In nature there are numerous examples of cheaters that subvert a normally mutualistic interaction with a symbiotic partner (Bronstein, 2001). The majority of mycoheterotrophic (MH) plants studied thus far cheat one of the most widespread mutualisms on earth – the mycorrhizal symbiosis. Because the mycorrhizal symbiosis has evolved in three major lineages of the fungal kingdom (Fig. 1), it is no surprise that each of these diverse groups of fungi has been infiltrated by MH plants (Fig. 1). In fact, there are only two orders within the Glomeromycota – a group of obligate plant symbionts and one class of Ascomycota that contains some mycorrhizal fungi – that have not (yet!) been found to associate with mycoheterotrophs (Fig. 1). In addition to MH plants that have cheated the mycorrhizal mutualism, there are also examples of MH plants which gain carbon directly from saprotrophic fungi that break down and assimilate complex organic substrates.
Of the three major fungal lineages that have been exploited by MH plants, the Glomeromycota is the most ancient mycorrhizal group and supports the greatest number of MH species (Leake, 1994). These fungi form arbuscular mycorrhizae (AM), the most common mycorrhizal association of plants in general (Smith & Read, 2008). Interestingly, the oldest group of mycorrhizal fungi also support perhaps some of the oldest lineages of MH plants, such as club mosses, ferns and whisk ferns, many of which have fully MH gametophytes (Winther & Friedman, 2007, 2008, 2009; Leake et al., 2008). Arbuscular mycorrhizal MH plants studied thus far have been found to associate mainly with fungi in the ‘Glomus A’ clade. However, it remains unknown if fungi in this group are also dominant mycobionts of autotrophic plants, in which case the apparent preference of MH plants for these fungi would not be unexpected.
Within the Agaricomycetes, all orders that contain mycorrhizal fungi have been exploited by MH plants (Fig. 1). In particular, fungi that normally form mutualistic ectomycorrhizae (EM) with many tree species have been repeatedly taken advantage of by MH species in the family Orchidaceae and in the subfamily Monotropoideae (Ericaceae; Taylor et al., 2002; Bidartondo, 2005; Fig. 1, Supporting Information Table S1). Owing to the diversity of fungal guilds that associate with mycoheterotrophs (Fig. 1, Table S1) it does not appear that any particular lineage of mycorrhizal fungi are more prone to be cheated than any other, and the ability of mycorrhizal fungi to provide sufficient carbon transfer from autotrophic plants to MH plants is phylogenetically widespread (Leake & Cameron, 2010).
The Hymenoscyphus ericae complex is perhaps the most widespread example of a mycorrhizal lineage that has not yet been found to host MH plants. This exception is striking, for two reasons. First, fungi within the H. ericae complex are abundant mycorrhizal symbionts with members of Ericaceae across huge areas of the globe (Smith & Read, 2008), and there is now mounting evidence that the plant partners of these fungi are not limited to Ericaceae, and can form mycorrhizae with some tree species (Curlevski et al., 2009; Grelet et al., 2009, and references therein). Second, even though none of the ericaceous plants that associate strictly with fungi from the H. ericae complex are known to be mycoheterotrophic, the majority of species in the closely related and potentially ancestral group Monotropoideae are fully mycoheterotrophic.
An apparently parallel evolutionary path to mycoheterotrophy has arisen among nonphotosynthetic orchids in the tribes Diseae, Gastrodieae and Vanilleae, which associate strictly with normally saprotrophic fungi (Table S1). To date, MH orchids associated with saprotrophic fungi have been found in multiple countries, including tropical forests of the Caribbean, Taiwan, Myanmar, India and Australia as well as in more temperate regions of Japan and Korea. Similar to MH plants that associate with mycorrhizal fungi, those that depend on saprotrophic fungi usually show a preference for particular fungal hosts (Yamato et al., 2005; Ogura-Tsujita & Yukawa, 2008; Ogura-Tsujita et al., 2008; Martos et al., 2009). So far there is no evidence of MH plants outside Orchidaceae that have exploited saprotrophic fungi in this way; however, there remain many unstudied groups of MH plants, especially in the tropics.
Within this phylogenetically and physiologically widespread set of fungal hosts of MH plants is nested another pattern – specificity; with few exceptions, individual MH plants are specialized on single families, genera, or even species of fungi (Taylor et al., 2002; Bidartondo, 2005). Sometimes what appear to be less specific patterns of fungal preferences are caused by the existence of cryptic species of MH plants that each associate with certain fungal taxa (Taylor et al., 2004). Two recent studies of MH orchids in the tropics appear to be exceptions to the specificity rule. In the study by Roy et al. (2009), two species of MH orchids in the tribe Neottieae were found to associate with multiple EM fungi, and Martos et al. (2009) found two species of MH orchids in the tribes Orchidoideae and Epidendroideae that associate with various litter-decaying or wood-decaying basidiomycetes. Another example of a generalist MH species outside the Orchidaceae is Pyrola aphylla (Ericaceae), which has recently been found to associate with numerous EM fungi throughout the plant’s distribution in western USA (Hynson & Bruns, 2009).
