Introduction

In mycorrhizal symbioses, both the phyto- and mycobiont usually associate with a broad range of partners (Molina et al. 1992; Bruns et al. 2002), but there are numerous exceptions to this rule. In addition to the mycoheterotrophic plants (Orchidaceae, Ericaceae and other families) that have specialized on very narrow groups of fungal species (Bidartondo 2005; Hynson and Bruns 2010), certain fully autotrophic plants from different ecosystems display high specificity towards fungi. In ectomycorrhizal (EcM) associations, such specific interactions are apparent in a tropical liana species Gnetum africanum (Gnetaceae; Bechem and Alexander 2012a), a tropical tree species Pisonia grandis (Nyctaginaceae; Chambers et al. 2005), and an entire tree genus Alnus (Betulaceae; Molina et al. 1992; Tedersoo et al. 2009). P. grandis typically inhabits guano-rich soils in coral islands and associates with only one to three species of fungi from the /tomentella-thelephora lineage (Chambers et al. 2005; Suvi et al. 2010; Hayward and Horton 2012). Alnus spp. prefer riparian and early successional habitats in subarctic to warm temperate ecosystems, where they associate with both Actinobacteria and a restricted set of EcM fungi (Molina et al. 1992; Pritsch et al. 1997). Specificity for mycobionts in autotrophic plants has been often ascribed to a specific ecological niche or habitat (Molina et al. 1992), but recent findings argue against this notion in several plant taxa. In particular, apart from Alnus, other actinorhizal plants such as Dryas and Cercocarpus (Rosaceae) host a broad range of fungal symbionts that largely overlap with mycobionts of neighboring trees and shrubs (Harrington and Mitchell 2005; Ryberg et al. 2009; Bjorbækmo et al. 2010; McDonald et al. 2010). Other members of the Nyctaginaceae family associate with multiple fungal species from several phylogenetic lineages (Tedersoo et al. 2010b). In P. grandis, specificity is evident throughout the steep gradient of soil nitrogen concentration (Hayward and Horton 2012).

The genus Gnetum comprises 30–33 species of lianas and understorey trees and has a pantropical distribution (Won and Renner 2006). Within Gnetum, EcM symbiosis was first described in G. africanum in the Democratic Republic of Congo, where conspicuous yellow root tips were associated with fruit bodies of Scleroderma sp. through rhizomorph connections (Fassi 1957). Similar associations were later confirmed in varieties of G. africanum in Cameroon (Onguene 2000) and other Gnetum species in South America (St. John 1980; Singer et al. 1983; Moyersoen 1993). Recently, sequence analysis of EcM root tips from natural and planted G. africanum confirmed that only a few closely related species of Scleroderma associate with G. africanum (Bechem and Alexander 2012a). Experimental inoculation trials revealed that only isolates of Scleroderma sp. were able to form EcM with G. africanum, whereas the non-native Pisolithus and Paxillus species failed to infect rooted cuttings (Bechem and Alexander 2009). Scleroderma sp. provides mineral phosphorus and nitrogen to the associated G. africanum and thereby stimulates growth of the cuttings (Bechem 2011; Bechem and Alexander 2012a, b). Moreover, G. africanum cannot establish without its root symbiotic fungi in plantations (Bechem and Alexander 2012b). Leaves of G. africanum are very popular and relatively expensive source of food in Equatorial Africa. Because of overexploitation of natural populations and difficulties in establishing plantations, this species has been lost from many natural habitats near human settlements. Unlike G. africanum, G. gnemon constitutes an understorey tree in lowland rain forests in the Indo-Malay and West Pacific regions. Throughout its range, both the leaves and flowers and fruits of G. gnemon are widely used for food (Cadiz and Florido 2001). Cutting trees for access to edible parts has resulted in decline of natural populations of this tree species. The present study addresses diversity of root symbionts of G. gnemon. We hypothesized that G. gnemon has a very narrow range of mycobionts as observed for G. africanum and as suggested for South American Gnetum species.

