Evidence for novel and specialized mycorrhizal parasitism: the orchid Gastrodia confusa gains carbon from saprotrophic Mycena

Yuki Ogura-Tsujita, Gerhard Gebauer, Toshimasa Hashimoto, Hidetaka Umata, Tomohisa Yukawa


We investigated the physiological ecology of the Asian non-photosynthetic orchid Gastrodia confusa. We revealed its mycorrhizal partners by using molecular identification and identified its ultimate nutritional source by analysing carbon and nitrogen natural stable isotope abundances. Molecular identification using internal transcribed spacer and large subunit nrDNA sequences showed that G. confusa associates with several species of litter- and wood-decomposer Mycena fungi. The carbon and nitrogen isotope signatures of G. confusa were analysed together with photosynthetic plant reference samples and samples of the ectomycorrhizal epiparasite Monotropa uniflora. We found that G. confusa was highly enriched in 13C but not greatly in 15N, while M. uniflora was highly enriched in both 13C and 15N. The 13C and 15N signatures of G. confusa were the closest to those of the fruit bodies of saprotrophic fungi. Our results demonstrate for the first time using molecular and mass-spectrometric approaches that myco-heterotrophic plants gain carbon through parasitism of wood or litter decaying fungi. Furthermore, we demonstrate that, several otherwise free-living non-mycorrhizal, Mycena can be mycorrhizal partners of orchids.


1. Introduction

More than 80 per cent of land-plant families are estimated to be mycorrhizal (Trappe 1987; Wang & Qiu 2006), and although mycorrhizal associations are typically mutualistic and generalistic, hundreds of non-photosynthetic plants are known to nutritionally exploit their target mycorrhizal fungi (Bidartondo 2005). Recent studies have shown that such ‘myco-heterotrophic’ plants (MHPs) associate with narrow clades of ectomycorrhizal fungi, such as Thelephoraceae (Taylor & Bruns 1997; McKendrick et al. 2000), Russulaceae (Taylor & Bruns 1999; Girlanda et al. 2006), Sebacina (McKendrick et al. 2002; Selosse et al. 2002) and Tricholoma (Bidartondo & Bruns 2001; Yokoyama et al. 2005) or arbuscular mycorrhizal fungal clades (Bidartondo et al. 2002; Merckx & Bidartondo 2008), that are simultaneously mycorrhizal with neighbouring photosynthetic plants (e.g. mycorrhizal trees). This has been called mycorrhizal ‘epiparasitism’ or ‘cheating’ (Björkman 1960; Cullings et al. 1996; Bidartondo et al. 2002) because MHPs obtain photosynthetic products from neighbouring autotrophic plants via a mycorrhizal network, instead of supporting the carbon needs of mycorrhizal networks as most other plants do (Smith & Read 2008).

While mycorrhizal epiparasitic plants are now universally recognized to be ecologically and evolutionarily diverse, direct mycorrhizal ‘parasitism’ is also suspected among MHPs. Several myco-heterotrophic orchids, mainly distributed in Asia, are shown to associate with free-living litter- or wood-decomposer fungi. By contrast with mycorrhizal epiparasites, direct mycorrhizal parasites would have to bring free-living fungi into the mycorrhizal symbiosis to obtain nutrition. An Asian non-photosynthetic orchid, Gastrodia elata, has long been reported to associate with the pathogenic wood-decay fungus Armillaria (Kusano 1911; Xu & Mu 1990; Kikuchi et al. 2008a,b) that was also reported from the non-photosynthetic orchid Cyrtosia septentrionalis (Hamada 1939; Cha & Igarashi 1996). Species of the genus Gastrodia and its related genus Didymoplexis are suspected to associate with several decomposer fungi such as Armillaria, Marasmius and Fomes (Burgeff 1932; Campbell 1962, 1964). The world's largest non-photosynthetic plant, Erythrorchis ochobiensis, is reported to form mycorrhizas with a wide range of wood-rotting fungi (Hamada & Nakamura 1963; Umata 1995, 1997, 1998a,b). Mycorrhizal parasitism is also supported by recent work in the non-photosynthetic orchid Epipogium roseum with a litter-decomposing Coprinaceae spp. in laboratory conditions (Yamato et al. 2005) and Eulophia zollingeri associates with another litter decomposer, Psathyrella cf. candolleana (Ogura-Tsujita & Yukawa 2008). However, it is still unclear whether these plants obtain nutrition from truly saprotrophic fungi (SF) under field conditions because some fungal groups that were not known to include ectomycorrhizal lineages, such as Tulasnella (Bidartondo et al. 2003), are now known to form ectomycorrhizas with trees and targeted by non-photosynthetic plants. Furthermore, there is no conclusive evidence for plant specificity towards litter- or wood-decomposing fungi; previous studies have examined a small number of individuals from a small number of nearby locations.

