ITS) of nuclear the Biology Center of the Academy of Sciences of the ribosomal DNA (nrDNA) was amplified using primers Czech Republic in Ceske Budejovice. Briefly, the sam- ITS1F and ITS4 (pair #1, specific to fungi). Weak PCR ples were rinsed in 0.1 M cacodylate buffer, postfixed in products were used as a template for a second PCR 1% osmium tetroxide for 4 h, dehydrated in an ascend- using primers ITS1 and ITS4 as in Tesitelova et al. ing acetone series and embedded in Spurr resin. Sam- (2013). All extracts were also amplified with primers ples showing vital pelotons on semi-thin sections were ITS1OF and ITS4Tul (pair #2) to detect additional Tulas- selected for ultrathin cutting. Sections were stained with nella lineages (Taylor & McCormick 2008) and primers uranyl acetate and lead citrate and examined in JEOL ITS3Seb and TW13 (pair #3) to obtain sebacinoid ITS2 JEM 1010 electron microscope operating at 80 kV. and part of the large subunit (28S) of nrDNA (Selosse et al. 2007). All PCR products obtained with pair #1 and double-banded PCR products obtained with pairs #2 Seed sowing and #3 were cloned using pGEM-T Vector System I To investigate fungi triggering germination, ripe, natu- (Promega, Madison, WI, USA). At least six positive rally pollinated capsules were collected in July 2008 clones per cloning and unique products from PCR #2 from ten N. ovata adults (ten capsules per plant) from and #3 were sequenced as in Tesitelova et al. (2012). six meadow and six forest sites, and from 10–20 N. cor- Operational taxonomic units (OTUs) were delimited data adults (five capsules per plant) from six forest sites based on a 97% similarity threshold over ITS using TO- (Table S1, Supporting information). The seeds from each PALI v2.5 (Milne et al. 2004). Putative fungal species group of six sites were pooled in order to avoid a geno- identity and ecology were assigned using BLASTn type effect due to different parental origins (i.e. three search against the NCBI database (http:// seed mixtures in total), that is N. cordata, N. ovata from www.ncbi.nlm.nih.gov). One representative per OTU meadows and N. ovata from forests, so that any ecotype and site was deposited in GenBank. Only OTUs from effect could be assessed in N. ovata. Seeds were kept at rhizoctonia or ECM taxa were considered potentially 4 °C before sowing. About 60 seeds of N. cordata and 85 mycorrhizal, based on current knowledge of orchid seeds of N. ovata were placed on a 42-lm nylon mesh mycorrhizal fungi (see Dearnaley et al. 2013), and used (Silk and Progress Ltd., Brnenec, Czech Republic) in for statistical analyses.

Plant phylogenetic markers single analysis under a partitioned model with Cepha- lanthera longibracteata, C. damasonium and Epipactis at- To obtain the relative phylogenetic positions of the rorubens as out-groups (GenBank Accession nos used studied Neottia species with N. nidus-avis and N. smallii, are in Fig. S1, Supporting information). Neottia sequences were obtained from GenBank or amplified from roots (or scales in N. nidus-avis) using primers ITS1P and ITS4 for plant ITS region (Selosse Stable isotope analysis et al. 2002a), primers 18S-F (Medlin et al. 1988) and 18S- Samples for isotopic analysis were collected at two R (Xiao et al. 1999) for part of the 18S subunit of N. ovata (F3-S and F6-S) and two N. cordata forest sites nrDNA, primers for mitochondrial cytochrome oxidase (C1-S and C5-S; Table S1, Supporting information) in subunit 3 (cox3) by Duminil et al. (2002), and TabC and 2008. Five leaf samples from Neottia adults (differing TabD (Taberlet et al. 1991) for trnL(UAA) plastid intron. from those used for fungal barcoding), PMH orchid Epi- PCR conditions were as described in Tesitelova et al. pactis helleborine (only at F3-S site), which associates (2013), with the exception of annealing temperatures, with ECM fungi (Bidartondo et al. 2004), and two to that is 50 °C for 18S, 58 °C for cox3 and 52 °C for trnL three species of nonorchid autotrophic plants per site (UAA), and elongation increased to 120 s for 18S. were collected in early June (F3-S, F6-S, and C1-S, Neot- tia flowering) or in late July (C5-S, N. cordata fruiting). Phylogenetic analyses Because of their absence during plant collection, two to Phylogenetic relationships were analysed using Bayes- five sporocarps of up to three saprotrophic and ECM 0 ian inference. For Sebacinales, the conserved 5 part of fungal species were collected at the same locations in 28S nrDNA from one representative per OTU at each September 2008 (no sebacinoids were found). All leaves site and orchid developmental stage (protocorm/adult) were collected at similar light level (moderate shade) was used. The sequences obtained and their close BLAST and distance from soil (between 10 and 15 cm, but less matches, Sebacinales sequences from GenBank covering in N. cordata owing to its tine growth habit) to avoid the sequence diversity from both Clades A and B (Weiß isotopic bias due to photosynthesis rate and use of CO2 et al. 2011), and two out-group taxa, in total 150 from soil respiration. Samples were homogenized in 2- sequences, were aligned using algorithm E-INS-i in MA- mL Eppendorf tubes in a ball mill MM200 (Retsch FFT v6 (Katoh et al. 2005) and manually adjusted, yield- Gmbh, Haan, Germany). Total N concentrations (%) ing a 719-bp-long alignment. Model GTR with and contents of 13C and 15N were measured by an autocorrelated discrete gamma (AdGamma; Yang 1995) online continuous flux isotope ratio mass spectrometer was selected as the best substitution model by evaluat- (IRMS Delta S, Finnigan MAT, Bremen, Germany) at ing AIC of different models in KAKUSAN4 v2 (Tanabe the Technical Platform of Functional Ecology (OC081, 2011). Two runs for 5 9 106 generations with sampling INRA, Nancy, France). Standard deviations of repli- 13 every 1000th generation were conducted in MRBAYES cated standard samples were 0.040 for C and 0.208 for v3.1.2 (Huelsenbeck & Ronquist 2001) under the best 15N. model assumption. The first 25% of trees were dis- carded as a burn-in, and a majority-rule consensus tree Statistical analyses was built from the remaining trees. Branches with a Bayesian posterior probability (BPP) below 0.95 were EstimateS (Colwell 2006) was used to produce rarefac- regarded as poorly supported. For Tulasnellaceae phy- tion curves of mycorrhizal OTUs in N. cordata and logeny, sequences were selected in an analogous way N. ovata in order to test whether the whole diversity resulting in 50 sequences plus three out-group taxa. The was sampled. Differences in d13C, d15N content and N 0 conserved region of nrDNA 5.8S and a short 5 part of concentrations among species within sites were tested 28S nrDNA were aligned as described above, yielding a by one-way ANOVA followed by a post hoc Tukey HSD 258-bp-long alignment. As the best substitution model, test (a = 0.05) in R v3.0.2 (R Development Core Team HKY85 + AdGamma was selected and further analy- 2013). Species composition of mycorrhizal fungi in sed as described above. N. ovata and N. cordata was analysed using R package For Neottia phylogeny, only the parsimony-informa- vegan v2.0-9 (Oksanen et al. 2013) if not stated other- tive ITS, 18S and trnL(UAA) intron were used (cox3 wise. Each combination of site and developmental stage produced only a single parsimony-informative position (adult/protocorm) entered the analyses as an indepen- within our samples). Each gene was analysed separately dent observation. Occurrence of fungal OTUs at each as above. Because they produced nonconflicting topolo- site was summed over all investigated adults or proto- gies (<95% support in case of incongruence; Fig. S1, corms and replaced by presence–absence data because Supporting information), the loci were combined into a of unequal sampling effort per site (one to four adults

