Do mycorrhizal symbioses cause rarity in orchids?


  • Ryan D. Phillips,

    Corresponding author
    1. Botanic Garden and Parks Authority, Kings Park and Botanic Garden, Fraser Avenue, West Perth, WA 6005, Australia
    2. School of Plant Biology, The University of Western Australia, Nedlands, WA 6009, Australia
    3. Evolution, Ecology and Genetics, Research School of Biology, 116 Daley Rd, The Australian National University, Canberra, ACT 0200, Australia
    Search for more papers by this author
  • Matthew D. Barrett,

    1. Botanic Garden and Parks Authority, Kings Park and Botanic Garden, Fraser Avenue, West Perth, WA 6005, Australia
    Search for more papers by this author
  • Kingsley W. Dixon,

    1. Botanic Garden and Parks Authority, Kings Park and Botanic Garden, Fraser Avenue, West Perth, WA 6005, Australia
    2. School of Plant Biology, The University of Western Australia, Nedlands, WA 6009, Australia
    Search for more papers by this author
  • Stephen D. Hopper

    1. School of Plant Biology, The University of Western Australia, Nedlands, WA 6009, Australia
    2. Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3AB, UK
    Search for more papers by this author

Correspondence author. E-mail:


1. While rarely tested, the rarity of a species may be linked to the rarity of symbiotic partners. The requirement of many terrestrial plants to form a symbiosis with mycorrhizal fungi may limit the distribution and abundance of plant species. Here, in just the second test of the role of mycorrhiza in host species rarity, we investigate the influence of mycorrhizal specificity, distribution and ecological requirements in the genus Drakaea (Orchidaceae).

2.In situ seed baiting was used to resolve the distribution of mycorrhizal fungi for three common and two co-occurring rare species of Drakaea. Mycorrhizal fungi were isolated from wild adult orchids and protocorms and the ITS nrDNA regions sequenced. An in vitro study tested the range of Drakaea species of which each fungus can support germination.

3. All Drakaea species studied used a narrow monophyletic clade of Tulasnella endophyte for protocorm, seedling and adult plant stages. In situ seed baiting revealed that germination of Drakaea is largely restricted to the same microhabitat as that of the orchid. However, within this habitat Drakaea exhibit comparable germination to other Western Australian orchid genera. Rare and common Drakaea exhibited no difference in germination rates and both germinated in suitable habitat not currently occupied.

4.Synthesis. In contrast to the previous study of the role of mycorrhiza in plant rarity, there was no evidence that mycorrhizal specificity contributed to rarity in Drakaea. However, the formation of mycorrhiza being mostly restricted to a specific microhabitat may limit the abundance of Drakaea in some landscapes. In Drakaea, the highly specific pollination system of sexual deception may contribute to rarity. A trend towards high specificity in orchid mycorrhizal associations in southern Australia is hypothesized to result from the predominance of old landscapes affording the opportunity for specialization on a single or few mycorrhiza(s) best adapted to the landscape conditions.


While declines in species through human modifications of landscapes have been predicted to result in numerous extinctions of their symbiotic partners (Koh et al. 2004; Dunn et al. 2009), the effects of the intrinsic rarity on symbiotic partners are poorly understood. Mycorrhizas are symbiotic associations between a fungus and plant where networks of fungal hyphae hosted by underground roots or stems radiate into the soil and provide nutrients to the plant in exchange for carbon (Newsham, Fitter & Watkinson 1995). Over 90% of angiosperm species form mycorrhizal associations, including ecologically important species in many plant communities (Read, Koucheki & Hodgson 1976; Allen 1991; Newsham, Fitter & Watkinson 1995; Brundrett 2009). Mycorrhizal fungi can affect the diversity of plant communities (Van der Heijden et al. 1998) and potentially the distribution of plant species (Swarts et al. 2010). As such, mycorrhizal interactions may contribute towards intrinsic or human-mediated rarity in some plants. While the rarity of species has been linked to a wide range of ecological traits (Schemske et al. 1994; Bevill & Louda 1999), only one previous study has tested the role of mycorrhizal fungi in causing rarity of the host (Swarts et al. 2010). Terrestrial orchids are notable for their obligate, often specific relationship with mycorrhizal fungi, making them useful test species of the hypothesis that mycorrhiza are involved in plant species rarity.

The Orchidaceae is an exceptionally diverse family (c. 26 000 species; Royal Botanic Gardens, Kew 2009) renowned for their dependency on often specific mycorrhizal and pollinator symbioses (Nilsson 1992; Smith & Read 2008; Peakall et al. 2010). All terrestrial orchids investigated so far require a partnership with a mycorrhizal endophyte for germination, progression to the seedling stage and annual growth (Rasmussen & Rasmussen 2009). Dependence on the fungus is complete at the achlorophyllous protocorm stage for carbon and nutrient uptake (Rasmussen & Rasmussen 2009). Most species establish photosynthesis but remain reliant on the fungus for carbon and nutrients to varying extents (Cameron, Leake & Read 2006; Rasmussen & Rasmussen 2009). While levels of specificity are highly variable between species (e.g. Bonnardeaux et al. 2007; Bidartondo & Read 2008; Jacquemyn et al. 2010; Swarts et al. 2010), orchids tend to form a mycorrhizal relationship with a relatively narrow phylogenetic breadth of fungi in comparison with other types of mycorrhizal relationships (Smith & Read 2008). Further, not all fungi used by the orchid are equally effective at supporting germination and progression beyond the seedling stage (Otero, Bayman & Ackerman 2005; Bidartondo & Read 2008). In orchids the presence of an appropriate fungus and its ability to support germination and growth of the orchid will be critical for the persistence and proliferation of orchid populations.

Mycorrhiza can lead to host rarity through limiting distribution, restricting the potential range of usable habitats and through scarcity within suitable habitat. These interactions could occur either from the absence of a mycorrhiza or the inability to form a symbiosis in certain environments. It is hypothesized that all of these scenarios are more likely to arise when the orchid forms a relationship with a small range of fungi. Previous studies of orchid mycorrhizas have shown that some species use a small number of fungal species as delineated by DNA sequencing (e.g. McCormick, Whigham & O’Neill 2004; Irwin, Bougoure & Dearnaley 2007; Huynh et al. 2009; Jacquemyn et al. 2010; Roche et al. 2010; Swarts et al. 2010). However, the distinction must be made between phylogenetic specificity and ecological specificity. Several orchid species use numerous closely related fungal species and have narrow phylogenetic specificity (e.g. Taylor & Bruns 1997; Shefferson et al. 2007; Ogura-Tsujita & Yukawa 2008), though the use of several fungal species may still provide ample opportunity to exploit different edaphic environments and geographical regions. Here, we refer to high mycorrhizal specificity as the use by an orchid of one or few endophytes.