The natural abundance of 13C and 15N in organisms is a powerful tool for the elucidation of nutritional sources. Along food chains isotope values of heterotrophic consumers are generally enriched relative to their ultimate sources and they are closest to those of their immediate source. Mycorrhizal epiparasitic plants show highly enriched isotope signatures relative to autotrophic plants (Gebauer & Mayer 2003; Bidartondo et al. 2004) and their signatures are most like those of coassociated ectomycorrhizal fungi (Trudell et al. 2003). In their biomass, litter- and wood-decomposing fungi are generally enriched in 13C (Kohzu et al. 1999; Boström et al. 2008) but depleted in 15N relative to ectomycorrhizal fungi (Gebauer & Taylor 1999; Kohzu et al. 1999; Trudell et al. 2004). Thus, isotope signatures can be used to distinguish between mycorrhizal fungi living in humus and decomposing fungi feeding on litter and wood. We hypothesized that the isotope signatures of direct mycorrhizal parasitic plants targeting litter- and wood-decomposing fungi would be enriched in 13C but depleted in 15N relative to the indirect mycorrhizal epiparasites targeting mycorrhizal fungi, and that the direct mycorrhizal parasites' isotope signature would be the closest to that of litter- and wood-decomposing fungal tissues.

In this study, we tested this hypothesis with the Japanese myco-heterotrophic orchid Gastrodia confusa. There are ca 50 species of Gastrodia distributed through Siberia and Japan to Malesia, Australasia and Madagascar, where several species are endangered. They are all terrestrial non-photosynthetic plants from moist and sheltered forests (including G. elata that is cultivated commercially by collecting tubers from the wild that are then grown in special facilities). Gastrodia confusa is a rare plant of bamboo forests in Japan and Taiwan. We examined: (i) the mycorrhizal fungi of G. confusa at ten distantly located sites to identify their symbionts, and (ii) the abundance of 13C and 15N stable isotopes to reveal its nutritional sources. The mycorrhizal fungus of this orchid has never been identified before, but Tashima et al. (1978) isolated fungi from its mycorrhizas that were able to induce seed germination in vitro.

2. Material and methods

(a) Molecular identification of mycorrhizal fungus

A total of 44 plants of G. confusa was sampled from 10 sites (5–1000 km apart) in Japan from 2005 to 2007. For the study site and the number of samples see the electronic supplementary material A and B. All sites were dense bamboo forests dominated by arbuscular mycorrhizal Phyllostachys heterocycla and/or Phyllostachys bambusoides, which have been widely naturalized in Japan, with shaded and sparse herbaceous understories. Microscopic observation revealed that the mycorrhizal fungus mainly colonizes long-branched roots derived from the underground tuber and tissues that have a dark brownish colour. Accordingly, fresh and brownish mycorrhizal root tissues were used for fungal identification. Collected roots were washed in water and kept at −80°C until use. Voucher specimens of G. confusa were deposited in the herbarium of the National Museum of Nature and Science (TNS773656–773660, 8500020–8500022, 8500024, 8500028, 8500041, 8500042, 8500055, 8500060).

