sign in sign up
<<
Home , Amylocorticiales, Anomoporia, Anomoporia bombycina, Anthurus, Botryobasidiaceae, Botryobasidium

fungal ecology 6 (2013) 256e268

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/funeco

Evolution of nutritional modes of (, ) as revealed from publicly available ITS sequences

Vilmar VELDREa, Kessy ABARENKOVa, Mohammad BAHRAMa, Florent MARTOSc,1, ~ Marc-Andre SELOSSEb, Heidi TAMMa, Urmas KOLJALGa, Leho TEDERSOOa,d,* aInstitute of Ecology and Earth Sciences, University of Tartu, 40 Lai, 51005 Tartu, Estonia bCentre d’Ecologie Fonctionnelle et Evolutive, CNRS, UMR 5175, 1919 Route de Mende, 34293 Montpellier cedex 5, France cUniversitedeLaReunion, Peuplements Vegetaux et Bioagresseurs en Milieu Tropical (UMR C53), Equipe Dynamiques ecologiques au sein des ecosystemes naturels, 15 Avenue Rene Cassin, BP 7151, 97715 Saint-Denis cedex 9, France dNatural History Museum of Tartu University, 46 Vanemuise, 51046 Tartu, Estonia article info abstract

Article history: Fungi from the Ceratobasidiaceae have important ecological roles as pathogens, Received 1 September 2012 saprotrophs, non-mycorrhizal endophytes, orchid mycorrhizal and ectomycorrhizal sym- Revision received 5 March 2013 bionts, but little is known about the distribution and evolution of these nutritional modes. Accepted 6 March 2013 All public ITS sequences of Ceratobasidiaceae were downloaded from databases, annotated Available online 24 April 2013 with ecological and taxonomic metadata, and tested for the non-random phylogenetic Corresponding editor: distribution of nutritional modes. Phylogenetic analysis revealed six main within Havard€ Kauserud Ceratobasidiaceae and a poor correlation between molecular phylogeny and morphologi- calecytological characters traditionally used for . Sequences derived from soil Keywords: (representing putative saprotrophs) and orchid clustered together, but remained distinct from pathogens. All nutritional modes were phylogenetically conserved in the Ceratobasidiaceae based on at least one index. Our analyses suggest that in general, Evolutionary ecology autotrophic orchids form root symbiosis with available Ceratobasidiaceae isolates in soil. Ectomycorrhiza-forming capability has evolved twice within the Ceratobasidiaceae and it Phylogeny had a strong influence on the evolution of mycoheterotrophy and host specificity in certain orchid taxa. Saprotrophepathogen continuum ª 2013 Elsevier Ltd and The British Mycological Society. All rights reserved. Thanatephorus

Introduction decomposers of dead organic matter, provide mineral nutri- tion to as mycorrhizal symbionts, or devastate Fungi play a fundamental role in nutrient and carbon cycling populations as phytopathogens. Despite their major eco- in terrestrial ecosystems and sediments. They act as logical and economic importance, the taxonomic and

* Corresponding author. Institute of Ecology and Earth Sciences, University of Tartu, 40 Lai, 51005 Tartu, Estonia. Tel./fax: þ372 7376222. E-mail address: [email protected] (L. Tedersoo). 1 Current address: School of Life Sciences, University of KwaZulu-Natal, Private Bag X01, Scottsville, Pietermaritzburg 3209, South Africa. 1754-5048/$ e see front matter ª 2013 Elsevier Ltd and The British Mycological Society. All rights reserved. http://dx.doi.org/10.1016/j.funeco.2013.03.004 Evolution of nutritional modes in Ceratobasidiaceae 257

ecological complexity of many important taxa remains poorly currently considered a peripheral member of the Can- understood because of the ephemeral habit, and the lack of tharellales, although its phylogenetic relations to other fruit bodies in most fungal groups. members of this group are not well resolved (Moncalvo et al. The family Ceratobasidiaceae (Cantharellales, Basidiomy- 2006; Hibbett et al. 2007). cota) consists of the two closely related sexual genera, Cera- Most attention to the members of the Ceratobasidiaceae tobasidium and Thanatephorus along with their Rhizoctonia has been sparked by their role as widespread soil-borne crop asexual forms. They form one of such ‘cryptic’ fungal groups pathogens. Their necrotrophic capability is remarkably non- that play important ecological roles as crop pathogens, orchid specific and affects a multitude of plant taxa. There is a con- mycorrhizal symbionts, saprotrophs and endophytes tinuum between the necrotrophic and saprotrophy (Parmeter 1970; Sneh et al. 1996; Roberts 1999). These fungi in the Ceratobasidiaceae. Most strains possess some sapro- spend most of their life cycle in the morphologically simple trophic capability, but aggressive pathogens are poor sapro- asexual (i.e. ‘anamorphic’) stage during which they can only trophic competitors and depend on nutrients acquired from be observed macroscopically as irregular sclerotia or, for living plant tissue (Papavizas 1970). By contrast, strains that phytopathogens, as necrotic lesions in the tissues of a tre- have lost their ability to infect and cause serious damage to mendous range of host plants. Even when the sexual (i.e. living organs, may function as commensal or even mutualistic ‘teleomorphic’) stage occurs, it represents an inconspicuous, endophytes (i.e. fungi that grow diffusely in tissues, without fragile, web-like layer of generative hyphae covering living causing any visible symptoms; Sen et al. 1999) and increase leaves and stems of hosts, or plant debris (Roberts 1999). Dif- their hosts’ resistance to pathogenic strains (Sneh 1998). ficulties with the induction of fruiting in culture and the lack However, little is known about the frequency and role of of morphological characters to distinguish among biological Ceratobasidiaceae endophytes (Sen et al. 1999). have considerably complicated taxonomic and eco- As a further major interaction, the Ceratobasidiaceae logical research on these fungi. includes a large number of taxa that form orchid mycorrhiza To provide a practical method for identification, plant (OrM). Orchids have an unusual relationship with their pathologists implemented a tractable test based on number of mycorrhizal fungi compared to other mycorrhizal plants: they nuclei per cell and anastomosis compatibility in co-culture receive all nutrients, including carbon, from their fungi during instead of morphology-based binomial taxonomy (Carling their heterotrophic germination (Smith & Read 2008; 1996; Sharon et al. 2008). Traditionally, strains of binucleate Dearnaley et al. 2012). At the adult stage, most orchids develop Rhizoctonia (BNR; binucleate strains are occasionally placed in photosynthetic capability, but clearly continue to benefit from the anamorphic Ceratorhiza) and multinucleate Rhi- their mycobionts by receiving mineral nutrients. Some spe- zoctonia (MNR) are considered anamorphs of Ceratobasidium cies associated with Ceratobasidium allow a net carbon flow and Thanatephorus, respectively, although a few species of from adult orchids to the (Cameron et al. 2006, 2008), Thanatephorus are binucleate (Roberts 1999). Both the MNR and while in some other species, adult orchids still obtain carbon BNR are separated into anastomosis groups (AG), including 13 from fungi (Selosse & Roy 2009; Yagame et al. 2012). Thus, the MNR AGs (AG-1 to AG-13) and 16 BNR AGs (AG-A to AG-S, balance between costs and benefits for fungi remains poorly excluding AG-J, AG-M and AG-N; Sharon et al. 2008), some of understood in OrM (Dearnaley et al. 2012), and we treat here which are further divided into anastomosis subgroups based the association between autotrophic orchids and their on more detailed analysis of anastomosis compatibility, mycorrhizal fungi as symbiotic along the mutualismeparasi- morphology, pathogenicity and other criteria. All multi- tism continuum (Egger & Hibbett 2004). As a side finding in the nucleate AGs correspond to the teleomorphic species T. research on OrM, it was discovered that some Ceratobasidia- cucumeris (anamorph ). The binucleate AG-A, ceae isolates also form ectomycorrhiza (EcM) (Warcup 1991; AG-B(o), AG-D, AG-P and AG-Q correspond to Ceratobasidium Yagame et al. 2008, 2012; Bougoure et al. 2009), and indeed cornigerum, whereas AG-Ba and AG-Bb correspond to some Ceratobasidiaceae have been sporadically reported in Ceratobasidium setariae. However, teleomorphs of the remain- community analyses of EcM fungi (e.g. Rosling et al. 2003). ing binucleate AGs remain unknown (Roberts 1999). However, the phylogenetic and biogeographic distribution of Teleomorphs of Ceratobasidiaceae are characterized by these relatively uncommon EcM groups remains poorly aseptate basidia (a derived, apomorphic feature), but also self- understood (Tedersoo et al. 2010). replicating and large indeterminate sterigmata Development of molecular methods has led to rapid (ancestral features; Roberts 1999). Numerous minor genera unravelling of the systematics and ecology of microbes in the have been described within the Ceratobasidiaceae, but later past few decades. Barcoding with the Internal Transcribed synonymized with either Ceratobasidium (characterized by a Spacer (ITS) region of the ribosomal DNA locus has become a single layer of basidia arising from horizontally branching standard means of identification in fungi (Schoch et al. 2012). hyphae of <10 mm in diameter) or Thanatephorus (charac- ITS sequences offer suitable resolution for identification of the terized by multiple layers of basidia arising from vertically Ceratobasidiaceae strains and provide an invaluable tool for branching hyphae of >10 mm in diameter; Roberts 1999). differentiating AGs and their subtypes both from cultured Roberts (1999) included the genera and Scotomyces in strains, soil and plant tissue (Johanson et al. 1998; Sharon et al. the Ceratobasidiaceae, but molecular phylogenetic analyses 2006). Phylogenetic analyses suggest that most AGs are place both taxa outside Ceratobasidiaceae (Larsson 2007; K-H. monophyletic, but still correspond to complexes of molecular Larsson, pers. comm. 02.08.2012). We, therefore, consider the and ecological species (Gonzales et al. 2001, 2006; Sharon et al. Ceratobasidiaceae family to consist of the two genera Cerato- 2006, 2008). AG subtypes usually have different ecological and Thanatephorus only. The Ceratobasidiaceae is niches and their within-group ITS sequence similarity is over 258 V. Veldre et al.

