Ectomycorrhizas from a Lower Eocene angiosperm forest


Author for correspondence:
Alexander R. Schmidt
Tel: +49 551 397957


  • The development of mycorrhizal associations is considered a key innovation that enabled vascular plants to extensively colonize terrestrial habitats. Here, we present the first known fossil ectomycorrhizas from an angiosperm forest.
  • Our fossils are preserved in a 52 million-yr-old piece of amber from the Tadkeshwar Lignite Mine of Gujarat State, western India. The amber was produced by representatives of Dipterocarpaceae in an early tropical broadleaf forest. The ectomycorrhizas were investigated using light microscopy and field emission scanning electron microscopy. Dissolving the amber surrounding one of the fossils allowed ultrastructural analyses and Raman spectroscopy.
  • Approx. 20 unramified, cruciform and monopodial-pinnate ectomycorrhizas are fossilized adjacent to rootlets, and different developmental stages of the fossil mycorrhizas are delicately preserved in the ancient resin. Compounds of melanins were detectable in the dark hyphae.
  • The mycobiont, Eomelanomyces cenococcoides gen. et spec. nov., is considered to be an ascomycete; the host is most likely a dipterocarp representative. An early ectomycorrhizal association may have conferred an evolutionary advantage on dipterocarps. Our find indicates that ectomycorrhizas occurred contemporaneously within both gymnosperms (Pinaceae) and angiosperms (Dipterocarpaceae) by the Lower Eocene.


Mycorrhizas are ubiquitous in terrestrial ecosystems. Up to 90% of all vascular plants live in a mutualistic association with fungi (Malloch et al., 1980). One selective advantage of mycorrhizal symbioses is an increase in the plant’s uptake of phosphorus and nitrogen. Additionally, some fungal partners protect plants against droughts and diseases such as microbial soilborne pathogens. At the same time, the fungus gets a relatively constant and direct access to carbohydrates. This symbiotic relationship is considered to be a key innovation of early land plants that enabled them to extensively colonize terrestrial habitats (Cairney, 2000; Wang et al., 2010).

Various classes of mycorrhizas have evolved over the course of time (Brundrett, 2002). Arbuscular endomycorrhizas are the oldest and most abundant ones (Cairney, 2000) and are recorded since the Early Devonian (Remy et al., 1994). Presently, > 70% of all angiosperms build such endomycorrhizal associations, while only 2% build ectomycorrhizal ones (Brundrett, 2009). Within the gymnosperms, ectomycorrhizas are only known from Pinaceae and from the genus Gnetum (Brundrett, 2009).

The evolution of different classes of mycorrhizas was influenced not only by changing environmental conditions, but also by the appearance of possible new fungal symbionts (Cairney, 2000; Hibbett & Matheny, 2009). Genes required for the formation of arbuscular mycorrhizas have been found in all embryophyte lineages (Wang et al., 2010). This suggests that early land plants had the potential to form arbuscular endomycorrhizas, and that extant plants that do not form this kind of mycorrhiza have either lost or suppressed the genes involved.

Several studies suggest that various ectomycorrhizas evolved independently, at least once in the Pinaceae, and additionally in several disparate lineages of angiosperms (Fitter & Moyersoen, 1996; Hibbett & Matheny, 2009). However, the geographic origins and subsequent spread of ectomycorrhizal associations are still unclear. Since ectomycorrhizas are most widespread today in boreal and temperate forests, Alexander (2006) addressed the question of whether ectomycorrhizal associations arose in these environments, and only later moved into tropical latitudes, or whether ectomycorrhizas arose independently in the tropics. Until now, the only fossil evidence for ectomycorrhizas had been reported from the roots of Eocene Pinaceae on Vancouver Island (LePage et al., 1997), and these fossils may suggest an origin for ectomycorrhizas in the northern latitudes.

Here, we present the first fossil evidence of ectomycorrhizas from an early tropical rainforest that has Gondwanan affinities. The fossil ectomycorrhizas are enclosed in Early Eocene (52 million yr old) Indian amber that was produced by a tropical angiosperm tree of the family Dipterocarpaceae (Rust et al., 2010). The Indian amber’s chemistry is quite distinct from that of most other fossil resins, and it is weakly cross-linked by comparison (Dutta et al., 2009, 2011; Mallick et al., 2009). For this reason, we were able to dissolve the amber surrounding one mycorrhizal system and apply ultrastructural analyses to the ectomycorrhiza’s surface. Raman spectroscopy revealed compounds of melanins in the dark hyphae.