The development of the seed baiting technique (Rasmussen & Whigham 1993; Brundrett et al. 2003), where packets of orchid seed are buried below the soil surface and later scored for germination, has enabled investigation of the distribution of mycorrhizal fungi. Germination of baits requires both the presence of the mycorrhizal fungus and suitable environmental conditions for the formation of the symbiosis. So far, most studies have focused on the small scale (over 10s to 100s of metres) of the distribution of fungi within orchid populations. Baiting studies in several habitats have demonstrated that the fungi are more widespread than the orchid within populations and that increases in population size is usually limited by seed arrival rather than the absence of suitable microsites (Batty et al. 2001a; McKendrick et al. 2002; Diez 2007; Jacquemyn et al. 2007). The distribution of fungi within habitat patches is correlated with variations in edaphic properties such as soil moisture, organic content, pH and nutrient levels (Batty et al. 2001a; Diez 2007). These environmental correlates demonstrate the possibility for environmental conditions limiting the abundance of mycorrhizal fungi and ultimately the orchids that rely upon them. However, resolution of the role of mycorrhiza and the interaction with environmental variables in orchid rarity at multiple spatial scales requires baiting in areas of both occupied and unoccupied suitable habitat.

In the diverse orchid flora of the South-West Australian Floristic Region (sensuHopper & Gioia 2004), Drakaea has among the highest levels of intrinsic rarity (Phillips et al. 2010) with five of the nine extant species rare and endangered and one species presumed extinct (Hopper & Brown 2007). The mycorrhizal ecology of the genus is largely unknown, though some species are known to use a specific fungal morphotype that is violet-pink and mucid (Ramsay, Dixon & Sivasithamparam 1986). A series of hypotheses were tested to determine if mycorrhizas control intrinsic rarity in Drakaea: (i) rare Drakaea use a smaller number of fungal species than common Drakaea; (ii) rare Drakaea have a lower germination frequency in situ than common Drakaea; (iii) Drakaea have a lower germination frequency than co-occurring orchid genera; (iv) Drakaea fungi are more widespread than the orchid within and between sites; and (v) the distribution of fungi is related to microhabitat.

Materials and methods

Study species and study sites

All species of Drakaea are geophytic herbaceous perennials that produce a single leaf at the beginning of each autumn that senesces during spring (Hopper & Brown 2007). The fungal infection resides within the collar region just below the soil surface at the base of the leaf (Ramsay, Dixon & Sivasithamparam 1986). The genus is almost entirely confined to well-drained grey sands, with the largest populations for most species occurring in low-lying open areas of woodland or regenerating disturbances in jarrah forest (Hopper & Brown 2007). A single flower is produced annually that is pollinated by sexual deception of thynnine wasps (Stoutamire 1974; Peakall 1990; Hopper & Brown 2007)

This study focused on two communities of Drakaea, each comprising one rare species and co-occurring common congeners. The rare Drakaea elastica is endemic to the Swan Coastal Plain where it co-occurs with the comparatively common and widespread Drakaea glyptodon (Hopper & Brown 2007). On the Swan Coastal Plain, Drakaea tend to be confined to the open, low-lying areas associated with Kunzea glabrescens Toelken and are largely absent from the surrounding Banksia woodland (Hopper & Brown 2007). The rare Drakaea micrantha is widely scattered through apparently suitable habitat in the southern jarrah forests, where it co-occurs with the common D. glyptodon, Drakaea livida and Drakaea thynniphila (Hopper & Brown 2007). In this region most large Drakaea populations occur in disturbed areas such as regenerating clearings and along fire-breaks (Hopper & Brown 2007). Both study regions are primarily gently undulating sandplains where relatively subtle topographic changes lead to changes in vegetation community. Brown, Thomson-Dans & Marchant (1998) list both D. elastica and D. micrantha as Declared Rare Flora under the Wildlife Conservation Act (1950).

Seed collection details

Seeds for baiting studies were collected from one or few sites for each species within the study area (Table 1), a range of approximately 330 km from north to south. For species where seed was collected from more than one site, seed batches were combined and mixed prior to use. Seeds used for the in vitro experiment were collected from a single site for each species (Table 1), with the exception of D. glyptodon. Seed of D. glyptodon was collected from geographically separated sites (Northcliffe and Ruabon, Fig. 1) with these seed batches maintained separately. Stems with a pollinated flower were picked just prior to dehiscence, approximately 4 weeks after pollination. Stems were placed in vials with water in the laboratory and the seed capsules picked and placed in envelopes at the commencement of dehiscence. Seed was dried in the envelopes at 15 °C and 15% relative humidity for 3 months before being placed in vials and refrigerated at 4 °C.

Table 1.   Locations of study sites for the Drakaea mycorrhizal baiting study. Region: SCP, Swan Coastal Plain; SJF, southern jarrah forest; DE, Drakaea elastica; DG, Drakaea glyptodon; DL, Drakaea livida; DM, Drakaea micrantha; DT, Drakaea thynniphila. X, present at site; Y, fungal baiting was undertaken at the site for the common–rare experiment; M, fungi baiting was undertaken at the site for the microhabitat experiment; N, fungal baiting was not undertaken at the site
SiteFungal isolatesFungi baitingRegionDEDGDLDMDT
  1. *sites where seed was collected from for the in vitro experiment.

  2. †sites where seed was collected for the in situ seed baiting experiments. Specific details of sites have been withheld because of the rarity of some of the orchids involved in the study.