1 Photosynthetic, associated with Sebacinales Clade B 1 1

mycorrhizal fungi unknown

0.61 PMH or MH, 1 associated with Sebacinales Clade A

0.04

Fig. 1 Bayesian phylogenetic tree of Neottia genus based on a concatenation of ITS, 18S and trnL(UAA) intron, with photosynthetic Neottia ovata, N. cordata and N. smallii, putatively photosynthetic and likely partially mycoheterotrophic (PMH) N. camtschatea, and nonphotosynthetic mycoheterotrophic (MH) N. nidus-avis in adulthood. Bayesian posterior probabilities are denoted above the branches. The tree is rooted with Cephalanthera spp. and Epipactis atrorubens. or two to six protocorms). Sites F9, F11 and C12 were the effects of Neottia species and habitat type on OTU com- omitted from all analyses because of exclusive occur- position in adults. Distance-based RDAs (dbRDA) were rence of singleton OTUs (i.e. OTUs present only at one used to test the effects of habitat and Neottia species on dis- site). Singleton OTUs were excluded from all analyses tances inferred from an ultrametric phylogenetic tree of except for nonmetric multidimensional scaling (NMDS). Sebacinales. Significance of predictors was tested by Monte An NMDS analysis based on the Jaccard dissimilarity Carlo permutation tests with 999 permutations. matrix was used to visualize variation in mycorrhizal The effects of seed origin (seeds from meadow vs. for- fungi of Neottia from different habitats (meadow vs. forest N. ovata) and of seed sowing habitat (forest vs. mea- est). To take into account the spatial distribution of popu- dow) on the number and size of protocorms per seed lations, the pairwise distances between sites (site C9 was packet were analysed at the nine sites with protocorms omitted due to absence of coordinates) were obtained (Table S2, Supporting information) by means of linear using geographical distance matrix generator (http://bi- mixed-effect models with ‘site’ as a random factor. Both odiversityinformatics.amnh.org/open_source/gdmg/) numbers and sizes were log-transformed before the and transformed into spatial eigenvectors using principal analysis in R package NLME v3.1 (Pinheiro et al. 2013). coordinates of neighbourhood matrix (PCNM). Signifi- cant PCNM vectors for OTU composition were forward- Results selected at a = 0.05 with R package packfor v0.0-8 (Dray et al. 2013) and used in further analyses. Phylogenetic relationships between studied species Variation in all mycorrhizal OTU composition or specifi- cally Sebacinales OTU composition associated with Neottia The two leafless species (N. nidus-avis and N. camtschatea) adults attributable to orchid species identity, habitat type formed a well-supported monophyletic clade in single- (forest/meadow) or spatial PCNM vectors was partitioned (Fig. S1, Supporting information) and multilocus analyses using the function varpart. The variation partitioning was (Fig. 1). Neottia ovata placed at the base of the genus in the based on a redundancy analysis (RDA) with Hellinger multilocus analysis with low BPP. In the single-locus standardization of the data (Legendre & Gallagher 2001). analyses, the position of N. ovata was poorly supported Another Hellinger-standardized RDA was used to test the and varied from the base of the genus (18S), a polytomy effect of developmental stage (protocorms vs. adults; factor with the leafless and the other green Neottia species (trnL ‘site’ as a covariate) on mycorrhizal OTU composition at (UAA) intron) to the base of the clade with the leafless seven sites with analysed protocorms. Hellinger-standard- species (ITS; Fig. S1, Supporting information). Meadow ized RDA with PCNM vectors as covariates (to account for (n = 4) and forest (n = 2) N. ovata adults did not differ in the effect of geographical distance) was further used to test ITS.

