Although fossil liverworts have been described from the Aptian of Spain and that of Antarctica, no fossil record exists for mycorrhizae. Marchantiophytes (liverworts) belong to the paraphyletic Bryophyta. They are divided into Marchantiopsida and Jungermanniopsida [72]. In the living clade, the colonization by the mycorrhizal fungus is restricted to the gametophyte. The occurrence of symbiosis between hepathics and fungi has been known for over a century; these early works have been reviewed by Boullard [7] and have been confirmed for the last 30 years [61]. Two types of associations are recognized according to whether the fungus possesses septate or no septate hyphae: •

glomeromycetous infections of the arbuscular mycorrhizal type occur in many liveworts: this is reported for thalloid [20], [65] and [79] as well as leafy species [12];

Frahm [24] reported the first record of a fossil hornwort from Dominican amber, but to date no fossil record is available for mycorrhizae. The mycotrophic status of living anthocerotophytes is also uncertain. Inter- and intracellular fungal endophytes were found in several anthocerotophytes species [64], but this clade has been considered as devoid of fungal infection until the observation of fungal infection in Phaeoceros laevis (L.) Prosk. [44]. The gametophyte of P. laevis is associated with an aseptate fungus that forms extracellular thick-walled hyphae, intercellular thin-walled hyphae and intracellular arbuscules; vesicles are also present.

There are very few descriptions of fossil bryophytes and no data concerning mycorrhization. For extant mosses, the largest group of bryophytes are considered as being non mycorrhizal, but often contain endophytic hyphae. Read et al. [61] stated that it is certainly of physiological interest that the mosses resist colonization by mycorrhizal fungi.

The fossil relationships between fungi and plants suggested by Kidston and Lang [36] in the Rhynie chert have now been documented [8], [63] and [85]. Arbuscular mycorrhizal symbioses may be deduced based on the presence of hyphae, vesicles, intracellular arbuscules and spores. The fungus Glomites rhyniensis is known from Aglaophyton sporophyte, and from the young gametophyte Lyonophyton rhyniensis. A continuous dark band in the cortex of Aglaophyton contains the arbuscules; the first recognisable arbuscules occur in young aerial axes in the gametophyte.

As for Aglaophyton, Boullard and Lemoigne [8] reconsidered the possible fossil relationship suggested by Kindston and Lang [36] about Rhynia. They proved that Rhynia elaborated mycothallus by pointing typical arbuscular mycorrhizal structures.

The presence of endomycorrhizal fungi has been reported from fossil lycophytes. Krings et al. [39] and [40] describe rhizomatous axes of Nothia aphylla, an early plant from the Rhynie chert with possible affinities in the zosterophyllophytes (the sister group to lycophytes), that host a fungus reminiscent of Glomites rhyniensis (Glomeromycota), the endomycorrhizal fungus of Aglaophyton major. The infection pathway in N. aphylla is distinctly different from that seen in A. major. The fungus enters the plant as an intracellular endophyte and then becomes intercellular in the cortex. The authors hypothesized that N. aphylla displays an alternative mode of colonisation by endomycorrhizal fungi, perhaps related to the peculiar anatomy of the lower portion of the rhizomatous axis (i.e. the rhizoidal ridge).

Palaeomyces asteroxylii has been observed in the inner cortex of rhizomes and basal parts of stems of Asteroxylon. It formed aseptate hyphae, which gave rise to fine branches and terminal vesicles. The similarity of this fungus and mycorrhizal fungi has been reported by several authors [8], [30], [36], [51], [52] and [59].

A few works on Carboniferous floras described endomycorrhizae in rhizomes or roots of different plants. Weiss [87] observed this type of association showing digestion phases on root or rhizome of a plant having affinity to the Lycopodiales, and several authors noted the occurrence of mycorrhizal fungi (Endogonaceae) in Carboniferous coal balls from Europe and America (see [86]).

Lycophytes were the first plants with roots [36]. They have a separate gametophyte phase without roots and a sporophyte with roots and microphylls. The living llycophytes comprise Lycopodiales, Selaginellales, and Isoetales. The gametophytes of Lycopodium can be either subterranean and achlorophyllous, or surface living and chlorophyllous, depending on the species, and both are invaded by fungi. Boullard [6] considered that these gametophytes generally formed mycothalli. An association showing some characteristics of AM without arbuscules and with very fine coiled hyphae has been observed in the subterraneum gametophyte of Lycopodium clavatum [66]. Arbuscule-like structures have also been described in chlorophyllous gametophytes of Lycopodium cernuum [17]. Lycopodium and Selaginella sporophytes develop AM associations, whereas Isoetes until now appear to lack this type of association.

Fungal reproductive structures have been found within aerial axes of Psilophyton dawsonii restricted to the inner cortical regions [82]. These fungal structures are compared to other fossil and extant spores and a possible mycorrhizal status is discussed.

Several occurrences of mycorrhizal associations have been reported from the axis of a putatively primitive member of the Cladoxylales (W. Stein, pers. commun.). Fungal hyphae and vesicles occur in the cortex, and extraradical spores arise from intraradical hyphae, or are disarticulated in the matrix.

