1.mycological research 110 (2006) 12571270available at www.sciencedirect.comjournal homepage: www.elsevier.com/locate/mycresDiverse tulasnelloid fungi form mycorrhizas with epiphytic orchids in an Andean cloud forest Juan Pablo SUAREZa,b,*, Michael WEIb, Andrea ABELEb, Sigisfredo GARNICAb, Franz OBERWINKLERb, Ingrid KOTTKEb a cnica Particular de Loja, San Cayetano Alto s/n C.P. 11 01 608, Loja, EcuadorCentro de Biologa Celular y Molecular, Universidad Te bEberhard-Karls-Universita Tubingen, Botanisches Institut, Spezielle Botanik und Mykologie, Auf der Morgenstelle 1, t D-72076 Tubingen, Germanyarticle info abstractArticle history: The mycorrhizal state of epiphytic orchids has been controversially discussed, and the Received 3 May 2006state and mycobionts of the pleurothallid orchids, occurring abundantly and with a high Received in revised form number of species on stems of trees in the Andean cloud forest, were unknown. Root sam- 7 August 2006ples of 77 adult individuals of the epiphytic orchids Stelis hallii, S. superbiens, S. concinna and Accepted 12 August 2006Pleurothallis lilijae were collected in a tropical mountain rainforest of southern Ecuador. Ul- Published online 31 October 2006 trastructural evidence of symbiotic interaction was combined with molecular sequencing Corresponding Editor:of fungi directly from the mycorrhizas and isolation of mycobionts. Ultrastructural analy- John W. G. Cairney ses displayed vital orchid mycorrhizas formed by fungi with an imperforate parenthesomeand cell wall slime bodies typical for the genus Tulasnella. Three different Tulasnella isolates Keywords:were obtained in pure culture. Phylogenetic analysis of nuclear rDNA sequences from cod- Heterobasidiomycetes ing regions of the ribosomal large subunit (nucLSU) and the 5.8S subunit, including parts of Molecular phylogenythe internal transcribed spacers, obtained directly from the roots and from the fungal iso- Pleurothallidinaelates, yielded seven distinct Tulasnella clades. Tulasnella mycobionts in Stelis concinna were Southern Ecuador restricted to two Tulasnella sequence types while the other orchids were associated with up Tropical mountain rain forestto six Tulasnella sequence types. All Tulasnella sequences are new to science and distinct Ultrastructure from known sequences of mycobionts of terrestrial orchids. The results indicate that tulas-nelloid fungi, adapted to the conditions on tree stems, might be important for orchidgrowth and maintenance in the Andean cloud forest. 2006 The British Mycological Society. Published by Elsevier Ltd. All rights reserved. IntroductionWilliamson 1972), but a high infection rate was reported from canopy-dwelling orchid species in Florida (Benzing Although most land plants are associated with symbiotic 1982). Different degrees of infection including non-infected fungi forming mycorrhizas or mycorrhiza-like associations,roots were observed in epiphytic orchids in Ecuador (Ber- many epiphytes live without such associations, e.g. mosses, mudes & Benzing 1989). Goh et al. (1992) found high fungal col- many liverworts, bromeliads, and ferns (Kottke 2002). Find- onization in the epiphytic orchid Dendrobium crumenatum from ings on the mycorrhizal state of epiphytic orchids were con-a natural stand in Singapore, but only low or no mycorrhiza- troversial. Only sporadic fungal colonization was found tion in orchids from nurseries. Rivas et al. (1998) and Pereira in a number of epiphytic Malaysian orchids (Hadley &et al. (2005) reported intense colonization of epiphytic orchids* Corresponding author.E-mail address: jpsuarez@utpl.edu.ec 0953-7562/$ see front matter 2006 The British Mycological Society. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.mycres.2006.08.004

2. 1258 J. P. Suarez et al.in Costa Rica and Brazil, respectively. All investigators stated minimized. The genera Stelis and Pleurothallis belong to that fungal colonization was restricted to roots attaching tosubtribe Pleurothallidinae, the largest subtribe in the tribe the substrate; aerial roots were not colonized.Epidendreae of Orchidaceae (Luer 1986a,b), which is widely dis- Identication of root-associated fungi was mostly achieved tributed in tropical America. These two genera include 485 by isolation on sterile media (Rasmussen 2002). However, the Pleurothallis and 465 Stelis species reported until now for Ecua- distinction between endophytic fungi inhabiting only the ve- dor (L. Endara, pers. comm.). Many of these epiphytes are en- lamen or the root surface and reliably mycorrhiza-formingdemic species of Ecuadorian tropical forests. Only a few of fungi colonizing the cortical tissue was mostly unclear (Cur-them are in culture so far. The rapid loss of habitats requires rah et al. 1997; Pridgeon 1987). Xylaria (Ascomycota) was fre- an understanding of the symbiotic relationships in order quently isolated (Bayman et al. 1997, Tremblay et al. 1998), to support conservation efforts for these orchids. According but was never proven experimentally or demonstrated by ul- to Hamilton et al. (1995) approximately 90 % of the Northern trastructure to form mycorrhizas with orchids. Fungal isola- Andean forests have been already destroyed. Consequently, tion from pelotons as a more selective approach has been the orchids and their fungi might be lost in the near future successfully attempted in terrestrial orchids (Warcup & Talbot if not taken into culture. As the mycorrhizal state and myco- 1967, 1971, 1980, Bougoure et al. 2005). In cases where sexual bionts of the epiphytic Pleurothallidinae were unknown, no stages could be achieved, the isolated fungi were determined advice could be given to laboratories interested in orchid cul- as basidiomycetes belonging to the Sebacinales, Tulasnellalesturing or to local forest management aiming to rehabilitate or Ceratobasidiales (Warcup 1981; Warcup & Talbot 1967,the tropical mountain forest with its epiphytic orchid diversity 1971, 1980). Tulasnella anamorphs (Epulorhiza) were isolated,(see: http://www.bergregenwald.de of which this work is a e.g. from epiphytic Epidendrum conopseum in Florida (Zettler part). We therefore started with light- and transmission elec- et al. 1998), epiphytic Epidendrum rigidum, Polystachia concreta tron microscopic investigation of the selected orchid species, (Pereira et al. 2003), and terrestrial Oeceoclades maculata from and continued with DNA isolation and sequencing of the Brazil (Pereira et al. 2005). DNA sequencing supported the most frequently observed fungal group, the Tulasnellales. In presence of Tulasnella, Sebacinales and Ceratobasidium inparallel, isolation of mycelia was carried out, yielding several Cypripedium spp. from the temperate Northern HemisphereTulasnella isolates. We were especially interested to see (Shefferson et al. 2005). Molecular tools were also used to iden-whether the Tulasnellales present as mycobionts of epiphytic tify fungal isolates obtained from pelotons (Bougoure et al. orchids in the tropical mountain rain forest were distinct 2005) or by direct DNA isolation from pelotons (Kristiansenfrom those