Capel YSCPX    
Doman Rd YSCP     
Peppermint Grove YSCP     
RuabonDGYSCP X*†X  
Thomas Rd YSCP     
Canebrake YSJF X X*† 
DenbarkerDGNSJF XX  
Mowen cuttingDMYSJF X X† 
Mowen 22 (M) YSJF XXX 
Mowen 29 (M) YSJF XXX 
NorthcliffeDGYSJF X*†X  
N WalpoleDGYSJF XX  
Rainbow CaveDLNSJF XX*†  
S Manjimup (M) YSJF X†X  
S Rocky GullyDMYSJF X XX*†
WildernessDGNSJF X†   
W WalpoleDTNSJF X†  X†
Figure 1.

 Location of study sites for a mycorrhizal baiting study of common and rare Drakaea and collection sites of mycorrhizal fungi for DNA sequencing. Grey dots represent sites where baiting was undertaken with D. elastica (rare) and D. glyptodon (common). Black dots represent sites where baiting was undertaken with D. glyptodon (common), D. livida (common), D. micrantha (rare) and D. thynniphila (common). Black triangles are sites where fungi were collected but no seed baiting was undertaken. Letters in parentheses are the species where mycorrhizal fungi sequences were obtained for that site. DG, Drakaea glyptodon; DE, Drakaea elastica; DL, Drakaea livida; DM, Drakaea micrantha; DT, Drakaea thynniphila.

Fungal isolation and sequencing

Fungi were isolated from adult plants across the range of each species within the study region during 2007, including several sites where fungal baiting was undertaken (Fig. 1). Samples were collected from between one and seven sites from up to seven individuals per site. Fungi were also isolated from protocorms generated during the in situ seed baiting study. Sequencing of isolates from both adult plants and protocorms generated during the in situ seed baiting permits confirmation of the fungi present for the entirety of the life cycle. In vitro germination tests the ability of a fungus to form a symbiosis with an orchid, though it should be noted that in vitro specificity can be broader than in situ specificity (Masuhara & Katsuya 1994). Fungi were isolated from adult plants using the method described in Ramsay & Dixon (2003). The infected collar is washed under tap water before undergoing a series of three rinses in sterile water. Under a dissecting microscope the collar is dissected to release the pelotons from the cortical cells. A serial dilution is undertaken where pelotons are pipetted into a series of large droplets of sterile water. Pelotons were plated out onto Soil Solution Equivalent (SSE) medium with 1% streptomycin sulphate added to reduce bacterial contamination. SSE media were made according to the following: for 2 L, 0.4 g MES buffer, 1400 mL of reverse osmosis water, 200 mL of stock solutions A, B and C, 4 g of sucrose and 16 g of agar [stock solution A: 0.4 g L−1 NH4NO3, 0.0136 g L−1 KH2PO4, 0.61 g L−1 MgCl·6H20 and 0.058 g L−1 NaCl; stock solution B: 0.861 g L−1 CaSO4·2H20; stock solution C: 0.073 g L−1 FeEDTA(Na)]. The pH was adjusted to 5.5 by adding HCl or KOH. The medium was cooked using a microwave and autoclaved for 20 min at 121 °C and 1.05 kg cm−2 at 1.03–1.38 bar. After autoclaving, a total of 20 mL of streptomycin (1 g in 70 mL) was added using a syringe and a sterile 22-μm Millipore filter attachment. Three weeks after isolation, a subset of pelotons for each species were scored for presence or absence of growth. At this point randomly selected individual pelotons were subcultured onto separate SSE plates. After a further 4 weeks, single hyphal tips were subcultured onto PDA medium (3.9 g potato dextrose agar powder and 6.0 g agar to 500 mL of deionized water) to be grown-on for sequencing.

Fungal DNA was extracted based on the method of Gardes et al. (1991). There was some variation in the pattern of fungal growth from plate to plate, but growth was generally most prolific along and just under the surface of the agar. Blocks of agar with a thick concentration of surface hyphae were cut and sliced parallel to the upper surface of the agar. DNA was extracted from these upper slices. About 0.8 mL of fungi and agar was ground with a pestle in a 1.5-mL Eppendorf tube until only very fine chunks of tissue were visible. One millilitre of lysis buffer (2% CTAB, 0.1 M Tris-HCl, 1.4 M NaCl, 0.02 M EDTA) and 5 μL of RNAase were added to the ground tissue. Extracts were incubated for 40–60 min at 60–65 °C. Samples were centrifuged for 10 min at 16 000 g and the supernatant was transferred to a new Eppendorf tube. Proteins were extracted by adding 600 μL of 24:1 chloroform: isoamyl alcohol and shaking for 45–60 min. The extracts were spun at 11 600 g for 10 min before the aqueous (upper) phase was transferred to a new Eppendorf tube. Seven hundred microlitres of isopropanol was added and the extract was incubated at 20 °C for 30 min. After 20 min of centrifuging at 16 000 g, the supernatant was pipetted off and the pellet washed in 500 μL of 70% ethanol. After removing the ethanol and allowing the pellet to dry, the DNA was resuspended in 60 μL of TE buffer.

Nuclear DNA from internal transcribed spacers 1 and 2 and the intervening 5.8S subunit of the ribosomal gene (hereafter referred to as ITS) was initially amplified using the primers ITS1 (F) (5′-TCCGTAGGTGAACCTGCGG-3′) and ITS 4 (F) (5′-TCCTCCGCTTATTGATATGC-3′; White, Bruns & Taylor 1990) in a Corbett Cg1-96 thermocycler (Qiagen, Venlo, The Netherlands). Following confirmation that the fungi involved were Tulasnella, we switched to the Tulasnella-specific primer ITS4-Tul (5′-CCGCCAGATTCACACATTGA-3′; Taylor & McCormick 2008) to increase products from the PCRs. The reaction mix for PCRs contained 5 μL of fungal DNA, 10 μL of 5× polymerization buffer, 6 μL of 25 nM MgCl, 12.8 μL of sterile H2O, 8 μL of ITS1F (3.2 μM), 8 μL of ITS4-Tul (3.2 μM) and 0.2 μL of TAQ polymerase. After a 2-min denaturation at 95 °C, amplifications consisted of 36 cycles. Each cycle consisted of a 1-min denaturation at 95 °C before an annealing step of 2 min at 50 °C and extension for 1 min at 72 °C. The PCR was completed with a final extension phase of 8 min at 72 °C. Products from PCRs were purified using the Agencourt AMPure PCR Purification System by Beckman Coulter (Beckman Coulter Inc., Brea, CA, USA). Sequencing was undertaken by Macrogen Inc. (Seoul, Korea).