Materials and Methods

Amber piece no. TAD 248 was found in situ in the Tadkeshwar Lignite Mine of Gujarat State, western India, which outcrops Early Eocene shallow marine sediments. The amber-bearing strata have been assigned to the Ypresian (52 million yr old) based on shark teeth, foraminiferans and dinoflagellates (Rust et al., 2010).

The original 4 × 3 × 2 cm piece of amber was divided into two smaller pieces (TAD 248 a and b) in order to better access the inclusions. The amber pieces were ground and polished manually using a series of wet silicon carbide papers (grit from FEPA P 600 to 4000 (25.8–5 μm particle size), firm Struers) and examined under incident (Carl Zeiss Stemi 2000) and transmitted light microscopes (Carl Zeiss AxioScope A1) equipped with Canon 450D digital cameras. Sometimes incident and transmitted light were used simultaneously. Some images were obtained from several optical sections using the software package HeliconFocus 5.0 (Kharkov, Ukraine) for a better illustration of the three-dimensional inclusions.

For scanning electron microscopy, a c. 20 mm3 block containing an ectomycorrhizal system was removed from amber piece TAD 258a using a dental drill. The tiny amber block was placed on a microscopic slide and dissolved using several drops of a mixture of toluene and 70% ethanol (10 : 1) as described by Rust et al. (2010). The remaining microfossils were washed several times with a few drops of 70% ethanol. The obtained fragments of the hyphal mantle were then placed on a carbon-covered scanning electron microscope mount using a wet hair from a superfine brush, sputtered by gold/palladium (2 × 120 s at 20 mA, 10 nm coat thickness) using an Automatic Sputter Coater (Canemco Inc., Quebec, Canada) and examined under a field emission scanning electron microscope (Carl Zeiss LEO 1530).

Raman spectra were recorded from extracted dark hyphae using a Horiba Jobin Yvon LabRam-HR 800 UV micro-Raman spectrometer. The spectrometer has a focal length of 800 mm. For excitation, the 488 nm line of an Argon Ion Laser (IMA 106020B0S, Melles Griot, Carlsbad, CA, USA) with a laser power of 20 mW was used. The laser was dispersed by a 600 l mm−1 grating on a CCD detector with 1024 × 256 pixels, yielding a spectral resolution of 0.43 cm−1. An Olympus BX41 microscope equipped with an Olympus LMPlanFl 100 × objective with a numerical aperture of 0.8 focused the laser light onto the sample. The confocal hole diameter was set to 100 μm. The acquisition time was varied between 10 and 300 s for a spectral range of 100–5000 cm−1. By using different filters, the power of the laser was reduced to 0.1, 10 and 50% of its original power at the laser exit. For calibration of the spectrometer, a silicon standard with a major peak at 520.4 cm−1 was used. All spectra were recorded and processed using LabSpec version 5.19.17 (Jobin-Yvon, Villeneuve d’Ascq, France).

For permanent preparation, the pieces of amber were fully embedded in a high-grade epoxy (Buehler Epoxicure, Lake Bluff, IL, USA) under vacuum (see Nascimbene & Silverstein, 2001 for protocols). After curing, the resultant epoxy plugs surrounding each sample were cut and polished to create clear flat surfaces close to the amber and its inclusions.

Both amber fragments are currently housed in the amber collection of the Division of Invertebrate Zoology of the American Museum of Natural History, New York. All Indian amber pieces will finally be deposited in the amber collection of the INSA Project Geology at the University of Lucknow, India.


Description of the fossil ectomycorrhizas

Approximately 20 unramified, cruciform and monopodial-pinnate ectomycorrhizas are fossilized adjacent to rootlets of up to 180 μm in diameter (Fig. 1a–e, Supporting Information, Fig. S1a–c). The nonmycorrhizal parts of the absorbing roots are 300 μm to 8 mm in length and 60–130 μm in diameter. Unbranched mycorrhizas are 320 μm to 1.9 mm long and 90–140 μm in diameter (Figs 1b, S1c). Cruciform ectomycorrhizal systems (Figs 1c, S1b) are 200–310 μm (rarely up to 700 μm) long, and their two branches are 120–220 μm (rarely 500 μm) long and 70–100 μm in diameter. Monopodial-pinnate ectomycorrhizal systems (Figs 1a,d,e, S1a) mostly range between 350 and 550 μm in length, and their finger-like branches are 100–300 μm (rarely up to 530 μm) long and 60–90 μm wide. The monopodial-pinnate system in amber fragment TAD 248b (Figs 1e, S1a) is 1.3 mm in length and bears five finger-like branches, in which both of the most basal branches are bifurcated.