described for other habitats of the Northern Hemi- et al. 2001). A taxon distantly related to Laccaria, an ectomycor- sphere and Australia. This knowledge would help to decide if rhiza-forming fungus, was found in Dactylorhiza majalislocal or ubiquitous fungal isolates were appropriate for culti- (Kristiansen et al. 2001) in addition to Tulasnella. Ectomycor-vation of the local orchids, and would support evaluation of rhiza-forming mycobionts were also proven for non- loss of local fungi for rehabilitation of orchids in the tropical photosynthetic orchids by DNA isolations and sequencingmountain forest area. directly from mycorrhizas (Taylor & Bruns 1997, 1999; Taylor et al. 2003, Bidartondo et al. 2004, Selosse et al. 2004, Julou et al. 2005), thus widening the previous knowledge on orchidMaterials and methods mycobionts. Selosse et al. (2004) conrmed their molecular nding of Tu- Study site ber spp. (Ascomycota) as orchid mycobionts by ultrastructural demonstration of ascomycetous hyphae in the cortical cells ofThe study site is located on the eastern slope of the Cordillera El the orchid roots. Ascomycetes can be discerned from basidio- Consuelo in the northern Andes of southern Ecuador. The area mycetes by ultrastructure of the cell wall and the septal pore, of about 1000 ha belongs to the Reserva Biologica San Francisco and different groups of basidiomycetes can be distinguishedand borders the Podocarpus National Park in the north, half by the parenthesomes covering the dolipores (Wells & Ban-way between Loja and Zamora, Zamora-Chinchipe Province doni 2001); tulasnelloid fungi display characteristic slime bod- (3 58 S, 79 04 W). The tropical mountain rainforest covers the ies in the cell walls (Bauer 2004). In spite of these diagnostic steep slopes between 1850 and 2700 m a.s.l. Characteristic possibilities transmission electron microscopy has rarelyand most frequent trees are Melastomataceae, Rubiaceae, Laura- been used in orchid studies addressing fungal identity. How- ceae and Euphorbiaceae reaching a height of 25 m. Crown density ever, the previous work is encouraging (CurrahSherburne as measured by a spherical densitometer is 94 % on average, 1992; Andersen 1996) and minimizes errors resulting from only 7.5 % were open canopy (Homeier 2004). contamination during isolation of fungi or DNA directly from The richness and abundance of epiphytes is due to the mycorrhiza samples. In our study of the orchid mycobiontssemi- to sub-humid climate with rainfall during ten months of four epiphytic, pleurothallid orchid species in the Andeanand even more frequent fog combined with moderate temper- cloud forest of south Ecuador, we therefore combined ultra-atures (Richter 2003). Mean annual precipitation at 1950 m structural studies with DNA sequencing and isolation.a.s.l. is 2200 mm, annual mean temperature is 15,5C (14,4- Stelis concinna, S. hallii, S. superbiens, and Pleurothallis lilijae 17,5C). Precipitation increases with higher elevation and Foldats were selected because of the abundance and frequentreaches 4000 mm at 2600 m asl. Air humidity in two months owering of these small orchids in the tropical mountain is 96 % on average and does not fall below 70 % during the rainforest of the study area. Thus severe violations of thedrier season (Noske 2004). The high air humidity is especially orchid populations in this highly endangered forest could be important for stem epiphytes. 3. Diverse tulasnelloid fungi form mycorrhizas 1259Sampling Fungal isolationSampling was carried out at small paths at an altitudinal gra- Isolation of fungi was initiated the day of sampling. Colonized dient between 1850 and 2100 m a.s.l. Stelis hallii was found inroot pieces were surface-sterilized. Roots were rinsed in the forest covering the steep slopes between 1800 anddistilled water with some drops of liquid soap, immersed in 1900 m a.s.l., while S. superbiens and Pleurothallis lilijae wereethanol (70 %) for 30 s, immersed in Ajax chloro 20 % (house- collected in the forest covering the mountain ridge betweenhold bleach, sodium hypochlorite 5.25 %) for 10 min and 1900 and 2100 m a.s.l. Stelis concinna was restricted to the up- nally rinsed in sterile distilled water. The velamen was per part of the mountain ridge where the forest was less then removed using a stereo microscope, a thin blade and for- dense, with only 92 % crown density, and exposition to fre-ceps. Five square sections of 1-3 mm thickness were cut by quent and heavy winds. hand from the middle part of the root and transferred toRoots were collected continuously during three yearsa plate with MYP media (malt extract 7 g, peptone 1 g, and from 2003 until 2005 from a total of 77 owering individuals,agar agar 15 g l1) or MMNC media (modied Melin-Norkrans; 22 of S. hallii, 17 of S. superbiens, 25 of S. concinna, and 13 of Kottke et al. 1987; NaCl 0,025 g, KH2PO4 0,5 g, (NH4)2HPO4 0,25 g, Pleurothallis lilijae. All selected plants were epiphytes on CaCl2 0,05 g, MgSO4 7H2O 0,15 g, FeCl3 (1 %) 1 ml, thiamin trunks or branches of standing trees at 50 cm to 200 cm1 ml, malt extract 5 g, glucose 10 g, caseinhydrolysate 1 g, above the forest oor. Distances between trees with ower- agar 20 g, riboavin 1 ml of 0.01 % solution, trace elements ing orchids varied between 50 cm and several metres (up to 10 ml according to Fortin and Piche 1979). No antibiotics 20 m). Identication of trees was not taken into consider- were added. ation. Roots of one owering individual orchid per tree stem were collected. One to four roots per plant individual were packed in aluminum foil to prevent desiccationDNA extraction, PCR and sequencing and transported to the laboratory the same day. As pre- investigation had shown that mycorrhizal fungi colonized Portions of 1-2 cm length of well colonized roots of which the only roots in contact with the stems, best when also covered velamen was removed were collected in cups the same day or by mosses or a minute humus layer, later on only such rootsdried and kept on silica gel for later DNA isolations. DNA was were selected. Root samples were processed the day ofextracted from the fresh or dried mycorrhizal tissue and from collection as pre-investigation had revealed a fast loss offungal mycelium of our own isolates using a Plant Mini Kit vitality in the symbiotic fungi. Vouchers of the orchid(Qiagen, Hilden, Germany). A rst attempt to PCR amplify ge- specimens were deposited in the Herbarium of UTPL, Loja, nomic DNA was carried out from mycorrhizal tissue using Ecuador, including owers xed in ethanol. Vouchers of universal fungal primer combinations ITS1F/ITS4, ITS1F/NL4, the mycorrhizas were embedded in resin and deposited inNLMW1/LR5, NLMW1/TW14 and ITS1F/TW14 (details con- the Herbarium of Tubingen University (TUB). cerning the primers used are given in the Electronic AppendixA). Several PCR products were obtained and sequenced. DNAisolated from fungal cultures was amplied using the primer Light and transmission electron microscopy combination ITS1F/NL4 or ITS1/NL4. Nested PCR was con-ducted to specically amplify DNA from tulasnelloid