97e98 %; AG subtypes could be, therefore, considered as Checker (Nilsson et al. 2010a) revealed 19 reverse comple- molecular operational taxonomic units (MOTUs; Vilgalys & mentary and ten chimeric sequences. The reverse comple- Cubeta 1994; Gonzales et al. 2001; Sharon et al. 2006). This mentary sequences were re-orientated and kept in the level of ITS sequence similarity roughly matches the barcod- analyses. All chimeric sequences, 361 sequences representing ing gap in most groups of Basidiomycota and other taxonomic groups or , 26 highly divergent (Schoch et al. 2012). The International Nucleotide Sequence sequences that potentially belong to Ceratobasidiaceae and Databases (INSD) have accumulated over 3 000 ITS sequences 946 partial (>100 bp missing) or low-quality sequences were of the Ceratobasidiaceae that provide a rich reference material excluded. Sequences with low read quality or of chimeric for molecular identification and inferring evolutionary nature were tagged accordingly in the INSD copy of the UNITE hypotheses. database as implemented in the PlutoF workbench Here, we utilize the wealth of accumulated ITS sequence (Abarenkov et al. 2010b). The remaining 2 257 nearly full- data to address the evolution of ecological strategies, biogeo- length ITS sequences formed the ‘full conservative’ dataset graphic patterns and host specificity in the Ceratobasidiaceae. that is central in the subsequent analyses. Because AG-D-II In particular, we hypothesized that both the pathogenic and and AG-H were only represented by notoriously low-quality EcM isolates are phylogenetically clustered but distinct from or incomplete sequences, a ‘liberal’ dataset including an other modes of nutrition, assuming that these biotrophic additional 285 ITS sequences was also analyzed. In addition, strategies require specific genetic adaptations. We also tested all 44 nuclear rDNA Large Subunit (LSU) sequences (at least a null hypothesis that both the OrM and soil-derived sapro- 900 bp) of the Ceratobasidiaceae were retrieved from INSD and trophic isolates are evenly distributed across the Ceratobasi- UNITE. These were supplemented with a reference dataset of diaceae family and we expected that orchids preferentially 95 sequences covering the Cantharellales and most other associate with non-pathogenic isolates. Based on the results orders of the to determine a suitable out- of several case studies that focused on specific host taxa or group for the ITS-based analysis. geographic regions (e.g. Kuninaga et al. 1997; Waterman et al. Metadata on isolation source, interacting taxon, locality 2011), we predicted that several Ceratobasidiaceae MOTUs and AG were compiled for all full-length ITS sequences from display strong patterns in biogeography and host specificity. INSD and associated publications (Table S1). Isolation source and substrate formed a basis for the statistical metadata analysis. The main sources of isolation included diseased Materials and methods tissue of crop plants (implying pathogenic interaction), roots of orchids (OrM interaction) and soil (see below), followed by Data assembly ectomycorrhizal root tips (EcM interaction) and healthy non- mycorrhizal root or symptomless above-ground tissue of All available full-length ITS sequences belonging to the Cera- wild plants (endophytic interaction; Sen et al. 1999). Some tobasidiaceae were retrieved from INSD in Jun. 2010. A direct non-mycorrhizal isolates from crop plants were found to be search by taxonomic assignment to the family Ceratobasi- non-pathogenic or even beneficial in experimental studies, diaceae or the genera Ceratobasidium, Thanatephorus, Rhizocto- and were therefore treated as endophytes. Soil-derived nia or Ceratorhiza in GenBank (3 167 sequences), search for sequences were separated into entries obtained from soils matching unidentified sequences (299 additional sequences) of: (1) crop fields by plant pathologists in studies targeting using the web tool emerencia (Nilsson et al. 2005) and the 1 000 pathogens; or (2) natural or semi-natural ecosystems best BLAST matches for highly deviating sequences AJ633124 addressing soil fungal communities. The first category was (49 matching sequences) and AB000014 (one matching included among pathogens, while the latter group was con- sequence) retrieved a total of 3 516 INSD entries. In addition, sidered suggestive of saprotrophy, although we anticipate 84 sequences were included from the UNITE database that pathogenic and EcM strains may also be present in the (Abarenkov et al. 2010a). All sequences were aligned with soil of natural habitats. All annotated metadata are publicly MAFFT (Katoh & Toh 2008) and inspected by use of the pro- available in the UNITE database and through a link-out func- gram Seaview (Gouy et al. 2010) to identify short and low- tion in the European Nucleotide Archive (ENA). quality ITS sequences. Non-Ceratobasidiaceae taxa, other genes, reverse complementary, chimeric and low-quality Phylogenetic analyses sequences were detected as outlined in Tedersoo et al. (2011). Sequences of Ceratobasidiaceae were often mis- The full-length ITS sequences ranged from 580 to 680 labelled or misidentified in INSD. Of 3 188 sequences named as nucleotides. The final alignments subjected to phylogeny Ceratobasidiaceae, 51 ITS sequences (1.6 %) belonged to other reconstruction were created with PRANKSTER (Loytynoja€ & families within Cantharellales or to other orders of Agar- Goldman 2005) using default parameters, and corrected icomycetes, especially (incl. 28 sequences of Bjer- manually. Two segments of ITS1 (ca. 15e30 and ca. 20e60 bp, kandera adusta; Table S1). Conversely, 10 ITS sequences varying among sequences) and one segment of ITS2 (ca. belonging to the Ceratobasidiaceae were wrongly assigned to 10e20 bp) were omitted from phylogenetic analyses, because other groups of fungi or other kingdoms in INSD. In these were highly variable and could not be reliably aligned addition, 249 Ceratobasidiaceae sequences were assigned to across the whole dataset. The length of individual sequences the kingdom Fungi or Basidiomycota without further was thereby reduced to 540e580 nucleotides. The final full specification. Visual inspection and the programs Reverse alignment spanned 1 178 sites, of which 571 were variable and Complementary Checker (Nilsson et al. 2010b) and Chimera 444 were informative. For Maximum Likelihood (ML) and Evolution of nutritional modes in Ceratobasidiaceae 259