Figure 1.

Light-microscopic photomicrographs of the ectomycorrhizal systems of Eomelanomyces cenococcoides gen. et spec. nov. from Eocene amber of India. (a) Overview of a rootlet with three monopodial-pinnate ectomycorrhizal systems and one cruciform system (located on the right) with surrounding mycelium (TAD 248a). The system in the middle right represents the holotype (shown in greater detail in Fig. 1(d)). (b) An unramified ectomycorrhiza which has been later removed from the amber for ultrastructural and Raman analyses (see Figs 2, 3 and Supporting Information, Fig. S2; TAD 248a). (c) Cruciform ectomycorrhizal system (TAD 248a). (d) Monopodial-pinnate ectomycorrhizal system forming microsclerotia. This system represents the holotype of E. cenococcoides gen. et spec. nov. (TAD 248a). (e) Large monopodial-pinnate ectomycorrhizal system with five finger-like branches forming microsclerotia. The arrowheads show additional furcation of the basal branches (TAD 248b). (f) Microsclerotia formed at the surface of the basal branch of the system shown in the lower left of Fig. 1(a) (TAD 248a). (g) Large microsclerotium exposed at the surface of the system shown in Fig. 1(e) (TAD 248b). (h) Dark hypha with two septa (arrowheads) extending from the system shown in Fig. 1(b) (TAD 248a). Bars, 500 μm (a); 100 μm (b–e); 20 μm (f–h).

Different developmental stages of the ectomycorrhizas are preserved in the piece of amber. Young ectomycorrhizas show dark pseudoparenchymatous mantles from which numerous irregularly septate dark pigmented hyphae of 1.2–3.3 μm in diameter extend (Figs 1b,c,h, 2a–c, S1b,c). Their walls are 0.2–0.3 μm thick, and iris diaphragms are possessed at the septa (Fig. 2g) whereas clamp connections are absent. Compounds of melanins were detectable in these dark hyphae using Raman spectroscopy. Some of the peaks were assignable to the key monomers of eumelanin: hydroquinone, indolequinone and semiquinone (Fig. 3). Some of these hyphae form chlamydospore-like inflated distal hyphal ends which are clavate or broad fusiform to lemon-shaped and 12–16 μm long and 6.5–9.7 μm wide (Figs 2c,d, S2d). Short forked flat hyphae 7–15 μm long, 2.4–5 μm wide and c. 0.5 μm thick (Figs 2e, S2a), as well as short young hyphae 2.5–3 μm in diameter (Fig. S2a), are sometimes visible at the surface of the pseudoparenchymatous mantle. Dense hyphal systems extend in all directions into the clear translucent amber (Figs 1a,c, S1b,c), suggesting that some ectomycorrhizas were still alive when initially embedded. Sometimes several hyphae form simple rhizomorphs that are mostly c. 10 μm in diameter or thinner, seldom reaching 75 μm (Figs 4a,b, S2b). Generally, hyphae exhibit thick cell walls (Figs 2f,g, S2e) and are frequently coated by a toluene-insoluble substance (Fig. S2b,c). Single hyphae within the rhizomorphs are 1–3 μm wide. The dark hyphae of the mycelium are often coated by light circular structures possessing a rough surface (Fig. S1d). Hyphae are absent around older ectomycorrhizal systems; instead, numerous spherical to ovoid microsclerotia (hardened mycelia serving as dormant stages) are formed at their surface (Figs 1d–g, S1g). The microsclerotia are mostly 35–40 μm long and 25–30 μm wide, sometimes reaching 55–60 μm in length and 50 μm in width. Small ones are only 15–20 μm in size. Microsclerotia are also formed in the nearby hyphal systems (Figs 4d, S1f). Clavate short hyphal ends 15–43 μm long and 5.5–6.5 μm wide are regularly formed in the mycelium (Figs 4c, S1d,e). Sometimes they appear at regular distances of c. 450–550 μm apart at the supporting hyphae. The otherwise thick walls of the hyphae become thinner and almost disappear in these branches (Figs 4c, S1e).