fungi, Light microscopy was used to select material with fungal coils.as the ultrastructural analysis had revealed these fungi fre- Transversal sections were cut from the middle part of each quently in the cortical tissue of all the orchid species under in- root sample by hand using a razor blade. Sections were vestigation. The rst amplication was carried out with the stained by Methyl blue 0.05 % solution (C. I. 42780, Merck) in primer combination ITS1F/TW14 or ITS1/TW14 and the sec- lactic acid for 10 min on microscopic slides. The samplesond, using template obtained in the rst PCR in dilutions of were examined in fresh lactic acid at 100- to 1000-fold magni- 101, 102 and 103, with the primer combinations ITS1/ cation (Leitz SM-LUX or Zeiss Axioskop 2).ITS4-Tul for the internal transcribed spacers (ITS1, 5.8S nu- Root pieces of 1 cm length of all the samples displaying clear ribosomal gene and ITS2) and NLMW1/LR5, ITS4-TulR/ high frequency of vital looking hyphal coils, 56 in total andLR5 and 5.8S-Tul/NL4 for the 5 part of the nuclear large sub- at least ten of each species, were xed in 2.5 % glutaralde- unit ribosomal DNA (nucLSU). Primers ITS4-Tul and ITS4- hyde-formaldehyde in Srensen buffer (Karnovsky 1965), TulR target a Tulasnella-specic sequence at the 3 end of post-xed in 1 % osmium tetroxide for 1 h, dehydrated in anITS2. The Tulasnella-specic primer 5.8S-Tul (5-TCATTCGAT acetone series and at embedded in Spurrs resin low viscos- GAAGACCGTTGC-3) designed for this study targets a specic ity, longer pot-life formulation (Spurr 1969). Semithin sections sequence at the 5 end of the 5.8S rDNA. were cut from the embedded samples, stained with 0.6 % neo-PCR conditions were as follows: initial denaturation at fuchsin crystal-violet, mounted in Entellan, and observed in 94C for 3 min; 35 cycles, each cycle consisting of one step the light microscope. 20 samples with apparently vital of denaturation at 94C for 30 s; annealing depending of the hyphae, originating from different plant individuals, were primer combinations for 45 s and extension at 72C for selected for ultrathin cutting. Sections were mounted on For-1 min; a nal extension at 72C for 7 min was performed to mvar-coated copper grids and stained with 1 % uranyl acetate nish the PCR. The PCR reaction volume was 50 ml, with con- (40 min) and lead citrate (12 min). Sections were examined us- centrations of 1.5 mM MgCl2, 200 mM of each dNTP (Life Tech- ing transmission electron microscopes Zeiss TEM 902 or Zeiss nologies, Eggenstein, Germany), 0.5 mM of each of the primers TEM109.(MWG-Biotech, Ebersberg, Germany), 1U Taq polymerase (Life 4. 1260 J. P. Suarez et al.Technologies, Eggenstein, Germany), with an amplication MrModeltest, version 2.2 (Nylander 2004) involving four buffer (Life Technologies, Eggenstein, Germany). incrementally heated Markov chains over four million gener- In every PCR a control including PCR mix without DNA ations and using random starting trees. Trees were sampled template was included. Success of the PCR amplicationsevery 100 generations resulting in a total of 40000 trees from was tested in 0.7 % agarose, stained in a solution of ethidium which the last 24000 were used to compute a 50 % majority bromide 0.5 mg ml1. PCR products were puried using the rule consensus tree. Each analysis was repeated to check QIAquick protocol (Qiagen). Cycle sequencing was conducted the reproducibility of the results (Huelsenbeck et al. 2002). using BigDye version 3.1 chemistry, and sequencing was An accumulation curve of clades vs number of collected indi- done on an ABI 3100 Genetic Analyzer (Applied Biosystems,viduals from the four orchid species was computed with Esti- Foster City, CA). Both strands of DNA were sequenced.mateS (Version 7.5, R. K. Colwell, unpubl.). Sequence editing was performed using Sequencher version We determined the proportional differences between se- 4.5 (Gene Codes, Ann Arbor, MI). The sequences obtainedquences within each clade of the nucLSU D1/D2 in order to de- in this study are available from GenBank under accession ne sequence types. We compared the number of Tulasnella numbers DQ178029-DQ178118 (Table 1). sequence types within single and between different orchid We also included in this study sequence data from Tulas- species. The proportional differences between sequences nella reference strains kindly provided by the National Insti- were pooled into ve tables (Electronic Appendix B). tute of Agrobiological Sciences (NIAS), Japan, which were previously isolated from Australian orchids and determined by J. H. Warcup (WarcupTalbot 1967, 1971).ResultsPhylogenetic analysesMicroscopical and ultrastructural features of the mycorrhizasWe used BLAST (Altschul et al. 1997) against the NCBI nucleo-Fungal pelotons were present in nearly all cross-sections of tide database (GenBank; http://www.ncbi.nlm.nih.gov/) to de- roots sampled directly from the tree bark. No fungal pelotons tect published sequences with a high similarity to the nucLSUwere observed in aerial roots. This observation was conrmed sequences obtained from the Ecuadorian epiphytic orchids.by sampling roots of another 65 epiphytic Stelis and Pleurothal- For thorough phylogenetic analysis of the Tulasnella se- lis orchids, indicating that the roots became colonized only quences we analyzed nucLSU and ITS-5.8S alignments includ- where the fungi contacted the bark or the thin humus layer. ing the closest BLAST matches together with the sequencesPelotons were distributed throughout the cortex, with no dif- from the Warcup Tulasnella reference isolates (see above)ference between cortical layers. Vital, blue staining and col- and other sequences from Tulasnellaceae and related groups lapsed, slightly yellow coloured pelotons were visible in the retrieved from GenBank.same cells suggesting that cells became re-infected several Sequences were aligned using the G-INS-i or L-INS-i strat- times. According to the light microscopical observations, egy as implemented in MAFFT v5.667 (Katoh et al. 2005). Duemany fungal pelotons were found collapsed after the plants to the heterogeneity of the Tulasnella sequences we had to ex- had been kept one night in the laboratory. Abundant hyphae clude considerable portions of the nucLSU sequences for phy- colonized the velamen. logenetic analysis. Even the 5.8S ribosomal region, considered TEM observations conrmed the known fungal-root inter- as universally conserved, exhibited a remarkable heterogene- action in orchid mycorrhizas. Hyphae of more or less equal di- ity as was already mentioned by Bidartondo et al. (2003). As ameter were surrounded by the plant plasma membrane, the expected, the ITS1 and ITS2 rDNA could not be aligned over plant vacuole forming small compartments or a network of the whole data set. Therefore, we used the 5.8S region to cal- small vacuoles (Fig 1). Degenerating hyphae were attached culate phylogenetic trees of a wider phylogenetic spectrum to collapsed pelotons (Fig 2). Alive hyphae contained abundant and produced several other phylogenetic analyses including glycogen granules (Figs 1, 