(n = 87) (n = 26)

Thelephoraceae (1.6%, 2 OTU, n = 2) (0.8%, 1 OTU, n = 1) Russulaceae (0.8%, 1 OTU, n = 1) (11%, 1 OTU, n = 4) (2.5%, 2 OTU, n = 3) Atheliaceae (0.8%, 1 OTU, n = 1) Russulaceae Ceratobasidiaceae (2.5%, 1 OTU, n = 1) (2.5%, 3 OTU, n = 3) Ceratobasidiaceae (2.5%, 1 OTU, n = 1) Tulasnellaceae (15%, 9 OTU, n = 18) Tulasnellaceae (11%, 3 OTU, n = 4)

Sebacinales Sebacinales (76%, 28 OTU, n = 87) (73%, 7 OTU, n = 26)

Fig. 2 Incidence of rhizoctonia and ectomycorrhizal fungi in Neottia ovata (adults and protocorms) and N. cordata (adults); n, number of individuals where the taxon was found.

Germination course and identity of fungal symbionts populations). In N. camtschatea, an OTU related to the ECM genus Geopora was found along with a Sebacinales Seeds of N. cordata showed minimal germination (< OTU. TEM exclusively revealed septal pores typical of 0.2%), very high mortality (> 98%) already after one Sebacinales in root pelotons, displaying imperforate pa- year in soil (over all sites) and no protocorms even at renthesomes with straight margins and electron-trans- sites with N. cordata adults. In N. ovata, maximum ger- parent layers surrounding an electron-dense layer mination (on average 8.5% of germinating nonmycorrhi- (Fig. 3). zal seeds and mycorrhizal protocorms over all sites) Phylogenetic analysis placed the Sebacinales OTU was observed in 2010 after 2 years in soil with 66 vital from N. camtschatea within Clade A (Figs 4, S3, Support- and 8 decayed protocorms (in 29 of 312 packets) devel- ing information) close to symbionts from MH orchids oped only at sites with N. ovata adults (see Table S2, Neottia nidus-avis and Hexalectris spicata. Conversely, most Supporting information for details). OTUs from N. cordata and N. ovata (protocorms and Fungi were analysed in 62 adults and 38 protocorms of adults) positioned in Clade B with no preference for any N. ovata from 28 and seven sites, respectively, 26 adults specific subclade (Fig. 4). OTUs SEB 19, 27 and 31 clus- of N. cordata from 13 sites and one adult of N. camtscha- tered at the base of Clade A with low support (BPP = tea. Amplification succeeded in all adult samples and in 0.61) and had no high similarity to any GenBank 29 protocorms, resulting in 678 sequences divided into 60 sequences. Interestingly, five different ca. 400-bp-long in- OTUs of rhizoctonia and ECM fungi and 47 OTUs of trons were detected in the three phylogenetically unre- putative endophytes (mainly Helotiales and Chaetothyri- lated OTUs SEB 1, 15 and 20 (Fig. 4), inserted between ales; Table S1, Supporting information). No plateau was the ITS1F and ITS1 priming sites. The OTUs varied not reached for rhizoctonia and ECM fungi (see rarefaction only in intron sequence (SEB 1 and 20) but also in its curves in Fig. S2ab, Supporting information), so that the presence (SEB 1; Fig. S3, Supporting information). whole fungal diversity was not sampled. Adults of the three Neottia species provided a total of 32 Sebacinales Effect of Neottia species, developmental stage and OTUs. At least one Sebacinales OTU was found in each habitat on mycorrhizal symbionts adult individual except for one N. ovata that contained no rhizoctonias (Table S1, Supporting information). Seb- OTU diversity normalized by number of samples in acinales OTUs were also present in all but three N. ovata N. ovata adults was ca. twice that in N. cordata for both protocorms (seven OTUs including three shared with Sebacinales and all mycorrhizal OTUs (Fig. S2ab, Sup- N. ovata adults). Of the 36 Sebacinales OTUs, 64% were porting information). Neottia ovata and N. cordata shared found only at one site each, but within sites at least one no OTUs, except for two Tulasnellaceae OTUs (TUL 2 OTU was frequently shared among individuals (Table S1 and 7; Fig. S4, Supporting information). Accordingly, and Fig. S3, Supporting information). Tulasnellaceae (10 N. ovata populations clustered away from those of N. cor- OTUs; Fig. S4, Supporting information), Ceratobasidia- data in NMDS (Fig. 5A), and the species significantly dif- ceae (4 OTUs) and various ECM fungi sometimes co- fered in mycorrhizal OTU composition (RDA, F = 7.95, occurred with Sebacinales (Fig. 2, Table S1, Supporting P < 0.001), even when only forest populations were com- information) among which only an OTU close to Tylos- pared (F = 7.96, P < 0.001). Congruently, of the environ- pora fibrillosa was repeatedly found (in four N. cordata mental predictors analysed by variance partitioning,

(A) (B)

Fig. 3 Mycorrhizae of Neottia cordata (A, C) and N. ovata (B, D). (A) Light microscopy of root cross-section of N. cordata. cC: collapsed hyphal pelotons, iC: intact living pelotons, s: starch grains. (B) Transmission electron microscopy of hyphae with dolipores (p) sur- rounded by plant plasma membranes (arrowheads) in a root of N. ovata, bar: 2 lm. (C, D) Ultrastructure of Sebacinales dolipore and straight parenthesome (arrow) in N. cordata (C) and N. ovata (D), bar: 0.5 lm.