DNA was extracted from root samples, at least one per plant, using the DNeasy Plant Mini Kit (Qiagen, Valencia, CA, USA). Fungal internal transcribed spacer (ITS) sequences of nrDNA were amplified with the primer combination ITS1F/ITS4 or ITS1F/ITS4B (White et al. 1990; Gardes & Bruns 1993). For phylogenetic analysis, the large subunit (LSU) nrDNA sequences of each fungal ITS type were amplified using primer combinations LR0R/LR5 (Moncalvo et al. 2000) or LR0R/LR3 (Vilgalys & Hester 1990). PCR amplification and sequencing were carried out as described by Ogura-Tsujita & Yukawa (2008). An additional sequencing primer, LR3R (Hopple & Vilgalys 1999), was used for LSU sequences. PCR products that were difficult to sequence directly were cloned using the pGEM-T Vector System II (Promega, Madison, WI, USA). GenBank accession numbers (http://www.ncbi.nlm.nih.gov) of ITS sequences were AB454381–AB454413. DNA sequences were analysed using a BLAST search (Altschul et al. 1997) against the NCBI sequence database (National Center for Biotechnology Information, GenBank) to find the closest sequence matches in the database.

Large-subunit sequences from mycorrhizal roots (908 bp) were used for phylogenetic analysis. LSU sequences of fungal types I, II and III (electronic supplementary material C) were obtained from individuals from site S2 and of type IV from S10. LSU sequences of Mycenaceae from GenBank were also added to the analysis by referring to Moncalvo et al. (2000, 2002), Walther et al. (2005) and Matheny et al. (2006), and sequences of Tricholoma matsutake, Catathelasma ventricosum, Entoloma prunuloides and Lyophyllum decastes were used as outgroup taxa. DNA sequences were aligned using ClustalX (Thompson et al. 1997), followed by manual adjustment. Phylogenetic analyses were conducted with PAUP* v. 4.0.b6 (Swofford 2001). Distance trees were obtained using the neighbour-joining (NJ) method (Saitou & Nei 1987) with a Kimura two-parameter correction (Kimura 1980). For assessing the relative robustness for branches, the bootstrap method (Felsenstein 1985) was used with 1000 replicates.

(b) Stable isotope abundance analysis

Samples for isotope analysis were collected from Gastrodia site S6 (electronic supplementary material A and C) and a neighbouring reference site in September 2007. At this location, G. confusa grows in the floor of dense bamboo (P. heterocycla) forest with Piper kadsura, Gynostemma pentaphyllum and Thelypteris acuminata patchily covering the understorey. An evergreen broadleaf forest dominated by the ectomycorrhizal tree Lithocarpus edulis neighbours the bamboo forest, where an ectomycorrhizal epiparasite plant, Monotropa uniflora, grows on the forest floor with Damnacanthus indicus, P. kadsura and Trachelospermum asiaticum. Since M. uniflora is known to be an epiparasite associated with the ectomycorrhizal fungal family Russulaceae (Bidartondo & Bruns 2001; Young et al. 2002; Yang & Pfister 2006), this species was collected as a reference for comparison. Five 1 m2 plots were selected in both Gastrodia and Monotropa sites; each plot included orchids or monotropoids and at least two autotrophic reference plant species. Flower stalks of a non-photosynthetic plant, leaves of three to two reference autotrophic plants and soil samples from the organic layer were taken from each of the five plots at each site. The reference plants for the Gastrodia site were P. kadsura, G. pentaphyllum and T. acuminata and those for the Monotropa site were D. indicus, P. kadsura and T. asiaticum. Five samples were collected for each plant with the exception of P. kadsura (n=3) at the Monotropa site and G. pentaphyllum (n=4). A total of 14 sporocarps of saprotrophic fungal species (four Marasmiellus, three Hydropus, two Gymnopus, one Conocybe, one Gymnopilus, one Lepiota, one Paneolus, one unknown), which mostly occurred on decayed bamboo stems (DS), leaves or roots, were also collected within 30 m of the Gastrodia population. The sporocarps of representative species were deposited as dried herbarium specimens (TNS-F-18424–18428). DNA was extracted from all sporocarps, and ITS was sequenced with the same methods described earlier. DS and leaves, which are expected to be the main substrate for the SF, were also collected from the five Gastrodia plots.