3 and 5). The hyphae formed septa, subsets of related sequences, for which we used portions ofclamps were not observed. (Fig 3). The septa showed dolipores the ITS1 and ITS2 regions in addition to the 5.8S sequences. with imperforate, dish-shaped parenthesomes with slightly The alignments used can be obtained from TreeBASE (http:// recurved margins (Fig 6). These tulasnelloid parenthesomes www.treebase.org/) under accession number S1629. were observed in all the 20 mycorrhizas analyzed by TEM. Oc- Neighbour-joining (NJ) and a Bayesian likelihood approachcasionally, the hyphal walls were split into two layers and a - were used to estimate the phylogenetic relationships. Thebrillar or slimy mass appeared between the two layers (Figs 4 neighbour-joining analysis was performed in PAUP* (Swoffordand 5, arrows). This phenomenon became very prominent in 2002) using the BIONJ modication of the NJ algorithm to ageing cultures and the slime was then strongly osmiophilic accomplish the observed high genetic variability in the se-(not shown). The combination of this type of parenthesomes quences used (Gascuel 1997). DNA substitution models and in- and the slime bodies in the cell walls was conrmed for dividual model parameters were estimated using the Akaikeall the investigated mycorrhizas and in the Tulasnella isolates. information criterion (AIC) as implemented in Modeltest, ver-The recurved ends of the parenthesome were only detected by sion 3.7 (PosadaCrandall 1998). For the Bayesian approach serial sectioning, since the appearance of the parenthesomes based on Markov chain Monte Carlo (MCMC) we used varied among the sections and may appear attened or bowed MrBayes, version 3.0b4 (HuelsenbeckRonquist 2001). Each in a steeper angle. In three samples we additionally found at, dataset was analyzed using the DNA substitution models esti- imperforate parenthesomes, indicating sebacinoid fungi (Wil- mated using the Akaike information criterion (AIC) inliamsThiol 1989; not shown). In one sample a dome-shaped 5. Diverse tulasnelloid fungi form mycorrhizas1261 Table 1 List of sampled individuals from which tulasnelloid sequences were obtained. Letters and numbers behind thespecies names correspond to species, orchid individual, and root (superscript). Superscript b marks a second sequenceobtained from the same root sample. Clades A-G correspond to the MCMC phylogenetic analysis. The two rDNA regionsfrom the same root listed in each line originate from a single PCR ampliconOrchid speciesnucLSU nrDNA ITS-5.8ScladeGenBank accession no. cladeGenBank accession no. Pleurothallis lilijae C2.1.1 A DQ178035 A DQ178099Pleurothallis lilijae C2.1.2 E DQ178067Pleurothallis lilijae C2.1.3 A DQ178040 A DQ178100Pleurothallis lilijae C2.15 E DQ178080Pleurothallis lilijae C2.17 A DQ178102Pleurothallis lilijae C2.21 E DQ178079Pleurothallis lilijae C2.1MN A DQ178034 A DQ178098Pleurothallis lilijae C2.5MN7F DQ178047 F DQ178069Pleurothallis lilijae C2.MN1 D DQ178063 D DQ178116Pleurothallis lilijae C2.MN5 F DQ178049 F DQ178070Pleurothallis lilijae C2MN2E DQ178068 E DQ178081Pleurothallis lilijae C2MN6B DQ178045Stelis concinna 7.6 A DQ178108Stelis concinna 7.7 A DQ178106Stelis concinna 7.8 A DQ178091Stelis concinna 7.13.2A DQ178109Stelis concinna 7.13.3A DQ178107Stelis concinna 7.13.4A DQ178110Stelis concinna 7.14.2A DQ178112Stelis concinna 7.18.3 A DQ178043 A DQ178095Stelis concinna 7.18.4E DQ178082Stelis concinna 7.19.1A DQ178094Stelis concinna 7.19.3 A DQ178032 A DQ178093Stelis concinna 7.20.1 A DQ178030 A DQ178096Stelis concinna 7.20.2 A DQ178042 A DQ178088Stelis concinna 7.20.3 A DQ178033 A DQ178090Stelis concinna 7.20.4 A DQ178041 A DQ178089Stelis concinna 7.21.1E DQ178075Stelis concinna 7.21.2E DQ178076Stelis concinna 9.2 A DQ178111Stelis concinna 9.3 cultureG DQ178029 G DQ178029Stelis concinna 9.6 A DQ178084Stelis concinna 9.7 A DQ178092Stelis concinna 9.8A DQ178038 A DQ178097Stelis concinna 9.9A DQ178031 A DQ178086Stelis hallii 1.1B DQ178044 B DQ178113Stelis hallii 1.2E DQ178065Stelis hallii 1.2b D DQ178051Stelis hallii 1.4 G DQ178118Stelis hallii 1.6 B DQ178114Stelis hallii 1.7D DQ178050Stelis hallii 1.8 A DQ178085Stelis hallii 1.11 culture E DQ178066 E DQ178066Stelis hallii 1.15 D DQ178057Stelis hallii 1.16 D DQ178055Stelis hallii 1.17A DQ178103Stelis hallii 1.18 A DQ178037 A DQ178104Stelis hallii 1.18bD DQ178053Stelis hallii 1.19 D DQ178059 E DQ178073Stelis hallii 1.19b E DQ178071Stelis hallii 1.21E DQ178072Stelis hallii 1.21b E DQ178077Stelis hallii 1.23 D DQ178060Stelis superbiens C3.5.2 culture A DQ178036 A DQ178036Stelis superbiens C3.5.3E DQ178083Stelis superbiens C3.5.4E DQ178078Stelis superbiens C3.9.2 A DQ178039 A DQ178087Stelis superbiens C3 MN3 D DQ178058 D DQ178117Stelis superbiens C3.MN4 C DQ178046 C DQ178115(continued on next page) 6. 1262 J. P. Suarez et al. Table 1 (continued)Orchid speciesnucLSUnrDNA ITS-5.8S clade GenBank accession no.clade GenBank accession no. Stelis superbiens C3.1MND DQ178056Stelis superbiens C3.2MND DQ178052 ADQ178105Stelis superbiens C3.3MND DQ178054Stelis superbiens C3.4MND DQ178061 ADQ178101Stelis superbiens C3.4MN4 D DQ178062Stelis superbiens C3.5MN5 E DQ178064 EDQ178074Stelis superbiens C3.5MN5bF DQ178048 parenthesome was found that displayed coarse perforationsgrowth rates contrasting with the relative fast growth rate and might thus putatively be assigned to Ceratobasidium (Cur-observed in the fungi isolated. Ascomycetes closest to Crypto- rahSherburne 1992; not shown). No simple-pored ascomy-sporiopsis were the most frequently isolated fungi. cetes were found in the cortical tissue of the 20 investigated samples, although they were present in the velamen (notMolecular identication and phylogenetic analysis shown).of mycorrhiza-associated fungiFungal isolation and molecular identication of isolates The combinations of universal fungal primers yielded PCRproducts preliminarily identied by BLAST searches as closest Fungal growth was observed in only 44 plates out of 108 used to Cryptosporiopsis, Fusarium, Trichoderma (Ascomycota) and for fungal isolation, each one containing ve root pieces. FourBjerkandera, Antrodiella (Basidiomycota). Tulasnella sequences fungal cultures were obtained from a total of 13 plates of Steliswere infrequently obtained, only the primer combination hallii, 15 from 36 plates of Stelis superbiens, 22 from 55 plates of NLMW1/LR5 yielding few PCR products. Primer combinations Stelis concinna, and three from four plates of Pleurothallis lilijae including ITS1F and ITS4 failed to amplify Tulasnella DNA as mycorrhizas. A preliminary molecular identication of thewas already reported by Bidartondo et al. (2003). Sequences fungal isolates was carried out by BLAST searches againstof Sebacinales, basidiomycetes involved in a broad range of the GenBank nucleotide database retrieving the most similarmycorrhizal associations (Wei et al. 2004), were also available sequences (data not shown). The isolated fungi detected, but at lower frequence than Tulasnellales sequences. were mainly ascomycetes closest to Xylaria, Hypoxylon andNo cultures of Sebacinales were obtained from