Neottia species identity had the strongest effect on the thermore, the two habitats did not differ with respect to mycorrhizal OTU composition in adults (Fig. 5B). In con- the number (linear mixed-effect model, likelihood ratio trast, no correspondence between Neottia species and test, LR = 0.014, P = 0.91) and size (LR = 1.67, P = 0.196) phylogenetic position of associated Sebacinales OTUs of N. ovata protocorms in seed packets. Within sites, no was found (dbRDA, F = 1.54, P = 0.139; Fig. 4). difference in germination was observed between N. ovata At the seven sites where fungi in both N. ovata proto- seeds originating from meadow and forest habitats (LR = corms and adults were analysed, these stages differed 0.03, P = 0.862 for the number and LR = 0.54, P = 0.47 for in mycorrhizal OTU composition (RDA, F = 4.35, size of protocorms). P < 0.05) despite sharing several OTUs, likely due to the exclusive presence of SEB 1 in adults and frequent Stable isotope analysis occurrence of SEB 5 in protocorms (Fig. 4). Protocorms also featured lower diversity of rhizoctonia and ECM At the four forest sites, fungal sporocarps were mostly OTUs (but not markedly in Sebacinales; Fig. S2c, Sup- enriched as compared with autotrophic plants in 13C porting information), although this may be caused by and 15N (Fig. 6). Of the two Neottia species, only N. cor- more aggregated distribution of protocorms as com- data plants were slightly enriched in 13C compared with pared with adults, owing to germination limited on surrounding autotrophic plants, but not significantly. only one to three microsites per site. Both Neottia species differed significantly in 13C abun- Habitat type (forest vs. meadow) had little effect on dance from PMH Epipactis helleborine and fungal sporo- fungal diversity (Fig. S2ab, Supporting information). It carps (Fig. 6). Similarly, the abundance of 15N in both significantly influenced only the composition of all Neottia species was not significantly different from that mycorrhizal OTUs in N. ovata adults (RDA, F = 2.63, in autotrophic plants and saprotrophic Mycena, but was P < 0.05); but if considered separately, Sebacinales were significantly below the 15N abundance in E. helleborine not influenced (F = 1.89, P = 0.087), also in terms of their and ECM fungi. In all, this suggested that these Neottia phylogenetic position (dbRDA, F = 0.82, P = 0.48). Fur- species did not receive significant MH nutrition from

Fig. 4 Cladogram of Bayesian majority-rule consensus tree (based on part of 28S nrDNA) showing position of Sebacinales operational taxonomic units (OTUs) found in Neottia spp. (with the exception of SEB10 for which no 28S was ampliﬁed). Circles and squares indicate protocorms and adults, respectively, and are open or closed to indicate meadow or forest habitats. Numbers of symbols indicate the number of sites where the OTU was detected (full phylogram in Fig. S3, Supporting information). Bayesian posterior probabilities are denoted above the branches. The tree is rooted with Exidia glandulosa and Geastrum saccatum. Putative ecology of taxa from GenBank: ECM – ectomycorrhizal, EEM – ectendomycorrhizal, ENP – endophytic, ERM – ericoid mycorrhizal, LA – liverwort associated, OM – orchid mycorrhizal. Country codes follow international ISO codes.

(A) (B)

Spatial vectors

0.04 0.02 0.14 (0.04) (0.07)( (0.13)

0.02 0.03 (0.02) (0.07) NMDS2

Habitat 0.02 (0.02) −2 0 2 4

−4 −2 0 2 4 NMDS1

Fig. 5 (A) Nonmetric multidimensional scaling (NMDS) plot of mycorrhizal OTUs in adults and protocorms of Neottia ovata and adults of N. cordata from forest and meadow habitats. (B) Proportion of variation (adjusted R2) explained by the effects of spatial distances, Neottia species (N. ovata, N. cordata), habitat types (forest, meadow), and their combinations on community composition of rhizoctonia and ectomycorrhizal OTUs (or Sebacinales only, in parentheses) analysed by variation partitioning.

ECM fungi. Both Neottia species had high N concentra- Suarez et al. 2008; Kartzinel et al. 2013). Three N. ovata tions (Fig. S5, Supporting information): N. ovata was in OTUs were grouped at the base of Clade A, where the highest range for autotrophic or PMH plants, but endophytic lineages are increasingly found (Weiß et al. less N-concentrated than sporocarps, while N. cordata 2011), but their ecology remains unclear. This situat- had N concentrations significantly higher than that of ion differs from the MH N. nidus-avis which (i) is more autotrophic plants and similar to or higher than that of specific and (ii) targets ECM Sebacinales from Clade A most sporocarps. (McKendrick et al. 2002; Selosse et al. 2002a,b), a preference also known in the North American MH Hexa- lectris spicata (Kennedy et al. 2011). Sebacinales from Discussion N. camtschatea, nested within Clade A close to N. nidus- avis symbionts, may be also ECM. Fungal associations in Neottia Besides Sebacinales, other rhizoctonia (mainly Tulas- A large diversity (60 OTUs) of putative mycorrhizal nellaceae and rarely Ceratobasidiaceae) and ECM fungi fungi was found in Neottia species, but Sebacinales were occasionally found (Fig. 2). Of the latter group, dominated (found in 77% of protocorms and 99% of the frequent Tylospora spp., common ECM fungi of adults; Fig. 2) regardless of habitat type or geographical north-temperate spruce forests, are hitherto unknown localization. Only Sebacinales were identified by TEM from orchids. Additional rhizoctonias, ECM, or endo- observation (Fig. 3), which supports them as dominant phytic fungi frequently co-occur with a dominant sym- associates of Neottia. Sebacinales are divided into two biont in rhizoctonia-associated orchids (Dearnaley et al. clades that differ ecologically: Clade A includes ECM 2013), but their roles and potential mycorrhizal abilities species associated with trees and (P)MH orchids, while remain unknown. Clade B species are mycorrhizal mainly in Ericaceae, At the fungal genus/family level, the fungal commu- green orchids (where they represent one of the rhizocto- nity was similar in both N. ovata and N. cordata (Fig. 2), nia clades) or liverworts (e.g. Selosse et al. 2007; Weiß although with little overlap in the OTU composition and et al. 2011). Additionally, both clades have endophytic no sharing of the dominant Sebacinales (Fig. 4), perhaps abilities (Selosse et al. 2009; Weiß et al. 2011; Garnica due to divergent habitat preferences. This contrasts with et al. 2013). Although the whole diversity of Sebacinales several other orchid genera where related species mostly and ECM fungi was not sampled and new OTUs may share mycorrhizal OTUs (e.g. Jacquemyn et al. 2010; appear (Fig. S2, Supporting information), the most com- Swarts et al. 2010; Kennedy et al. 2011; Tesitelova et al. mon associates of N. ovata and N. cordata are likely Seb- 2012). Diverging OTU composition and/or species-spe- acinales from Clade B (Fig. 4). Dominant association cific habitat conditions could explain the surprisingly with Sebacinales Clade B was hitherto reported only low N. ovata germination at sites with N. cordata adults. outside Europe, for instance for Australian Caladenia By contrast, overall poor germination of N. cordata at all (Swarts et al. 2010) or some tropical epiphytes (e.g. sites questions the quality of sown seeds.

