Evolution of mycoheterotrophy in plants: a phylogenetic perspective

From an evolutionary point of view, mycoheterotrophy represents one extreme end in the mutualism–parasitism continuum of the mycorrhizal symbiosis. Strictly speaking, mycoheterotrophs are fully dependent on fungi for carbon metabolites and have usually lost chlorophyll as a result. Exploitation of fungi by plants has evolved independently many times among Embryophytes. Phylogenetic relationships of mycoheterotrophs provide an insight into the evolution of mycoheterotrophy by comparison of the degree of adaptation to a mycoheterotrophic mode of life between achlorophyllous and related green taxa. This February 2010 Virtual Special Issue of New Phytologist contains a comprehensive collection of papers on mycoheterotrophic plants, illustrating the significant progress that has been made in our understanding of their biology (http://www.newphytologist.com/view/0/virtspecissues.html; Selosse & Cameron, 2010). However, in many cases the phylogenetic context of the mycoheterotrophic mode of life studied remains poorly known and consequently our understanding of the evolution of mycoheterotrophy is still sparse. The vast majority of land plants are mycorrhizal and therefore it may not be surprising that a mycoheterotrophic mode of life has evolved, almost without exception, at least once in all of the major land plant lineages (Fig. 1). Mycoheterotrophs are unknown in the small relictual hornworts (∼100 spp.) and in the mosses (∼12 000 spp.). The latter can be explained by the fact that all mosses are nonmycorrhizal. Some species of the moss genus Buxbaumia are often described as achlorophyllous and thus mycoheterotrophic (Leake, 1994; Bidartondo, 2005). However, despite their reduced habit, all Buxbaumia species contain chlorophyll (A. Vanderpoorten & T. Madsen, pers. comm.) and, just like other mosses, Buxbaumia is nonmycorrhizal (Duckett et al., 2004). In liverworts at least one species (Aneura mirabilis) is fully mycoheterotrophic (Bidartondo et al., 2003). Another closely related species awaits investigation (Crum & Bruce, 1996). The gametophytes of most lycophytes and some fern families are nonphotosynthetic and mycorrhizal, pointing towards a mycoheterotrophic mode at this stage of their life (Winther & Friedman, 2009). For at least one fern species (Schizaea fluminensis) it has been suggested that the sporophyte is also mycoheterotrophic but this requires further investigation (Bidartondo, 2005). Some experts consider the parasitic gymnosperm Parasitaxus usta as a mycoheterotroph and argue that the plant obtains carbon from fungi but water from a parasitic interaction with a host plant (Feild & Brodribb, 2005). However, the identity and the exact role of the fungi involved in this interaction remain undetermined. Lineages of plants that have evolved mycoheterotrophy. Land plant phylogeny based on Palmer et al. (2004) and APG III (2009) for angiosperm relationships with lineages containing mycoheterotrophic species are indicated in black. All families with mycoheterotrophic species are shown next to the clade to which they belong, with the number of mycoheterotrophic species shown in parenthesis. Burm., Burmanniaceae; Thism., Thismiaceae. 1In these species the gametophytes are fully mycoheterotrophic, but the sporophytes contain chlorophyll. 2These species (belonging to Actinostachys and Schizaea) have only mycoheterotrophic gametophytes, but Schizaea fluminensis may have a fully mycoheterotrophic life cycle. 3Parasitaxus usta is both mycoheterotrophic and directly parasitic according to Feild & Brodribb (2005). 4As far as is known, all Orchidaceae are initial mycoheterotrophs. Perhaps best known are angiosperm mycoheterotrophs. Outside of monocots there are at least seven independent origins of mycoheterotrophy: one in Polygalaceae (Epirixanthes); two or three in Ericaceae (Monotropeae, Pterosporeae and Pyrola aphylla); and four in Gentianaceae (Voyria, Voyriella parviflora, Cotylanthera and Sebaea oligantha), which account for a total of 46 species. The remaining angiosperm mycoheterotrophs are all monocots. Within monocots at least 43 lineages evolved a fully mycoheterotrophic mode of life independently and there are no less than 411 fully mycoheterotrophic