root samples Cryptosporiopsis and less often basidiomycetes closest to Bjer-(Kottke et al. 2007). kandera, Polyporus and Tulasnella. Three cultures were identi-The total number of investigated roots was 134, consider- ed as Tulasnella. These Tulasnella cultures exhibited slowing that three or four roots were collected from each of the77 orchid individuals. Tulasnelloid fungi were detected in 84samples (63 %), including the PCR products obtained by thetulasnelloid specic primer combinations without successfull Fig 1 Ultrastructure of the cortical tissue of Stelis concinna root displaying alive hyphae (h) of equal diameter inFig 2 Degenerating hyphae adjoining collapsed hyphae active host cell (c). (v) Small compartments of orchid cell(ch) in an active cortical cell of Stelis concinna root. vacuoles. Bar [ 1 mm.Bar [ 1 mm. 7. Diverse tulasnelloid fungi form mycorrhizas1263 Fig 3 Branched hypha displaying septa (arrow) without Fig 5 Hypha in cortical root tissue of Stelis concinna clamp formation in root cortical tissue of Stelis concinna. displaying brillar slime between cell wall layers Bar [ 1 mm. (arrowhead) and a doliporus with imperforate, slightly dish-shaped parenthesomes (arrow). Bar [ 0.5 mm.sequencing. The nested PCR conducted in order to selectively amplify Tulasnella DNA using the primer combination ITS1/ 5.8S overview tree (Electronic Appendix C) in such a way TW14 in the rst amplication and the primer combinations that we obtained best consistency with the rooted LSU tree ITS1/ITS4-Tul for the ITS-5.8S region and ITS4-TulR/LR5 or(Fig 7). Portions of ITS1 and ITS2 were added to the 5.8S align- 5.8S-Tul/NL4 for a part of the LSU region, respectively, in the ment where phylogenetic analysis was restricted to suitable second PCR yielded PCR products for the majority of samples.subsets of sequences detected in the 5.8S analysis, nally PCR success was higher with DNA extracted from fresh root resulting in an increase of phylogenetic resolution for these samples and lower with DNA from dried roots.subsets (Figs 8 and 9). The phylogenetic analyses of nucLSU and ITS-5.8S se-Our analysis of proportional differences between se- quences yielded consistent results. Seven clades, which inquences within each clade of the nucLSU D1/D2 yielded 13 the following we refer to as clades A to G, were retrievedTulasnella sequence types. We treated sequences as belonging from the analyses of both ribosomal regions (Fig 7 and Elec- tronic Appendix C). BIONJ (trees not shown) and MCMC yielded similar groupings of Tulasnella clades. Only small var- iations were present in the clade support values. As men- tioned above, the 5.8S tree (Electronic Appendix C) was less resolved than the nucLSU tree. However, the unexpected het- erogeneity displayed by the 5.8S data set made it difcult to nd a suitable outgroup sequence. Therefore, we rooted the Fig 6 Close-up of a median section through the doliporus. The parenthesomes consist of two electron-dense Fig 4 Square section of active hypha of Tulasnella, dis-membranes bordering an internal electron transparent playing mitochondria (m), glycogen rosettes, and brillar zone and show slightly recurved borders (arrows). slime between cell wall layers (arrowheads). Bar [ 1 mm.Bar [ 0,3 mm. 8. 1264 J. P. Suarez et al. Fig 7 Phylogenetic placement of Tulasnella sequences from Stelis hallii, Stelis superbiens, Stelis concinna and Pleurothallis lilijae inferred by MCMC analysis of nuclear rDNA coding for the 5 terminal domain of the large ribosomal subunit (nucLSU). Numbers on branches designate neighbor-joining bootstrap values / MCMC estimates of posterior probabilities (only values exceeding 50 % are shown). Note that genetic distances cannot be directly correlated to branch lengths in the tree, since highly diverse alignment regions were excluded for tree construction. The tree was rooted with Multiclavula mucida AF287875. 9. Fig 8 Phylogenetic placement of Tulasnella sequences, clades A-C, from Stelis hallii, Stelis superbiens, Stelis concinna and Pleurothallis lilijae inferred by MCMC analysis of nuclear ITS-5.8S rDNA. Numbers on branches designate neighbor-joining bootstrap values / MCMC estimates of posterior probabilities (only values exceeding 50 % are shown). Note that genetic dis- tances cannot be directly correlated to branch lengths in the tree, since highly diverse alignment regions were excluded for tree construction. The tree was rooted with Tulasnella sequences from clade D from the analysis of 5.8S rDNA (Electronic Appendix C). 10. 1266 J. P. Suarez et al. Fig 9 Phylogenetic placement of Tulasnella sequences, clades E and F, from Stelis hallii, Stelis superbiens, Stelis concinna and Pleurothallis lilijae inferred by MCMC analysis of nuclear ITS-5.8S. Numbers on branches designate neighbor-joining bootstrap values / MCMC estimates of posterior probabilities (only values exceeding 50 % are shown). Note that genetic distances cannot be directly correlated to branch lengths in the tree, since highly diverse alignment regions were excluded for tree construction. The tree was rooted with the Tulasnella sequence from the Warcup isolate T. violea DQ520097. 11. Diverse tulasnelloid fungi form mycorrhizas 1267to the same sequence type when proportional differences recurved margins consist of two electron-dense membranes were 1 %. Clades D and E comprised four sequence types,bordering an internal electron transparent zone as was al- while the other clades consisted only of one sequence typeready shown by Andersen (1996) for Rhizoctonia repens N.Ber- each (Fig 7, Electronic Appendix B). A direct comparison be-nard Epulorhiza repens (syn.), the anamorph of Tulasnella tween 5.8S-ITS and nucLSU phylogenies was hampered by deliquescens (syn. T. calospora sensu WarcupTalbot 1967 de the fact that publicly available sequence data were restrictedRoberts 1999). The combination of this parenthesome type to either one or the other of these two DNA regions for the and the slime bodies in the cell walls was previously described vast majority of specimens studied so far (Table 1).from mycorrhiza-like associations of the liverwort Aneura pin-Mycobionts of the same sequence type were shared among guis housing Tulasnella (Ligrone et al. 1993), and apparently is orchid species, e.g. sequence types 2, 6, and 12. Tulasnella se-the best structural indication of Tulasnella (Bauer 2004). The quences from all four studied orchid species were present inascomycetes that we isolated from the roots were not found sequence type 1 (Fig 7). Tulasnellas belonging to different se- in the cortical tissue of the orchids, but colonized only the ve- quence types were detected in mycorrhizas from the same lamen. Therefore these ascomycetes cannot be considered as plant and even from the same root piece (Table 1). Six differentmycorrhiza-forming fungi. Without specic primers Tulasnella Tulasnella sequence types were found in mycorrhizas of S. halliiis nearly undetectable (Bidartondo et al. 2003). The use of both and S. superbiens and ve in P. lilijae, while only two sequenceTulasnella-specic primers ITS4-Tul