Until the recent and detailed electron microscope studies by Carafa et al. (2003) and Duckett et al. (2006a) our knowledge of the endophytes in these two genera was limited to early light microscope investigations (Goebel 1891; Grun 1914; Lilienfeld 1911). These suggested that fungal associations in Haplomitrium Dedecek and Treubia might be different from those found elsewhere in land plants and so this proved to be.

Electron microscopy confirmed the presence of both extra and intracellullar hyphae in Treubia (Fig. 1). The former forms pseudoparenchyma-like structures within the mucilage-filled intercellular spaces in the thalli (Fig. 1D, E). In the older thalli these hyphae produce spores with highly distinctive multilamellate walls (Fig. 1F). The intracellular fungal zone lies in the lower part of the thalli and comprises coils of fine hyphae some of which terminate in short-lived fungal lumps (Fig. 1B, C). Exactly the same kind of intracellular cytology is found in the epidermal and sometimes the subepidermal cells in the subterranean mucilageinvested ‘roots’ of Haplomitrium. Spores with multilamellate walls like those in Treubia can also be seen in this mucilage. Unlike all other bryophyte-fungus associations where hyphal entry is via the rhizoids, in Treubia the rhizoids remain uninfected and entry is via mucilage-filled spaces between the epidermal cells. In Haplomitrium, which lacks rhizoids, hyphae penetrate directly through the epidermal cell walls.

The presence of an intercellular phase for the infections in Treubia draws parallels with the same in Lycopodium L. (Duckett & Ligrone 1992) and even more strikingly with the Devonian plant Nothia aphylla Lyon ex Høeg. (Krings et al. 2007a, b). Given this distinctive ultrastructure and the fact that Treubia and Haplomitrium are now considered to have had a very long evolutionary history in the land plant tree of life, it is extremely disappointing that the only published molecular study to date (Ligrone et al. 2007) revealed the presence of a Glomus in Haplomitrium chilensis R.M.Schust. closely allied to the Glomus Tul. & C.Tul. Group A endophytes found in Conocephalum Hill, Fossombronia Hazsl. and Pellia Raddi. However, a much more extensive sequencing programme of the Treubia and Haplomitrium endophytes now in progress (Bidartondo & Duckett 2010, unpublished data) is revealing that these two genera, plus several primitive taxa in the thalloid liverworts, lack Glomus but contain zygomycetous endophytes, a group of fungi recently resolved as ancestral to the glomeromycetes (James et al. 2006). Thus, we now have the first glimpses of what may well turn out to be a hitherto unsuspected spectrum of novel fungal symbioses very early in the ancestry of land plants.