Among the similarities between the liverworts and vascular plants (Tracheophyta; Fig. 1), the mycorrhizal symbiosis is perhaps the least expected. Vascular plants associate intimately with soil fungi: a limited colonization of roots by the fungus builds a dual organ, the mycorrhiza, that allows nutrient exchanges. The fungus exploits plant photosynthates and provides mineral resources for its host. However, nonvascular plants, such as liverworts, also form various associations with fungi, imitating the mycorrhizas: Russell & Bulman report on new advances in our understanding of this symbiosis on pp. 567–579 in this issue.
Liverworts (6000–8000 species) belong to the (presumably) paraphyletic Bryophyta (Fig. 1) and consist of a reduced sporophyte growing on a free-living gametophyte. Although some debate still exists (Goffinet, 2000; Nishiyama et al., 2004), most phylogenies place liverworts as the most basal extant land plants (Dombrovska & Qiu, 2004; Groth-Malonek et al., 2004). Liverworts are classically divided into two subclades (Fig. 1 and Table 1) whose monophyly is now questioned (He-Nygren et al., 2004). Jungermanniopsida mostly have small shoots with leafy expansions, but some show a simple thalloid organization; Marchantiopsida are thalloid, and some have a complex structure, including a lower storage parenchyma, a green aerenchyma with stomata-like pores and sometimes a hydrophobic cuticle. In addition, some Jungermanniales have subterranean axes bearing rhizoids, with positive gravitropism reminiscent of roots (Duckett et al., 1991). Among the similarities between liverworts and tracheophytes, the mycorrhizal symbiosis is certainly the least expected one.
|Jungermanniales||B or A|
|Metzgeriales||G, A or B (M in Aneuraceae)|
Mycorrhizal relationships vary among tracheophytes: most of them associate with biotrophic fungi, the Glomeromycetes (Schüßler et al., 2001), to form arbuscular mycorrhizas diagnosed by intracellular arbuscular hyphae (Smith & Read, 1997). This association is ancestral in tracheophytes (Selosse & Le Tacon, 1998), because Siluro-Devonian fossils already display arbuscular mycorrhizal (AM) associations, for example Rhynia or Aglaophyton (Fig. 1; Boullard & Lemoigne, 1971). As vascular plants diversified, so did their interaction with symbiotic fungi. Perhaps as an adaptation to various soil conditions, some taxa switched to asco- and basidiomycetes that form mycorrhizas with extracellular colonization (ectomycorrhizas, for example on temperate trees) or cell penetration by hyphal pelotons (in Ericaceae and orchids) (Smith & Read, 1997). At a functional level, carbon flux reversed in some lineages of achlorophyllous plants recovering carbon from their fungi (mycoheterotrophy). All these transitions are highly homoplasic, i.e. occurred several times (Selosse & Le Tacon, 1998).
Liverworts and their fungi
Reports of colonization of liverworts by fungi date back to the 1800s (Stahl, 1949; see Boullard, 1988 for a review). Investigations of ultrastructural features and fungal identities resumed in the 1980s. The colonization is restricted to the vegetative thallus, except for the meristematic apices, and results in a symbiotic structure that is homologous to neither a root nor a cormus. It is therefore best called a ‘mycothallus’ (Boullard, 1988). Colonization of gametophytes does not in itself distinguish them from tracheophytes, as gametophytes of some Lycophyta and Filicophyta (Fig. 1) also form mycothalli (Read et al., 2000). However, characteristics of true mycorrhizas appear at closer inspection.
Ascomycetes colonize swollen rhizoids of some leafy Jungermannopsida species (Table 1; Duckett et al., 1991; Turnau et al., 1998), sometimes forming ingrowth pegs into surrounding cells (Boullard, 1988; Kottke et al., 2003). Their dense intracellular growth resembles ascomycetous pelotons occurring in ericoid mycorrhizas. Indeed, strains of the ericoid partner Hymenoscyphus ericae were isolated from Cephaloziella exiliflora (Chambers et al., 1999), and H. ericae strains can colonize liverwort rhizoids in vitro (Read et al., 2000). Reciprocally, ascomycetous isolates from rhizoids form ericoid mycorrhizas (Duckett & Read, 1991).