and 5.8S-Tul that target types were detected in mycorrhizas of S. concinna (Fig 7).ITS2 and 5.8S, respectively, increased the success of Tulasnella amplication, but a complete dataset of all tulasnelloid fungi associated with the investigated orchids can currently not be Discussionguaranteed. The accumulation curve of clades vs. number of collected individuals was not saturated (Electronic Appendix D). The importance of orchid mycorrhizal fungi and their role inTulasnella was conrmed as a genus with a range of orchid orchid seed germination are known since Bernards observa-mycorrhiza forming species (Kristiansen et al. 2004; Ma et al. tions (Bernard 1909). In natural conditions, the protocorm2003; McCormick et al. 2004; Shefferson et al. 2005; Warcup does not develop further unless it receives an exogenous car- 1971, 1981, 1985; WarcupTalbot 1967, 1971, 1980). Tulasnella bohydrate supply from a suitable mycorrhizal fungus (Smithcalospora was proposed by Hadley (1970) as an universal orchidRead 1997). Although not demonstrated in nature so far, symbiont considering the capacity to establish in vitro symbi- this dependence will also account for germination of epiphyticotic associations with a broad range of orchid species. There orchids. Our nding of constant colonization of roots in con- are, however, taxonomic problems concerning the species tact with the bark supports the view of Benzing (1982) that concept in Tulasnella (Roberts 1999). The T. calospora strains the role of the fungi may be crucial also for the adult epiphytic (T. deliquescens de Roberts 1999) isolated by Warcup from ter- orchids. The most important benet for the orchid may be to restrial Australian orchids (WarcupTalbot 1967) belong to retain the fungus in order to assure further seed germination.two separate clades according to our nucLSU tree (Fig 7). The Saprotrophic capabilities, documented for several Tulasnellastrains are even more clearly distinguished in the 5.8S-ITS species (Roberts 1999), could explain the growth of Tulasnellatree (Fig 8), clustering with Tulasnellas detected in terrestrial on tree bark, including the capacity of fruiting close to colo- orchids from a wide range of localities, such as Spathaglottis nized roots (J.P.S., pers. obs.), and promotion of seed germina-plicata from Singapore and Epipactis gigantea from California. tion. We observed decline of vitality of hyphae after only oneOur molecular analyses indicate that T. calospora is likely to night of plant storage in the laboratory, independent ofcomprise several distinct species. Taxonomic problems ac- whether or not the samples were cooled. The decline of vital- count also for T. violea and T. asymmetrica, as T. violea accession ity was rst indicated by loss of stainability of the pelotons, AY293216 appears close to T. pruinosa AF518662 in the nucLSU which appeared yellow instead of blue in the light microscope.phylogenetic tree, while the Warcup T. violea isolate from the TEM revealed very rare occurrences of vital pelotons, but orchid Thelymitra sp. clustered separately; the Warcup abundant collapsed hyphae in this material. No isolates ofT. asymmetrica strains (T. pinicola de Roberts 1999) isolated Tulasnellales were obtained from the stored samples and PCR from the orchid Thelymitra sp. fall into two groups (Fig 7). On amplication success was very low, although only fully tur- the other hand Tulasnellas with quite different trophic strat- gescent roots were processed. It is rather unlikely that theegies are displayed as closely related in the nucLSU tree, e.g. fast decline of hyphal vitality was linked to carbon shortage the mycobiont of the myco-heterotrophic liverwort Cryptothal- as plenty of starch grains were visible in the root tissue andlus mirabilis (AY192482) that also forms ectomycorrhizas with large amounts of glycogen were found in the vital hyphae. Dis-Pinus and Betula (Bidartondo et al. 2003), and T. irregularis ruption of the extraradical mycelium or increase of plant (AY243519), a Warcup isolate from Dendrobium dicuphum, an defense reaction could be involved (Hadley 1982). So far, the Australian orchid. Attempts are currently undertaken to col- reason for this fast decline is unclear but it might erroneouslylect and describe Tulasnellas occurring on the bark of the imply the impression of low hyphal colonization rate. sampled trees in our research area and to induce basidia for- Our combination of ultrastructural research and DNA se- mation in the isolate cultures. Morphological description quencing revealed that Tulasnella species were regularly asso-and determination combined with sequence typing appeared ciated with the epiphytic Stelis and Pleurothallis species. Weas promising tools to further clarify relationships between successfully isolated three of the associated Tulasnellas, these poorly studied mycobionts. identity proven by DNA sequences and septal porus type. Irrespective of the taxonomic problems, available se- The at bell-shaped, imperforate parenthesomes with slightlyquence data from nucLSU and ITS-5.8S made it possible to 12. 1268 J. P. Suarez et al.compare the seven Tulasnella clades of symbionts of epiphytic strains by the National Institute of Agrobiological Sciences orchids studied here with named Tulasnella species and my-(NIAS), Japan, is also acknowledged. corrhizal Tulasnellas mostly from terrestrial orchids. Tulas- nellas of the studied epiphytic orchid species were distinct Supplementary data from so far known Tulasnellas associated with terrestrial or- chids. Tulasnelloid fungi associated with the terrestrial tem- Supplementary data associated with this article can be found, perate orchids Cypripedium spp. (subfamily Cypripedioideae) in the online version, at 10.1016/j.mycres.2006.08.004. and Dactylorhiza majalis (AY634130) (subfamily Orchidoideae) were displayed in a basal position with respect to the Tulas- nella sequence types from Stelis and Pleurothallis in the nucLSUreferences phylogeny (Fig 7). In the 5.8S nrDNA phylogeny, the tulasnel- loid fungi associated with Dactylorhiza majalis (AY634130), Orchis purpurea (AJ549121), Spathaglottis plicata (AJ313457), Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, Miller W, Ophrys sphegodes (AJ549122) (all subfamily Orchidoideae) and Lipman DJ, 1997. Gapped BLAST and PSI-Blast: a new genera- Cypripedium fasciculatum (AY966883) appeared in a basal posi-tion of protein database search programs. Nucleic Acids Re-search 25: 33893402. tion relative to Tulasnella sequence types from Stelis and Pleu- Andersen TF, 1996. A comparative taxonomic study of Rhizocto- rothallis (Electronic Appendix C). Molecular phylogeneticnia sensu lato employing morphological, ultrastructural and analyses of the family Orchidaceae are consistent in respect molecular methods. Mycological Research 100: 117128. to a basal position of Cypripedioideae and Orchidoideae com-Bauer R, 2004. Basidiomycetous interfungal cellular interactions pared to Epidendroideae (e.g. Cameron et al. 1999). Switches a synopsis. In: Agerer R, Piepenbring M, Blanz P (eds), Frontiers from terrestrial to