Portal de Periódicos da CAPES

The evolution of mycorrhiza‐like associations in liverworts: an update

2005; Wiley; Volume: 167; Issue: 2 Linguagem: Inglês

10.1111/j.1469-8137.2005.01471.x

1469-8137

Ingrid Kottke, Martin Nebel,

Lichen and fungal ecology

Although liverworts do not have roots, many of them are associated with mycorrhiza-forming fungi (see compilation in Nebel et al., 2004). The cellular structures of the associations and the fungi involved provide good arguments for the hypothesis that these are mycorrhiza-like associations (Read et al., 2000; Kottke et al., 2003), although physiological experiments revealing nutrient exchange are still missing. Because of convenience, the associations may even be termed 'mycorrhizas' (Brundrett, 2004), but to avoid unfruitful discussion about terms we prefer to speak of 'mycorrhiza-like associations'. A recent commentary on the symbiotic fungal associations of liverworts entitled 'Are liverworts imitating mycorrhizas?' (Selosse, 2005) asked whether liverworts established the symbosis by obtaining the fungi through host shifts from mycorrhizas of higher plants, which would automatically include a rather recent evolution. The overview presented by us (Nebel et al., 2004), however, indicated strong congruency between the evolution of liverworts and their specific symbiotic fungi, tempting us to hypothesize that 'the symbiotic fungal associations of liverworts are the possible ancestors of mycorrhizae' (Nebel et al., 2004). The term 'ancestors' was not meant in a strict phylogenetic sense, as there is no heritage and thus no phylogeny of the symbiosis; instead, the symbiosis has to be individually established each time. The term should mean that liverworts developed the symbiosis with fungi before the true mycorrhiza, the root–fungal symbiosis, evolved, and that the fungi originally switched from the gametophytes of liverworts to the roots of the sporophytes of tracheophytes and not vice versa. The extant situation might, however, include both transfer directions. The most recent molecular phylogenies of liverworts (Davis, 2004; He-Nygren et al., 2004) appear to substantiate our hypothesis. We therefore present an overview (Fig. 1) updated on the basis of the phylogenetic tree of liverworts given by Davis (2004), which was obtained from a 12-gene backbone data matrix and in which most nodes appeared to be highly supported (Davis, 2004). The revision of several main clades therein was also supported by results from other authors (Forrest & Crandall-Stotler, 2004; He-Nygren et al., 2004). The overview (Fig. 1) reveals much more clearly than before (Nebel et al., 2004) that: (1) all the basal groups of liverworts are associated with Glomeromycota; (2) the more derived liverworts clades lost the mycorrhiza-like association with Glomeromycota at one common event probably connected to change of the habitat from terrestrial to epiphytic; and (3) the symbiotic state was re-gained at least twice and independently by species growing terrestrial on rotten wood or humus, but this time with Basidiomycota and/or Ascomycota. Occurrence of mycorrhiza-like associations in the main clades of liverworts. The phylogenetic tree of liverworts was based on the maximum likelyhood tree obtained from a 12-gene backbone data matrix by Davis (2004). The figure has been redrawn, simplified and slightly modified. (Treubia is included in a tentative position.) For further explanations, see text. The new molecular concept of the liverworts phylogeny relieves us from several problems we noticed in the older version (Nebel et al., 2004). These problems were the position of Blasia as a taxon without a fungal symbiont, the occurrence of different symbiotic fungal associations (Glomeromycota and Basidiomycota) in the former Order Metzgeriales, and the question of dating the loss of the symbiotic state in the leafy liverworts as basal or derived. In the recent liverworts phylogeny, Blasia was transferred to the Complex thalloid clade (Davis, 2004; Forrest & Crandall-Stotler, 2004; He-Nygren et al., 2004). In the phylogenetic tree of Davis (2004), Blasia, Sphaerocarpos and Riccia, which have no fungal symbionts, appear each as sister groups to the Glomeromycota-associated Marchantia, Dumortiera and Monoclea, indicating a common symbiotic ancestor and the multiple, independent loss of the symbiosis. It would be difficult to accept a