monocots. The Orchidaceae contains the largest number of fully mycoheterotrophic species, at least 210, which represent more than 30 independent occurrences. Significantly, as far as is known, all Orchidaceae (∼20 000 spp.) are completely dependent on fungal carbon during the early development of the seedling and are thus ‘initial’ mycoheterotrophs. In addition, a number of chlorophyll-containing orchids have been shown to be partly dependent on mycorrhizal fungi during their adult stage, a nutrition strategy that is known as partial mycoheterotrophy or mixotrophy. Partially mycoheterotrophic species have also been shown to exist in Ericaceae (Selosse & Roy, 2009). Recent developments in the field, particularly the discovery of partial mycoheterotrophic species closely related to ectomycorrhizal mycoheterotrophs, have shown the need for solid phylogenetic hypotheses to study the evolution of mycoheterotrophy in plants (Selosse & Roy, 2009). Unfortunately, these hypotheses are not available for the majority of mycoheterotrophic clades, and the identification of relatives of mycoheterotrophic plants has proven to be a taxonomic and phylogenetic challenge in many cases. Many mycoheterotrophic plants are rare or difficult to find, and in extreme cases particular species are only known from one or two collections (Maas et al., 1986). Obtaining study material is therefore often the first obstacle to be tackled when trying to unravel the evolutionary history of these intriguing plants. In addition, in parallel to parasitic plants, mycoheterotrophic plants have evolved convergent adaptations in their morphology and anatomy as a result of their peculiar mode of life. In general, mycoheterotrophs are characterized by small ‘dust’ seeds with undifferentiated embryos. Leaves are typically scale-like or absent entirely, and the vascularization of the stems is often reduced. Stomata are mostly absent from above-ground parts. The subterranean organs of mycoheterotrophic plants are typically highly modified. Structures with absorptive functions may be of relatively minor importance for mycoheterotrophic plants because this function is performed by the fungal symbiont. Root systems often show a trend towards a decrease in surface area: roots are short and thick and lack root hairs. Some species produce rhizomatous tubers and lack roots entirely (Leake, 1994). As a result of these convergences, unrelated mycoheterotrophs often share a similar habit and consequently many families with mycoheterotrophic members have been thought to be closely related. When morphological characters experience convergent evolution as a result of adaptations to a particular mode of life, molecular data offer a promising solution for inferring phylogenetic affinities. Indeed, the phylogenetic position of many mycoheterotrophic groups has been successfully inferred using DNA data, sometimes with surprising results, which in turn urged a re-evaluation of morphological characters (e.g. the placement of Triuridaceae in Pandanales; Rudall & Bateman, 2006). However, phylogenetic reconstructions in plants often rely heavily on data from the chloroplast genome. In parallel to what is observed in directly parasitic plants, the chloroplast genome of mycoheterotrophs may be significantly reduced in size. As a result of the relaxation of purifying selection, genes involved in the photosynthetic apparatus may be lacking or are highly divergent (Barrett & Freudenstein, 2008; Freudenstein & Senyo, 2008). However, few studies have examined mycoheterophs from a molecular evolution perspective. Whether because of the rarity of the plants or the absence of genetic sequence, mycoheterotrophic species are regularly absent in chloroplast DNA data sets. Moreover, nuclear and mitochondrial data of putative relatives of mycoheterotrophic plants are often not available, and thus a considerable sampling and sequencing effort is needed to infer phylogenies with nonchloroplast DNA data sets. Furthermore, nuclear and mitochondrial substitution rates of mycoheterotrophic plants are often greatly elevated. Whether these rate accelerations are the result of a small effective population size, selective molecular constraints, ecological niche changes, or other causes remains unclear. However, extreme