Bayesian phylogeny reconstruction, RAxML 7.2.8 (Stamatakis implemented in Phylocom (Webb et al. 2008). To address 2006) and MrBayes 3.1 (Huelsenbeck & Ronquist 2001) were phylogenetic distinctness of nutritional modes, we calculated respectively used. Only RAxML was able to handle the full the pairwise Nearest Taxon distances (NTD) for all these eco- dataset with reasonable speed. In all RAxML analyses, logical guilds. Differences between observed and randomized GTR þ G þ I evolutionary model and 1 000 fast bootstrap rep- NTD that exceeded 2 SD of the randomized dataset were licates were used. To use both methods for comparison to considered statistically significant. In addition, pairwise Uni- confirm validity of tree topology and to reduce tree size for frac distances were calculated in the Fast Unifrac server convenient display, the dataset was collapsed into 288 entries (http://bmf2.colorado.edu/unifrac/) by use of 999 permuta- by clustering the sequences with BLASTclust (Biegert et al. tions. Unifrac distance metric measures the phylogenetic 2006) based on 99 % sequence similarity and 90 % coverage distance among communities by calculating the proportional criteria (99 % threshold was used instead of 97 %, to keep all length of the tree branches that lead to descendants from each anastomosis subgroups in separate clusters). The longest single community but not from both communities, and tests sequence of each cluster served as a representative sequence. whether this distance is different from a random draw from Phylograms were constructed from this ‘collapsed con- the communities (Hamady et al. 2010). All these distance servative’ dataset with both RAxML and MrBayes (burn- metrics were calculated with and without the pathogen-rich/ in ¼ 2 000; evolutionary model GTR þ G þ I as revealed from mainly-MNR to assess the robustness of results. Sig- Modeltest; Posada & Crandall 1998). All Bayesian analyses in nificance values of the Unifrac pairwise distances were sub- this study were run for 10 million generations sampled every jected to reduction of false discovery rate by use of 1 000 generations. Burn-in value was determined according to BenjaminieHochberg FDR correction. at which generation the log likelihood scores reached sta- Based on the ML phylogram of the ‘full conservative’ tionary level. To include all ecological metadata in a single dataset, we generated ancestral state reconstructions for tree, a phylogram was constructed directly from the ‘full major clades and well-supported subgroups as implemented conservative’ dataset with RAxML. Patristic distances (pair- in the Ape package of R (Paradis et al. 2004). We particularly wise total path length between two terminals) were exported focused on the evolution of EcM habit within the Ceratobasi- from this full dataset tree using PDAP package of Mesquite diaceae, because EcM sequences were mainly concentrated in 2.75 (Midford et al. 2003; Maddison & Maddison 2009), and two subclades. All nutritional modes were allowed to change subjected to statistical analyses regarding the nutritional to any other state at different probabilities, because no a priori mode of sequenced isolates (see below). Trees were visualized model exists. together with metadata using the online tool iTOL (Letunic & Biogeographic patterns and host specificity of the most Bork 2007; http://itol.embl.de/), and are publicly available as common MOTUs and major clades were assessed based on shared projects of the user ‘CeratobasidiumThanatephorus’ both the full and non-redundant datasets. Because of highly (case sensitive). For improved readability, clade names of the biased sequence availability in different countries and hosts, Ceratobasidiaceae are labelled based on the phylogenetic we sought the patterns of endemicity on a continental scale distribution of known AGs preceded by a slash (Moncalvo et al. and host specificity at the plant tribe (OrM) and family 2002). (pathogens) levels. The LSU sequences were aligned with PRANKSTER using default parameters and minimal manual correction. The LSU alignment spanned 1 622 sites, of which 706 were variable and Results 517 were informative. LSU phylograms were generated both with RAxML and MrBayes (GTR þ G þ I evolutionary model as Taxonomic and phylogenetic patterns revealed from Modeltest) to assess the monophyly of the Ceratobasidiaceae and to obtain an outgroup for rooting ITS The Bayesian (ASDSF ¼ 0.011) and ML analyses revealed sim- phylograms. All alignments are available in TreeBase (acces- ilar tree topology in the LSU-based phylogenetic recon- sion TB2:S13952). struction. The LSU-based phylogeny confirmed monophyly of To remove pseudoreplicates from the analyses of dis- the Ceratobasidiaceae within Cantharellales (Fig 1). The tribution of nutritional modes, we excluded redundant clade was inferred as a to the sequences from each study by keeping the longest repre- Ceratobasidiaceae, but with no statistical support. Within the sentative sequence from each MOTU (barcoding threshold of Ceratobasidiaceae, Thanatephorus (syn. Uthatobasidium) fusis- 97 % sequence similarity and 90 % of coverage) per study, porus formed a basal branch that was supported in both the resulting in 508 representative sequences. The Net Related- Bayesian (PP ¼ 1.0) and ML (BS ¼ 95) analyses. Only the /BD ness Index (NRI, standardized mean of distances between all clade was not represented by LSU sequences, but it was nested pairs of sequences from compared categories) and the Nearest among four other clades that were only partly supported Taxon Index (NTI, standardized mean of distances between (PP ¼ 0.70; BP ¼ 75) as a sister group to the /fusisporus clade pairs of the nearest neighbours) were calculated based on based on the ITS analysis (Fig 2). Therefore, we rooted the ITS patristic phylogenetic differences of an ultrametric tree to test phylogram at the /fusisporus clade. The topology of the LSU clustering and overdispersion of nutritional modes on larger phylogram was congruent with the ITS phylogram. The and smaller phylogenetic scales, respectively (Webb 2000). /mainly-MNR clade was monophyletic and nested within the Sampling species as random draws from the phylogeny BNR clades in both the LSU- and ITS-based analyses. without replacement served as a null model and statistical In the ‘collapsed conservative’ ITS dataset, phylograms significance was calculated based on 999 permutations as revealed in both the ML and Bayesian analyses (ASDSF ¼ 0.019) 260 V. Veldre et al.

AY463475 mesenterica TREMELLOMYCETES 96 AY586693 ansatum 100 AY463403 sp class AY701526 cornea 96 EU118672 Stypella papillata 94 AY586653 Elmerina holophaea AF506492 Auricularia mesenterica AF506493 Exidia glandulosa AY586654 Exidiopsis calcea AF506476 sernanderi 100 AY756071 Minimedusa pubescens Cantharellales 97 AM259216 Sistotrema confluens core group AF347095 repandum (, 77 81 AJ606042 Sistotrema alboluteum AF506473 , EU118616 cinerea ) 100 DQ089013 botryosum EU118607 Botryobasidium subcoronatum EU118629 curtisii AY586657 Haplotrichum conspersum 100 AY585831 Gloeotulasnella cystidiophora 84 100 AB369929 sp Tulasnellaceae AB369928 Epulorhiza sp AB369933 Epulorhiza sp 95 UDB013030 Thanatephorus fusisporus AF518664 Thanatephorus fusisporus /fusisporus DQ369859 Thanatephorus cucumeris AG-3 AF354064 Thanatephorus cucumeris AG-3 78 AF354078 Thanatephorus cucumeris AG-5 AF354066 Thanatephorus cucumeris AG-8 AF354070 Thanatephorus cucumeris AG-2-BI DQ097887 Thanatephorus cucumeris AG-2-2 selalle AF354069 Thanatephorus cucumeris AG-8 DQ097888 Thanatephorus cucumeris AG-8 AF354119 Thanatephorus cucumeris AG-8 rahtnaC AF354068 Thanatephorus cucumeris AG-8 AF518655 Thanatephorus cucumeris DQ917658 Thanatephorus cucumeris AF354072 Thanatephorus cucumeris AG-4-HGII 100 AF354077 Thanatephorus cucumeris AG-4-HGIII AF354073 Thanatephorus cucumeris AG-4-HGII

AF354074 Thanatephorus cucumeris AG-4-HGII eaecaidisabotareC AF354076 Thanatephorus cucumeris AG-4-HGIII /mainly- AF354075 Thanatephorus cucumeris AG-4-HGIII MNR AF354081 Ceratobasidium sp CAG-4 AF354096 Thanatephorus cucumeris AG-6 AF354058 Thanatephorus cucumeris AG-1-IC AF354060 Thanatephorus cucumeris AG-1-IA AF354059 Thanatephorus cucumeris AG1-IB AF354084 Ceratobasidium sp CAG-7 AF354083 Ceratobasidium sp CAG-6 100 AF354080 Ceratobasidium sp CAG-3 AF354061 Thanatephorus cucumeris AG-6 AF354062 Thanatephorus cucumeris AG-6 AF354111 Thanatephorus cucumeris AG-10 AF354071 Thanatephorus cucumeris AG-10 AF354079 Thanatephorus cucumeris AG-11 AF354065 Thanatephorus cucumeris AG-9 AF354063 Thanatephorus cucumeris AG-2-1

AF354092 Ceratobasidium sp AG-A SETECYMOCIRAGA AF354091 Ceratobasidium sp AG-K /AK 83 DQ097889 Ceratobasidium sp AG-G AF354093 Ceratobasidium sp AG-L /GLO AF354094 Ceratobasidium sp AG-O 78 FJ207506 Ceratobasidiaceae sp 92 DQ520098 Ceratobasidium sp FO38200 AY293171 Ceratobasidium sp GEL5602 /CHI AY634127 Ceratobasidiaceae sp 100 AY505557 Piriformospora indica 100 EU625999 Sebacina vermifera 100 DQ520103 Craterocolla cerasi AY745701 Tremellodendron sp 78 AY586711 Sarcodon imbricatus

99 AY586658 gracilipes ssalc AY586635 Bankera fuligineoalba EU118674 Tomentellopsis bresadoliana AY586636 Boletopsis grisea AY586726 macrospora 93 AY586652 maculata 96 EU118639 fuciformis AY463401 roseum AY885164 100 EU118636 argillacea EU118637 Jaapia ochroleuca Jaapiales AY586679 subillaqueatum AY586629 bombycina EU118610 borealis AY586628 subincarnatum 96 AY586639 Byssocorticium pulchrum AY586632 decipiens Atheliales AY463480 Tylospora asterophora AY586662 Hygrophorus olivaceoalbus AY586685 Clitocybe nebularis EU118620 Cristinia helvetica AY586680 Mallocybe agardhii AY586681 Inocybe fibrosa EU118622 Cystidiodontia laminifera AY586646 Clavaria fumosa EU118673 atrotomentosa 93 AY586715 Suillus luteus AY586645 Chroogomphus rutilus AY586723 Tylopilus felleus AY586659 aurantiaca EU118643 mollusca 97 EU118657 unica 82 EU118653 sordida EU118665 hydnoides Polyporales EU118668 fimbriatum AF506432 aspellum 100 AF506413 volemus AF506462 nauseosa AF506489 lenta AF506396 ovinus 100 AY586714 Subulicystidium sp AY586720 Trechispora nivea Trechisporales 95 EU118621 laeve 90 EU118631 alutaria 71 AY586665 rubiginosa AY586722 subulatus EU118628 clavatus 70 AF336259 stoloniferum 100 AY885165 Phallus hadriani Hysterangiales 88 AJ406479 Anthurus archeri AF139943 Aseroe arachnoidea EU118641 dendroidea AY586682 himantia Gomphales 88 AF139975 Sphaerobolus stellatus 0.1 100 AF336251 rufescens AF287859 Geastrum saccatum