Samples were dried at 105°C, ground to a fine powder and stored in a desiccator with silica gel until analysed. Relative nitrogen and carbon isotope abundances of the samples were measured using a dual-element analysis mode with an elemental analyser coupled to a continuous flow isotope ratio mass spectrometer as described in Bidartondo et al. (2004). Measured abundances are denoted as δ values that were calculated according to the given equation δ15N or δ13C=(Rsample/Rstandard−1)×1000‰, where Rsample and Rstandard are the ratios of heavy isotope to light isotope of the samples and the respective standard. Standard gases (nitrogen and carbon dioxide) were calibrated with respect to international standards by using the reference substances N1 and N2 for N isotopes and ANU sucrose and NBS 19 for C isotopes, provided by the International Atomic Energy Agency (Vienna, Austria).

The data were tested for statistical differences using the Kruskal–Wallis non-parametric test followed by a post hoc Mann–Whitney U-test with an adjusted significance level according to Holm (1979). The autotrophic reference plants were treated as one group after confirming insignificant differences among the data of each species.

3. Results and discussion

(a) Fungal identification

Our results from the molecular identification of mycorrhizal fungi revealed that G. confusa consistently associates with Mycenaceae fungi. A total of 106 roots from 44 individuals from 10 sites were analysed and fungal ITS sequences were successfully amplified from 105 roots. The ITS sequences obtained from 94 roots of 41 individuals had high-DNA sequence homology with those species of the basidiomycete Mycena by BLAST analysis, and grouped into four types (electronic supplementary material C). The most frequent types were I and II, which were found at eight and six sites, respectively. Types III and IV were rare and found at only two and one site(s), respectively. The fungal DNA sequences from single individuals were identical, with the exception of four individuals simultaneously associated with two types. Three individuals lacked Mycena ITS sequence but alternatively generated sequences that closely matched those of Marasmiellus ramealis (DQ450030; 90–93% identical) or Clitocybura (DQ192179; 84% identical). The Marasmiellus-like ITS sequence from orchid roots was completely identical to the sequences from five sporocarps used in isotope analysis. Thus, there is the possibility of a geographical mosaic of mycorrhizal specificity in G. confusa (sensu Thompson 2005).

To clarify the phylogenetic relationships among the four Mycena types, the LSU sequences from each type were analysed together with those from GenBank (figure 1). All LSU sequences were monophyletic with those of fungi belonging to Mycenaceae with bootstrap support 96 per cent. The LSU sequences from G. confusa roots were paraphyletic among the sequences of Mycenaceae fungi that are known as free-living litter and wood decomposers (Maas Geesteranus 1992; Cannon & Kirk 2007). Although, several Mycena species, such as Mycena osmundicola and Mycena orchidicola, were reported to be mycorrhizal in orchids (Fan et al. 1996; Xu & Fan 2001), this is the first cultivation independent molecular report to reveal that Mycenaceae species are mycorrhizal partners of any plant.

Figure 1

Phylogenetic relationships of the mycorrhizal fungi of G. confusa (black circles) based on LSU ribosomal DNA sequences of Mycenaceae available in GenBank. GenBank accessions are shown in parentheses. The numbers shown to the right of black circles indicate the fungal ITS sequence type shown in electronic supplementary material C. Phylogenetic analysis was conducted using NJ with 1000 bootstrap replicates (values of more than 60% are near branches).