Equisetum is the surviving genus of the well-represented arborescent sphenophytes from Carboniferous period. Fossil mycorrhizae have not yet been discovered in this clade. Mycorrhizae are unknown in the photosynthetic gametophytes of Equisetum and the mycorrhizal status of the sporophyte is still controversial. First, the genus was classified as non-mycotrophic, but Koske et al. [37] showed that the infection depends upon associated plants and/or soil moisture conditions. However, their conclusions are based on a small sample. According to Dhillion [16], typical vesicular-arbuscular mycorrhizal structures occurred in the roots of the sporophytes of Equisetum and it is reported that plants growing in wet (hydric) habitats had lower levels of colonization by AM mycorrhizal fungi than plants growing in mesic/dry-mesic habitats. The habitat is considered to play putatively an important role in determining the degree of mycotrophy in these plants.

Andrews and Lenz [2] observed symbiotic fungus in the cortical cells of a fern stem, possibly a rhizome of Scleropteris illinoiensis from the Middle Pennsylvanian of Illinois. They called this association mycorrhizome, based on the comparison between dense tangles of mycelium or ovoid bodies and living mycorrhizae. Colonization by fungal hyphae have been observed in the inner cortex of Botryopteris antiqua (Strullu-Derrien, personal observation); the occurrence of coils and vesicles is consistent with the notion of ‘Paris’-type mycorrhiza.

Mycorrhizal symbiosis is of general occurrence in living filicophytes, except Marsilaceae and Salviniaceae [6] and [30]. By studying the arbuscular mycorrhizae of pteridophytes in New Zealand, Cooper [14] reached practically the same conclusion. Berch and Kendrick [4] in southern Ontario and Zhao [89] in Yunnan (southwestern China) observed that the arbuscular mycorrhizal colonization was low in sporophytes of leptosporangiates ferns, whereas it was relatively high in the eurosporangiates. According to Brundrett [9], ferns like Ophioglossum and Botrychium have sporophytes with relatively thick roots, which are consistently mycorrhizal, while Filicales that have more fine roots with long roots hairs, show limited or inconsistent mycorrhization. On the other hand, Gemma et al. [27] observed that the highest proportion of arbuscular mycorrhizae among the 89 species studied in Hawaï concerned Dicksoniaceae, Dryopteridaceae, and Lindsaeaceae. The occurrence in the eurosporangiates was low.

Two different growth forms exist among gametophytes of ferns: green epiterrestrial forms and achlorophyllous subterranean forms [61] and [67]. Boullard [6] suggested that the dependence of fern gametophytes on mycotrophism decreased with increasing the photosynthetic ability.

Arbuscular mycorrhizal associations have been reported for the gametophytes of Psilotum and Tmesipteris [6] and [55]; cortical cells are occupied by dense coils of aseptate hyphae with some of which bearing terminal vesicles. Rhizomes and roots of the sporophytes are generally colonized. More recently, Duckett and Ligrone [18] observed glomeromycotean fungi in sporophytes and gametophytes of Psilotum and Tmesipteris that are entirely intracellular, and colonization closely resembles the ‘Paris’-type mycorrhiza; large-trunk hyphae, arbuscules and degeneration in the host cells are described.

Other patterns have been presented for mycorrhization in ferns. Schmid and Oberwinkler [68] observed associations with coils in the roots of Gleichneniaceae, which are formed by an unidentified ascomycete. Cooper [14] reported the occurrence of ectomycorrhizae in a fern, but as it is shown by the illustrated root, it is probably the root of a Nothofagus that is concerned [9].

To date no data are available for mycorrhization of Progymnosperms and Pteridosperms.

Fossil arbuscular mycorrhizae occurred in silicified roots from the Triassic of Antarctica; Stubblefield et al. [83] and [84] observed nonseptate fungal hyphae within and between cells of the central cortex and swellings comparable to chlamydospores and vesicles. Arbuscules are also present nearly filling the host cell. Example of a mixed colony comprised of fungi with affinities to the Glomineae and the Gigasporineae are given by Phipps and Taylor [58] also from the Triassic of Antarctica. The fossil gigasporinean mycorrhiza shows irregular intercellular and intracellular hyphae coiled within the cells and arbuscules with thick trunks and narrow branches. In the glominean form, hyphal diameter is more uniform, with coiling rarely present and arbuscules showing thin trunks. Vesicles may be lateral or terminal and spores are not found.

Like fossil members, living cycads show arbuscular mycorrhizae in their roots. Muthukumar and Udaiyan [50] reported that root colonization is a typical ‘Arum’ mycorrhizal type; however various levels of colonization exist among the species. It has been suggested that the endomycorrhizal fungi in cycads have a special function, i.e. to promote the fixing of nitrogen by the endosymbiotic cyanobacteria [23].

Fossil mutualistic associations have not yet been shown in Ginkgophytes. In living Ginkgo biloba Bonfante-Fasolo and Fontana [5] described arbuscular mycorrhizal symbioses; intercellular hyphae and vesicles are rare while intracellular coils and intercalary arbuscules are abundant in the inner cortex. According to Smith and Smith, Gingko should be included in a ‘near-Paris’ or ‘intermediate’ type.