Figure 2.

Scanning electron micrographs of Eomelanomyces cenococcoides gen. et spec. nov. from Eocene amber of India (TAD 248a). (a,b) Fragments of the pseudoparenchymatous mantle of the ectomycorrhiza shown in Fig. 1(b). (c) Surface of the pseudoparenchymatous mantle showing extending hyphae with initial formation of chlamydospore-like inflated distal hyphal ends. (d) Lemon-shaped inflated distal hyphal end of a short hypha extending from the mantle. (e) Surface of the pseudoparenchymatous mantle with short forked flat hyphae. (f) Hypha showing thick walls at cross-break. (g) Broken hypha exposing a septum with an iris diaphragm. Bars, 20 μm (a,b); 5 μm (c–e); 1 μm (f,g).

Figure 3.

Raman spectrum of a single hypha of the ectomycorrhiza shown in Fig. 1(b). The coloured pattern underneath the spectrum represents the peaks analysed by Gauss–Lorentz function. The peak at 1426 cm−1 can be assigned to hydroquinone, those at 1340, 1532 and 1572 cm−1 to indolequinone, and the one at 1290 cm−1 to semiquinone. The main peak at 1606 cm−1 represents carbon bonding.

Figure 4.

Light-microscopic photomicrographs of the mycelium of Eomelanomyces cenococcoides gen. et spec. nov. surrounding rootlets and ectomycorrhizal systems from Eocene amber of India (TAD 248a). (a) Branched rhizomorphs. (b) Detail of Fig. 4(a). Numerous hyphae composing the rhizomorph are visible. (c) Two hyphae each forming a clavate short hyphal end. The walls of the hyphae thin and almost disappear in these branches. (d) Two microsclerotia formed at a branched hypha. Bars, 100 μm (a, b, d); 10 μm (c).

Preservation of the ectomycorrhizas is excellent, allowing description of the mycobiont as Eomelanomyces cenococcoides gen. et spec. nov. (see the next section, ‘Taxonomic summary’). E. cenococcoides is a fungus containing melanin and developing ectomycorrhizas as black pseudoparenchymatous mantles on the surface of absorbing roots of the host. Hyphae with iris diaphragms at the septa extend outward from this mantle. In this regard, the fossil is similar to the recent anamorphic genus Cenococcum, but distinguished by the high variability in the branching of the ectomycorrhizal systems and by the regular formation of microsclerotia.

Taxonomic summary

Eomelanomyces cenococcoides Beimforde, Dörfelt et A. R. Schmidt gen. et spec. nov. (Figs 1, 2 and 4, S1, S2).

Descriptio: Fungus anamorphus cum substantia ‘melanin’ et ectomycorrhizam formans in plantis. Systema mycorrhizas non ramosa vel cruciformis aut monopodialiter pinnata. Rami frequenter situ in dextero angulo. Tunica mycorrhizae in superficie est pseudoparenchymatica cum cellulis planis, 60–140 μm in diametro, frequenter cum hyphis ramosis, coloratis, non regularibus septis, 1.2–3.3 μm in diametro. Ex tunica pseudoparenchymaticae hyphae eminentes cum septis. Septa cum simplicibus centralibus cavis ut in genere recentem Cenococcum. Nonnullae hyphae apices formantes ad similitudinem chlamydosporibus, usque ad 8 × 5 μm in diametro. Hyphae conjunctae in chordam myceliae ut in simplicibus rhizomorphis. In aetate mycorrhizae sine vividis ramosis hyphis eminentibus autem cum multis microsclerotiis ovoideis, c. 35–50 × 25–35 μm in diametro.

Typus: In resina fossile ex India, collectio numerus AMNH TAD 248; Systema ectomycorrhizae est spectata in Fig. 1(d) est holotypus.

Etymologia: Eo, eos: Eocaen; melanos: nigrum. Epitheton speciei propter similitudinem cum recenti genere anamorpho Cenococcum.


Assignment of the mycobiont

Although the fungal mantle is excellently preserved, the root tissue of the host plant decayed in the amber (likely as a result of taphonomic conditions affecting the preservation of woody tissues). Consequently, we could not document the Hartig net in which the mycobiont penetrates the intercellular spaces of the host.

