Molecular Evidence for the Early Colonization of Land by Fungi and Plants

See allHide authors and affiliations

Science  10 Aug 2001:
Vol. 293, Issue 5532, pp. 1129-1133
DOI: 10.1126/science.1061457


The colonization of land by eukaryotes probably was facilitated by a partnership (symbiosis) between a photosynthesizing organism (phototroph) and a fungus. However, the time when colonization occurred remains speculative. The first fossil land plants and fungi appeared 480 to 460 million years ago (Ma), whereas molecular clock estimates suggest an earlier colonization of land, about 600 Ma. Our protein sequence analyses indicate that green algae and major lineages of fungi were present 1000 Ma and that land plants appeared by 700 Ma, possibly affecting Earth's atmosphere, climate, and evolution of animals in the Precambrian.

Plants, animals, and fungi are well adapted to life on land, but the first colonists faced a harsh physical environment (1,2). The establishment of terrestrial eukaryotes may have been possible only through associations between a fungus and a phototroph (3, 4). The most widespread of these symbioses today are lichens and arbuscular mycorrhizae. The former consist of cyanobacteria or green algae and ascomycotan (or more seldom, zygo- or basidiomycotan) fungi, and the latter join a plant with a glomalean fungus (4). It is unclear when land was successfully colonized by eukaryotes or how the environment was affected. However, the importance of symbiosis in this process is suggested by evidence of arbuscular mycorrhizae in the earliest fossil fungi (460 Ma) and in some of the earliest land plants (4–6). Lichens are known from the Devonian (400 Ma) (7), although Precambrian lichens and other fungi have been proposed (4, 8). Unfortunately, most fungi and primitive plants do not preserve well in the fossil record, leaving open the possibility of an earlier, unrecorded history (1). Molecular clock estimates for the evolution of fungi, based on a well-studied ribosomal gene, have suggested a late Precambrian (600 Ma) colonization of land (9), but until now, abundant data from nuclear protein-coding genes have not been analyzed.

We assembled and aligned amino acid sequences of potentially orthologous groups from available fungi in the National Center for Biotechnology Information (GenBank) protein sequence databases (10). Of those, 119 proteins were found to be suitable for estimating fungal divergence times. For clustering purposes, we followed a widely used taxonomic arrangement (11,12). Molecular clock methods were used to date divergences between Basidiomycota and Ascomycota, Archiascomycetes and other Ascomycota, Hemiascomycetes and filamentous Ascomycetes, Ustilaginomycetes and Hymenomycetes, Zygomycota (Mucorales) plus Blastocladiales and Basidiomycota plus Ascomycota, and Neocallimasticales versus all other groups. We also estimated the divergence between the human pathogenic yeast Candida albicans and the bakers' yeast Saccharomyces cerevisiae, which is of importance for genetic and genomic studies of these species. The estimate between the filamentous ascomycotan groups Plectomycetes and Pyrenomycetes provides an internal time constraint within the fungal fossil record. Plectomycetes include the model fungus Aspergillus and many other human and animal pathogens such as Histoplasma and Coccidioides, whereas Pyrenomycetes include the model Neurospora and many plant pathogens such as Fusarium (head scab of wheat, Panama disease of banana), Cryphonectria (chestnut blight), andMagnaporthe (rice blast). Glomales, all other Zygomycota plus Blastocladiales (chytrids), and Neocallimasticales (chytrids) were recognized as monophyletic groups (11, 12).

Several groups of fungi, and comparisons among groups, lacked sufficient sequences for time estimation and thus were not considered: Glomales, Urediniomycetes, and the divergences between Zygomycota (Mucorales) and Blastocladiales and between Blastocladiales andNeocallimastix. Because of the widespread symbiotic relationships between fungi and plants (including green algae), we also obtained divergence time estimates for a green alga (Chlorophyta,Chlamydomonas) versus higher plants (Embryophyta) and a moss (Bryophyta, Physcomitrella) versus vascular plants (Tracheophyta, Arabidopsis, and other taxa) (13). These taxa were selected because of availability of nonchloroplast nuclear protein sequences, permitting animal-based calibration.