epiphytic habit or back were found to be in Basidiomycote Mycology. IHW-Verlag, Eching, Germany, pp. major driving forces in radiation and specialization of orchids325337. Bayman P, Lebron LL, Tremblay RL, Lodge DJ, 1997. Variation in (Cameron 2002, 2005), and beside pollinator relationships my-endophytic fungi from roots and leaves of Lepanthes (Orchida- corrhizal interactions are now recognized crucial for orchid ceae). New Phytologist 135: 143149. evolution (Taylor et al. 2003). However, more species need to Bermudes D, Benzing DH, 1989. Fungi in neotropical epiphyte be sampled including terrestrial orchids of the study site to ar-roots. BioSystems 23: 6573. rive at convincing conclusions about coevolution between or-Bernard N, 1909. Levolution dans la symbiose. Les orchidees et chids and their mycobionts.leur champignons commenseaux. Annales des Sciences Nature- Our results show differences in the number of Tulasnella lles Botanique, Paris 9: 1196. Benzing DH, 1982. Mycorrhizal infections of epiphytic orchids symbionts associated with one orchid species. Six Tulasnellain southern Florida. American Orchid Society Bulletin 51: sequence types were associated with one individual orchid618622. species of S. hallii and S. superbiens, ve in P. lilijae, but only Bidartondo MI, Bruns TD, Wei M, Sergio C, Read D, 2003. Spe- two sequence types were found with S. concinna. We cannot ex-cialized cheating of the ectomycorrhizal symbiosis by an epi- clude the possibility that the differences in numbers of Tulas-parasitic liverwort. Proceedings of the Royal Society of London, nella symbionts will vanish when a higher number of orchid Series B, Biological Sciences 270: 835842. Bidartondo MI, Burghardt B, Gebauer G, Bruns TD, Read DJ, 2004. specimen and roots will be examined. However, preferencesChanging partners in the dark: isotopic and molecular evi- for fungal partners have been demonstrated in other epiphyticdence of ectomycorrhizal liaisons between forest orchids orchids (Otero et al. 2002, 2004). In several cases we found thatand trees. Proceedings of the Royal Society of London, Series B, one orchid individual was associated with Tulasnellas fromBiological Sciences 271: 17991806. more than one clade even in the same root segment. Obvi-Bougoure JJ, Bougoure DS, Cairney WG, Dearnaley DW, 2005. ITS- ously, diverse Tulasnellas form mycorrhizas with the green, RFLP and sequence analysis of endophytes from Acianthus, epiphytic, pleurothallid orchids in the Andean cloud forest. Caladenia and Pterostylis (Orchidaceae) in southeasternQueensland. Mycological Research 109: 452460. Whether these distinct fungi are crucial for seed germination Cameron KM, Chase MW, Whitten WM, Kores PJ, Jarrell DC, needs to be veried experimentally. In case of the Rhizocto-Albert VA, Yukawa T, Hills HG, Goldman DH, 1999. A nias, seed germination was stimulated by non-optimal myco- phylogenetic analysis of the Orchidaceae: evidence from bionts, but symbionts that were not fully compatible resultedrbcL nucleotide sequences. American Journal of Botany 86: in high seedling mortality (Rasmussen 2002). Our analyses208224. indicate that efcient rehabilitation of epiphytic orchids in Cameron KM, 2002. Intertribal relationships within Orchidaceae nature and recruitment in the nursery probably requires theas inferred from analyses of ve plastid genes. In Abstracts ofBotany 2002, Madison, WI, USA, abstract 116. Also available at usage of distinct Tulasnella species as orchid mycobionts.website http:// www.botany2002.org/section12/abstracts/33.shtml. Cameron KM, 2005. Leave it to the leaves: a molecular phyloge-netic study of Malaxideae (Epidendroideae, Orchidaceae). American Acknowledgements Journal of Botany 96: 10251032. Currah RS, Sherburne R, 1992. Septal ultrastructure of some This research was generously supported by the Deutsche For-fungal endophytes from boreal orchid mycorrhizas. Mycologi- schungsgemeinschaft (DFG project FOR 402). We thank thecal Research 96: 583587. Currah RS, Zelmer CD, Hambleton S, Richardson KA, 1997. Fungi Fundacion Cientca San Francisco for providing research fa- from orchid mycorrhizas. In: Arditti J, Pridgeon AM (eds), Or- cilities, Lorena Endara for help in orchid identication, andchid Biology: Reviews and Perspectives, VII. Kluwer Academic Paulo Herrera for help in laboratory work. The supply of fungalPublishers, Dordrecht, pp. 117170. 13. Diverse tulasnelloid fungi form mycorrhizas1269 Fortin JA, Piche Y, 1979. Cultivation of Pinus strobus root-hypocotyl Luer CA, 1986b. Icones Pleurothallidinarum III, Systematics of explants for synthesis of ectomycorrhiza. New Phytologist 83: Pleurothallis (Orchidaceae). Monographs in Systematic Botany 20: 109119.1109. Gascuel O, 1997. BIONJ: An improved version of the NJ algorithm Ma M, Tan TK, Wong SM, 2003. Identication and molecular based on a simple model of sequence data. Molecular Biology phylogeny of Epulorhiza isolates from tropical orchids. Myco- and Evolution 14: 685695.logical Research 107: 10411049. Goh CJ, Sim AA, Lim G, 1992. Mycorrhizal associations in some McCormick MK, Whigham DF, ONeill J, 2004. Mycorrhizal diver- tropical orchids. Lindleyana 7: 1317.sity in photosynthetic terrestrial orchids. New Phytologist 163: Hadley G, 1970. Non-specicity of symbiotic infection in orchid 425438. mycorrhiza. New Phytologist 69: 10151023.Noske N, 2004. Effekte anthropogener Sto rung auf die Diversita t Hadley G, 1982. Orchid Mycorrhiza. In: Arditti J (ed), Orchid Biology:kryptogamischer Epiphyten (Flechten, Moose) in einem Reviews and Perspectives II. Cornell University Press, Ithaca, NY,Bergregenwald in Sudecuador. Dissertation, Gottingen. pp. 83118. Nylander JAA, 2004. MrModeltest 2.2. Program distributed Hadley G, Williamson B, 1972. Features of mycorrhizal infection by the author. Evolutionary Biology Centre, Uppsala in some Malayan orchids. New Phytologist 71: 11111118. University. Hamilton LS, Juvik JO, Scatena FN, 1995. The Puerto Rico tropical Otero JT, Ackerman JD, Bayman P, 2002. Diversity and host cloud forest symposium Introduction and workshop synthe- specicity of endophytic Rhizoctonia-like fungi from tropical sis. Ecological Studies 110: 119.orchids. American Journal of Botany 89: 18521858. Homeier J, 2004. Baumdiversitat, Waldstruktur und Wachstums- Otero JT, Ackerman JD, Bayman P, 2004. Differences in mycor- dynamik zweier tropischer Bergregenwalder in Ecuador und rhizal preferences between two tropical orchids. Molecular Costa Rica. Dissertationes Botanicae 391: iv, 1207. Ecology 13: 23932404. Huelsenbeck JP, Ronquist FR, 2001. MrBayes: Bayesian inferencePereira OL, Rollemberg CL, Borges AC, Matsuoka K, of phylogenetic trees. Bioinformatics 17: 754755.Kasuya MCM, 2003. Epulorhiza epiphytica sp. nov. isolated Huelsenbeck JP, Larget B, Miller RE, Ronquist F, 2002. Potentialfrom mycorrhizal roots of epiphytic orchids in Brazil. applications and pitfalls of Bayesian inference of phylogeny. Mycoscience 44: 153155. Systematic Biology 51: 673688. Pereira OL, Kasuya MCM, Borges AC, Fernandes de Araujo E, 2005. Julou T, Burghardt B, Gebauer G, Berveilleir