(b) Isotope signatures

The results of stable isotope abundance analysis strongly support the hypothesis that G. confusa gains carbon from litter and/or wood SF. Isotope signatures were quite different between directly parasitic and epiparasitic MHPs, especially in 15N. The δ13C and δ15N values of G. confusa were significantly higher than those of autotrophic reference plants (figure 2a), but the differences were greater in 13C (enrichment factor ϵGastrodia−reference plants=10.2±1.3‰) than 15N (ϵGastrodia−reference plants=2.9±0.5‰). The mycorrhizal epiparasite M. uniflora was highly and significantly enriched in both 13C (ϵMonotropa−reference plants=7.5±0.8‰) and 15N (ϵMonotropa−reference plants=10.2±1.0‰) relative to photosynthetic plants (figure 2b). The enrichment factors found in M. uniflora are identical to those found in all previously examined epiparasitic MHPs, including a variety of orchids and Monotropa (Zimmer et al. 2007, 2008; Preiß & Gebauer 2008). However, these enrichment factors are significantly different from those found in G. confusa. Since litter- and wood-decaying fungi are generally enriched in 13C but depleted in 15N relative to ectomycorrhizal fungi living in humus, the higher δ13C than δ15N in SF might be reflected in the isotope signature of G. confusa. In fact, the δ13C and δ15N values of G. confusa were the closest to those of the litter- and wood-decaying saprotrophic fungal species in our habitat (figure 2a). The sporocarps of the SF were significantly enriched in 13C but not in 15N compared with their substrates, i.e. decayed bamboo litter (BL) and stems. Because total nitrogen contents of BL (1.10±0.21 mmol gdw−1) and decayed stems (0.21±0.07 mmol gdw−1) are markedly low, no N fractionation may be attributed to these severely N limited substrates.

Figure 2

Mean (±1 s.d.) δ13C and δ15N values in flower stalks of two myco-heterotrophic species (G. confusa and M. uniflora), leaves of five autotrophic species (open circles), sporocarps of saprotrophic fungi (SF, open squares), bamboo litter (BL, open triangles), decayed bamboo stems (DS, upward-pointing open triangle) and in soil samples collected from (a) a dense bamboo forest (G. confusa) and (b) its neighbouring evergreen broadleaf forest, (M. uniflora, filled triangle). Plant species: Di, Damnacanthus indicus; Gp, Gynostemma pentaphyllum; Pk, Piper kadsura; Tha, Thelypteris acuminata; Tra, Trachelospermum asiaticum.

(c) Conclusion

Our results demonstrate for the first time by using molecular and mass-spectrometric approaches that MHPs gain carbon through parasitism of wood- or litter-decaying fungi. Plant exploitation of fungi turns out to fall into two types, i.e. epiparasitic (tripartite, with a living photosynthetic plant as ultimate host and an intermediary mycorrhizal fungus) and parasitic (bipartite, without the need for a living plant). These two types of lifestyles have evolved independently in non-photosynthetic plants that cannot provide photosynthates to their mycorrhizal fungi and instead extract carbon (and nitrogen) from fungi. The epiparasites break into ancient arbuscular mycorrhizal and ectomycorrhizal symbioses, but the direct parasites recruit free-living fungi into novel orchid mycorrhizal symbioses. In this way, both the two ecologically dominant ways in which fungi obtain carbon—mycorrhization and decomposition—are susceptible to non-photosynthetic plants. Remarkably, these plants are able to evolve high specificity towards their fungi in the context of highly diverse mycorrhizal and saprotrophic fungal communities.


The authors thank Teruo Katsuyama, Shiro Kobayashi, Katsutoshi Maruno, Hiroko Nakayama, Munetaka Nishi, Kazuhiro Suzuki, Schuichiro Tagane, Tadayuki Tsujita and Tetsukazu Yahara for collecting; Christine Tiroch and Hiroko Abe for technical assistance; Kentaro Hosaka for valuable advices. Financial support was received from grants-in-Aid from the ministry of Education, Sports, Culture, Science and Technology of Japan (No. 17370033) and the Kyushu University Foundation.


    • Received August 31, 2008.
    • Accepted October 15, 2008.


View Abstract