Sequences were visually aligned in Se-Al v2.0 (Rambaut 2002). A sequence of the ITS region of a mycorrhiza from Chiloglottis validaHM196801 (Orchidaceae) was included as an outgroup (Roche et al. 2010) along with the most similar sequences from Genbank. A phylogeny of the mycorrhiza associated with Drakaea was created using a parsimony analysis in paup 4.0 (Swofford 2002). A heuristic search was undertaken using TBR branch swapping and 100 random addition sequences, saving a maximum of 200 trees per search. A bootstrap analysis using 1000 replicates of SPR branch swapping was used to give an estimate of the robustness of clades. A maximum-likelihood analysis with bootstrapping was undertaken in RaxML (Stamatakis, Hoover & Rougemont 2008).

In vitro cross specificity of endophytes

Germination trials were conducted using isolates from between four and six plants of each species except for D. thynniphila. Fungi of D. glyptodon were used from two sites, Northcliffe and Ruabon, which are in different biogeographic regions (Phillips 2010). For each isolate, seed from D. elastica, D. micrantha and D. glyptodon (Ruabon population) was germinated on three subcultures and seed from D. glyptodon (Northcliffe population) and D. livida was germinated on another three subcultures. Germination was undertaken on oats agar, comprising of 2.5 g of finely ground oats and 8 g of agar per litre of water and adjusted to pH 5.5. Seeds confined in packets of 90-μm mesh (approximately 150 seeds per packet) were sterilized in 1% bleach for 35 min before being rinsed in three batches of sterile water for three periods of 15 min. Seeds were spread on the surface of the plate of agar in separate portions of the plate. Seeds were scored for germination after 10 weeks according to Batty et al. (2001b), modified from Ramsay, Dixon & Sivasithamparam (1986): 0 – unimbibed seed; 1 – imbibed seed with cracked testa; 2 – germination and development of trichome initials; 3 – enlargement of protocorms, formation of mycorrhiza and initiation of leaf primordium; 4 – further enlargement of protocorm with first green leaf; 5 – seedling with first green leaf and initiation of dropper.

Seed baiting

Investigations of the ecology of mycorrhiza in the field were undertaken during 2007 using the seed baiting technique developed by Rasmussen & Whigham (1993) and modified by Brundrett et al. (2003). In this technique, seeds are confined within packets of fine mesh buried vertically immediately below the soil surface. The germination of seed is used to detect the presence of mycorrhiza capable of forming a symbiosis with the orchid. Several species can be baited with simultaneously by placing the seed in separate compartments within the bait. Each bait was made of 90-μm mesh, with 2 × 2 cm compartments made using a heat sealer, with each compartment containing the seeds of a different species. Approximately 150 seeds per species were used per seed bait.

An experiment was undertaken to test the hypothesis that Drakaea fungi are confined to the open sandy areas that the orchid primarily occurs in. This experiment was conducted in the southern jarrah forest (Fig. 1, Table 1). At each site 30 seed baits were randomly laid within an open sandy area containing Drakaea and 30 baits outside the open sandy area. Baits outside the open area were distributed every 12 degrees, between 1 and 10 m from the margin of the open sandy area. Six species were used in the seed baits, all of which are common on sandy soils in the southern jarrah forest. The comparators were D. glyptodon, D. livida, Caladenia flava R. Br., Elythranthera brunonis (Endl.) A.S. George, Paracaleana nigrita (J. Drummond ex Lindley) Blaxall and Thelymitra crinita Lindley. Drakaea glyptodon, D. livida and P. nigrita all occur predominantly in open sandy areas while C. flava, E. brunonis and T. crinita all occur in a variety of forested habitats (Brown et al. 1998). For each species, Mann–Whitney U-tests with sites as replicates were used to test for a difference in abundance between the open sandy area and the surrounding forest. Analyses were undertaken in spss 11.0 (IBM Corporation, Somers, NY, USA).

In two different ecosystems, we tested if rare Drakaea use rarer fungi than common Drakaea. In both cases all sites were in suitable Drakaea habitat as defined by Hopper & Brown (2007) and the extensive field surveys in Phillips (2010), but not all of the study species occur at each site (Table 1). Nine sites were used in thickets of K. glabrescens to test if the rare D. elastica has lower germination frequency than the common D. glyptodon. Six sites were used in the southern jarrah forest to test if the rare D. micrantha has lower germination frequency than the common D. glyptodon, D. livida and D. thynniphila. A Student’s t-test (K. glabrescens thickets) and a one-way anova (jarrah forest) using sites as replicates were used to test for differences in germination frequency between species.

Seed baits were planted at the beginning of June, coinciding with the first heavy rains, and removed in the last week of September/first week of October, coinciding with the drying of the soil during spring. Seed baits were scored for germination by classification into the five categories defined by Batty et al. (2001b). Successful formation of a symbiosis was considered to be when individuals reached stage three, where the protocorm is enlarged with many trichomes and pelotons are present. The presence or absence of germination to stage 3 was used to calculate the proportions of seed baits where mycorrhiza capable of supporting seedlings occurs (the germination frequency). All proportions were arcsin square-root transformed prior to analyses.

To test if sites with the orchid present had higher abundance of mycorrhiza capable of supporting germination, Student’s t-tests were conducted between sites with and without the orchid present for D. elastica, D. glyptodon, D. livida and D. micrantha. A meta-analysis using the results of the t-tests was used to test if there was an overall significant relationship across the four species. Probabilities were summed across sites and tested for significance using the method of Rosenthal (1984).

To test if the mycorrhiza used by Drakaea have lower germination frequency than other south-west Australian genera, we compared germination of Drakaea in situ with the other genera used in this study and results presented in the literature (Brundrett et al. 2003; Collins et al. 2007; Swarts 2007; Swarts et al. 2010). From the Collins et al. (2007) study, only sites in natural bushland were included. From the Swarts (2007) and Swarts et al. (2010) studies, only the sites within the distribution of the orchid species were included. Mann–Whitney U-tests were used to test for significant differences in germination frequency between each Drakaea species and the other south-west Australian species that have been studied.