D, Damesin C,Morphological and molecular characterization of mycorrhizal Selosse MA, 2005. Mixotrophy in orchids: insight from a com-fungi isolated from neotropical orchids in Brazil. Canadian parative study of green individuals and non-photosynthestic Journal of Botany 83: 5465. individuals of Cephanlanthera damasonium. New Phytologist 166:Posada D, Crandall KA, 1998. Modeltest: testing the model of DNA 639653.substitution. Bioinformatics 14: 817818. Karnovsky MJ, 1965. A formaldehyde glutaraldehyde xation ofPridgeon AM, 1987. The velamen and exodermis of orchid roots. high osmolarity for use in electron microscopy. Journal of Cell In: Arditti J (ed), Orchid Biology: Reviews and Perspectives, IV. Biology 27: 137138.Cornell University Press, Ithaca, NY, pp. 140192. Katoh K, Kuma K, Toh H, Miyata T, 2005. MAFFT version 5: im-Rasmussen HN, 2002. Recent developments in the study of orchid provement in accuracy of multiple sequence alignment. mycorrhiza. Plant and Soil 244: 149163. Nucleic Acids Research 33: 511518. Richter M, 2003. Using epiphytes and soil temperatures for eco- Kottke I, Guttenberger M, Hampp R, Oberwinkler F, 1987. An in climatic interpretations in Southern Ecuador. Erdkunde 57: vitro method for establishing mycorrhizae on coniferous tree161181. seedlings. Trees 1: 191194.Rivas M, Warner J, Bermudez M, 1998. Presencia de micorrizas en Kottke I, 2002. Mycorrhizae Rhizosphere determinants of orqudeas de un jardn botanico neotropical. Revista de Biologa plant communities. In: Waisel Y, Eshel A, Kafka U (eds), Tropical 46: 211216. Plant Roots: The Hidden Half, 3rd edn. Marcel Dekker, Roberts P, 1999. Rhizoctonia-forming Fungi: a taxonomic guide. Royal New York, pp. 919932.Botanic Gardens, Kew. Kottke I, Haug I, Preuing M, Setaro S, Suarez JP, Wei M,Selosse M-A, Faccio A, Scappaticci G, Bonfante P, 2004. Chloro- Nebel M, Oberwinkler F, 2007. Guilds of mycorrhizal fungi phyllous and achlorophyllous specimens of Epipactis micro- and their relation to trees, ericads, orchids and liverworts in phylla (Neottieae, Orchidaceae) are associated with a neotropical mountain rain forest. Basic and Applied Ecology,ectomycorrhizal septomycetes, including trufes. Microbial in press. Ecology 47: 416426. Kristiansen KA, Freudenstein JV, Rasmussen FN, Rasmussen HN,Shefferson RP, Wei M, Kull T, Taylor L, 2005. High specicity 2004. Molecular identication of mycorrhizal fungi in Neuwie- generally characterizes mycorrhizal association in rare ladys dia veratrifolia (Orchidaceae). Molecular Phylogenetics and Evolution slipper orchids, genus Cypripedium. Molecular Ecology 14: 33: 251258.613626. Kristiansen KA, Taylor DL, Kjoller HN, Rasmusen N,Smith SE, Read DJ, 1997. Mycorrhizal Symbiosis, 2nd edn. Academic Rosendahl S, 2001. Identication of mycorrhizal fungi fromPress., San Diego, California. single pelotons of Dactylorhiza majalis (Orchidaceae) using Spurr AR, 1969. A low viscosity epoxy resin embedding medium single-strand conformation polymorphism and mitochon- for electron microscopy. Journal of Ultrastructure Research 26: drial ribosomal large subunit DNA sequences. Molecular Ecol-3143. ogy 10: 20892093.Swofford DL, 2002. PAUP*4.0 Phylogenetic Analysis Using Ligrone R, Pocock K, Duckett JG, 1993. A comparative ultrastruc-Parsimony (*and Other Methods). Sinauer Associates., tural study of endophytic basidiomycetes in the parasitic Sunderland, MA. achlorophyllous hepatic Cryptothallus mirabilis and the closely Taylor DL, Bruns TD, 1997. Independent, specialized invasions of related allied photosynthetic species Aneura pinguis (Metzger-ectomycorrhizal mutualism by two nonphotosynthetic or- iales). Canadian Journal of Botany 71: 666679. chids. Proceedings of the National Academy of Sciences, USA 94: Luer CA, 1986a. Icones Pleurothallidinarum I, Systematics of45104515. Pleurothallidinae (Orchidaceae). Monographs in Systematic Bot-Taylor DL, Bruns TD, 1999. Population, habitat and genetic cor- any 15: 181. relates of mycorrhizal specialization in the cheating orchids 14. 1270 J. P. Suarez et al. Corallorhiza maculata and C. mertensiana. Molecular Ecology 8: Williams PG, Thiol E, 1989. Ultrastructural evidence for the17191732.identity of some multinucleate Rhizoctonias. New Phytologist Taylor DL, Bruns TD, Szaro TM, Hodges SA, 2003. Divergence in112: 513518.mycorrhizal specialization with Hexalectris spicata (Orchida-Zettler LW, Delaney TW, Sunley JA, 1998. Seed propagation of theceae), a nonphotosynthetic desert orchid. American Journal of epiphytic green-y orchid, Epidendrum conopseum. SelbyanaBotany 90: 11681179. 19: 249253. Tremblay RL, Zimmerman JK, Lebron L, Bayman P, Sastre I,Axelrod F, Alers-Garca J, 1998. Host specicity andlower reproductive success in the rare endemic Puertofurther readingRican orchid Lepanthes caritensis. Biological Conservation 85:295304. Warcup JH, 1971. Specicity of mycorrhizal association in some Cullings KW, 1994. Molecular phylogeny of the MonotropoideaeAustralian terrestrial orchids. New Phytologist 70: 4146.(Ericaceae) with a note on the placement of the Pyroloideae. Warcup JH, 1981. The mycorrhizal relationships of AustralianJournal of Evolutionary Biology 7: 501516.orchids. New Phytologist 87: 371381. Gardes M, Bruns TD, 1993. ITS primers with enhanced specicity Warcup JH, 1985. Rhizanthella gardneri (Orchidaceae), its Rhizoctoniafor basidiomycetes: application to the identication of my-endophyte and close association with Melaleuca uncinatacorrhizae and rusts. Molecular Ecology 2: 113118.(Myrtaceae) in western Australia. New Phytologist 99: 273280. Sampaio JP, Wei M, Gadanho M, Bauer R, 2002. New taxa in the Warcup JH, Talbot PHB, 1967. Perfect states of RhizoctoniasTremellales: Bulleribasidium oberjochense gen. et sp. nov., Papil-associated with orchids I. New Phytologist 66: 631641.iotrema bandonii gen. et sp. nov. and Fibulobasidium murrhard- Warcup JH, Talbot PHB, 1971. Perfect states of Rhizoctoniastense sp. nov. Mycologia 94: 873887.associated with orchids II. New Phytologist 70: 3540. Taylor DL, 1997. The evolution of myco-heterotrophy and specicity Warcup JH, Talbot PHB, 1980. Perfect states of Rhizoctonias as-in some North American orchids. PhD thesis, University ofsociated with orchids III. New Phytologist 86: 267272.California at Berkeley, CA. Wei M, Selosse M-A, Rexer K, Urban A, Oberwinkler F, 2004. Vilgalys R, Hester M, 1990. Rapid genetic identication and map-Sebacinales: a hitherto overlooked cosm of heterobasidiomy-ping of enzymatically amplied ribosomal DNA from severalcetes with a broad mycorhizal potential. Mycological ResearchCryptococcus species. Journal of Bacteriology 172: 42384246.108: 10031010. White TJ, Bruns TD, Lee SB, Taylor JW, 1990. Amplication and Wells K, Bandoni R, 2001. Heterobasidiomycetes. In:direct sequencing of fungal ribosomal RNA genes for phylo-McLaughlin DJ, McLaughlin EG, Lemke PA (eds), The Mycota.genetics. In: Innis MA, Gelfand H, Sninsky JS, White TJ (eds),Vol. VII. Part B. Systematics and Evolution. Springer Verlag,PCR-Protocols and applications: A Laboratory Manual. AcademicHeidelberg, pp. 85120.Press, San Diego, pp. 315322.