Basidiomycetes colonize leafy and thalloid Jungermannopsida species (Table 1). They form intracellular hyphal pelotons that finally undergo a lysis in the host cells (Ligrone et al., 1993), in a way reminiscent of basidiomycetous hyphae in orchid mycorrhizas. Indeed, the basidiomycetes involved are mycorrhizal on tracheophytes: Aneuraceae species (Metzgeriales) appear to be colonized by ectomycorrhizal Tulasnella spp. (Bidartondo et al., 2003; Kottke et al., 2003); symbionts in Lophozia spp. and Calypogeia muelleriana (Jungermanniales) are related to Sebacinales, an orchid mycorrhizal taxon (Kottke et al., 2003). Notably, the achlorophyllous Aneuraceae Cryptothallus mirabilis is mycoheterotrophic and derives carbon from its Tulasnella partners (Bidartondo et al., 2003).
Finally, many liverworts form arbuscular associations with Glomeromycetes (Table 1; Boullard, 1988). This is reported from thalloid (Ligrone & Lopes, 1989) as well as leafy species (Haplomitrium; Carafa et al., 2003). In several instances, the fungus also has AM ‘abilities’, as shown by in situ hyphal links to arbuscular mycorrhizas (Turnau et al., 1999), or by synthesis of the association with an AM fungus (Read et al., 2000). In this issue, Russell & Bulman report on the first identification of arbuscular associates from Marchantia foliacea (Marchantiopsida) based on structural and molecular investigations. Parenchyma cells are colonized by an intracellular arbuscule-forming fungus that was identified by its ribosomal DNA sequences (18S and ITS). All sequences, recovered from 10 populations, clustered within the Glomus group A (sensuSchüßler et al., 2001). Such specificity is unexpected: although some host preference may exist in the field, as discussed by Russell & Bulman, arbuscular associations are often nonspecific (Smith & Read, 1997). During their investigations, Russell & Bulman found other fungal sequences in M. foliacea, which add to the various endophytic fungi already reported from liverworts (Davis et al., 2003; Jumpponen et al., 2003). This opens the Pandora's box of the ‘mycothallosphere’, i.e. microbes growing within and on liverworts.
Liverworts and early land plant symbioses
What is the evolutionary meaning of the resemblances between liverworts associations and mycorrhizas of tracheophytes, which often involve identical fungal partners? Resemblances between two lineages can arise in the three following ways.
- 1Symplesiomorphy: the shared trait (a true homology) is retained from a common ancestor.
- 2Convergence: the shared trait arose twice, in separate ancestors (the common ancestor did not have it).
- 3Reversion: the common ancestor had the shared trait, and transmitted it to one lineage, whereas the second lineage lost it and then acquired it again.
Convergence between liverworts and tracheophytes most likely accounts for similarities in associations involving asco- and basidiomycetes. Because of the distinct patterns of rhizoid colonization, Duckett et al. (1991) already proposed independent origins for ascomycetous associations in liverworts. In addition, even if basal basidiomycetes associate simultaneously with liverworts and tree roots (Nebel et al., 2004), mycorrhizal asco- and basidiomycete taxa may not be sufficiently old (Berbee & Taylor, 2001) to have colonized Ordovician embryophytes. Fungal host shifts may have shaped these similarities; indeed, opportunity of meeting (and creating symbiosis with) a new species is high, because (1) mycorrhizal fungi largely explore soil around their hosts and (2) a high concentration of fungal inoculum is required for horizontal transmission of the symbiosis. Even without such features, host shifts are common in the evolution of fungal symbioses with the termites and ants (Mueller & Gerardo, 2002). A similar scenario was proposed for the sharing of Xylaria endophytes by liverworts and angiosperms (Davis et al., 2003).
Because (i) arbuscular associations are ancestral in tracheophytes, (ii) some hornworts (Fig. 1) also have arbuscular mycorrhizas (Boullard, 1988; Schüßler, 2000) and (iii) Glomeromycetes date back to the Ordovician (Berbee & Taylor, 2001), it is commonly assumed that arbuscular associations in liverworts are inherited from the ancestor to all embryophytes. Many losses would have occurred in evolution, for example among liverworts (Table 1) or for the whole mosses (Fig. 1) that have no symbiotic fungi. Because of this, arbuscular liverworts are often viewed as living fossils of early land plant fungal symbioses (e.g. Boullard, 1988; Carafa et al., 2003). This is tempting, but time elapsed since the common ancestor is the same in all lineages and early-branching ones are not necessarily more ‘ancestral’. All extant species may mix ancestral and derived characters.