nonsymbiotic Blasia or Sphaerocarpos as the most ancestral member of the liverworts because this would imply a multiple establishment of the Glomeromycota associations at the basis of the liverworts. All the basal groups which are sister to the Complex thalloids now, the Haplomitriales and the Treubiales as well as the Simple thalloid I (Pelliaceae, Fossombroniaceae, Pallaviciniaceae), are associated with Glomeromycota. Thus, the hypothesized ancestral situation can, most likely, be indicated by 'gain of symbiosis with Glomeromycota' for a common ancestor of the monophyletic liverworts (Fig. 1). The new molecular liverworts phylogeny separates the former Simple thalloids (Orders Fossombroniales, Metzgeriales) into two independent clades, the Simple thalloid I and the Simple thalloid II (Davis, 2004; see also He-Nygren et al., 2004). The Simple thalloid I contain only Glomeromycota-associated species (Pellia, Fossombronia, Petalophyllum, Pallavicinia, Jensenia, Symphyogyna) and few with no fungal symbiosis (Hymenophyton, Podomitrium; M. Nebel, unpublished results). The Simple thalloid II comprise the nonsymbiotic Metzgeriaceae, the nonsymbiotic genus Pleurozia and additionally the Aneuraceae, which form a symbiosis with Tulasnella species (Basidiomycota) (Bidartondo et al., 2003; Kottke et al., 2003). The more detailed molecular phylogenetic tree of Davis (2004), in which Metzgeria and Apometzgeria appear basal to Aneura, suggests that the Simple thalloid II lost the mycorrhizal state first, and later established a new symbiosis with Basidiomycota. It is tempting to speculate that the gain of Tulasnella, a member of the Basidiomycota with some saprophytic capabilities, supported the development of Aneura and Riccardia on rotten wood, and consequently supported the evolution of the myco-heterotrophic state of Cryptothallus. In the latter case, Tulasnella forms additionally ectomycorrhizas with Betula, Cryptothallus thus appearing as an epiparasite on Betula, gaining carbon via the fungus (Bidartondo et al., 2003). The leafy liverworts comprise two sister clades in the new molecular phylogenetic tree, the Leafy I and the Leafy II, representing an early dichotomy according to Davis (2004). According to our previous data collection (Nebel et al., 2004), the Leafy I contain only nonsymbiotic species. Interestingly, molecular phylogeny places this group of epiphytic species basal to the Leafy II (Davis, 2004; He-Nygren et al., 2004), whereas classical systematics considered the species-rich epiphytics to be a 'crown group' (Schuster, 1984; Crandall-Stotler & Stotler, 2000). Thus, the molecular results would support a hypothesized loss of the symbiotic state as one main event by a common ancestor of the Simple thalloid II and the Leafy liverworts (Fig. 1). This event might have been linked to the change of the habitat from terrestrial to epiphytic. The Leafy II contain many species that are symbiotically associated with Ascomycota and/or Basidiomycota, and all these species grow terrestrial or semiterrestrial on rotten wood or humus. Where the fungi were identified, they came out as members of the Hymenoscyphus ericae aggr. (Ascomycota) or as Sebacina species (Basidiomycota) (see literature in Nebel et al., 2004). Species with mycorrhiza-like associations occur in all the three subclades (A, B, C) of the phylogenetic tree presented by Davis (2004). The re-gain of the symbiosis, thus, was most likely by a common ancestor. It is not possible so far to indicate definitely the node of the gain, as the symbiotic state of Ptilidium is unclear. According to our re-investigation, Ptilidium is not associated with mycorrhizal fungi, and neither is Schistochila; no data are available for Temnoma. These three genera form the basal branches of the Leafy II according to Davis (2004); however, their position is critically discussed (Davis, 2004). We tentatively position the re-gain of the mycorrhiza-like state at the main node of the Leafy II (Fig. 1). So far, only about 5% of the species in the Leafy II have been investigated for symbiotically associated fungi. It is therefore too early to obtain a definite picture of the distribution of Sebacina and Hymenoscyphus. However, it is remarkable that both are equally present in Leafy II C, the most basal clade, but the Ascomycota predominate in Leafy II B and A, the more derived clades (Cephaloziineae, Lepidoziineae). The