−22 c (A)c (B) (C)c (D) c c b b −24 b b

−26 a −28 b a a C (‰)

13 a δ −30 a a a ab a b a a −32 a −34

d 15 c 10 d c d 5 d d b N (‰) b 15 δ 0 b b c a a a ab a a ac a ab −5 a b −10

Fig. 6 d13C and d15N values of plants and fungi at two Neottia ovata (A: F3-S; B: F6-S) and two N. cordata (C: C5-S; D: C1-S) forest sites. Different letters indicate statistically signiﬁcant differences (Tukey HSD, P < 0.05). Autotrophic plants – Senecio ovatus, Acer plat- anoides, A. pseudoplatanus, Vaccinium myrtillus and Picea abies; partially mycoheterotrophic orchid – Epipactis helleborine; saprotrophic fungi – Marasmius peronatus and Mycena sp.; ectomycorrhizal fungi – Xerocomus chrysenteron, Lactarius deterrimus, Cortinarius ﬂexipes and C. croceoconus.

Habitat type (meadow vs. forest) had little influence to the fact that even rhizoctonia-associated MH proto- on the composition and phylogenetic position of Sebaci- corms are hardly distinguishable from autotrophs (in nales OTUs associated with N. ovata (Figs 4 and 5). Seb- 13C and especially in 15N; Selosse & Martos 2014; acinales tended to cluster phylogenetically within Stockel€ et al. 2014). Two nonexclusive explanations for disturbed, meadow, or environmentally specific habitats this isotopic content include: different pathways for N (Garnica et al. 2013). However, some orchids may and C nutrition in rhizoctonia-associated protocorms opportunistically associate with available fungi, without may entail isotopic fractionations different from (P)MH significant influence of environmental conditions (Kartz- orchids; or rhizoctonias (whose natural isotopic enrich- inel et al. 2013). The observation that seeds from mea- ments are unknown due to production of thin and dow sites germinated in forest sites and vice versa small sporocarps hardly usable for isotopic analyses) (Table S2, Supporting information), as well as identical are less enriched than ECM fungi, for example due to plant ITS sequences, further argues against the existence different nutrient sources, so that mycoheterotrophy in of different N. ovata ecotypes over its ecological range. rhizoctonia-associated orchid may remain isotopically undetectable. Although our measurements may mean either autotrophy or a PMH based on rhizoctonias, they Trophic strategies of Neottia spp. exclude any PMH based on ECM fungi, which would N. ovata and N. cordata were isotopically indistinguish- entail 15N and 13C enrichments as described especially able from surrounding autotrophs, similarly to most in Neottieae (e.g. Gebauer & Meyer 2003; Julou et al. rhizoctonia-associated orchids (Gebauer & Meyer 2003; 2005; see E. helleborine on Fig. 6). This contrasts, for Selosse & Roy 2009; Merckx 2013). Thus, their nutrition example, with green species from the orchid genus fits the plesiomorphic state for orchids: the correspond- Cymbidium that simultaneously associate with rhizocto- ing heterotrophy level remains currently unclear, owing nia and ECM fungi (Ogura-Tsujita et al. 2012) and

