A View of Specificity in Orchid Mycorrhizae Using Molecular Symbiont Identification

Taylor, L., and Bruns, T.D.

Dept. of Plant Biology and Dept. of Environmental Science, Policy and Management. University of California, Berkeley CA, USA

Poster presented at The Fifth Internationl Mycological Congress
August, 1994 Vancouver, British Columbia, Canada

Specificity in orchid mycorrhizae has been controversial for more than 50 years. In vitro seed germination studies show a lack of specificity while isolations from field samples show higher specificity. We believe that associations as found in nature are more revealing. We have divided the issue of specificity into the components of habitat influence and active choice by the partners. Habitat influence was addressed by sampling four co-occuring native terrestrial orchids over a wide geographic range. Active choice was tested by sampling neighboring orchids of different species. Symbiont identification, which is critical to specificity studies, has been a major difficulty in orchids as with other mycorrhizal types. We have developed molecular techniques which allow discrimination of orchid endophytes directly from infected tissue. These techniques have allowed us to concretely identify for the first time an ectomycorrhizal fungus with which an achlorophylous orchid associates. Cephalanthera austinae was found to associate with several species in the Thelephoraceae. We also found several symbionts associated with each of the other orchid species. However, neighboring orchids of different species never shared symbionts. Further, there was no overlap in symbiont patterns between even closely related orchid species. Samples of Corallorhiza maculata taken over one year did not provide any evidence of seasonal turnover. Our results strongly support the existence of specificity in orchid mycorrhizae. Habitat was also seen to impact the symbiosis, as the distribution of particular fungal patterns was correlated with geography, forest type and elevation.

We are studying specificity in orchid mycorrhizae by combining molecular methods for fungal identification with sampling under natural conditions. Our goals are to test several hypotheses concerning specificity of orchids toward fungal symbionts and to understand the autecology of selected orchid-fungus interactions.

Our sampling strategy was designed to test two hypotheses put forth by opponents of specificity. J.T. Curtis (1937, 1939) proposed that biases in one orchid species toward association with a particular fungal species are due to similar habitat preferences of the two organisms rather than to innate features of the symbiosis. He further hypothesized that a particular orchid species will likely associate with different fungi when growing in different habitats. This is the first hypothesis we have tested. Harvais and Hadley (1967) and Hadley (1970) concluded from their studies that specificity was low in orchid mycorrhizal associations, but suggested that this could be further tested by examining orchids of different species where they co-occur. The second hypothesis we have tested is their prediction that neighboring orchids of different species will share the same fungal symbiont.

In order to test hypothesis 1, we have sampled four target orchids species Corallorhiza maculata (Rafinesque) Rafinesque, Corallorhiza mertensiana Bongard, Cephalanthera austinae (A. Gray) Heller, and Goodyera oblongifolia Rafinesque over a wide geographic range. To test hypothesis 2, we have sampled any orchids of different species found growing together. We chose the four species above as principal targets because their ranges overlap significantly, thereby increasing the odds of finding "neighbors." We have also analyzed seasonal turnover by sampling a local population of Corallorhiza maculata throughout one year.

The rhizoctonia symbionts of most orchids are difficult to discriminate or identify due to absence of useful morphological features and infrequent fruiting. We have employed the molecular technique of polymerase chain reaction (PCR) amplification of the internal transcribed spacer (ITS) of the nuclear ribosomal repeat and subsequent restriction fragment length polymorphism (RFLP) analysis for fungal symbiont discrimination (Liu and Sinclair, 1992). This procedure is rapid, and distinguishes Basidiomycetes at approximately the level of the biological species (Gardes and Bruns, 1993; Gardes, personal communication), and is the first technique to be used directly on orchid mycorrhizal tissue without fungal isolation.

Orchids were sampled by locating flower spikes, and digging up a small portion of root or rhizome. The overstory tree species was noted. If neighboring orchids of different species were located, the distance between samples was noted. In the study of seasonal turnover, samples were taken in January, March, June, July, September, and December from a single large individual of C. maculata and from a stand of nearby individuals of the same species. The fresh roots and rhizomes were scrubbed clean with a brush, blotted dry, then kept frozen. DNA was extracted from frozen tissue using an SDS buffer, chloroform extraction, and treatment with Gene Cleanš. Plant ITS sequences were amplified with the primer pair (CTTTATCATTTAGAGGAAGGAG) ITS 1P/ITS4 and AN annealing temperature of 52 C, while fungal ITS sequences were amplified with the primer pair ITS 1F/ITS 4 or ITS 1F/ ITS 4B and annealed at 53 C. The utility of these primer combinations for amplification of the desired plant or fungal DNA from mycorrhizal extracts is demonstrated in Figure 1. Eight to 10 uL of the resulting PCR products were digested in a total volume of 15 uL with the enzymes Alu 1, Mbo 1, Hinf 1 or Taq 1 according to manufacturers instructions. The digested DNAs were electrophoresed at 180 mV for 1 hour in 2% NuSieveš Agarose/1% Ultrapureš Agarose, stained with ethidium bromide, viewed and photographed on a transilluminator. The symbiont "types" "A" through "BB" were defined based on unique combinations of RFLP patterns from at least two enzymes. All Corallorhiza and Goodyera samples were amplified with ITS 1F/ITS 4 and characterized for Alu I and Mbo I RFLP patterns. Cephalanthera austinae samples were amplified and characterized primarily with the primers ITS1F/ITS 4B and the enzymes Alu I and Hinf I, though several samples were characterized with ITS 1F/ITS 4 and Alu I. Further details of these methods will be presented elsewhere.

