Rhynie chert: a window into a lost world of complex plant–fungus interactions

Using material from a block of 400-million-year-old Rhynie chert about half as big as a brick, Krings et al. (this issue; pp. 648–657) studied fungi populating 250 rhizomes from Nothia aphylla, an early vascular plant. As Taylor et al. (2004) pointed out, most palaeobotanists choose to study beautifully preserved plant material. Krings et al. have made one of the first studies where researchers instead examined a large number of creeping, underground rhizomes in various stages of fungus-induced distortion or degradation. Modern plants distinguish friends from foes and respond differently to each. Krings et al. sought evidence of active plant responses to fungi in the silicified cellular anatomy of colonized N. aphylla rhizomes. Even in the Devonian, early in land plant evolution, N. aphylla cells responded differently to each of three different fungal colonists. Depending on the fungus, cells in the N. aphylla rhizomes increased the thickness of, and pigmentation in, their cell walls, increased their cell size, or encased fungal hyphae in plant cell wall material. All these morphological responses fall within the range of those shown by modern plants to invasion (Beckman, 1980). ‘Some difficult puzzles are posed by this Rhynie material, such as, how were the spores from fungi dispersed when they were packed into host cells?’ Before returning to Devonian fungus–plant interactions, we will consider which lineages of extant terrestrial organisms were already present in the Devonian, based on fossils and on phylogenetic evidence; our conclusions are summarized in Fig. 1. For the fungi and protist outgroups, we estimated ages using sequences from James et al.'s (2006) multigene, 214-taxon phylogeny and four calibration points. As calibration points for the fungi, we assumed that the common ancestor of the Glomeromycota, the Ascomycota and the Basidiomycota originated after the origin of terrestrial plants, ~500 Ma based on dispersed spores (Gray, 1985). We assumed that the Glomeromycota stem lineage was at least as old as 460 Myr based on Glomeromycota-like spores from the Ordovician (Redecker et al., 2000). We fixed the time of the divergence of monocots from dicot plants at 168 Ma (Magallón & Sanderson, 2005) and that of the first divergences among the multicellular animals at 695 Ma (Douzery et al., 2004). Relative ages of fungi and plants. A land plant age estimate is from Magallón & Sanderson (2005); the animal age is from Douzery et al. (2004), and the fungal and outgroup ages were estimated for this Commentary using the James et al. (2006) six-gene data set and the molecular clock inference program r8s. Triangles represent groups that included the species listed and appeared monophyletic in a Bayesian analysis. The James et al. (2006) alignment (available at http://www.aftol.org/alignments/comb_prot_ex.nex) consisted of 2108 amino acid positions from protein-coding genes (elongation factor 1-alpha and the largest and second largest subunits of DNA-dependent RNA polymerase II) and 2244 nucleotide positions (ribosomal DNA large, small, and 5.8S subunits). We reduced James et al.'s (2006) original 214 taxa to 25. To infer a phylogeny with branch lengths, we used MrBayes 3.1.2 (Huelsenbeck et al., 2006), allowing the program to estimate the proportion of invariant sites and the gamma shape parameter separately for the DNA and amino acid regions, and to select an appropriate model of amino acid substitution for the protein portion of the alignment. We used two independent runs of 500 000 generations each, sampling trees every 1000 generations. To estimate posterior probabilities for a consensus topology with branch lengths, we used a set of 500 trees, 250 per run, sampled after 250 000 generations, at which time the log likelihoods for each run had converged and reached a plateau. We estimated divergence times by the penalized likelihood method with a truncated Newton algorithm, using the computer package r8s, version 1.70 (available from M. Sanderson at http://ginger.ucdavis.edu/r8s/) with the consensus topology and branch lengths from the Bayesian input tree. As an outgroup, we chose Toxoplasma gondii, a taxon that does not appear within the plant, animal and fungal lineages. Following Bayesian analysis, we removed T. gondii from the resulting tree, leaving a tree with 24 taxa, and a basal trichotomy with branch lengths, as required by the r8s computer programs. We tested for possible alternative equally likely estimates for node ages with the ‘Checkgradient’ option of the penalized likelihood program and set the analysis for 10 replicates under different starting conditions, using ‘set num_time_guesses’. Penalized likelihood can compensate for substitution rate