Fig 1 e Large subunit maximum likelihood phylogram demonstrating the phylogenetic placement of Ceratobasidiaceae among Agaricomycotina. serves as outgroup. Values above branches indicate bootstrap support ‡70 %, while thick branches indicate high posterior probabilities (PP ‡ 0.95) as revealed from a parallel Bayesian analysis. Bar indicates 0.1 substitutions per site. Evolution of nutritional modes in Ceratobasidiaceae 261

Unkn Pathog OrM Sapr Endoph EcM A AF472302 11 AF472303 11

100 AF504008 11 suropsisuf/ DQ028824 11 75 AF503973 25 86 DQ093780 11 11 FJ809766 11 100 EU218895 11 11 86 GQ223450 11 AY634129 11 100 UDB008679 11 UDB008680 110 AJ242890 12 AJ242896 1322 11 AG-I DQ061931 11 DQ672286 12 89 DQ028808 11 93 100 EF090498 424 EF090492 11 DQ028800 11 DQ028797 11 DQ028798 11 DQ028799 11 99 DQ028818 12 DQ028817 27 100 AJ242882 22 84 EU730859 12 96 DQ356407 11 DQ356409 11 70 AJ419929 2211 11 AJ716305 22 11 11 89 100 AM901722 11 11 93 GU985226 12 AB290020 11 AG-C 85 72 AM901962 11 100 UDB008709 11 UDB008720 11 97 11 UDB008678 IHC/ EU645602 11 AJ419928 11 EF536968 11 AJ242901 33571111 AG-I DQ421054 11 UDB008719 11 94 AB290021 11 GU446633 1111 AG-C AM697947 11 AM697948 22 UDB008710 11 DQ520098 11 AY634126 515 UDB008661 123 100 AM231372 11 FJ552883 35 AJ633124 22 100 DQ273373 11 88 100 FJ660483 11 EcM2 98 FN298243 33 DQ493565 14 (/ceratobasidium2; FJ807984 11 EU645652 11 Tedersoo et al., 2010) 79 GU083181 210 91 AJ419931 111111 EF100192 11 FJ237067 11 DQ093652 11 FM866376 11 98 UDB008670 12 GQ175312 15 15 100 GQ175305 18 18 FJ688125 11 DQ028791 37 DQ028796 11 FJ231393 13 100 AB198709 27 AB198714 11 AG-D-III DQ278930 4757 98 AB198699 11 AB198702 11 AG-D-I 87 AB198703 11 80 AY618223 11 AB449170 236 100 UDB008676 11 UDB008674 12 89 AJ318421 18 96 GQ221863 11 75 FJ440188 13

FJ440200 11 DB/ 74 AB286930 3323 AG-Ba 92 DQ102430 13 AG-B(o) AJ427400 11 AF503964 213 DQ084013 11 DQ307249 3648 AG-Bb AF503998 457 96 AF472288 15 AF503970 12 98 FJ788720 11 EU516900 11 EU516785 11 EU810056 1111 82 DQ028823 11 FJ788669 12 82 DQ028822 11 AJ427403 11 FJ788668 11 FJ788674 11 UDB008717 11 FJ788673 16 78 EF433959 11 EU516903 11 DQ672300 11 77 EU090897 11 97 AY634163 11 100 DQ672269 11 99 DQ672278 11 85 AY970109 12 DQ182419 11 73 FJ613066 11 11 FM866369 12 94 DQ084001 11 AJ318442 333263131111 100 AJ242875 2333 19 AG-G EU195338 11 AM040889 18 UDB008725 13 AJ242898 2545 DQ223780 11 AG-L AB286936 1123 11 97 FJ362331 11 AG-O 79 AM697940 13

AF472285 11 /GLO 100 AB432931 11 UDB004860 11 100 GQ175295 13 13 GQ175292 19 19 GQ175304 11 11 EcM1 76 AB432932 11 AB432928 11 (/ceratobasidium1; UDB004466 11 FR731600 13 Tedersoo et al., 2010) AB303048 116 FR731598 19 FR731599 12 94 FJ788681 18 100 AF200518 13 11 AF200514 24 100 FJ763573 12 FJ763577 2 FJ763578 11 AY927354 22 93 EU591775 11 AJ242903 71116581 1 2 3 72 75 DQ102426 11 DQ102425 11 AG-A

DQ279055 1111 KA/ DQ672314 11 77 DQ672313 11 73 DQ672315 11 EF429313 1111 AY927338 19 100 AJ242887 24913 11 AG-K 94 DQ672328 11 AG-A /mainly-MNR clade (continued) 0.05

Fig 2 e ITS maximum likelihood phylogram of the reduced conservative dataset, ancestral states of selected nodes and frequency of occurrence in different nutritional modes. The phylogram was rooted at the /fusisporus clade. Values above branches indicate bootstrap support ‡70 %, while thick branches indicate high posterior probabilities (PP ‡ 0.95) as revealed from a parallel Bayesian analysis. Bar indicates 0.05 substitutions per site. Circular diagrams below branches indicate the probability of nutritional modes being ancestral: red, pathogenic(pathog); black, saprotrophic(sapr); blue,orchidmycorrhizal (OrM); yellow, ectomycorrhizal (EcM); purple, endophytic (endoph); white, unknown (unkn). The table of counts of nutritional modes is based on the full conservative dataset e the first sub-column excluding and second sub-column including redundant sequences (99 % similarity within a study). 262 V. Veldre et al.

Unkn Pathog OrM Sapr Endoph EcM 100 EU810045 19 B EU810026 121 AJ000202 11 AB275641 39 AG-13 AF153806 13 AF354099 2347 AG-12 97 DQ301760 1113 AG-7 DQ279022 11 GU937740 11 AJ301901 11 71 DQ885780 2333 DQ279017 11 AG-R DQ278936 11 11 84 DQ279019 11 AF354084 11 AG-S 88 AJ318435 12 94 GU937739 22 AF153784 1113 AG-6 GU937738 11 DQ672268 1122 11 AG-Fb FJ788709 2324 79 AF354101 1212 AF354102 1123 AG-6 AF153783 111112 AY433813 11 AY634121 1111 94 DQ672312 11 DQ672311 11 FJ746906 1627 AG-7 81 AY586167 11 DQ093646 3581811 AG-E AY684922 10 23 9 30 1 3 90 AM901911 11 AG-5 AF478452 11 77 FJ435099 11 73 AB019025 35 AG-2-3 64 U57740 12 100 AF354114 2212 AG-11 EU591766 11 100 EU730848 23 EU730840 28 EU730809 16 AF153802 22 AG-11 DQ356413 23410 FJ788716 34 AG-8 87 FJ851083 11 AM697936 12 AM697938 12 AM697934 12 100 AF354110 11 98 FJ492108 24 AB054876 11 AG-2-BI AB054875 11 U57885 1111 EU591761 12 23 15 89 98 U57729 11 96 U57722 11 U57727 1111 AB000030 11 U57744 11 RNM-y EU244841 11 AG-2-1 U57880 11 81 U57882 11 lniam/ U57883 11 U57881 11 FJ553367 11 DQ913036 1114 FJ480916 12 99 AF308623 11136 EU591791 33 13312 83 U57734 25 96 DQ278980 11 AG-2-2 100 FJ492149 11 FJ492143 11 100 FJ435097 2346 EU244844 11 AG-9 83 EF017212 11 11314 76 1 1 DQ421056 11 DQ421057 1126 22 AG-3 EU480292 212 86 FJ788707 15 96 FJ788711 11 92 FJ746909 37 DQ356408 3424 AG-10 DQ421055 11 DQ021449 11422 AG-4-HGIII 92 U19954 12 DQ102447 10 20 18 74 EU375545 11 AG-4-HGI FJ492106 11 HM117643 11 97 100 UDB008690 15 94 UDB008714 11 EU218892 22 EU135906 11 UDB008721 11 FJ788721 14 FJ492085 11 97 AJ242888 9281390 AG-4-HGII AF222795 1111 AF354081 11 DQ408294 11 DQ102441 15 AY684921 11 82 DQ102434 11 AG-Fa DQ279014 11 DQ102433 48 11 100 EF197798 216 89 EF197799 11 AG-1-ID 89 AJ868444 1627 AG-1-IF 99 AF067641 11 16 6 21 77 AB000025 1112 AG-1-IB EU591762 7163 6 AG-1-IC GU937735 3323 GU937736 11 AB286941 22 AG-P 90 DQ278931 11 AJ301899 11 DQ173058 11 AY665171 11 73 AB479196 1111 AG-1-IA + FJ440209 1111 GU570159 65421195 AG-1-IE FJ667265 1112 AF354060 11

Fig 2 e (continued).