To reduce extrapolation error, we used multiple external calibrations from older divergences among animal phyla and kingdoms (plants, animals, fungi) derived from an analysis of 75 nuclear proteins calibrated with the vertebrate fossil record (14,15). We estimated times using two methods: the multigene approach whereby times are averaged across genes (16,17) and the average-distance approach involving concatenation of distances among genes (18, 19). We used a relative rate test to examine rate variation among taxa and a gamma correction to account for rate variation among sites (20). Rate variation was relatively low, with most comparisons averaging less than 10% difference between branches (Fig. 1). As would be expected, the branch length data for fungi, involving longer sequences (average protein length, 446 amino acids), showed less variation than branch length data for green algae and land plants, which involved shorter sequences (averaging 291 and 102 amino acids, respectively).

Figure 1

Branch length differences in rate-constant proteins (those passing the relative rate test). Each datum indicates the sequence divergence (branch lengths) of two branches (A and B) compared in relative rate tests, with branch A being the first named group of each pair. The calibration pairs are plant versus animal (PA), animal versus fungi (AF), plant versus fungi (PF), nematode versus arthropod/vertebrate (NAV), and arthropod versus vertebrate (AV). (A) Fungi: calibrations. (B) Fungi: comparisons; chytrids (Neocallimasticales) versus other groups (NF), Mucorales/Blastocladiales versus Basidiomycota/Ascomycota (MBF), Basidiomycota versus Ascomycota (BA), Ustilaginomycetes versus Hymenomycetes (UH), Archiascomycetes versus other Ascomycota (AE), Hemiascomycetes versus filamentous Ascomycota (HF), Candida albicans versus Saccharomyces cerevisiae (CS), Plectomycetes versus Pyrenomycetes (PP). (C) Algae: calibrations. (D) Algae: comparisons; chlorophytan green algae versus embryophytes (higher plants) (CE). (E) Land plants: calibrations. (F) Land plants: comparisons; bryophytes (mosses) versus tracheophytes (vascular plants) (BT). Trend lines for rate constancy (slope = 1) are shown for reference. Mean branch length ratios (sum of A branches/sum of B branches) are indicated for each pair (in parentheses).

Divergence time estimates for nearly all of the major divergences within fungi are deep within the Precambrian, 1458 to 966 Ma (Table 1 and Fig. 2). These times are significantly older than previous estimates based on the small-subunit nuclear ribosomal gene, which place the same splits at 660 to 370 Ma (9). Similarly, our dates for theCandida-Saccharomyces divergence (841 Ma) and the Pyrenomycetes-Plectomycetes divergence (670 Ma) are older than previous ribosomal gene estimates (140 and 310 Ma, respectively) (9). The closeness in divergence times of the major groups, with overlapping confidence intervals, resulted in one case of topological inconsistency: The Mucorales/Blastocladiales versus Basidiomycota/Ascomycota divergence is slightly more recent than the Basidiomycota versus Ascomycota divergence, although the difference is not significant.

Figure 2

Histograms of divergence times from rate-constant (black) and rate-variable (white) proteins. Rate-variable proteins are those rejected by the relative rate test. (A) Archiascomycetes versus other Ascomycota (88 proteins). (B) Hemiascomycetes versus filamentous Ascomycota (68 proteins). (C) Basidiomycota versus Ascomycota (52 proteins). (D) Chlorophytan green algae versus higher plants (48 proteins). (E) Mosses versus vascular plants (54 proteins). M, mode; m, mean (rate constant).

Table 1

Divergence time estimates.

View this table:

Given these time estimates and assumed fungal phylogeny (11,12), we can infer that Glomales originated after chytrids diverged from the other groups, but before Basidiomycota split from Ascomycota, about 1400 to 1200 Ma. Previous estimates are 590 Ma (9) and 462 to 353 Ma (21). This corresponds to the appearance of terrestrial fungi, and pushes back the origin of the first terrestrial eukaryotes, and earliest possible mycorrhizal associations, by about 800 million years (My). These data also indicate that lineage diversification within Ascomycota and Basidiomycota began shortly after divergence of those two major groups. The green algal divergence (1061 ± 109 Ma) establishes a minimum time when green algae existed, and the bryophyte divergence (703 ± 45 Ma), between two terrestrial lineages, is a minimum molecular clock estimate for the colonization of land by plants. The latter was unexpected considering the absence of fossil land-plant spores, containing decay-resistant sporopollenin, from the Precambrian.