variation among lineages if ‘smoothing’ is optimized to take advantage of statistical correlation of rates within lineages. We optimized ‘smoothing’ at 5600 to correct for autocorrelation (r8s, version 1.70, user's manual, available from M. Sanderson at http://ginger.ucdavis.edu/r8s/). This degree of smoothing was high and it indicated a large amount of substitution rate variation across lineages. Such variation was a reminder that estimates for ages of fungal groups have a large and unknown amount of error resulting from inconsistent substitution rates as well as from possible errors in calibration or in estimation of phylogeny (Taylor & Berbee, in press). The resulting phylogeny (Fig. 1) was consistent with the tree from the complete 214 taxa from James et al. (2006) in showing the Glomeromycota, zygomycetes and Chytridiomycota as paraphyletic to one another, but the Bayesian support for this branching order was not overwhelming. Alternative trees with reasonably high likelihoods showed a monophyletic clade that included the mycorrhizal Glomeromycota as well as saprobic species such as Rhizopus oryzae and commensals such as Smittium culisetae, a grouping also found by Liu et al. (2006). Whether these fungi were paraphyletic or monophyletic, phylogenetics and fossil evidence supported their presence in the Devonian, early in the history of vascular plants (Taylor et al., 2004). With an estimate of fungal divergences in time (Fig. 1), we can see that the extant fungi most similar to the N. aphylla fungi are from basal lineages (Fig. 1). These include zygomycetes, which share aseptate hyphae with some of the fossils, and Chytridiomycota water molds, which, like others of the fossils, have spherical thalli. However, fungi are not the only plant pathogens with a long evolutionary history. Oomycetes and plasmodiophoromycetes are nonfungal protists that were likely present in the Devonian (Fig. 1) and could account for some of the N. aphylla‘fungal’ remains. The oomycetes originated from a photosynthetic ancestor also common to brown algae and diatoms; oomycetes convergently evolved fungus-like hyphae and an absorptive mode of nutrition. Oomycetes cause disease in many land plants (late blight of potatoes is one such example) and, like the N. aphylla fungi and living zygomycetes, their hyphae lack regular septa. Because of the lack of available sequences, plasmodiophoromycetes were not included in Fig. 1, but they are biotrophic pathogens related to foraminifera (Archibald & Keeling, 2004) that infect a variety of hosts, including oomycete fungi and vascular plants. In plant cells, plasmodiophoromycetes form zoosporangia as well as thick-walled resting spores which are released when the plant roots disintegrate. They, like Chytridiomycota, could be responsible for spherical thalli within plant material. A strength of the Krings et al.'s paper is that each of the three fungi was found many times. In the absence of identifying characters, Krings et al. referred to each kind of fossil fungus by number. ‘Nothia aphylla fungus no. 1’ consisted of narrow, aseptate hyphae and wider ‘spores’ and, although this was the least frequent of the fungi, it was found in 15% of the axes, or 37 of the 250 examined. The fungus caused damage in plant tissue, judging from the disintegrated host cells lying below the region of cells filled with fungal hyphae and spores. Evidence that the host was alive at the time of infection is provided by a layer of thicker, darkened cell walls that separated the zones with and without fungus no. 1 in the infected N. aphylla rhizome. The formation of thickened cell walls and darkening of tissues are common responses of living plants to fungal invaders. Other evidence that the host was alive at the time of fungal colonization is provided by a layer of granular material covering hyphae found in conducting cells and a thickened and twisted rhizoid that was colonized by fungus. Some difficult puzzles are posed by this Rhynie material, such as, how were the spores from fungi no. 1 and no. 2 dispersed when they were packed into host cells and, in the case of fungus no. 1, attached to hyphae? In modern ecosystems, spores of Ascomycota, Basidiomycota and zygomycete fungi, and also of oomycetes, form on the outside of their plant substrate, often responding to environmental cues such as light, surface features, and oxygen availability. In these fungi, dispersal is effected by wind, water or animal movement. In the Devonian samples, large numbers of spherical spore-like structures were packed within cells of decaying tissues. Modern fungi immersed in underground plant tissue release motile spores capable of moving through viscous decaying matter, or they release spores that swim away once the plant material has been lost to decay. Could