© 2015 John Wiley & Sons Ltd 1132 T. TESITELOVA ET AL. feature 13C and 15N abundances intermediate between example from non-ECM to ECM Tulasnellaceae in the rhizoctonia-associated orchids and ECM-associated MH Aneuraceae liverworts (Bidartondo et al. 2003). An asso- orchids (Motomura et al. 2010): while these green Cym- ciation with ECM Ceratobasidiaceae occurs in a PMH bidium species perform PMH based on ECM fungi, the orchid species of Platanthera (although we lack data on two investigated Neottia do not. the only fully MH P. saprophytica and most green spe- Indeed, N. ovata and N. cordata associated mainly with cies; Yagame et al. 2012) and in the fully MH orchid rhizoctonias dominated by Sebacinales Clade B that does genus Rhizanthella (Bougoure et al. 2009). not form ECM with exception of few ECM subclades Full mycoheterotrophy evolved repeatedly and differ- (Hynson et al. 2013; Tedersoo & Smith 2013). Only a few ent adaptations were likely involved in this process recorded Sebacinales may belong to these ECM clades, (Roy et al. 2013). Notably, in the orchid family only ca. namely OTU SEB 36 from N. ovata placed within an ECM one-third of genera with MH species also include green clade close to Pyrola asarifolia symbionts (AB669633; Ha- species, but such genera have so far been little investi- shimoto et al. 2012; clade/serendipita2 in Tedersoo & gated (Merckx 2013). Our data show that the genus Smith 2013), and OTUs SEB 2 to 6 and 29, frequent in Neottia constitutes an interesting evolutionary model, N. ovata, close to an ECM symbiont of Fagus sylvatica especially if species from the genus South-East Asian (HQ154321; Garnica et al. 2013; Fig. 4). Although we can- diversity centre are covered, to study the evolution of not draw a firm conclusion regarding the nutrition of MH, possibly in tight association with Sebacinales. Sebacinales associated with N. ovata and N. cordata, Neot- Finally, the occurrence of rhizoctonia-associated species tia isotopic content at least supports that they did not in Neottia, as in some Epipactis species (Bidartondo et al. receive C and N from such ECM fungi. However, the 2004), complicates the simple scenario where the ances- possibility that N. ovata sometimes obtains some C from tor of the tribe Neottieae is considered PMH and ECM ECM fungi (Sebacinales or other) may explain the associated (Selosse & Roy 2009). There could be some reported variation in stable isotope abundances, ranging reversion from such an ancestral ECM association to from abundances of autotrophs to those typical for PMH association with rhizoctonias, or the Neottieae ancestor orchids (Gebauer & Meyer 2003; Abadie et al. 2006; Ted- may even be itself associated with rhizoctonia (but this ersoo et al. 2007; Liebel et al. 2010). In the future, analyses would raise the question of why so many species of photosynthetic abilities and isotopic content in N. ovat- shifted to ECM association among Neottieae). More a from open meadows may further reveal the exact auto- investigations on Neottieae and Neottia spp. are still trophic ability in these species. required to elucidate the evolution to MH nutrition and possible reversions on this way. Evolution to mycoheterotrophy Acknowledgements A shift from rhizoctonia to other saprotrophic or ECM symbionts has been proposed as a pre-adaptation of The authors thank Melanie Roy, Jean-Philippe Anglade, Gerard evolution to full MH in orchids, especially in Neottieae Joseph, Angelika and Heinz Baum, and Herve Christophe for (Selosse & Roy 2009; Roy et al. 2013). Although we help with material collection, Jakub Tesitel, Gabriela Rıhova, Milena Novakov a, Lucie Jonatov a, Tomas Tyml and Lenka investigated a minor part of the genus Neottia, the leaf- Bucinska for scientific or technical help, David Marsh for lan- less N. nidus-avis and N. camtschatea associated with guage corrections, and Dirk Redecker and four anonymous ref- ECM Sebacinales Clade A clustered phylogenetically erees for their useful comments. The work was supported by together (Fig. 1). Although isotopic analyses are pend- the Czech Science Foundation (project 31-14-21432S), and MAS ing, the leafless, but greenish N. camtschatea is likely was funded by the ‘Societe francßaise d’Orchidophilie’. PMH. As the rhizoctonia-associated N. ovata and N. cordata occupy a more basal or paraphyletic position References (Fig. 1), it is likely, until more detailed exploration of € this genus, that the association with rhizoctonias from Abadie J-C, Puttsepp€ U, Gebauer G, Faccio A, Bonfante P, Sel- Sebacinales Clade B and the corresponding nutrition osse M-A (2006) Cephalanthera longifolia (Neottieae, Orchida- ceae) is mixotrophic: a comparative study between green are ancestral in Neottia. The evolution to MH would and nonphotosynthetic individuals. Botany-Botanique, 84, have proceeded by keeping a preference for Sebaci- 1462–1477. nales, which remains to assess for the ca 50 other Neot- Bidartondo MI, Bruns TD, Weiß M, Sergio C, Read DJ (2003) tia species. The change from rhizoctonias to ECM Specialized cheating of the ectomycorrhizal symbiosis by an symbionts in evolution of MH orchids often involves a epiparasitic liverwort. Proceedings of the Royal Society of Lon- shift to totally unrelated ECM or saprotrophic fungi don. Series B: Biological Sciences, 270, 835–842. (reviewed in Dearnaley et al. 2013; Merckx 2013), but Bidartondo MI, Burghardt B, Gebauer G, Bruns TD, Read DJ (2004) Changing partners in the dark: isotopic and molecular sometimes occurs within the same fungal taxon, for