Leafy II A include also several species for which a symbiosis with fungi was not found (Herbertus, Lepicolea, epiphytic Plagiochila species, Trichocolea). These observations indicate a secondary loss of the symbiotic state in the 'crown group'. When we consider the systematic position of the associated fungi, the coincidence in evolution is striking. The Glomeromycota unambiguously comprise the most basal group of symbiotic fungi of land plants (Schüßler et al., 2001), and they form the only group that is associated with the basal groups of liverworts. The Basidiomycota associated with the more derived liverworts belong to the Orders Sebacinales and Tulasnellales. The Sebacinales, according to recent molecular phylogeny, is the most basal group of the Hymenomycetes in Basidiomycota (Weiß & Oberwinkler, 2001; Weißet al., 2004). The Tulasnellales were placed within the Cantharelloid clade (Hibbett & Thorn, 2001), the most basal clade in the Homobasidiomycetes of Basidiomycota (Larsson et al., 2004). Sebacina and Tulasnella species are also grouped with other basal Basidiomycota as 'Heterobasidiomycetes', all of them containing a doliporus with imperforate parenthesomes (Weißet al., 2004). No other, more derived symbiotic Basidiomycota have been detected in liverworts so far, although these could be easily recognised at the ultrastructural level because of their perforate parenthesomes. The restriction of liverworts to basal groups of symbiotic fungi can, in our opinion, be best explained by an 'old heritage'. There is good molecular support for the liverworts as the most ancestral land plants (Qiu et al., 1998; Groth-Malonek et al., 2004). Fossil spores of Glomeromycota and of liverworts were recently dated to the Ordovician (Redecker et al., 2000; Wellmann et al., 2003). Although there are up to now no fossils of mycorrhiza-like associations before the Lower Devonian, the more ancestral liverworts probably already etablished the symbiosis with Glomeromycota in the Ordovician. The gain of the symbiotic state with Sebacina, Tulasnella and Hymenoscyphus ericae occurred later in geological times, indicated from the more derived liverworts clades and the derived positions in the phylogenetic tree of fungi (Schüßler et al., 2001). Because there are no fossils and no molecular clock dating these fungi, it is highly speculative, but we might suggest that this event should be dated to the time before the Pangaea continent was split in parts. This would help to explain the worldwide occurrence of Sebacina, Tulasnella and the Hymenoscyphus ericae aggr. (Leotiales) as symbionts of liverworts (Allen et al., 2003; Kottke et al., 2003; Nebel et al., 2004, and literature therein). A host switch from higher plant mycorrhizas to the liverworts cannot be excluded a priori. However, it could apply for the late gain of Sebacina and Tulasnella symbionts (Fig. 1), as suggested by Selosse (2005). Marchantia foliaceae, according to molecular sequences, shared a Glomus species with Podocarpus sp. in the surroundings (Russel & Bulman, 2004). Turnau et al. (1999) observed hyphal connections between liverworts and plant roots. A share of the symbiont between Cryptothallus and Pinus/Betula was demonstrated in field material and in pure cultures (Bidartondo et al., 2003). The symbiotic state has to be established individually each time during the life cycle of a plant. Thus, the independent phylogeny of plants and fungi will allow new combinations. We should probably not expect that the liverworts are only associated to the oldest lineages of Glomeromycota and the most basal Sebacina or Tulasnella species. Our argumentation for an originally old establishment of the mycorrhiza-like associations in liverworts does not exclude a possible recent and ongoing exchange of the symbiotic fungi. However, the majority of the more than 5000 ectomycorrhiza-forming fungi is composed of Homobasidiomycetes (Basidiomycota) or Pezzizales (Ascomycota). It is rather difficult to believe that just Sebacina and Tulasnella would have recently switched from higher plants to liverworts. Even in the neotropical mountain rain forest, liverworts are associated with Tulasnella and Sebacina (Nebel et al., 2004). Ectomycorrhizal trees are extremely rare there, and these few Nyctaginaceae do not form ectomycorrhizas with Tulasnella and Sebacina but with Thelephora, Russula and Lactarius (Haug et al., 2005).