We assign the ectomycorrhizas of amber piece no. TAD 248 to a single fossil species because all (including different developmental stages) are arranged close to each other on the rootlets. Some photomicrographs of Fig. 1 and S1 may suggest variation in colour of the ectomycorrhizas and adjacent hyphae because different intensities of transmitted and incident light were used. Lighter colour is also caused by a refractive nanometre-sized space that sometimes appears between the amber inclusion and the surrounding resin. This gas-filled space probably originated by shrinkage of solidifying resin and desiccation of the inclusion during fossilization.

We consider E. cenococcoides to belong to the Dothidiomycetes (Ascomycota) because of its dark melanized hyphae, the formation of a mycorrhizal mantle, the regular formation of microsclerotia, and the similarity of this fossil to the extant anamorphic genus Cenococcum.

The septal porus is an iris diaphragm, which is not swollen like the dolipore of the Agaricomycotina (see Fig. 2g). Furthermore, clamp connections are absent. The dark colour of the pseudoparenchymatous mantle’s surface and of the hyphae is similar to the extant anamorphic genus Cenococcum, whose teleomorph is a species of the Dothideomycetes. The only extant species of this genus, Cenococcum geophilum Fr., forms mostly unbranched mycorrhizas with modern hosts of the Spermatophyta.

A particular feature of the mycelium surrounding the ectomycorrhizal systems is the regular occurrence of clavate short hyphal ends with very thin walls (Figs 4c, S1d,e). It remains unclear if these structures were for nutrient uptake.

Search for fossil melanin in E. cenococcoides

Raman studies of melanin from modern samples are reported by several research groups for different melanin-containing substances (Samokhvalov et al., 2004), as well as density-functional calculations of the three melanin monomers (Powell et al., 2004).

The Raman spectrum of the dark hyphae of E. cenococcoides revealed two very broad bands which are centred at 1354 and 1576 cm−1 (Fig. 3). This spectrum has a great similarity to the typical spectrum of amorphous carbon, with peaks at 1350 and 1550 cm−1 (Robertson, 1986). These two peaks are caused by vibrations of carbon atoms arranged in a graphitic-like structure. However, owing to the molecular structure of melanin, other vibrational modes involving oxygen, hydrogen and nitrogen should be visible. When analysing the spectrum with a Gauss Lorentz function, several underlying peaks could be identified between 1000 and 1600 cm−1. Based on the work of Powell et al. (2004), we were able to assign several of these peaks to hydroquinone, indolequinone and semiquinone, the key monomers of eumelanin. In the calculated spectrum of Powell et al. (2004) the peaks are very narrow, whereas our spectra, as well as those of Capozzi et al. (2005), show broad bands. This may be caused by the high fluorescence background, and by the degradation of the melanin molecules as a result of the Eocene age of our sample. Although eumelanins are not identical to fungal melanins (Butler & Day, 1998), the vibrational modes detectable with Raman spectroscopy likely do not show essential differences (see Cappitelli et al., 2005).

Previous studies exclusively report fossil melanins or melanosomes from animals, for example, melanin from Triassic ammonites (Mathur, 1996), and fossil melanosomes from Jurassic to Eocene feathers (Vinther et al., 2008; Li et al., 2010). Thus, our Raman spectrum provides the first indication of fossil fungal melanins in amber, and in fact the first record of melanin in a fossil fungus.

Probable assignment of the host tree

The ectomycorrhizas reported here were fossilized in a single piece of clear translucent amber. The only syninclusions of the fossil ectomycorrhizas are several lepidopteran scales. We assume that the resin originally filled a small hole inside the litter or soil horizon of the forest floor that contained vital ectomycorrhizas. The resin may have been released by the roots or dropped from above by one of the trees.

Today, arbuscular endomycorrhizas typically predominate in tropical forests (Janos, 1983; McGuire, 2007). However, several ecologically important tropical plant families, notably the amber-producing Dipterocarpaceae, are ectomycorrhizal (Lee, 1998). If present, ectomycorrhizal angiosperm trees generally make up a large portion of the total forest area, and can sometimes even dominate tropical rainforests (Connell & Lowman, 1989; McGuire, 2007), as dipterocarps do today in parts of Southeast Asia. The chemical analysis of the amber from the Tadkeshwar Mine revealed a class II or dammar-type resin (Dutta et al., 2009, 2011; Mallick et al., 2009; Rust et al., 2010), a cadinene-based polymer, which is produced primarily by trees in the Dipterocarpaceae (Van Aarssen et al., 1994; Anderson & Muntean, 2000). Independent evidence for the presence of Lower Eocene Dipterocarpaceae was recently obtained from fossil pollen grains that were found in the same sediments (Dutta et al., 2011). In addition, fossil wood samples attached to amber pieces showed microanatomical affinity to Dipterocarpaceae, including amber-filled resin canals, further substantiating the botanical source of this amber (Nascimbene et al., 2010; Rust et al., 2010).