The factors leading to specificity among MH plants remain unclear, but there are emerging data that may shed some light on the role of specificity. For instance, Julou et al. (2005) and others have proposed that for a plant to transition from a primarily autotrophic to a MH lifestyle there must first be a shift in the fungal associates of the plant, then selection of a specific fungal host, followed by the loss of photosynthesis and a transition to full mycoheterotrophy (Bidartondo et al., 2004). In Julou et al.’s (2005) study, this first step was exemplified by green orchids that have shifted from associating with Rhizoctonia species, a polyphyletic genus of fungi that are common symbionts of most orchids, to a suite of EM fungi. They also found that albino mutants of these normally green orchids associated with a diversity of EM fungi, and, based on the albino orchids’ carbon and nitrogen stable isotope values, they appeared to be fully mycoheterotrophic (Leake & Cameron, 2010). This, along with the recent findings of Hynson & Bruns (2009) and Roy et al. (2009), indicate that the loss of photosynthesis in mycoheterotrophs is not strictly contingent on fungal specialization, and apparently the ordering of steps towards the evolution of MH plants is not fixed.
Two obvious questions remain: why specialize? And what plant–fungal interactions determine the fungal host of adult mycoheterotrophs? Similar to many parasites, fungal specificity among mycoheterotrophs may be favored so that the plant can fine-tune its physiology to maximize the benefits from its fungal host over the course of its life cycle (Leake & Cameron, 2010). The cost of this fine-tuning to the mycoheterotroph may be that it prevents broad host jumps and severely limits the expansion of its distribution, while the cost to the fungal host remains unknown (Bidartondo, 2005). An alternative and perhaps not mutually exclusive explanation for why most adult MH plants have specific fungal partners may be that the plant has been rejected by the pool of available fungi present at a site and is thus engaged in a co-evolutionary arms race with fungi (Taylor, 2004; Bidartondo, 2005). However, the rarity of MH plants in most settings makes it seem unlikely that they would create much selective pressure on fungi. In addition, if an arms race were the primary factor that determines the fungal host of mycoheterotrophs, then one would expect to find more examples of host switches in distant plant populations or where there are encounters with a ‘naïve’ fungus, such as after a long-distance dispersal event (Bidartondo & Bruns, 2005). This type of host switching was demonstrated by Bidartondo & Bruns (2005) among Monotropa uniflora individuals, but the switches were only to fungal congeners.
An arms race model might also be expected to lead to co-cladogenesis, where the phylogenies of MH plants and their fungal hosts would become concordant over evolutionary time. Merckx & Bidartondo (2008) documented the best example of such phylogenetic concordance but with an interesting, and incongruent, twist. The phylogeny of MH plants in the genus Afrothismia (Thismiaceae) track the phylogeny of their fungal hosts in the Glomus A clade fairly well. However, molecular clock estimates show that this tracking could not have occurred as a result of co-speciation, because the diversification in the Glomus A clade took place 70 million yr before the existence of Afrothismia (Merckx & Bidartondo, 2008). This means that even though Afrothismia speciation tracked the phylogeny of Glomus taxa, Glomus taxon speciation was unrelated to that of Afrothismia species!
Before we bury the arms race model, however, we need to consider how MH plants interact with close relatives of their fungal hosts. Several studies have shown that seeds of MH plants germinate in response to their specific fungal hosts, but they will also germinate in response to closely related fungi (McKendrick et al., 2000; Leake et al., 2004; Bidartondo & Bruns, 2005). These ‘mistakes’ occur in the field, and in some cases seedling development starts, yet no mature plants associated with the wrong fungus have been found (McKendrick et al., 2000; Leake et al., 2004; Bidartondo & Bruns, 2005). This means that at some stage of development the plants associated with the wrong fungi either die or manage to switch to their correct symbiont. Death seems to be the more likely possibility, but the question remains of whether this is a plant response that is initiated by the fungus, supporting the arms race model, or simply a physiological mismatch. It may soon be possible to differentiate between these two possibilities by examining fungal gene expression when associated with compatible and incompatible MH plants.
There remain numerous plant lineages that contain fully, and potentially facultative, MH species that have yet to be studied in detail, especially in the tropics. Many recent studies have revealed new information on the identities of the fungi capable of hosting mycoheterotrophs, but it remains a mystery how and why particular fungi become targets. Extant populations of MH plants are examples of successful subversions of the fungal community; however, in nature there are settings where the fungal hosts for mycoheterotrophs are present, but the plants are not. Thus, there remain many unseen and unexplained factors that are potentially protecting fungal communities and networks from cheaters. We have revealed many of the players, but who or what are acting as the referees remains a mystery.