produced comparable results that differed only in statistical described AGs, but contained a sequence of the binucleate support (Fig 2). Phylograms of the ‘full conservative’ dataset teleomorph T. fusisporus (DQ398957). The /mainly-MNR, /AK displayed slight differences in tree topology and weaker and /CHI clades were supported in the Bayesian analysis branch support, perhaps due to a greater amount of noise (PP 95 %), whereas the /mainly-MNR, /AK, /CHI and /fusis- owing to accumulation of erroneous base calls in larger porus clades received bootstrap support over 70 % in the ML datasets. analysis. Thus, the monophyly of the /BD and /GLO clades is All ITS-based phylogenetic analyses resulted in similar tree poorly supported despite their consistent formation in all topology and revealed six major clades (Fig 2). All MNR AGs different analyses and datasets. were clustered in one well-supported clade (/mainly-MNR) We found no clear distinction between the teleomorph together with some BNR AGs (AG-E, AG-F, AG-P, AG-R, AG-S). genera Thanatephorus and Ceratobasidium (Fig S1). While most The remaining BNR sequences were divided between other Thanatephorus fruit-bodies (predominantly T. cucumeris) were clades, four of which included known AGs (/AK, /BD, /CHI placed in the /mainly-MNR clade, fruit-bodies with Thanate- and /GLO). The /fusisporus clade was not associated with any phorus morphology also appeared in the binucleate /GLO Evolution of nutritional modes in Ceratobasidiaceae 263

(T. theobromae) and /fusisporus clades (T. fusisporus). The 23 (14.6 %) were represented by only two sequences. Damaged majority of Ceratobasidium fruit bodies (including most C. cor- plant tissue, orchid roots, soil, EcM root tips, and healthy nigerum specimens) belonged to the BNR clades, but four shoot tissue or non-mycorrhizal root tissue were the domi- sequences labelled as C. cornigerum were scattered in the nant isolation source in 52 (33.1 %), 47 (29.9 %), 14 (8.9 %), 14 /mainly-MNR clade (among AG-E, AG-P, AG-R and AG-4-HGI). (8.9 %) and 7 (4.5 %) MOTUs, respectively (Table 1). The main Most of the AGs were monophyletic, including AG-1, AG-4, isolation source of 13 (8.2 %) MOTUs remains unknown, AG-B and AG-D that in turn contained monophyletic subgroups whereas sequences of ten MOTUs (6.4 %) were obtained from (BS > 70; Fig 2, Fig S2). By contrast, the AG-Fa formed a distinct multiple sources in equal abundance (usually a combination monophyletic group (BS ¼ 82) near the base of the /mainly-MNR of pathogenic, OrM and/or soil-derived sequences). The ‘lib- clade, whereas AG-Fb constituted a subgroup of the multi- eral’ dataset was divided into 252 MOTUs, including an addi- nucleate AG-6. AG-2 was polyphyletic as a whole, but each of its tional 84 singletons, eight doubletons, and three taxa with subgroups was monophyletic. Although AG-C, AG-I and AG-7 more than two sequences. have not been separated into subgroups, members of these The six major clades of the Ceratobasidiaceae differed in AGs were distributed in seemingly unrelated subclades within the available taxonomic information and relative sampling the /CHI (AG-C, AG-I) and /mainly-MNR (AG-7) clades. AG-S and depth. The /mainly-MNR clade comprised 68.4 % of sequences, AG-Q were represented by only one and two sequences, but only 38.9 % of MOTUs, indicating relative oversampling of respectively, and therefore the monophyly of these AGs cannot harmful pathogens by plant pathologists. Anastomosis be assessed. A large proportion of the /mainly-MNR clade was grouping information was available for 64.0 % of MOTUs in the covered with known AGs (Fig 2B), while the majority of termi- /mainly-MNR clade, but only for 16.4 % of species in the BNR nal taxa of the BNR clades lacked anastomosis grouping clades taken together. No AGs were assigned to members of information (Fig 2A). the /fusisporus clade. All nutritional modes except endophytes (0.1 < P < 0.5) Ecological patterns were significantly phylogenetically clustered compared to the null model based on the NRI index and inclusion or exclusion Based on metadata, putative pathogens, OrM symbionts, of the /mainly-MNR clade (P < 0.05 in all cases; Table 2). Based saprotrophs and EcM symbionts were represented by 1 129 on NTI index, all nutritional modes except pathogens (P > 0.5) (50.0 %), 401 (17.8 %), 69 (3.1 %) and 73 (3.2 %) sequences, were significantly phylogenetically clustered. In all analyses, respectively (Table 1). The putative ecology of 597 (26.5 %) OrM fungi displayed the strongest phylogenetic clustering sequences remained unknown due to the lack of metadata among all modes of nutrition. The NTD pairwise distance about the isolation source. In total, 82.9 % of sequences of metric revealed several phylogenetic dissimilarities among putative pathogens were affiliated to the /mainly-MNR clade. the nutritional modes (Table 2). OrM sequences differed from Only 14.3 % of sequences representing other ecological strat- pathogens (NTD ¼20.63; P < 0.001) and EcM fungi egies belonged to this clade. Sequences of putative EcM fungi (NTD ¼3.21; 0.01 < P < 0.05), but not from endophytes and clustered either in the /CHI or in the /GLO clade. A group of saprotrophs (0.1 < P < 0.5). EcM fungi were significantly dif- Australian EcM isolates sampled by Bougoure et al. (2009) was ferent from pathogens (NTD ¼19.15; P < 0.001) and sapro- exceptional as each of these strains comprised two ITS copies, trophs (NTD ¼3.36; 0.01 < P < 0.05), whereas endophytes one clustering with other EcM sequences in the /GLO clade differed significantly only from pathogens (NTD ¼5.59; and another clustering with non-EcM sequences in the /BD 0.001 < P < 0.01). Saprotrophs were phylogenetically clearly clade. Phylogenetic analyses and ancestral state recon- distinct from pathogens (NTD ¼8.18; 0.001 < P < 0.01). The structions suggested that EcM habit has evolved twice in the Unifrac pairwise distance metric corroborated these findings Ceratobasidiaceae (P < 0.01), but revealed no ancestral mode of phylogenetic differentiation (i.e. separation of commun- of nutrition with confidence, except that EcM habit is unlikely ities; Table 2). to be ancestral (Fig 2A). Biogeographic patterns and host specificity could be Based on sequence clustering at 97 % similarity cut-off, the addressed in sufficient detail only in the seven most common ‘full conservative’ dataset was divided into 157 MOTUs of MOTUs (found in more than five studies; representing AG-1- which 53 (33.8 %) were represented by a single sequence and IA, AG-2-1, AG-2-2, AG-4-HGI, AG-4-HGII, AG-5, AG-A) due to

Table 1 e Distribution of sequences and MOTUs (97 % similarity threshold; in parentheses) representing different nutritional modes among six major clades of Ceratobasidiaceae. For MOTUs, only the most common nutritional mode is scored EcM Endophyte OrM Pathogen Saprotroph Unknown Multiple equal Total

/mainly-MNR 0 (0) 3 (1) 47 (11) 936 (34) 26 (3) 531 (9) - (3) 1 543 (61) /AK 0 (0) 4 (0) 1 (0) 86 (4) 4 (0) 18 (0) - (0) 113 (4) /BD 13 (1) 2 (2) 151 (13) 34 (6) 5 (3) 24 (1) - (0) 229 (26) /CHI 15 (7) 7 (3) 90 (9) 21 (4) 25 (3) 13 (3) - (5) 171 (34) /GLO 45 (6) 14 (1) 46 (8) 52 (4) 8 (4) 9 (0) - (2) 174 (25) /fusisporus 0 (0) 0 (0) 24 (6) 0 (0) 1 (1) 2 (0) - (0) 27 (7) Total 73 (14) 30 (7) 359 (47) 1 129 (52) 69 (14) 597 (13) - (10) 2 257 (157) 264 V. Veldre et al.