Microhabitat preferences of Tulasnella

To test if Drakaea mycorrhiza is associated with a particular microhabitat that may limit the abundance of the fungus or mycorrhization, environmental variables were quantified for a 30 × 30 cm square surrounding the seed bait. The following variables were quantified: % cover of total leaf litter, % cover of leaf litter of Taxandria parviceps (Schauer) J.R. Wheeler & N.G. Marchant (jarrah forest only), % cover of eucalypt leaves (Eucalyptus marginata Sm. and Corymbia calophylla (Lindl.) K.D. Hill & L.A.S. Johnson), % cover of K. glabrescens leaf litter (K. glabrescens thickets only), % cover of Banksia leaf litter (B. menziesii R. Br. and B. attenuata R. Br.; K. glabrescens thickets only), % cover of herbs, % cover of moss, % cover of shrubs < 1, % cover of shrubs 1–2 m, % cover of shrubs > 2 m, leaf litter depth and % cover of debris. All percentage cover values were estimated visually with the aid of a grid. At each point the distance to and height of the nearest K. glabrescens was measured (K. glabrescens thickets only). With the exception of K. glabrescens, the presence of understorey plants within close proximity was not quantified because the relatively few species occurring in these open sandy sites are not mycorrhizal. For each variable at each site a Student’s t-test was conducted between baits with and without successful formation of symbiosis (stage 3 or higher). To test if there was an overall effect across sites, for each variable probabilities were summed across sites and tested for significance using the method of Rosenthal (1984). In addition, environmental variables were averaged across the 30 points at each site and linear regression analysis undertaken to test if any of the environmental variables were associated with variation in mycorrhizal abundance between sites. The additional variable of topographic relief was included, which was measured as metres a.s.l. using a Garmin GPSMAP 60CSx (Garmin International Inc., Olathe, KS, USA). Following both sets of statistical analysis Bonferroni corrections were undertaken. These approaches were preferred to Generalized Linear Modelling with subsequent ranking by Akaike Information Criterion because the high number of variables and models relative to the number of sites would increase the likelihood that models are ranked highly through chance alone or are poor predictors (Burnham & Anderson 2002).


DNA sequencing

A high proportion of pelotons grew after isolation for each of the five study species, suggesting that isolating fungi and growing them in culture has provided a comprehensive representation of the endophytes that occur within Drakaea. In all species, the percentage of pelotons that grew averaged in excess of 85%: D. elastica [86 ± 12 (SE; n = 6)], D. glyptodon (92 ± 5 [n = 14)], D. livida (100 ± 0 [n = 13)], D. micrantha [95 ± 5 (n = 6)] and D. thynniphila [94 ± 4 (n = 9)]. On PDA media the fungi formed slow-growing, orange-pink cultures.

The only sequences on Genbank with greater than 80% maximum identity and query coverage were from Tulasnella tomaculum (AY373296), T. violea (AY373293) and T. eichleriana (AY373292), all derived from McCormick, Whigham & O’Neill (2004). These three sequences were included in the alignment with the isolates from Drakaea. The final ITS alignment contained 860 characters, of which 98 were parsimony informative. All sequences derived from Drakaea mycorrhizal fungi belonged to Tulasnella. However, they were clearly distinct from the northern hemisphere species available on Genbank and the fungus isolated from Chiloglottis valida. The mycorrhiza from Drakaea, including adult plants and wild protocorms, formed a monophyletic clade with 100% bootstrap support in both the parsimony and maximum-likelihood analyses (Fig. 2). In a strict consensus tree, all nodes within this clade collapsed. Among the Drakaea mycorrhizal fungi, the most divergent sequences were 0.49% different, which is less than the 1% and 3% cut-offs previously used to define species in studies involving sequencing of mycorrhizal fungi (Tedersoo et al. 2003, 2008; Bidartondo & Read 2008).

Figure 2.

 Phylogeny of the mycorrhiza isolated from Drakaea. The tree represents one of 13 550 most parsimonious trees. All nodes within the Drakaea mycorrhiza clade collapse in the strict consensus tree. Numbers above the line are bootstrap values for parsimony analysis while numbers below the line are bootstrap values for a maximum-likelihood analysis. DG, Drakaea glyptodon; DE, Drakaea elastica; DL, Drakaea livida; DM, Drakaea micrantha; DT, Drakaea thynniphila; PROTO, fungi isolated from wild Drakaea protocorms. Genbank accessions: Tulsanella tomaculum (AY373296), T. violea (AY373293) and T. eichleriana (AY373292).

In vitro mycorrhizal trials

Germination trials revealed highly variable germination between subcultures taken from the same, single-peloton isolate. For example, for those isolates that did support germination, germination occurred in only 20 out of 47 plates. These differences are likely to have occurred through differences in the environmental conditions between Petri dishes, such as surface moisture on the agar, humidity and temperature. Given that some subcultures of isolates capable of supporting germination often did not, of the 61 plates where no germination was recorded, many of these are likely to be false negatives. By using 150 seeds per plate from the same seed batch, the presence or absence of germination on a plate is not considered to have arisen through differences in seed dormancy between replicates.

Isolates from all four species in the in vitro study supported germination. Fungal isolates from D. elastica and D. glyptodon supported germination of every Drakaea species, while isolates from D. livida and D. micrantha supported germination of one and four Drakaea species respectively (Table 2).

Table 2.   Germination of Drakaea seed on mycorrhizal fungi isolated from adult plants. Numbers refer to the number of plates on which germination was observed. Numbers in parentheses is the highest germination state observed (see Batty et al. 2001b). Seeds: DE, Drakaea elastica; DM, Drakaea micrantha; DG (R), Drakaea glyptodon (Ruabon Nature Reserve); DG (N), Drakaea glyptodon (Northcliffe); DL, Drakaea livida; DT, Drakaea thynniphila. TN, total number of subcultures tested. N, the number of subcultures that supported germination with that combination of seeds. Genbank numbers are given for those isolates where sequences have been obtained
D. elasticaLakelands HQ38675121 (4)1 (5)1 (3)11 (4)1 (5)1 (3)1
D. glyptodonNorthcliffe HQ38673852 (3)1 (3)021 (3)1 (3)1 (5)1
D. glyptodonNorthcliffe HQ3867356   0002 (3)2
D. glyptodonNorthcliffe 8b62 (4)infected1 (3)21 (3)1 (3)1 (3)1
D. glyptodonRuabon HQ38674561 (3)2 (5)2 (5)21 (3)1 (3)01
D. glyptodonRuabon 5601 (3)1 (3)12 (5)2 (5)2 (5)2
D. lividaRainbow Cave HQ38677161 (3)001   0
D. micranthaMowen Cutting HQ386763601 (3)01   0
D. micranthaMowen Cutting HQ38675341 (5)1 (5)1 (3)11 (3)001