evidence of ectomycorrhizal liaisons between forest orchids Leake JR (1994) The biology of myco-heterotrophic (‘sapro- and trees. Proceedings of the Royal Society of London. Series B: phytic’) plants. New Phytologist, 127, 171–216. Biological Sciences, 271, 1799–1806. Legendre P, Gallagher ED (2001) Ecologically meaningful Bougoure J, Ludwig M, Brundrett M, Grierson P (2009) Iden- transformations for ordination of species data. Oecologia, 129, tity and specificity of the fungi forming mycorrhizas with 271–280. the rare mycoheterotrophic orchid Rhizanthella gardneri. Liebel HT, Bidartondo MI, Preiss K et al. (2010) C and N stable Mycological Research, 113, 1097–1106. isotope signatures reveal constraints to nutritional modes in Colwell RK (2006) ESTIMATES: Statistical estimation of species orchids from the Mediterranean and Macaronesia. American richness and shared species from samples, version 8. http:// Journal of Botany, 97, 903–912. purl.oclc.org/estimates. McKendrick SL, Leake JR, Taylor DL, Read DJ (2002) Symbiotic Dearnaley JDW, Martos F, Selosse M-A (2013) Orchid mycorrhi- germination and development of the myco-heterotrophic zas: Molecular ecology, physiology, evolution and conserva- orchid Neottia nidus-avis in nature and its requirement for tion aspects. In: Fungal Associations, 2nd Edition. The Mycota locally distributed Sebacina spp. New Phytologist, 154, 233– IX (ed. Hock B), pp. 207–230. Springer-Verlag, Berlin and 247. Heidelberg, Germany. Medlin L, Elwood HJ, Stickel S, Sogin ML (1988) The character- Dray S, Legendre P, Blanchet G (2013) PACKFOR: forward selec- ization of enzymatically amplified eukaryotic 16S-like rRNA- tion with permutation (Canoco p.46). R package version 0.0- coding regions. Gene, 71, 491–499. 8/r109. http://R-Forge.R-project.org/projects/sedar/ Merckx VSFT (ed.) (2013) Mycoheterotrophy: The Biology of Plants Duminil J, Pemonge MH, Petit RJ (2002) A set of 35 consensus Living on Fungi, 1st edn. Springer, New York, NY. primer pairs amplifying genes and introns of plant mito- Milne I, Wright F, Rowe G, Marshall DF, Husmeier D, McGu- chondrial DNA. Molecular Ecology Resources, 2, 428–430. ire G (2004) TOPALI: software for automatic identification of Garnica S, Riess K, Bauer R, Oberwinkler F, Weiß M (2013) recombinant sequences within DNA multiple alignments. Phylogenetic diversity and structure of sebacinoid fungi Bioinformatics, 20, 1806–1807. associated with plant communities along an altitudinal gra- Motomura H, Selosse M-A, Martos F, Kagawa A, Yukawa T dient. FEMS Microbiology Ecology, 83, 265–278. (2010) Mycoheterotrophy evolved from mixotrophic ances- Gebauer G, Meyer M (2003) 15N and 13C natural abundance of tors: evidence in Cymbidium (Orchidaceae). Annals of Botany, autotrophic and myco-heterotrophic orchids provides insight 106, 573–581. into nitrogen and carbon gain from fungal association. New Ogura-Tsujita Y, Yokoyama J, Miyoshi K, Yukawa T (2012) Phytologist, 160, 209–223. Shifts in mycorrhizal fungi during the evolution of autotro- Gonneau C, Jersakov a J, de Tredern E et al. (2014) Photosynthe- phy to mycoheterotrophy in Cymbidium (Orchidaceae). Amer- sis in perennial mixotrophic Epipactis spp. (Orchidaceae) con- ican Journal of Botany, 99, 1158–1176. tributes more to shoot and fruit biomass than to hypogeous Oksanen J, Blanchet FG, Kindt R et al. (2013) VEGAN: community survival. Journal of Ecology, 102, 1183–1194. ecology package. R package version 2.0-9. http://CRAN.R- Hashimoto Y, Fukukawa S, Kunishi A et al. (2012) Mycohetero- project.org/package=vegan trophic germination of Pyrola asarifolia dust seeds reveals Pinheiro J, Bates D, DebRoy S, Sarkar D, the R Development convergences with germination in orchids. New Phytologist, Core Team (2013) nlme: linear and nonlinear mixed effects 195, 620–630. models. R package version 3.1-111. http://cran.r-project.org/ Huelsenbeck JP, Ronquist F (2001) MRBAYES: Bayesian inference web/packages/nlme/ of phylogenetic trees. Bioinformatics, 17, 754–755. R Development Core Team (2013) R: A Language and Environ- Hynson NA, Weiß M, Preiss K, Gebauer G, Treseder KK (2013) ment for Statistical Computing. R Foundation for Statistical Fungal host specificity is not a bottleneck for the germination Computing, Vienna, Austria. http://www.R-project.org/ of Pyroleae species (Ericaceae) in a Bavarian forest. Molecular Rasmussen HN, Whigham DF (1993) Seed ecology of dust Ecology, 22, 1473–1481. seeds in situ: a new study technique and its application in Jacquemyn H, Honnay O, Cammue BPA, Brys R, Lievens B terrestrial orchids. American Journal of Botany, 80, 1374–1378. (2010) Low specificity and nested subset structure character- Rasmussen HN, Johansen B, Andersen TF (1991) Symbiotic in ize mycorrhizal associations in five closely related species of vitro culture of immature embryos and seeds from Listera ov- the genus Orchis. Molecular Ecology, 19, 4086–4095. ata. Lindleyana, 6, 134–139. Julou T, Burghardt B, Gebauer G, Berveiller D, Damesin C, Selosse Roy M, Watthana S, Stier A, Richard F, Vessabutr S, Selosse M- M-A (2005) Mixotrophy in orchids: insights from a comparative A (2009) Two mycoheterotrophic orchids from Thailand study of green individuals and nonphotosynthetic individuals tropical dipterocarpacean forests associate with a broad ofCephalantheradamasonium.NewPhytologist,166,639–653. diversity of ectomycorrhizal fungi. BMC Biology, 7, 51. Kartzinel TR, Trapnell DW, Shefferson RP (2013) Highly Roy M, Gonneau C, Rocheteau A et al. (2013) Why do mixo- diverse and spatially heterogeneous mycorrhizal symbiosis trophic plants stay green? A comparison between green and in a rare epiphyte is unrelated to broad biogeographic or achlorophyllous orchid individuals in situ. Ecological Mono- environmental features. Molecular Ecology, 22, 5949–5961. graphs, 83,95–117. Katoh K, Kuma K, Toh H, Miyata T (2005) MAFFT version 5: Selosse M-A, Martos F (2014) Do chlorophyllous orchids het- improvement in accuracy of multiple sequence alignment. erotrophically use mycorrhizal fungal carbon? Trends in Plant Nucleic Acids Research, 33, 511–518. Science, 19, 683–685. Kennedy AH, Taylor DL, Watson LE (2011) Mycorrhizal speci- Selosse M-A, Roy M (2009) Green plants that feed on fungi: ficity in the fully mycoheterotrophic Hexalectris Raf. (Or- facts and questions about mixotrophy. Trends in Plant Science, chidaceae: Epidendroideae). Molecular Ecology, 20, 1303–1316. 14,64–70.