Table 2 e Phylogenetic clustering and pairwise phylogenetic distances of nutritional modes. Unifrac pairwise distances indicate phylogenetic dissimilarity and these are given to the left from the diagonal; nearest taxon (NT) pairwise distances are given to the right from the diagonal. For net Relatedness index (NRI), nearest taxon index (NTI) and NT, positive values denote phylogenetic clustering and negative values indicate overdispersion (or avoidance, NT). Significant values are highlighted in bold na NRI NTI Pairwise Unifrac\NTD (comdistNT) phylogenetic distances

EcM Endophyte OrM Pathogen Saprotroph Unknown

EcM 18 3.46 5.34 e 1.59 L3.21 L19.15 L3.36 L20.45 Endophyte 19 1.94 0.83 0.74 e 0.82 L5.59 1.24 L7.05 OrM 82 7.44 6.87 0.83 0.76 eL20.63 0.47 L18.43 Pathogen 216 1.58 2.64 0.93 0.84 0.84 eL8.18 L11.04 Saprotroph 35 3.57 2.67 0.83 0.76 0.74 0.83 eL9.62 Unknown 138 1.58 3.17 0.92 0.84 0.85 0.61 0.81 e

a The number of non-redundant sequences (one sequence per MOTUs per study).

limited sample size. All these MOTUs were distributed in genera Ceratobasidium and Thanatephorus are subtle and several continents and most of these displayed no evidence intermediate forms exist (Roberts 1999), so that the charac- for host specificity, infecting a wide range of plant families teristic features may reverse in Ceratobasidiaceae evolution. (Fig S1). Only AG-3 showed some host specificity, infecting Our data support that the simplest basidioma shape charac- predominantly species of the Solanaceae family, particularly teristic for Ceratobasidium (Roberts 1999) is likely ancestral and (Solanum tuberosum). Anastomosis groups that were at least evolved several times into the more complex Thana- relatively widely distributed in the phylogenetic tree (AG-C, tephorus basidioma shape. Although all MNR strains were AG-I, AG-2, AG-7) all had wide geographic distribution in the concentrated in the /mainly-MNR clade, a few BNR AGs (AG-E, Northern hemisphere, with sympatric occurrence of MOTUs. AG-F, AG-P, AG-R, AG-S) were deeply nested within this clade In OrM, there was no evidence for specificity between orchid (Gonzales et al. 2006; Sharon et al. 2008). The topology found tribes and MOTUs or major clades of the Ceratobasidiaceae suggests that the multinucleate organization is a derived (Fig S1). Because most orchid species were each represented feature that arose multiple times in the Ceratobasidiaceae by only a single study and geographic origin, we cannot evolution, and/or with many reversions to the binucleate address OrM specificity at orchid species level. At higher state. While there seems to have been one main shift from taxonomic level, most of the six major clades had cosmopol- binucleate to multinucleate organization, at least one more itan distribution and none of them were restricted to a single change is evident based on the sister relationship between continent or biome (Fig S1). AG-1 and AG-P. Although most AGs were monophyletic, large AGs usually consisted of multiple MOTUs that were not closely related (as low as 81 % ITS sequence similarity between Discussion different subgroups of the same anastomosis group; Gonzales et al. 2001; Sharon et al. 2006, 2008). The polyphyly of tele- Phylogenetic implications omorph genera and some AGs as well as multiple shifts in the number of nuclei imply that both the morphology-based and Phylogenetic analyses revealed that the Ceratobasidiaceae is anastomosis behaviour-based taxonomies lack species-level monophyletic within the Cantharellales and the /mainly-MNR resolution in the Ceratobasidiaceae. The limited number of clade is nested within BNR groups. Within Ceratobasidiaceae, known morphological characters available for fungi, espe- the /fusisporus clade was inferred as a basal branch given the cially for taxa with no or resupinate fruit-bodies, often results position of T. fusisporus isolates in the LSU phylogram. in lower taxonomic resolution when using morphology, as Therefore, we rooted the ITS phylogram at the /fusisporus compared to molecular features (Taylor et al. 2006). clade, but conservatively anticipate the unlikely possibility that the true rooting point lies within this clade. Use of Evolution of nutritional modes protein-encoding genes will be necessary for further in-depth phylogenetic reconstruction of the Ceratobasidiaceae Although represented in several major clades, both the (Gonzales et al. 2006). pathogenic, OrM, EcM, endophytic and putatively sapro- We found strong phylogenetic evidence for non- trophic isolates displayed phylogenetically clustered dis- monophyly in both the teleomorph genera and the nuclear tribution within the Ceratobasidiaceae. This indicates that count-based system of anamorphs. Species of the two tele- these ecological roles tend to be evolutionarily conserved. omorph genera Ceratobasidium and Thanatephorus were scat- Sequences derived from OrM and soil formed a phylogeneti- tered across several major clades, indicating that both taxa are cally coherent group, and the fact that they originated from polyphyletic and the morphological characters (nuclear state, studies with different aims and sampling protocols empha- septation of basidia, shape of sterigmata) used for the generic sizes this result. Assuming that soil-derived isolates mostly distinction are unconserved. Indeed, differences between the represent saprotrophs, it seems that orchids preferentially Evolution of nutritional modes in Ceratobasidiaceae 265

establish symbiosis with the less aggressive side of the indicating that EcM lifestyle has secondarily arisen twice in pathogen-saprotroph continuum in the Ceratobasidiaceae. the Ceratobasidiaceae. Tedersoo et al. (2010) referred to these One could conclude that association with pathogenic strains EcM fungal lineages as /ceratobasidium1 and /ceratobasi- could result in death of the orchids before they germinate or dium2, respectively, but provided no phylogenetic support for reach maturity and become available for sampling (Taylor this hypothesis. Present data indicates that the /ceratobasi- et al. 2003). However, some orchids germinate as success- dium1 lineage includes both EcM and OrM strains from SW fully with isolates from strongly pathogenic AGs as with those Japan, Thailand, Australia, Malaysia, Zambia and Madagascar, from non-pathogenic or weakly pathogenic AGs in vitro suggesting a subtropical and tropical distribution. The Japa- (Masuhara et al. 1993; Masuhara & Katsuya 1994; Pope & Carter nese (Yagame et al. 2008) and Australian (Bougoure et al. 2009) 2001). Alternatively, pathogenic isolates could be simply less OrM strains readily form EcM in experimental conditions and available for orchids in soil, because pathogens tend to be participate in tripartite interactions with mycoheterotrophic more limited to tissues and rhizosphere of infected plants. We (non-photosynthetic) orchids in nature. The evolution of EcM tested the latter hypothesis by separately considering only the habit in this lineage may have facilitated the development of MOTUs found from soil. There was no significant difference in mycoheterotrophy in several orchid taxa (Chamaegastrodia the association of OrM isolates with non-pathogenic or sikokiana and Rhizanthella gardneri), because mycohetero- pathogenic MOTUs (chi-square test: n ¼ 29; P ¼ 0.775), which trophic orchids usually associate with EcM fungi in temperate lends no support for an avoidance strategy. Therefore, our habitats. It is believed that EcM fungi provide a more stable analyses suggest that in general, orchids associate with and reliable carbon source compared to saprotrophic fungi available Ceratobasidiaceae strains in soil irrespective of their (Dearnaley et al. 2012). The /ceratobasidium2 lineage repre- pathogenicity. sents a collection of sequences from EcM root tips in several Sequences of plant pathogens were phylogenetically dis- community analyses in the boreal and temperate of tinct from sequences of all other nutritional modes. Phyloge- Europe, North America and Japan, suggesting a circumboreal netic differences between soil-derived sequences and distribution. Recently, Yagame et al. (2012) showed that Pla- pathogens may represent a trade-off between abilities to tanthera minor, a partially mycoheterotrophic (¼ mixotrophic) compete for nutrients in debris dispersed in soil and to attack green orchid, associates with both the /ceratobasidium1 (cf. living plant tissue. To our knowledge, phylogenetic distinct- types I1 and I2) and /ceratobasidium2 EcM lineages (I4). Both ness of saprotrophic and parasitic guilds has not been EcM lineages lack a sister group with well-defined nutritional explicitly tested at the genus or family level in other taxo- mode, and the ancestral state reconstruction was unable to nomic complexes of fungi, but this could be a common phe- resolve the ancestral mode of nutrition for them. nomenon in basidiomycetes as suggested by published phylograms (e.g. Cryptococcus: Findley et al. 2009). Within Biodiversity and biogeography Ceratobasidiaceae, pathogenic groups were potentially derived from putative soil saprotrophs in several instances Sequence clustering reveals that both the /mainly-MNR and (Sharon et al. 2006). One could argue that soil-derived isolates especially BNR clades of the Ceratobasidiaceae comprise high are non-pathogenic to most plants in natural conditions cryptic diversity of taxa that lack sequenced representative where they evolved, but may become aggressive against fer- cultures and fruit bodies. Culturing and anastomosis tests of tilized crop plants or when introduced to exotic habitats pathogenic isolates have been performed for decades by (Wingfield et al. 2001; Desprez-Lousteau et al. 2007). This is phytopathologists, but soil-derived and orchid root- certainly an uncommon phenomenon in the Ceratobasidia- associated strains have usually escaped these trials, ceae family, because pathogenic isolates are relatively well- although they are often isolated. Moreover, recent in situ sampled. molecular identification studies from roots and soil have Bougoure et al. (2009) reported that several cultured strains provided an unprecedented wealth of information about the contained two alleles (cf. ‘cistrons’) that shared only 80 % diversity of BNR clades, suggesting that many Ceratobasidia- ITS sequence similarity. While one of the alleles belonged to ceae taxa have never been obtained into pure culture. Both the the /ceratobasidium1 EcM lineage in the /GLO clade, the other ‘full conservative’ and ‘liberal’ datasets revealed a large allele was nested in the distantly related /BD clade, within number of MOTUs represented by only one or two sequences, other OrM and soil-derived isolates. This may represent a case indicating that part of the uncultured richness is yet to be of hybridization or transfer between an EcM fungus and a captured by molecular techniques. Because sequence analysis distantly related non-mycorrhizal fungus (Bougoure et al. of soil and root material is performed mostly in boreal and 2009). Some anastomosis groups (AG-1, AG-2, AG-B) encom- temperate regions, tropical soils probably contribute to a large pass lineages with only 81e83 % similarity (Sharon et al. 2006, proportion of the undetected richness and understanding of 2008), which may allow the coexistence of distantly related biogeographic patterns. For example, the ‘full conservative’ nuclei and thus create a possibility for recombination events dataset included 11 Ceratobasidiaceae MOTUs from 15 orchid (Xie et al. 2008). Such hybridization within sympatric, species in Reunion Island (Martos et al. 2012). The present anastomosis-compatible groups may result in shifts in host evaluation of biogeography and host specificity was restricted range and give rise to novel harmful pathogens (Brasier 2000; to the most common MOTUs of pathogens and OrM fungi and Desprez-Lousteau et al. 2007). to the six major clades. While the major clades have cosmo- Sequences isolated from EcM root tips fell consistently into politan distribution, the common species lack both host spe- two well-supported lineages in the BNR /GLO and /CHI clades. cificity and endemism that may be at least partly ascribed to Ancestral state reconstructions support these findings anthropogenic dispersal for crop pathogens. We cannot rule 266 V. Veldre et al.