Seed baiting

Comparison of germination frequency in open sandy areas and the surrounding jarrah forest demonstrated that Drakaea mycorrhizal fungi are largely confined to open sandy areas where the orchid typically occurs (Fig. 3). While for C. flava, E. brunonis, P. nigrita and T. crinita no difference in germination was detected between habitats (C. flava: P = 0.658, z = −0.44; E. brunonis: P = 0.513, z = −0.65; P. nigrita: P = 0.275, z = −1.09; T. crinita: P = 0.376, z = −0.88), for both Drakaea species, germination was significantly higher within the open sandy areas (D. glyptodon: P = 0.046, z = −1.99; D. livida: P = 0.037, z = −2.09).

Figure 3.

 Comparison of the mean proportion of seed baits with germination (± SE) in five genera of south-west Australian orchids in open sandy areas and the surrounding jarrah forest. CF, Caladenia flava subsp. flava; DG, Drakaea glyptodon; DL, Drakaea livida; EB, Elythranthera brunonis; PN, Paracaleana nigrita; TC, Thelymitra crinita. Dark bars: in open sandy areas, open bars: in sandy jarrah forest.

There was no significant variation in the frequency of germination of the four jarrah forest Drakaea (P = 0.926; F = 0.154; d.f. = 23; see Appendix S1 in Supporting Information). Likewise, a Student’s t-test of frequency of germination for the two Drakaea occurring in K. glabrescens thickets revealed no significant difference (P = 0.76; t = 0.317; d.f. = 8; see Appendix S1). For each Drakaea species there was no significant difference in germination between sites where the orchid did and did not occur. However, a meta-analysis revealed that across all species there was a significant trend of lower germination at sites where the orchid does not occur (P = 0.021; z = 2.039; Fig. 4).

Figure 4.

 The proportion of baits showing germination (± SE) in Drakaea species at sites with (white bars) and without (grey bars) the orchid species present. DG, Drakaea glyptodon; DE, Drakaea elastica; DL, Drakaea livida; DM, Drakaea micrantha.

Comparison of germination between Drakaea and other south-west Australian orchid genera demonstrated that, when comparing baiting studies in the orchid’s preferred habitat, Drakaea has a level of germination comparable to or higher than several common orchid genera (Table 3).

Table 3.   Comparison of germination in the wild of selected south-west Australian orchid genera. Germination is defined as the enlargement of protocorms, formation of mycorrhization and the initiation of leaf primordium. Values represent the mean proportion (± SE) of baits containing orchid seed that germinated, averaged across sites. From Collins et al. (2007), only sites of natural orchid habitat were included. From Swarts (2007) only sites within the distribution of each orchid species was included. Mann–Whitney U tests were conducted between each Drakaea species and member of the other genera. Capitals represent species where the germination frequency is significantly higher than Drakaea (P < 0.05), lower case is used for species where the germination frequency is significantly lower. A/a, Drakaea elastica; B/b, Drakaea glyptodon; C/c, Drakaea livida; D/d, Drakaea micrantha; E/e, Drakaea thynniphila
 Brundrett et al. 2003Collins et al. 2007;Swarts 2007;Present study
Caladenia arenicola Hopper & A.P. Brown0.155 ± 0.033E 0.054 ± 0.029 
Caladenia discoidea Lindley  0.041 ± 0.019 
Caladenia flava R. Br.0.133 ± 0.0600.112 ± 0.0360.102 ± 0.0130.084 ± 0.020
Caladenia huegelii H. G. Reichb.  0.112 ± 0.048 
Caladenia longicauda Lindley  0.022 ± 0.010abcde 
Caladenia thinicola Hopper & A.P. Brown  0.022 ± 0.022 
Drakaea elastica Lindley   0.093 ± 0.026
Drakaea glyptodon Fitzg.   0.092 ± 0.024
Drakaea livida J. Drummond   0.072 ± 0.025
Drakaea micrantha Hopper & A.P. Brown   0.073 ± 0.021
Drakaea thynniphila A.S. George   0.056 ± 0.013
Elythranthera brunonis (Endl.) A.S. George   0.076 ± 0.014
Microtis media R. Br.0.489 ± 0.066ABCDE0.017 ± 0.011ABD  
Paracaleana nigrita (J. Drummond ex Lindley) Blaxall   0.041 ± 0.010
Pterostylis recurva Benth. 0.033 ± 0.021  
Pterostylis vittata Lindley0.011 ± 0.011abcde   
Pyrorchis nigricans R. Br. 0.033 ± 0.017  
Thelymitra crinita Lindley   0.036 ± 0.012

Habitat preferences of Tulasnella

Regression analysis of environmental variables and the frequency of germination of Drakaea (combined across species) revealed no significant differences at P < 0.05 for jarrah forest or K. glabrescens thickets. The only variable approaching significance was a negative trend with low topographic relief in K. glabrescens thickets (P = 0.072; R2 = 0.389; Table 4). Similarly, a meta-analysis of t-tests comparing environmental variables between microsites with and without Drakaea mycorrhiza at each site, revealed no significant differences after the use of a Bonferroni correction (P = 0.05/10 for jarrah forest; P = 0.05/9 for Kunzea thickets; Table 5). Despite using the same narrow monophyletic clade of mycorrhiza, generally only a small number of the species in a bait actually germinated (Table 6), suggesting that mycorrhizal abundance varies at the scale of centimetres.