Selosse M-A, Bauer R, Moyersoen B (2002a) Basal hymenomy- Xiao L, Morgan UM, Limor J et al. (1999) Genetic diversity cetes belonging to the Sebacinaceae are ectomycorrhizal on within Cryptosporidium parvum and related Cryptosporidium temperate deciduous trees. New Phytologist, 155, 183–195. species. Applied and Environmental Microbiology, 65, 3386– Selosse M-A, Weiß M, Jany JL, Tillier A (2002b) Communities 3391. and populations of sebacinoid basidiomycetes associated Yagame T, Orihara T, Selosse M-A, Yamato M, Iwase K (2012) with the achlorophyllous orchid Neottia nidus-avis (L.) LCM Mixotrophy of Platanthera minor, an orchid associated with Rich. and neighbouring tree ectomycorrhizae. Molecular Ecol- ectomycorrhiza-forming Ceratobasidiaceae fungi. New Phytol- ogy, 11, 1831–1844. ogist, 193, 178–187. Selosse M-A, Setaro S, Glatard F, Richard F, Urcelay C, Weiß Yang Z (1995) A space-time process model for the evolution of M (2007) Sebacinales are common mycorrhizal associates of DNA sequences. Genetics, 139, 993–1005. Ericaceae. New Phytologist, 174, 864–878. Selosse M-A, Dubois M-P, Alvarez N (2009) Do Sebacinales commonly associate with plant roots as endophytes? Myco- logical Research, 113, 1062–1069. M.A.S., J.J. and T.T. designed the research; J.J., T.T., Stockel€ M, Tesitelova T, Jersakov a J, Bidartondo MI, Gebauer G M.A.S., M.K. and I.T. collected the samples; M.K. did (2014) Carbon and nitrogen gain during the growth of orchid the germination experiment; M.K., F.X.J., T.T. and J.K. seedlings in nature. New Phytologist, 202, 606–615. conducted laboratory work; T.T. and M.K. analysed the Suarez JP, Weiß M, Abele A, Oberwinkler F, Kottke I (2008) data; T.T. wrote the first draft of the study; and all Members of Sebacinales subgroup B form mycorrhizae with authors contributed to its final version. epiphytic orchids in a neotropical mountain rain forest. Mycological Progress, 7,75–85. Swarts ND, Sinclair EA, Francis A, Dixon KW (2010) Ecological specialization in mycorrhizal symbiosis leads to rarity in an endangered orchid. Molecular Ecology, 19, 3226–3242. Data accessibility Taberlet P, Gielly L, Pautou G, Bouvet J (1991) Universal primers for amplification of three non-coding regions of chloro- DNA sequences: GenBank accessions for fungi: – – plast DNA. Plant Molecular Biology, 17, 1105–1109. KJ188443 KJ188548, KJ188553 KJ188622, KJ413254; Tanabe AS (2011) Kakusan4 and Aminosan: two programs for plants: KJ023673–KJ023682, KJ188549–KJ188552. comparing nonpartitioned, proportional and separate models Sequence alignments, phylogenetic analysis input files for combined molecular phylogenetic analyses of multilocus and tree files: Dryad entry: doi:10.5061/dryad.033ds. – sequence data. Molecular Ecology Resources, 11, 914 921. Sampling site locations and overview of fungal Taylor DL, McCormick MK (2008) Internal transcribed spacer OTUs found in each Neottia individual with respective primers and sequences for improved characterization of basidiomycetous orchid mycorrhizas. New Phytologist, 177, GenBank accessions (Table S1, Supporting informa- 1020–1033. tion). Taylor AF, Fransson PM, Hogberg€ P, Hogberg€ MN, Plamboeck Carbon and nitrogen stable isotope values and N con- AH (2003) Species level patterns in 13C and 15N abundance centrations: Dryad entry: doi:10.5061/dryad.033ds. of ectomycorrhizal and saprotrophic fungal sporocarps. New Phytologist, 159, 757–774. Tedersoo L, Smith ME (2013) Lineages of ectomycorrhizal fungi revisited: foraging strategies and novel lineages Supporting information revealed by sequences from belowground. Fungal Biology Reviews, 27,83–99. Additional supporting information may be found in the online ver- Tedersoo L, Pellet P, Koljalg~ U, Selosse M-A (2007) Parallel sion of this article. evolutionary paths to mycoheterotrophy in understorey Eric- Fig. S1 Single-locus phylogenetic trees of Neottia genus. aceae and Orchidaceae: ecological evidence for mixotrophy in Pyroleae. Oecologia, 151, 206–217. Fig. S2 Rarefaction curves of operation taxonomical units. TesitelovaT,T esitel J, Jersakov aJ, Rıhova G, Selosse M-A (2012) Symbiotic germination capability of four Epipactis spe- Fig. S3 Bayesian phylogenetic tree of Sebacinales. cies (Orchidaceae) is broader than expected from adult ecol- Fig. S4 Bayesian phylogenetic tree of Tulasnellaceae. ogy. American Journal of Botany, 99, 1020–1032. Tesitelova T, Jersakova J, Roy M et al. (2013) Ploidy-specific Fig. S5 Nitrogen concentrations in plants and fungi from two symbiotic interactions: divergence of mycorrhizal fungi Neottia ovata and two N. cordata forest sites. between cytotypes of the Gymnadenia conopsea group (Or- chidaceae). New Phytologist, 199, 1022–1033. Table S1 Locations of Neottia populations and overview of Weiß M, Sykorov a Z, Garnica S et al. (2011) Sebacinales every- associated fungi. where: previously overlooked ubiquitous fungal endophytes. PLoS ONE, 6, e16793. Table S2 Seed sowing design and germination success.