rate heterogeneity may mislead phylogenetic inference methods by producing artificial clades, an error known as ‘long-branch attraction’ (Merckx et al., 2009). As a result of the problems outlined above, the phylogenetic relationships of many groups of mycoheterotrophic plants are still poorly known. The affinities of Corsiaceae, Thismiaceae and Triuridaceae remain elusive. Many mycoheterotrophic genera are yet to be included in phylogenetic analyses: for example, Cheilotheca (Ericaceae), Corsiopsis (Corsiaceae), Epirixanthes (Polygalaceae), Miersiella and Marthella (Burmanniaceae), Voyria (Gentianaceae), Kihansia, Peltophyllum, Seychellaria, Soridium and Triuridopsis (Triuridaceae), and several genera of Orchidaceae. The advent of molecular biology tools has made the isolation and identification of fungal associates of mycoheterotrophs much more tractable than with morphological approaches alone (Hynson & Bruns, 2010). It is now possible to reconstruct the phylogenetic patterns of both a group of mycoheterotrophs and their fungal associates to assess relative amounts of specialization, cophylogenetic evolution and host-shifting, making the mycoheterotroph system a useful model in studies of host–parasite relationships. Extreme host specialization towards narrow fungal lineages has been observed in many groups of mycoheterotrophs (Leake, 2004). In a lineage of arbuscular mycorrhizal mycoheterotrophs, this specialization process even resulted in a delayed cospeciation pattern (Merckx & Bidartondo, 2008). Among orchids, studies have revealed a general, although not universal, pattern of shift in fungal utilization from saprophytic/parasitic fungi to ectomycorrhizal species that is correlated with the transition from partial mycoheterotrophy to full mycoheterotrophy (Taylor et al., 2002). However, broad generalization may be premature, because recent studies examining tropical orchid groups have uncovered additional cases of the use of saprophytic fungi (Martos et al., 2009; Ogura-Tsujita et al., 2009). Such studies add to our understanding of the mycoheterotroph phylogeny, as well as expanding our knowledge on the evolutionary context in which diversification has occurred. Well-supported phylogenetic hypotheses are powerful tools for using to study the evolutionary physiology and ecology of mycoheterotrophy. Phylogenies also give insights into the biogeography and timing of the processes involved. Modern molecular clock techniques now allow us to evaluate the evolution of traits along a geological timescale but have been used only occasionally to study the origin of mycoheterotrophic lineages. In addition to the potential problems with DNA substitution rate heterogeneity, the paucity of fossil material for mycoheterotrophs contributes to the difficulty in reconstructing their evolutionary history. Despite these drawbacks, we now know that there are ancient lineages of mycoheterotrophic plants, and many lineages of mycoheterotrophs have radiated and achieved circumglobal geographic ranges (Bidartondo & Bruns, 2001; Merckx et al., 2008). As new material and data become available we will be able to identify the relatives of an increasing number of mycoheterotrophic plant lineages. This phylogenetic information will provide a solid framework to map ecological traits and study the distribution, diversification and divergence times of these enigmatic taxa. Combined with new physiological and ecological data on mycoheterotrophs and their relatives, the evolutionary context will allow for the identification of common patterns in the evolution of mycoheterotrophy in plant lineages (Leake & Cameron, 2010). In addition, efforts to sequence full plastid genomes of partially and fully mycoheterotrophic plants should be undertaken, both to resolve phylogenetic relationships and to obtain insights into the processes of plastid genome evolution of mycoheterotrophs. On a lower taxonomic level, studies on the population genetics of mycoheterotrophs are urgently needed to investigate morphological, phenological and molecular diversity in these groups. The authors thank Tom Madsen, Alan Smith, Nicole Hynson, Marc-André Selosse and Duncan Cameron for insightful discussions. V.M. is supported by the Fund for Scientific Research Flanders (FWO Vlaanderen).