Table 4.   Regression of environmental variables against germination frequency of Drakaea in open sandy areas of the southern jarrah forest and Kunzea glabrescens thickets of the Swan Coastal Plain. Values represent R2 for the regression. Numbers in parentheses are significance values
 Jarrah forestKunzea glabrescens thickets
% Leaf litter – total0.16 (0.432)0.032 (0.644)
% Leaf litter –Taxandria0.084 (0.577) 
% Leaf litter –Eucalyptus0.345 (0.119) 
% Leaf litter –K. glabrescens 0.128 (0.344)
% Leaf litter –Banksia 0.460 (0.213)
% Herb0.005 (0.898)0.008 (0.818)
% Moss0.0003 (0.973)0.227 (0.195)
% Shrubs < 10.054 (0.656)0.073 (0.481)
% Shrubs 1–20.012 (0.833)0.001 (0.986)
% Shrubs 2+0.045 (0.686)0.007 (0.827)
Leaf litter depth (mm)0.172 (0.413)0.072 (0.487)
Distance to K. glabrescens (m) 0.001 (0.957)
Height of K. glabrescens (m) 0.114 (0.374)
Rainfall 0.137 (0.326)
Topographic relief 0.389 (0.072)
Table 5.   Comparison of environmental variables between microsites with and without Drakaea germination. For each site t-tests were conducted between microsites with and without germination. A meta-analysis across sites was conducted using the method given in Rosenthal (1984). Values represent P-values, while numbers in parentheses are the Z-score
 Jarrah forestKunzea glabrescens thickets
% Leaf litter – total0.390 (−0.279)0.233 (−0.727)
% Leaf litter –Taxandria0.383 (0.299) 
% Leaf litter –Eucalyptus and Corymbia0.424 (−0.191) 
% Leaf litter –Kunzea 0.357 (−0.365)
% Herb0.236 (−0.720)0.163 (0.984)
% Moss0.054 (1.606)0.288 (0.560)
% Shrubs < 10.047 (1.67)0.290 (0.555)
% Shrubs 1–20.385 (0.296) 
% Shrubs 2+0.448 (−0.130)0.032 (1.849)
Leaf litter depth (mm)0.417 (−0.021)0.474 (−0.065)
Distance to K. glabrescens (m) 0.448 (0.131)
Height of K. glabrescens (m) 0.377 (0.315)
% Debris0.464 (−0.091) 
Table 6.   The number of species of Drakaea germinating in seed baits. Each seed bait contained multiple 2 × 2 cm compartments with seed of one species in each compartment. In the jarrah forest baiting was conducted using Drakaea glyptodon, Drakaea livida, Drakaea micrantha and Drakaea thynniphila. In the Kunzea glabrescens thickets D. elastica and D. glyptodon were used
Baits with germinationOne species germinatedTwo species germinatedThree species germinatedFour species germinated
Jarrah forest
 D. glyptodon8210
 D. livida7320
 D. micrantha8220
 D. thynniphila5310
 Jarrah forest combined28520
 K. glabrescens thickets318


Evidence from sequencing of the ITS region and germination studies has demonstrated that both the common and rare species of Drakaea use the same narrow monophyletic clade of mycorrhizal fungi. Maximum sequence divergence was 0.49%, which would lead all isolates to be classified as a single species based on the criterion of less than the 3% changes in ITS base pairs (Tedersoo et al. 2003, 2008) or the stricter criterion of < 1% (Bidartondo & Read 2008). However, different groups of fungi show different levels of sequence divergence between species, suggesting that these arbitrary thresholds only have a limited applicability (Nilsson et al. 2008). Because the fungi used by Drakaea are so dissimilar to any Tulasnella previously sequenced, use of phylogenetic methods to identify species will require much further sequencing of Tulasnella species to provide additional outgroups.

The same narrow monophyletic clade of Tulasnella recorded in all adult plants was also isolated from protocorms from the in situ baiting study for the three species sampled. Further, in vitro these fungi supported germination of multiple species of Drakaea, regardless of the orchid species from which the fungus was isolated. These lines of evidence demonstrate that the fungus can support the entire life cycle of Drakaea in the wild, although further protocorms need to be sampled to establish if other fungi are also present at this stage. However, the higher levels of specificity reported in seedlings compared with adult plants in other orchid species (Bidartondo & Read 2008) suggests that this may not be the case.

While not all fungal isolates supported germination in vitro, the isolates that did support germination were collected from a broad geographical range (Fig. 1), suggesting that variation in the ability of isolates to support germination is not contributing to the rarity or restricted distributions of D. elastica and D. micrantha. The variation in germination observed between isolates may occur through ecological or physiological differences between fungal individuals within a putative phylogenetic fungal species. (e.g. Huynh et al. 2009). Alternatively, this clade may contain multiple species of fungus that may be resolved with more variable markers.

It is evident that mycorrhizal relationships are not driving rarity in D. elastica and D. micrantha. In the cases of D. glyptodon and D. livida the use of a single widespread narrow monophyletic clade of fungi has enabled the orchid to occupy a wide distribution despite high specificity. Further, germination readily occurred at sites not occupied by the common and rare Drakaea, demonstrating the distribution of Tulasnella within suitable habitat is not limiting the formation of new orchid populations. Within habitat patches, Drakaea showed comparable or higher levels of germination in the field than other common south-west Australian genera (Table 3; Brundrett et al. 2003; Collins et al. 2007; Swarts 2007; Swarts et al. 2010), suggesting that scarcity of mycorrhiza within suitable habitat is not the cause of high intrinsic rarity in Drakaea.

The major limitation placed on the abundance of Drakaea by the formation of a mycorrhizal relationship with this clade of Tulasnella is that the formation of the symbiosis is largely confined to a specific microhabitat. The sole previous study testing the role of mycorrhiza in plant species rarity demonstrated that while germination was locally frequent, the formation of mycorrhiza played a role in the rarity of Caladenia huegelii (Orchidaceae) through limiting the orchid to a restricted geographical range (Swarts et al. 2010). While the present study supports previous studies by demonstrating that mycorrhizal specificity does not necessarily lead to host species rarity (McCormick, Whigham & O’Neill 2004; Jacquemyn et al. 2010), it highlights the potential for mycorrhiza-induced rarity through the symbiosis only forming in specific parts of the landscape. This may occur through either the absence of suitable mycorrhizal fungi from some habitats or an inability of the orchid to form a mycorrhizal symbiosis in some habitats. A limitation of the use of in situ seed baits is that selection and habitat specificity may be evident at later life stages. In particular, a subset of habitats that facilitate germination may not permit the orchid to reach adulthood and ultimately reproduce, suggesting that habitat specificity may play a greater role in orchid rarity than evident from baiting studies.