Materials and methods

Root samples of seedlings and saplings of G. gnemon were collected in three sites in lowland rain forests of the Morobe Province in Papua New Guinea. Bawituc (6.639° S; 147.909° E; ca 200 m above sea level (a.s.l.)) is a secondary rain forest site located 15 km northeast of the city of Lae. Oomsis (6.712° S; 146.807° E; 150 m a.s.l.) is a selectively cut primary forest located 30 km west of Lae. Kamiali (7.303° S; 147.130° E; 10 m a.s.l.) is a slightly disturbed primary forest located 50 km south of Lae. In all sites, some G. gnemon trees were killed by overexploitation and most of the remaining trees displayed signs of recent harvesting of leaves (i.e., partial defoliation). We uprooted nine seedlings or saplings of G. gnemon (three in Bawituc, one in Oomsis, and five in Kamiali) and wrapped a 10-cm fragment of a healthy long root with intact short roots in a plastic bag. The roots were washed and examined visually for the presence of a fungal mantle. All clusters of root tips were excised and stored in CTAB buffer (1 % cetyltrimethylammonium bromide, 100 mM Tris-HCl (pH 8.0), 1.4 M NaCl, and 20 mM EDTA). DNA was extracted from at least three EcM root tips of each tree individual using a DNeasy 96 Plant Kit (Qiagen, Crawley, UK) as recommended by the manufacturer. PCR was performed with primers ITS-OF and LB-W (Tedersoo et al. 2008) or ITS-OF and ITS4 (White et al. 1990). The PCR included 5 μl of 5× HOT FIREPol Blend Mastermix Ready to Load (Solis Biodyne, Tartu, Estonia), 3 μl of template DNA, 0.5 μl of each 20 μmol/ml primer, and 16 μl dsH20. PCR reactions were run under the following conditions: an initial 15 min at 95 °C, followed by 35 cycles of 95 °C for 30 s, 55 °C for 30 s, 72 °C for 1 min, and a final cycle of 10 min at 72 °C. PCR products were separated by electrophoresis through a 1.5 % agarose gel, purified using Exo-Sap enzymes (Sigma, St. Louis, MO, USA) and subjected to sequencing with primers ITS5 and/or ITS4 (White et al. 1990). Sequences were assembled into contigs and checked for quality by use of Sequencher 4.10 software (GeneCodes Corp., Ann Arbor, MI, USA). Species were delimited based on 97.0 % ITS sequence similarity threshold. All unique sequences from each individual plant were submitted to the International Sequence Databases and are available under accessions JX316729-JX316738. To address the phylogenetic placement of Scleroderma isolates from the root samples of G. gnemon relative to other sources, all 194 ITS sequences of Scleroderma that had information about the country of origin were downloaded along with their metadata from the copy of GenBank and UNITE databases over the PlutoF workbench (Abarenkov et al. 2010). Sequences covering <80 % of the ITS region and redundant sequences (identical and originating from the same study and country) were removed. Four sequences of Pisolithus spp. were used as outgroups for the alignment in MAFFT server (http://mafft.cbrc.jp/alignment/server/) and subsequent maximum likelihood phylogenetic analysis with fast bootstrapping (using default options) as implemented in RAxML (Stamatakis et al. 2008).

Results

Root samples from seven out of nine G. gnemon trees included EcM root tips as revealed by a combination of microscopic examination and molecular identification. Based on microscopic examination, EcM root tips were lacking in two seedlings that were collected from different sites. Sequence analysis of non-ectomycorrhizal root tips revealed colonization of putative endophytes or soil saprotrophs (not shown). In the root samples of other G. gnemon trees, only a single EcM fungal species was identified from each tree individual. This was supported by microscopic observation of a single dominant morphotype on the root tips of each tree. Morphotypes differed from each other by color and the amount of mycelium and rhizomorphs. In addition to white and yellow rhizomorphic types described in Bechem and Alexander (2012a), smooth pale and smooth black morphotypes were observed. From root tips of individual G. gnemon trees, two species of Scleroderma (/pisolithus-scleroderma lineage; cf. Tedersoo et al. 2010a) and a single species of the /tomentella-thelephora, /russula-lactarius, /sordariales, and /inocybe lineages were recovered from root tips of G. gnemon (Table 1). Species of the /inocybe lineage was found from two G. gnemon individuals at Bawituc. None of the recovered EcM fungal species were represented in a concurrent sampling effort of >250 specimens of EcM fungi throughout Papua New Guinea.