out the possibility that host specificity occurs at the strain Nationale de la Recherche (ANR program SYSTRUF), and level as shown for the FusariumeGibberella complex (Ma et al. F. Martos by the Region Reunion. 2010). In OrM fungi, the lack of specificity for plant groups is consistent with the facultative nature of this symbiosis for fungi, as revealed by the lack of phylogenetic fidelity on the fungal side (Martos et al. 2012). Appendix A. Supplementary data The use of only ITS sequences for inferring biogeographic and ecological questions has, however, a few limitations. Supplementary data related to this article can be found at First, tracing the origin and specificity of pathogens is best http://dx.doi.org/10.1016/j.funeco.2013.03.004. approached by use of population genetics techniques, because the ITS region has insufficient resolution at the population references level. Second, ITS-based phylogenetic trees often exhibit low phylogenetic resolution because of abundant insertions and deletions that are difficult to handle by alignment and phy- logenetic programs. Another source of error is the paucity of Abarenkov K, Nilsson RH, Larsson KH, Alexander IJ, Eberhardt U, metadata. Both phylogenetic uncertainty and missing meta- Erland S, Høiland K, Kjøller R, Larsson E, Pennanen T, Sen R, Taylor AFS, Tedersoo L, Ursing B, Vralstad T, Liimatainen K, data render the results of evolutionary ecology studies less Peintner U, Koljalg~ U, 2010a. The UNITE database for statistically supported due to greater noise to signal ratio and molecular identification of fungi e recent updates and future reduced sample size, respectively. However, such noise is perspectives. New Phytologist 186: 281e285. unlikely to bias the qualitative patterns when it is distributed Abarenkov K, Tedersoo L, Nilsson RH, Vellak K, Saar I, Veldre V, randomly along the phylogram. Parmasto E, Prous M, Aan A, Ots M, Kurina O, Ostonen I, Jogeva~ J, Halapuu S, Poldmaa~ K, Toots M, Truu J, Larsson K-H, Koljalg~ U, 2010b. PlutoF e a web based workbench for ecological and taxonomic research, with an online Conclusions implementation for fungal ITS sequences. Evolutionary Bioinformatics 6: 189. € Our global analysis of public ITS sequences of the Ceratoba- Biegert A, Mayer C, Remmert M, Soding J, Lupas A, 2006. The MPI Toolkit for protein sequence analysis. Nucleic Acids Research 34: sidiaceae family sheds light onto phylogenetic relations and W335eW339. distribution of ecological strategies within this large, ecologi- Bougoure J, Ludwig M, Brundrett M, Grierson P, 2009. Identity and cally and economically important fungal family. All major specificity of the fungi forming with the rare nutritional modes such as saprotrophy, pathogenic, OrM, EcM mycoheterotrophic orchid Rhizanthella gardneri. Mycological and endophytic interactions were phylogenetically conserved. Research 113: 1097e1106. Although pathogens have arisen multiple times independ- Brasier C, 2000. The rise of the hybrid fungi. Nature 405: 134e135. ently (Gonzales et al. 2001, 2006), they are phylogenetically Cameron DD, Johnson I, Read DJ, Leake JR, 2008. Giving and receiving: measuring the carbon cost of mycorrhizas in the distinct from most other functional guilds. Orchid root sym- green orchid, Goodyera repens. New Phytologist 180: bionts are phylogenetically overlapping with putative sapro- 176e184. trophs from soil samples, suggesting that saprotrophic strains Cameron DD, Leake JR, Read DJ, 2006. Mutualistic mycorrhiza in from natural soils are more easily accessible for orchids. EcM orchids: evidence from plantefungus carbon and nitrogen lifestyle has evolved separately in two major clades of Cera- transfers in the green-leaved terrestrial orchid Goodyera 171 e tobasidiaceae. Probably through improved carbon nutrition, at repens. New Phytologist : 405 416. least one of these events may have triggered loss of photo- Carling D, 1996. Grouping in Rhizoctonia solani by hyphal anastomosis reaction. In: Sneh B, Jabaji-Hare S, Neate S, synthesis in certain orchid taxa associating with these clades Dijst G (eds), Rhizoctonia Species: taxonomy, molecular biology, of Ceratobasidiaceae. Our study emphasizes that the Cerato- ecology, pathology and disease control. Kluwer Academic basidiaceae is both functionally and taxonomically highly Publishers, Dordrecht, The Netherlands, pp. 37e47. diverse, and that the classical morphological, cytological and Dearnaley JDW, Martos F, Selosse M-A, 2012. Orchid mycorrhizas: anastomosis investigations have limited power of resolution molecular ecology, physiology, evolution and conservation at both the species and genus levels. Further sampling in aspects. In: Hock BE (ed), Fungal Associations, 2nd edn. The Mycota IX Springer, Berlin, Germany. natural habitats, especially in tropical sites, will probably Desprez-Lousteau M-L, Robin C, Buee M, Courtecuisse R, enable us to recover many novel species and to address Garbaye J, Suffert F, Sache I, Rizzo DM, 2007. The fungal biogeographic patterns within the Ceratobasidiaceae. dimension of biological invasions. Trends in Ecology and Evolution 22: 472e480. Egger KN, Hibbett DS, 2004. The evolutionary implications of exploitation in mycorrhizas. Canadian Journal of Botany 82: Acknowledgements 1110e1121. Findley K, Rodriguez-Carres M, Metin B, Kroiss J, Fonseca A, This work was initially led by V. Veldre, but was finalized by Vilgalys R, Heitman J, 2009. Phylogeny and phenotypic Cryptococcus co-authors following the accidental death of the first author. characterization of pathogenic species and closely related saprobic taxa in the . Eukaryotic Cell 8: 353. We thank A. Stamatakis and A. Aberer for advice regarding Gonzales D, Carling DE, Kuninaga S, Vilgalys R, Cubeta MA, 2001. phylogenetic analyses. The bulk of this project was funded Ribosomal DNA systematics of Ceratobasidium and from Estonian Science Foundation grants PUT171, 8235 and Thanatephorus with Rhizoctonia anamorphs. Mycologia 95: 9286, and FIBIR. M.-A. Selosse is funded by the Agence 1138e1150. Evolution of nutritional modes in Ceratobasidiaceae 267