The restricted distribution of Drakaea fungi within the landscape raises the question of why these fungi are limited to such a specialized microhabitat. Drakaea fungi are among the slowest-growing of all mycorrhizal fungi associating with south-west Australian orchids. For example, while Pterostylis fungi can cover a 55-mm diameter Petri dish in only 3 days on SSE medium, Drakaea fungi rarely achieve full plate culture (K. Dixon et al. unpublished observation). The slow-growing Drakaea fungi may be out-competed in high-organic-matter environments leaving the fungi mostly restricted to open sandy sites with minimal organic matter. This hypothesis should be evaluated through experiments testing the competitive ability of Drakaea fungi with other soil fungi under varying organic carbon regimes. Alternatively, the fungus may occur in other microhabitats but not form a symbiosis with Drakaea, a hypothesis that could be evaluated by real-time PCR assays of soil (Landeweert et al. 2003).

Determining the environmental variables correlated with fungal presence may assist in understanding the interaction between the habitat requirements of Tulasnella and the limitation of Drakaea to a specific microhabitat. Previous studies have shown potassium and the presence of leaf litter (Batty et al. 2001a), soil moisture, organic matter and low pH (Diez 2007) to be correlated with orchid mycorrhiza distribution within habitat patches. In the present study, no correlation between environmental variables and orchid germination was detected at the scale of the 30-cm plot. However, in most cases there was pronounced variation in germination between species within baits suggesting that Drakaea fungi may primarily respond to variation in the environment at the scale of centimetres. Further, additional environmental variables not measured in this study may be important in controlling the distribution of the fungus.

At the landscape scale, correlation of environmental variables and the level of germination at each site could give an indication of environmental variables that exert an influence on variation in the abundance of Drakaea mycorrhizal fungi, their ability to support germination and ultimately maintenance of Drakaea populations. In the present study, none of the measured variables showed a significant relationship with the germination of Drakaea. That germination was higher at sites where the orchid occurs suggests that environment is playing a role in determining the distribution of orchid and/or fungus. In the analysis of the Drakaea in K. glabrescens thickets, low topographic relief was approaching significance, and further testing is needed to determine if low topographic relief does lead to increased germination of Drakaea. In south-west Australia, even subtle changes in topography are strongly correlated with soil moisture, which suggests that soil moisture may have the greatest effect on mycorrhization within areas of broadly suitable habitat. Such a relationship would conform to the observation that Drakaea in drier areas usually occur in lower-lying parts of the landscape (Hopper & Brown 2007; Phillips 2010).

Having shown that the formation of a specific mycorrhizal symbiosis does not appear to be important in causing rarity for D. elastica and D. micrantha, further work is required to resolve the causes of rarity in these species. At the landscape scale, unoccupied habitat with suitable mycorrhizal fungi demonstrates that dispersal limitation may play a role. Within habitat patches, because of the excess of microsites that contain mycorrhizal fungi, Drakaea populations are limited partially by the number of seed set events. Given that each Drakaea species is primarily pollinated by a single different species of thynnine wasp (Hopper & Brown 2007; Phillips 2010), if seed set is limiting population size and establishment, it can be attributed to the ecology and behaviour of a single pollinator species.

Mycorrhizal specificity in Drakaea in the context of other Orchidaceae

Orchids exhibit several contrasting patterns of mycorrhizal specificity, which have different implications for the ecology of the orchid. Some species have been shown to form relationships with several mycorrhizal fungi from a narrow phylogenetic range (Taylor & Bruns 1997; Ogura-Tsujita & Yukawa 2008). Alternatively, other orchids use a small range of fungi at one site but are capable of switching to different fungi in other parts of their distribution (McKendrick et al. 2002; Martos et al. 2009). In these circumstances the orchid has multiple species that it can form associations with and remains an ecological generalist. It also demonstrates the importance of sampling at multiple sites when determining the level of mycorrhizal specificity of a plant species. Alternatively, all Drakaea species investigated in this study use the same narrow monophyletic clade of Tulasnella across the study area, a distance of 330 km. The level of specificity detected in Drakaea, where each species uses a single narrow monophyletic clade with < 3% sequence divergence in the ITS region (Tedersoo et al. 2003), has previously been observed in Caladenia huegelii (Swarts et al. 2010) and Rhizanthella gardneri (Bougoure et al. 2009), other terrestrial species from south-west Australia, Liparis lilifolia in the eastern and mid-western USA (McCormick, Whigham & O’Neill 2004), Orchis mascula from western Europe (Jacquemyn et al. 2010) and some species of Chiloglottis in eastern Australia (Roche et al. 2010). The present study includes five out of the nine extant species of Drakaea (including both adult plants and protocorms) making this the highest level of specialization so far recorded in an orchid genus.

Review of the level of mycorrhizal specialization exhibited by orchids in southern Australia (Ramsay, Dixon & Sivasithamparam 1986; Ramsay, Sivasithamparam & Dixon 1987; Perkins & McGee 1995; Irwin, Bougoure & Dearnaley 2007; Swarts 2007; Bougoure et al. 2009; Huynh et al. 2009; Roche et al. 2010; Wright et al. 2010) suggests that the orchid flora of this region may have a relatively high incidence of mycorrhizal specialization compared with the terrestrial orchid floras in temperate Eurasia and North America. The prevalence of relatively old, stable landscapes in southern Australia (Hopper 2009) affords the opportunity for specialization on a single or few mycorrhiza(s) best adapted to the landscape conditions through minimal edaphic changes over extended periods of time. The hypothesis of higher mycorrhizal specialization in older landscapes needs to be investigated with more detailed comparative studies. Further, given the potential consequences of mycorrhizal specificity for rarity in orchids (Swarts & Dixon 2009; Swarts et al. 2010) and where they can occur in the landscape, this hypothesis highlights the need to investigate interactions of mycorrhizal ecology with landscape characteristics to understand both rarity of individual species and trends between floras.


R.D.P. would like to acknowledge the generous funding of the Australian Orchid Foundation, Holsworth Wildlife Research Endowment, the School of Plant Biology at The University of Western Australia and the BankWest Landscope Conservation Visa Card Trust Fund Grants. R.D.P. was supported by an Australian Postgraduate Award. The authors thank Nigel Swarts, Sacha Rouss, Matt Hyde, Bryony Retter and Steve Phillips for assistance in the laboratory. Rod Peakall and two anonymous referees provided comments that improved the final manuscript.