Indeed, some derived (i.e. secondarily evolved) features may occur in arbuscular liverworts. Firstly, specificity with Glomus group A observed by Russell & Bulman in M. foliaceae is derived, as group A originates late in Glomeromycetes phylogeny (Schüßler et al., 2001) – certainly later than the early land plants. (Specificity is also reported for liverworts associated with basidiomycetes (Nebel et al., 2004) and can be viewed as an evidence of host shifts performed by isolated fungal lineages.) Secondly, different patterns of arbuscular infections exist in liverworts: in Haplomitrium and Treubia, fungi enter intercellularly, through plant mucilage secretion (Carafa et al., 2003), whereas in Marchantiales, they enter intracellularly, through rhizoids (Ligrone & Lopes, 1989; Read et al., 2000; Russell & Bulman). Infection also differs in hornworts, where rhizoids are free of fungi (Schüßler, 2000). Therefore, extant features of arbuscular mycothalli do not allow unequivocal reconstitution of an ancestral association, because some (or all) features are derived. For similar reasons, it is even uncertain that all arbuscular associations in liverworts are symplesiomorphic with tracheophytes: at least some could be reversions or convergences.
So, do arbuscular mycothalli tell us anything about early land plants? Although the earliest undoubted fossil liverworts are Upper Devonian, late Silurian early land plant remains have structural affinities with liverworts, both in vegetative parts (Edwards et al., 1995) and spore walls (Wellman et al., 2003). However, note that the very earliest land plants are only known as fossil spores and predate by 60 Myr the Siluro-Devonian arbuscular associations (Boullard & Lemoigne, 1971). If one admits that liverworts are morphological analogues of early land plants, then fungal associations could have occurred, and should be searched for, in early land plants. Extant mycothalli suggest an ancestral predisposition to fungal symbioses in Embryophyta. Indeed, fungal symbioses were claimed to be common during land conquest: the enigmatic and widespread Devonian Spongiophyton and Prototaxites are likely symbioses between fungi and algae of unknown affinities (Selosse, 2002). Theoretically, fungal hyphae are well preadapted to three-dimensional exploitation of soils; similarly, algae, and possibly early land plants, are well preadapted to fluid exploitation (i.e. capture of light and atmospheric CO2). An early land plant–fungus symbiosis would thus be adapted to terrestrial life, at the soil–atmosphere interface (Selosse & Le Tacon, 1998). The finding of underground fossil parts of early land plants may verify such predictions – unfortunately, however, soils rarely undergo fossilization, due to erosion.
Evolutionary implications of similarities between liverworts and tracheophytes will certainly benefit from advances in Embryophyta phylogeny; recent papers challenge our view, even claiming monophyly for Bryophyta (Nishiyama et al., 2004). More curiously, the physiology of mycothalli remains a mystery. Except in the very particular mycoheterotrophic Cryptothallus mirabilis (Bidartondo et al., 2003), no work goes beyond fungal identification and/or morphological analysis. How does a mycothallus function? Is it a true functional imitation of a mycorrhiza?
The fungus probably recovers carbon, as suggested by glycogen accumulation in hyphae (Carafa et al., 2003) and biotrophic Glomus achieving their life cycle on hornworts (Schüßler, 2000). However, the fact that the fungus is mycorrhizal on other plants does not imply that it furnishes the same nutrients to the liverworts. The biotrophic interaction, as well as the reproducible and limited pattern of fungal colonization, as observed by Russell & Bulman, suggests that coevolution (in a broad sense) occurred, but gives no clues about the exchanges. This does not invalidate a neutral to weakly parasitic fungal behaviour (Read et al., 2000); equivalently, abuse by the plant, i.e. use of fungal carbon, is possible at least for Aneuraceae species related to the mycoheterotrophic C. mirabilis. As noted by Read et al. (2000), ‘there is an urgent need to test such hypotheses’. Claims that Marchantia nepalensis cannot reproduce sexually without fungi (Chauduri & Rajaram, 1926, in Boullard, 1988) or that increased colonization in polluted soils could be protective (Turnau et al., 1998) need careful investigation. All tools are now available to analyse exchanges: axenic thalli can be obtained from spores, and symbioses can be synthesized in vitro (Read et al., 2000; Schüßler, 2000). It is now possible to transpose experimental designs already used for investigation of mycorrhizal functioning. And, beyond exchanges, microcosms and in vitro cultures could ultimately assess the fitness of each partner, demonstrating which profits from the coexistence.
I acknowledge M.-C. Boisselier, V. Knoop, I. Kottke, Y.-L. Qiu, J. Russell, K. Turnau and especially B. Goffinet for their helpful comments and recent (or unpublished) papers.