Gonzales D, Cubeta MA, Vilgalys R, 2006. Phylogenetic utility of Nilsson RH, Abarenkov K, Veldre V, Nylinder S, de Wit P, indels within ribosomal DNA and beta-tubulin sequences Brosche S, Alfredsson JF, Ryberg M, Kristiansson E, 2010a. An from fungi in the Rhizoctonia solani species complex. Molecular open source chimera checker for the fungal ITS region. and Evolution 40: 459e470. Molecular Ecology Resources 10: 1076e1081. Gouy M, Guindon S, Gascuel O, 2010. SeaView version 4: a Nilsson RH, Veldre V, Wang Z, Eckart M, Branco S, Hartmann M, multiplatform graphical user interface for sequence Quince C, Godhe A, Bertrand Y, Alfredsson JF, 2010b. A note on alignment and phylogenetic tree building. Molecular Biology the incidence of reverse complementary fungal ITS sequences and Evolution 27: 221e224. in the public sequence databases and a software tool for their Hamady M, Lozupone C, Knight R, 2010. Fast UniFrac: facilitating detection and reorientation. Mycoscience 52: 278e282. high-throughput phylogenetic analyses of microbial Papavizas GC, 1970. Colonization and growth of Rhizoctonia communities including analysis of pyrosequencing and solani in soil. In: Parmeter JR (ed), Rhizoctonia solani, Biology PhyloChip data. The ISME Journal 4:17e27. and Pathology. University of California Press, Berkeley, CA, Hibbett DS, Binder M, Bischoff JF, Blackwell M, Cannon PF, USA, pp. 108e122. Eriksson OE, Huhndorf S, James T, Kirk PM, Lucking€ R, 2007. A Paradis E, Claude J, Strimmer K, 2004. APE: analyses of phylogenetics higher-level phylogenetic classification of the Fungi. and evolution in R language. Bioinformatics 20: 289e290. Mycological Research 111: 509e547. Parmeter JR (ed), 1970. Rhizoctonia solani, Biology and Pathology. Huelsenbeck JP, Ronquist F, 2001. MRBAYES: Bayesian inference University of California Press, Berkeley, CA, USA. of phylogenetic trees. Bioinformatics 17: 754e755. Pope EJ, Carter DA, 2001. Phylogenetic placement and host Johanson A, Turner HCC, McKay GJ, Brown AE, 1998. A PCR-based specificity of mycorrhizal isolates belonging to AG-6 and AG-12 method to distinguish fungi of the sheath-blight complex, in the Rhizoctonia solani species complex. Mycologia 93:712e719. Rhizoctonia solani, R. oryzae and R. oryzae-sativae. FEMS Posada D, Crandall KA, 1998. MODELTEST: testing the model of Microbiological Letters 162: 289e294. DNA substitution. Bioinformatics 14: 817e818. Katoh K, Toh H, 2008. Recent developments in the MAFFT Roberts P, 1999. Rhizoctonia-forming Fungi. A Taxonomic Guide. multiple sequence alignment program. Briefings in Royal Botanic Gardens, Kew, London, UK. Bioinformatics 9: 286e298. Rosling A, Landeweert R, Lindahl BD, Larsson K-H, Kuyper TW, Kuninaga S, Natsuaki T, Takeuchi T, Yokosawa R, 1997. Sequence Taylor AFS, Finlay RD, 2003. Vertical distribution of variation of the rDNA ITS regions within and between ectomycorrhizal fungal taxa in a podzol soil profile. New anastomosis groups in Rhizoctonia solani. Current Genetics 32: Phytologist 159: 775e783. 237e243. Schoch CL, Seifert KA, Huhndorf S, Robert V, Spouge JL, Larsson K-H, 2007. Re-thinking the classification of corticioid Levesque CA, Chen WFungal Barcoding Consortium, 2012. fungi. Mycological Research 111: 1040e1063. Nuclear ribosomal internal transcribed spacer (ITS) region as a Letunic I, Bork P, 2007. Interactive Tree Of Life (iTOL): an online universal DNA barcode marker for Fungi. Proceedings of the tool for phylogenetic tree display and annotation. National Academy of Sciences, USA. http://dx.doi.org/10.1073/ Bioinformatics 23: 127e128. pnas.1117018109. Loytynoja€ A, Goldman N, 2005. An algorithm for progressive Selosse M-A, Roy M, 2009. Green plants that feed on fungi: facts and multiple alignment of sequences with insertions. Proceedings of questions about mixotrophy. Trends in Plant Science 14:64e70. the National Academy of Sciences, USA 102: 10557e10562. Sen R, Hietala AM, Zelmer CD, 1999. Common anastomosis and Ma L-J, van der Does HC, Borkovich KA, Coleman JJ, Daboussi MJ, internal transcribed spacer RFLP groupings in binucleate Di Pietro A, Dufresne M, Freitag M, Grabherr M, Henrissat B, Rhizoctonia isolates representing root endophytes of Pinus 2010. Comparative genomics reveals mobile pathogenicity sylvestris, Ceratorhiza spp. from orchid mycorrhizas and a chromosomes in Fusarium. Nature 464: 367e373. phytopathogenic anastomosis group. New Phytologist 144: Maddison WP, Maddison DR, 2009. Mesquite: a modular system for 331e341. evolutionary analysis. Version 2.6. URL: http://mesquiteproject.org. Sharon M, Kuninaga S, Hyakumachi M, Sneh B, 2006. The Masuhara G, Katsuya K, 1994. In situ and in vitro specificity advancing identification and classification of Rhizoctonia spp. between Rhizoctonia spp. and Spiranthes sinensis (Persoon) using molecular and biotechnological methods compared with Ames, var. amoena (M. Bieberstein) Hara (). New the classical anastomosis grouping. Mycoscience 47:299e316. Phytologist 127: 711e718. Sharon M, Kuninaga S, Hyakumachi M, Naito S, Sneh B, 2008. Masuhara G, Katsuya K, Yamaguchi K, 1993. Potential for Classification of Rhizoctonia spp. using rDNA-ITS sequence symbiosis of Rhizoctonia solani and binucleate Rhizoctonia with analysis supports the genetic basis of the classical seeds of Spiranthes sinensis var. amoena in vitro. Mycological anastomosis grouping. Mycoscience 49:93e114. Research 97: 746e752. Smith SE, Read DJ, 2008. Mycorrhizal Symbiosis, 3rd edn. Academic Midford P, Garland T, Maddison W, 2003. PDAP Package of Mesquite. Press, London, UK. URL: http://mesquiteproject.org/pdap_mesquite. Sneh B, Jabaji-Hare S, Neate SM, Dijst G (eds), 1996. Rhizoctonia Martos F, Munoz F, Pailler T, Kottke I, Gonneau C, Selosse M-A, Species: taxonomy, molecular biology, ecology, pathology and 2012. The role of epiphytism in architecture and evolutionary disease control. Kluwer Academic Publishers, Dordrecht, The constraint within mycorrhizal networks of tropical orchids. Netherlands. Molecular Ecology 21: 5098e5109. Sneh B, 1998. Use of non-pathogenic or hypovirulent fungal Moncalvo J-M, Nilsson RH, Koster B, Dunham SM, Bernauer T, strains to protect plants against closely related fungal Matheny PB, Porter TM, Margaritescu S, Weiss M, Garnica S, 2006. pathogens. Biotechnology Advances 16:1e32. The cantharelloid clade: dealing with incongruent gene trees and Stamatakis A, 2006. RAxML-VI-HPC: maximum likelihood-based phylogenetic reconstruction methods. Mycologia 98:937. phylogenetic analyses with thousands of taxa and mixed Moncalvo J-M, Vilgalys R, Redhead SA, 2002. One hundred and models. Bioinformatics 22: 2688e2690. seventeen clades of euagarics. and Taylor DL, Bruns TD, Szaro TM, Hodges SA, 2003. Divergence in Evolution 23: 357e400. mycorrhizal specialization within Hexalectris spicata Nilsson RH, Kristiansson E, Ryberg M, Larsson KH, 2005. (Orchidaceae), a nonphotosynthetic desert orchid. American Approaching the taxonomic affiliation of unidentified Journal of Botany 90: 1168e1179. sequences in public databases e an example from the Taylor JW, Turner E, Townsend JP, Dettman JR, Jacobson D, 2006. mycorrhizal fungi. BMC Bioinformatics 6: 178. Eukaryotic microbes, species recognition and the geographic 268 V. Veldre et al.

limits of species: examples from the kingdom Fungi. Philosophical Webb CO, 2000. Exploring the phylogenetic structure of ecological Transactions of the Royal Society, Series B 361: 1947e1963. communities: an example for rain trees. American Tedersoo L, May TW, Smith ME, 2010. Ectomycorrhizal lifestyle in Naturalist 156: 145e155. fungi: global diversity, distribution, and evolution of Webb CO, Ackerly DD, Kembel SW, 2008. Phylocom: software for the phylogenetic lineages. Mycorrhiza 20: 217e263. analysis of community phylogenetic structure and trait evolution. Tedersoo L, Abarenkov K, Nilsson RH, Schußler€ A, Grelet G-A, Version 4.0.1. Available from URL: http://www.phylodiversity. Kohout P, Oja J, Bonito GM, Veldre V, Jairus T, Ryberg M, net/phylocom/. Larsson K-H, Koljalg~ U, 2011. Tidying up international Wingfield MJ, Slippers B, Roux J, Wingfield BD, 2001. Worldwide nucleotide sequence databases: ecological, geographical and movement of exotic forest fungi, especially in the tropics and sequence quality annotation of ITS sequences of mycorrhizal the Southern Hemisphere. Bioscience 51: 134e140. fungi. PLoS One 6. e24940. Xie J, Yanping Y, Jiang D, Li G, Huang J, Li B, Hsiang T, Peng Y, Vilgalys R, Cubeta MA, 1994. Molecular systematics and 2008. Intergeneric transfer of ribosomal genes between two population biology of Rhizoctonia. Annual Review of fungi. BMC Evolutionary Biology 8: 87. Phytopathology 32: 135e155. Yagame T, Orihara T, Selosse MA, Yamato M, Iwase K, 2012. Warcup JH, 1991. The Rhizoctonia endophytes of Rhizanthella Mixotrophy of Platanthera minor, an orchid associated with (Orchidaceae). Mycological Research 95: 656e659. ectomycorrhiza-forming Ceratobasidiaceae fungi. New Waterman RJ, Bidartondo MI, Stofberg J, Combs JK, Gebauer G, Phytologist 193: 178e187. Savolainen V, Barraclough TG, Pauw A, 2011. The effects of Yagame T, Yamato M, Suzuki A, Iwase K, 2008. Ceratobasidiaceae above- and belowground mutualisms on orchid speciation mycorrhizal fungi isolated from nonphotosynthetic orchid and coexistence. American Naturalist 177: E54eE68. Chamaegastrodia sikokiana. Mycorrhiza 18:97e101.