The Gujarat region of India during the Lower Eocene was characterized by a strong occurrence of evergreen angiosperms (Sahni & Kumar, 1974; Willis & McElwain, 2001). Gymnosperms occurred rarely, and only a few fossils from the families Araucariaceae, Podocarpaceae and Ginkgoaceae have ever been recovered (Salujha et al., 1967) – their extant representatives are all characterized by arbuscular mycorrhizas, like most other gymnosperms. Based on direct fossil evidence, the Cambay amber and associated sediments provide one of the earliest unequivocal Asian records of a diverse, broadleaf tropical angiosperm forest (Rust et al., 2010).

We consider the host of E. cenococcoides to be an angiosperm, because representatives of the family Pinaceae and the genus Gnetum, the only known extant gymnosperms forming ectomycorrhizas (Brundrett, 2002), have not been observed in the Eocene of India. Furthermore, besides dipterocarps, no other angiosperms with known ectomycorrhizal associations, past or present, have been identified from this deposit. We therefore propose that the amber-producing Dipterocarp is the probable host of E. cenococcoides.

Palaeobiogeographical implications

The palaeogeographic and temporal origin of Dipterocarpaceae and their association with ectomycorrhizal fungi have frequently been discussed. It is typically suggested that dipterocarps originated in eastern Africa or Madagascar and drifted northward on the Indian platform, reached Asia during the Eocene and spread (Dutta et al., 2011). Alternatively, an origin in Southeast Asia has been proposed and taken into account (Lakhanpal, 1970; Sasaki, 2006). However, the monophyly of the three subfamilies of Dipterocarpaceae and of the Sarcolaenaceae, along with their consistent association with ectomycorrhizal fungi, suggest that the potential to form ectomycorrhizas is an ancestral character of the Dipterocarpaceae family (Ducousso et al., 2004; Moyersoen, 2006). Ectomycorrhizal symbioses may have conferred a selective advantage for some tropical tree species (McGuire, 2007), even in early tropical broadleaf rainforests, and the high diversity and abundance of Dipterocarpaceae in Asia might be based on their potential to associate with ectomycorrhizal fungi.

Ectomycorrhizal associations are considered to be unstable evolutionarily dynamic associations that evolved independently in several major clades of fungi (Hibbett & Matheny, 2009) as well as several times within the angiosperm clade that includes Rosids and Asterids and within the Pinaceae (Fitter & Moyersoen, 1996; Hibbett & Matheny, 2009). Consequently, E. cenococcoides itself is not necessarily an ancestral mycobiont of its host. The only previously reported fossil record of ectomycorrhizas is actually from the roots of Eocene (c. 50 million yr old) Pinaceae from Vancouver Island (LePage et al., 1997). Our find provides evidence that angiospermous ectomycorrhizal associations in the Paleogene tropics occurred contemporaneously with gymnospermous ectomycorrhizal associations in the Nearctic.


The authors would like to thank David A. Grimaldi (New York), Matthias Gube (Jena), Heike Heklau (Halle), Jouko Rikkinen (Helsinki), Kerstin Schmidt (Jena) and Gerhard Wagenitz (Göttingen) for their helpful comments; Jes Rust (Bonn) for advice; and Wolfgang Dröse (Göttingen) and Dorothea Hause-Reitner (Göttingen) for assistance with histology and the field emission microscope. We are grateful to the anonymous reviewers for their constructive suggestions. H.S. and R.S.R. thank Ashok Sahni (Lucknow) for his unfailing encouragement and kind advice. H.S. would like to recognize Naresh Chandra Mehrotra, the director of the Birbal Sahni Institute of Palaeobotany (Lucknow), for his support of laboratory and field work. R.S.R. thanks the Department of Science and Technology, Government of India. This is publication number 65 from the Courant Research Centre Geobiology, funded by the German Initiative of Excellence.