Fossil mycorrhizae have been reported in cordaitean rootlets by several authors [29], [43], [54] and [90]. In Amyelon radicans, Osborn observed aseptate hyphae that have given rise to clumps in the inner cortical region; vesicles were also found. Halket confirmed these observations, but found more frequent septa in the hyphae. Later Harley [30] and Boullard and Lemoigne [8] corroborated Osborn's observations. Cridland [15] re-examined most specimens of cordaitean rootlets supposed to contain mycorrhizae and found that fungi are sometimes present, but not constant in occurrence, and seem to be parasites or saprophytes and not symbionts. Baxter [3] considered that the mycorrhizal nature of the fungi is open to question, because he never found any hyphae in the tissues of well-preserved specimens.

Cordaites are the sister group to Coniferophytes. Fossil mycorrhizae have been observed in a few members of the latter clade. Stockey et al. [76] described arbuscular mycorhizae from anatomically preserved roots of the Middle Eocene taxodiaceous conifer Metasequoia milleri. Mycorrhizal structures occur in the root cortex. Coiled hyphae are most common within cells of the inner cortical region, and these produce numerous, highly branched arbuscules. Another example of mycorrhizal association has been given by Cantrill and Douglas [10] on fossil roots with nodular and abbreviated lateral roots from the Lower Cretaceous of the Otway Basin, Victoria (Australia). Roots are placed together with foliage of Geinitzia tetragona sp. nov., which is possibly taxodiaceous.

The occurrence of fossil ectomycorrhizae in Coniferophytes has been first demonstrated by LePage et al. [41]. Ectomycorrhizae have been found among permineralized plant remains in the Middle Eocene Princeton chert of British Columbia. They are associated with roots of Pinus and show a Hartig net that extends into the endodermis, a pseudoparenchymatous mantle, and contiguous extramatrical hyphae. The mycorrhizal rootlets lack root hairs and dichotomize repeatedly to form large, coralloid clusters. Reproductive structures are absent. For the fungus, the authors suggested the comparison with the extant basidiomycete genera Rhizopogon and Suillus. Ectomycorrhizal associations have also been described by Fairon-Demaret et al. [21] on small roots that may belong to the taxodiaceous Glyptostroboxylon from the Eocene of Belgium (55 Myr). A mantle of hyphae surrounds each root frequently branched and a tiny Hartig net has been observed.

Except for all the Pinaceae and some genus in the Cupressaceae that form ectomycorrhizae, all living Coniferophytes in the northern hemisphere as well as in the southern hemisphere show arbuscular mycorrhizal associations [74] and [77].

No fossil mycorrhizae are known in Gnetophytes. Living Gnetophytes comprise the three genus Gnetum, Ephedra, and Welwitschia. Ectomycorrhizae have been observed in Gnetum by several authors [22], [53] and [75]. Collier et al. [13] described the tendency to be mycorrhizal for desert annual and perennial plants, and reported the occurrence of arbuscular mycorrhizae in Ephedra trifurca. Welwitschia mirabilis is also associated with arbuscular mycorrhizal fungi, but Jacobson et al. [31] noted that W. mirabilis is not mycorrhizal at sites of the Namib Desert where grasses are absent.

To date no fossil mycorrhizal record is available for angiosperms. However, the three basic types of mycorrhizae occur in extant angiosperms. Only a few orders (Ceratophyllales, Caryophyllales, Brassicales) are usually considered as non-mycorrhizal [74] and [77]. A review of mycorrhizae in living angiosperms with information on the mycorrhizal status of the various families is given by Brundrett [9].

The above analyses of the mycorrhizal status of embryophytes are summarized in Fig. 4. The mycorrhizal status is presented in relation with the different clades of embryophytes, including fossil and living plants, and it shows the distinction between paramycorrhizae and eumycorrhizae. While mycorrhization is documented very well for living plants, there are only a few studies for fossil plants and the examples reported are often controversial. Some examples awaiting further analyses are shown in Fig. 4 and some others are being studied (Nothia, Botryopteris, Amyelon, Glyptostroboxylon…).

Plants belonging to the lycophytes and monilophytes clades represent the group in which fossil and living members develop both paramycorrhizae and eumycorrhizae. It seems very important to study the fungal symbiosis on this turning point group. This may contribute to understand why lycophytes and monilophytes have now so few surviving arborescent genus and why the angiosperms are dominant in modern environments.

An important point to be taken into consideration is the wet habitats of the earliest plants and of plants growing in cold environments [81]. It has been generally accepted that AM associations exist in living plants growing in wet habitats. Additionally, dark pigmented septate endophytes are frequent colonisers of plant roots under these extreme environmental conditions, capable of forming mutualistic associations functionally similar to mycorrhiza [34], together with an important role of melanin synthesis for protection. AM associations in wet habitats as well as dark pigmented septate endophytes reported by only a few authors have to be taken into account for an overall study of the mycorrhization.