Table 1 Identification of EcM fungi associated with G. gnemon individuals at various sites in Papua New Guinea
Full size table

The maximum likelihood phylogram of the genus Scleroderma suggested that the two recovered species are relatively closely related, but not sisters to each other (Fig. 1). The analysis revealed that all isolates from Gnetum spp. in Africa, Malaysia, and Papua New Guinea belong to a narrow, well-supported (BS = 94) clade of closely related species. Scleroderma sp1 (UDB013025) is highly similar and probably conspecific with a root tip isolate of Dipterocarpaceae from Malaysia (GQ268591) that share 98.6 % ITS sequence similarity. Scleroderma sinnamariense is the only described species that has available ITS sequences and belongs to the clade that includes root tip isolates of Gnetum spp.

Fig. 1
figure 1

Maximum likelihood phylogram demonstrating the phylogenetic placement of mycobionts of G. gnemon from Papua New Guinea (PNG; in bold) among isolates of Scleroderma. Bootstrap values >50 are indicated above branches. Bar indicates 0.1 substitutions per site. The narrow clade associating with Gnetum spp. is highlighted

Full size image

Discussion

Contrary to our hypothesis, G. gnemon associated with multiple species of EcM fungi that belonged to several phylogenetic lineages such as /pisolithus-scleroderma, /inocybe, /russula-lactarius, /tomentella-thelephora, and /sordariales. Previously, only members of Scleroderma spp. were identified from root tips of African and Malaysian Gnetum spp. (Bechem and Alexander 2012a). The other fungal species associating with G. gnemon belong to lineages that are all commonly ectomycorrhizal with the Dipterocarpaceae and Fabaceae host trees in tropical forests (Tedersoo et al. 2007, 2011; Peay et al. 2010; Smith et al. 2011; Phosri et al. 2012). Our limited sampling effort does not enable us to address whether these fungi are randomly selected from the local pool of EcM fungi that associate with the species of Dipterocarpaceae, Intsia (Fabaceae), and Myrtaceae in these lowland rain forests. It is likely that these fungal species have a broad host range because Scleroderma spp. that associate with Gnetum also establish EcM symbiosis with other EcM host trees (Bechem and Alexander 2012a; see also Singer et al. 1983 and Onguene 2000 for potential mycorrhizal networks involving Gnetum).

The two Scleroderma species from G. gnemon belonged to a narrow clade that includes all sequenced isolates of Gnetum symbionts sequenced so far. This indicates that G. africanum exhibits high specificity for a few closely related Scleroderma spp. in the S. sinnamariense group (Bechem and Alexander 2012a), whereas G. gnemon hosts a broader range of mycobionts including some of these specific taxa. Because the genus Gnetum probably originates from South America and it subsequently dispersed to Africa and Southeast Asia (Won and Renner 2006), information on South American fungi is required to understand the evolutionary ecology and history of host specificity in EcM symbiosis involving Gnetum.

Nonetheless, our study indicates that specificity for mycobionts may strongly vary among congeneric plant species. This is in striking contrast with the genus Alnus, where all studied plant species share their mycobionts and display high specificity for these (Molina 1981; Põlme et al., unpublished). In the genus Pisonia, in-depth molecular investigations of EcM root symbionts are so far restricted to P. grandis. These studies recovered only one to three fungal species belonging to the /tomentella-thelephora lineage in coral cays of the Indian and Pacific oceans (Chambers et al. 2005; Suvi et al. 2010; Hayward and Horton 2012). Mycological surveys have recovered several EcM fungal species from the /pisolithus-scleroderma and /russula-lactarius lineages under canopies of other Pisonia species in Central America (Pegler 1983; Guzman et al. 2004), suggesting that these taxa may not have such a narrow specificity, although reciprocal association with a restricted set of partners seems to be a common phenomenon in the Nyctaginaceae family (Tedersoo et al. 2010b). However, host species require careful identification by molecular identification or tracing of roots because EcM plant species of Coccoloba and Neea may co-occur with the South American Pisonia species, whereas Intsia bijuga and some introduced EcM trees may co-occur with P. grandis in the Pacific islands.

In conclusion, G. gnemon does not, as suggested for G. africanum (Bechem and Alexander 2012a), specialize exclusively on Scleroderma mycobionts. Our results demonstrate for the first time that congeneric plant species may substantially differ in their specificity for ectomycorrhizal symbionts. Further sampling of Gnetum species in South America is required to address the evolution of host specificity in this gymnosperm genus.