spores in the N. aphylla rhizomes have germinated into crawling plasmodia, as in some stages in the plasmodiophoromycetes? Or, might they have directly released flagellated zoospores like modern oomycetes or Chytridiomycota, or released zoospores after the further disintegration of surrounding tissue, like modern plasmodiophoromycetes? Comparison with fungi associated with sporophytes of extant plants provides little help in identifying fungi no. 1 and no. 2. The closest taxonomic relationships of N. aphylla may be with the Zosterophyllophyta (Krings et al., 2007), in which case, its closest living relatives may be lycophytes, or club mosses. Unlike the Devonian fossil fungi, most fungi recorded from modern lycophytes have regularly septate hyphae and are members of Ascomycota and Basidiomycota (Farr et al., 1989). The oomycete reported from lycophyte sporophytes, Phytophthora cinnamomi (Farr et al., 1989), has aseptate hyphae but no other characters in common with the N. aphylla fossils. However, the gametophytes of several early diverging land plants have mycorrhizal, Glomeromycotan fungi that resemble fungus no. 3 in the N. aphylla rhizomes. Glomeromycota has the best continuous fossil record of any group of fungi, beginning with the 460 Ma Ordovician spores that serve in Fig. 1 as a calibration point marking the minimum age for the Glomeromycota stem lineage (Redecker et al., 2000). By the Devonian, fossil evidence for the Glomeromycota included structures that modern vesicular mycorrhizal fungi form in their hosts, such as arbuscules (the highly branched fungal hyphae within host plant cells that are perhaps the main site of plant–fungal nutritional exchange), vesicles, aseptate hyphae, and spores (Remy et al., 1994). As might be expected if it were a mutualistic mycorrhizal partner, fungus no. 3 was common, colonizing more than 95% of rhizome segments. As in modern fungal partners of lycophytes (Schmid & Oberwinkler, 1993) and other nonflowering plants such as the fern ally Psilotum nudum (Duckett & Ligrone, 2005) and the liverwort Marchantia (Russell & Bulman, 2005), the hyphae did not produce arbuscules. However, the hyphae of the fungus did produce vesicles or give rise to thick-walled spores. Somewhat surprisingly, hyphae from fungus no. 3 were encased with what Krings et al. interpreted as plant cell wall material. Encasement is a common response of a plant to invasion of its cells by a plant pathogen, for example, by the haustorium of a rust fungus or powdery mildew. An encasement layer may form around the hypha of a mycorrhizal fungus as it enters a cell, but hyphae of extant vesicular–arbuscular mycorrhizal fungi are not usually surrounded by a regular sheath of plant cell material (Schmid & Oberwinkler, 1993; Duckett & Ligrone, 2005). In extant nonflowering plants, mycorrhizal fungi usually produce coils of thin hyphae within host cells, but hyphal coils were missing from fungus no. 3 in N. aphylla. Possibly, the hyphal coils were missing because the Devonian fungi, although representing Glomeromycota, differed just as much from their closest living relatives as N. aphylla does from any living lycopod. Not only have plants changed since the Devonian, but fungi must also have changed, and not necessarily in faithful coevolutionary lock step with plants. If perfect coevolution were the rule, then the fungus that was the most recent common ancestor of one Glomeromycota species from a liverwort and a second Glomeromycota species from a flowering plant would date back to the Silurian origins of its host lineages. A sample of fungi from divergent hosts would then serve as a good sample of the fungi originally associated with ancient plants. Unfortunately, however, the Glomus species from the liverwort Marchantia are relatively recent and are nested among Glomus species from flowering plants (Russell & Bulman, 2005). Figure 1 suggests that the first division of the Glomeromycota into surviving lineages may have occurred well after early land plant radiations. Not only the Glomeromycota but also the Ascomycota and Basidiomycota coalesce to one Devonian species each. Although the diversity of Devonian fungi may have been less than what is presently found, it had to be greater than three species. The small sample size of surviving lineages from the larger pool of Devonian fungi limits what phylogenetic inference from living fungi can tell us about the ecological relationships of their ancestors. In light of Devonian diversity that surely has been lost, fungal fossils from Rhynie chert provide a remarkable window, in fact, the only window, into a lost world of complex fungus–plant interactions. Krings et al. have taken a great stride in beginning to interpret the view from that window.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.