The more recent estimates obtained in the ribosomal RNA gene study (9) largely are the result of using a more recent calibration (965 Ma) for the fungi-animal divergence, based on a molecular clock estimate tied to the vertebrate fossil record (22). However, that calibration was revised in a subsequent study to ∼1200 Ma (23), narrowing the difference. The calibration used here for the divergence of the three kingdoms (1576 Ma) derives from a more recent molecular clock study (14) also tied to the vertebrate fossil record, but based on refined vertebrate calibration dates, different methodology, and a different data set. A colonization of land by fungi deep in the Precambrian (>900 Ma) is inferred with either calibration (1200 or 1576 Ma) or data set (ribosomal RNA gene or nuclear proteins).

Because these time estimates are older than expected, it is of interest to know the effect of an internal (minimum) calibration. The oldest well-documented ascomycotan fossils are 400-My-old Rhynie chert perithecia, asci, and ascospores that resemble those of extant Pyrenomycetes (24). These fossils are about 60% as old as our estimate for the Plectomycetes-Pyrenomycetes divergence (670 Ma) and cannot be accounted for with the ribosomal estimate (310 Ma) (9). If a 60% rate adjustment is applied, the minimum times for the major fungal divergences are 875 to 580 Ma. A second fossil constraint involves red algae, which are dated to 1200 Ma (25). If red algae evolved on the plant lineage (26), the divergence of the three kingdoms (set here at 1576 Ma) (14) must be older than 1200 Ma. Applying this rate adjustment (76%) yields minimum time estimates of 1108 to 757 Ma for the major splits within fungi. Thus, Precambrian time estimates also are obtained with an internal calibration and red algal fossil constraint.

Most sequence data available for timing higher plant divergences come from organellar genomes. Unfortunately, because such genomes are absent, or differ greatly, between kingdoms, it is difficult or impossible to calibrate outside of plants. Thus, calibration usually is done with the plant fossil record, especially the liverwortMarchantia (450 Ma) (27). Our time estimate of 700 Ma for the divergence of moss and vascular plants, based on external calibration and nuclear proteins, indicates that some divergences among higher plants may be older than currently thought.

One of the first steps toward the colonization of land by eukaryotes may have been the formation of a lichen symbiosis, perhaps an endosymbiosis of a fungus and a unicellular cyanobacterium (28). Presumably, this would have led to other symbioses between fungi and phototrophs. Lichens and free-living cyanobacteria, often with bryophytic plants, form a biological crust in harsh terrestrial environments today (29) and may have done so in the Neoproterozoic (900 to 544 Ma) or even earlier, perhaps along with extremophilic animals such as tardigrades.

There is geochemical evidence for terrestrial ecosystems (prokaryotic) as early as 2600 Ma (30) and microfossil evidence 1200 to 800 Ma (2). However, despite speculation of Precambrian lichens and their impact (8, 31), there is no undisputed fossil evidence of terrestrial eukaryotes until the Ordovician (480 to 460 Ma), when land plants and fungi first appear (6, 32). If they arose earlier, as our data suggest, their potential effect on the environment and biota should be considered. In particular, two phenomena currently explained by other mechanisms may be the result of an early colonization of land by fungi and plants. One is a period of global glaciations (“Snowball Earth” events) 750 to 580 Ma (33) and the other is a Neoproterozoic rise in oxygen, possibly leading to the Cambrian explosion of animals (34) (Fig. 3).

Figure 3

Divergence time estimates for major groups of fungi, plants, and animals (Table 1). Thick horizontal bars at branch points are ±1 SE; narrow bars delimit 95% confidence intervals; thick bars on branches denote fossil record of fungi; solid circles are calibration points; open circle is internal (fungal) fossil constraint. H, Hemiascomycetes. The branching order of three groups (Ascomycota, Basidiomycota, Mucorales/Blastocladiales) is shown as unresolved for topological consistency. On the basis of branching order from other data (11, 12), glomalean fungi diverged after chytrids and before the basidiomycotan/ascomycotan divergence, ∼1400 to 1200 Ma.

Fungi can enhance weathering (31), which in turn can lead to lower CO2 levels and global temperatures (33, 35). In addition, the burial of terrestrial carbon, rich in decay-resistant compounds of land plants (36) and less dependent on abundance of phosphorus (37), would further affect global climate. Either or both of these mechanisms could explain lower global temperatures (episodic or general) and a rise in oxygen in the Neoproterozoic. Examination of sediments from this time period with appropriate methods (1, 6) may reveal fossil evidence of an early colonization of land by fungi and plants.

  • * To whom correspondence should be addressed: Department of Biology, 208 Mueller Laboratory, Pennsylvania State University, University Park, PA 16802, USA; e-mail: sbh1{at}


View Abstract

Navigate This Article