RESULTS: Specificity: Mycogeography: Epiparasitism: Comments

We were able to obtain diagnostic PCR generated RFLP patterns from 90% of Corallorhiza and Cephalanthera samples collected. Lack of amplification was usually because we did not obtain colonized tissue. We had a very low success rate in PCR amplification from certain orchid species (data not shown). We believe this is because the primers used do not match the sequences of some groups of rhizoctonia fungi. Success with Goodyera oblongifolia was less than 50%. Therefore, the results for this orchid probably do not represent the full range of symbionts.

We are in the process of generating RFLP patterns from culture collection reference strains of rhizoctonia fungi, with the hope that they can be matched with the patterns from unknown orchid mycorrhizae (data not shown). However, in screening 18 reference strains, we have yet to find a match with an unknown orchid pattern. We believe the lack of matches is because the reference strains screened represent only a small proportion of the distinct fungal entities in this speciose group of fungi. In screening the reference strains, we have found that species of Tulasnella and Sebacina do not amplify using the primer ITS 1F. These isolates do amplify with the conserved primers ITS 1/ITS 4. This may explain the lack of amplification in certain orchid species.


In our study, neighboring orchids of different species did not share the same fungus. Every orchid species studied had a unique range of fungal symbionts. Thus, a high degree of specificity exists in these orchids. This specificity is not simply a correlation between one orchid and one or several fungi. The absence of any overlap in fungal symbionts, even in sites where several orchid species are growing and there are several fungi capable of association with orchids present, indicates that a process of selection is occurring which determines the pairings we see in nature. We would like to learn more about this process of selection. An important question is when does selection occur in the ontogeny of the plant. Studies of seedling germination and subsequent development in nature may allow us to address this.

Our results disagree with Curtis' views concerning habitat and lack of specificity. However, his assertion that the symbionts found will vary depending on the environmental variables of the site is borne out by our data. As shown in Figure 3, there were strong geographic patterns in the distributions of symbiont types. Some patterns appear very localized, while others are widespread. For example, type M was found only once, in Wildcat Regional Park, while all six other individuals sampled in the park contained type D. Type C was notably found only in sites above 6000 feet elevation. It was also the only type found in these sites. Most Corallorhiza plants were found growing under mature conifers, particularly Pseudotsuga menziesii, and a variety of symbionts occurred in these sites. Every sample of Corallorhiza maculata collected under pure or mixed stands of Quercus was associated only with type D. Type D was not found in any other sites.

We have solid evidence that seasonal changes in symbiont identity does not occur in Corallorhiza maculata. However, other orchid species may operate differently. The perennial nature of the densely infected rhizomes of Corallorhiza maculata.would seem to make it difficult to "loose" a fungus, once established. Many orchid species, on the other hand, support only sporadic colonization and have an annual cycle of root or sinker production. Seasonal turnover is more likely in these species.

The conclusive identification of several mycorrhizae of Cephalanthera austinae as ectomycorrhizae-forming Thelephoroid species reveals an interesting new trophic niche for non- photosynthetic orchids. The diversity of patterns revealed in the morphologically homogeneous mycorrhizae of Cephalanthera austinae further demonstrates the sensitivity of this approach. Other "saprophytic" species have long been known to associate with pathogens such as Armillaria (Kusano, 1911) and rotters such as Marasmius (Burgeff, 1959). Warcup (1985, 1991) showed that the rhizoctonia endophyte of Rhizanthella gardneri forms ectomycorrhizae with the shrub Melaleuca uncinata, under which the orchid is always found. Therefore, three-way interactions are not undocumented in orchids. However, this is the first report of an orchid association with an ectomycorrhizal fungus in which the fungus has been clearly identified to a specific taxon. This mode of epiparasitic survival appears very similar to that of several Monotropoid species which associate with ectomycorrhizal Basidiomycetes (Kenneth Cullings, Jr., doctoral dissertation).

In considering coevolution, one must consider that carbon flows from fungus to plant, unlike other mycorrhizae. At present, it is not known whether the fungi gain any fitness from this bizarre association. Coevolution as strictly defined in terms of parallel phylogenies is not obeyed by orchid mycorrhizae. Corallorhiza trifida associates with a clamped Basidiomycete (Carla Zelmer, personal communication) while the congeneric Corallorhiza maculata and Corallorhiza mertensiana associate with rhizoctonia-like fungi which are only very distantly related to the clamped Basidiomycete. Our results showing that C. maculata and C. mertensiana do not share any symbionts are rather striking in the light of recent evidence that C. mertensiana is derived from a paraphyletic C. maculata clade (Freudenstein and Doyle, 1994). Thus, from this limited data, it would appear that "host switching" in the sense of the fungus as the host for the orchid, has happened repeatedly in orchid evolution.

Lee's home page the Bruns Lab Home Page