First cycad seedling foliage from the fossil record and inferences for the Cenozoic evolution of cycads.

The morphology of the early ontogenetic stages of cycad foliage may help resolve the relationships between extinct to extant cycad lineages. However, prior to this study, fossil evidence of cycad seedlings was not known. We describe a compression fossil of cycad eophylls with co-occurring fully developed leaves of adult specimens from the early Palaeocene ( ca 63.8 Ma) Castle Rock flora from the Denver Basin, CO, USA and assign it to the fossil genus Dioonopsis (Cycadales) based on leaf morphology and anatomy. The new fossil seedling foliage is particularly important because fully differentiated pinnate leaves of adult plants and the eophylls belong to the same species based on shared epidermal micromorphology, therefore, increasing the number of morphological characteristics that can be used to place Dioonopsis phylogenetically. Significantly, the seedling fossil has a basic foliage structure that is very similar to seedlings of extant cycads, which is consistent with a cycadalean affinity of Dioonopsis. Nevertheless, the set of morphological characters in the seedling and adult specimens of Dioonopsis suggests a distant relationship between Dioonopsis and extant Dioon. This indicates that extinct lineages of cycads were present and widespread during the early Cenozoic (Palaeogene) coupled with the subordinate role of extant genera in the Palaeogene fossil record of cycads.

Introduction

The evolution of cycads is a topic that fascinates an increasing number of researchers and avocational cycad enthusiasts. With 355 modern species in 10 genera [1], cycads have long been seen as relicts of a once flourishing and diverse group of plants [2]. Recent molecular phylogenetic studies, however, infer a late Palaeogene/Neogene origin and a late Miocene radiation for most extant genera [3,4] with only half of this diversity confirmed in the fossil record [5].

Fossils that show the early ontogenetic stages of plants may preserve important morphological characters to facilitate our understanding of the evolution of and relationships between extinct groups. These early stages, especially seedlings, are rarely fossilized. Among gymnosperms, seedlings have been reported only in few groups including scarce araucariaceous, taxodiaceous and pinaceous seedlings, and a few seedlings inferred as belonging to Welwitschia, ginkgophytes and glossopterids [6-14].

Herein, we present the first fossil record of a cycad seedling found in close association with a leaf flush of an adult cycad plant of the same species (figure 1). The leaves of two ontogenetic stages, the eophylls (the first foliage) and the fully differentiated pinnate leaves of adult plants, are documented by macromorphological and epidermal features and assigned to the extinct genus, Dioonopsis. Based on the gross morphological similarity of the fully differentiated leaves, Dioonopsis has traditionally been assigned to the lineage leading to the extant cycad genus, Dioon [15,16], which lacks a fossil record. The new fossil seedling sheds light on the evolutionary relationship between Dioonopsis and extant cycads (Dioon).

Figure 1.

Eophylls and fully developed leaves of Dioonopsis praespinulosa. Eophylls (a–d). (a) Two eophylls (DMNH15662). (b) Enlargement of leaflet, arrow shows anastomosis of veins. (c) Epidermal cell pattern on the adaxial side of a leaf; note isodiametric anticlinal walls. (d) Cyclocytic stomata on the abaxial side of a leaf. Fully developed foliage of adult plants (e–h). (e) Leaflet showing a broad base where it attaches to the rachis; arrow shows N-shaped vein anastomoses (DMNH8993). (f) Fully developed leaves (leaf flush); arrow indicates the leaflet that provided the epidermal details shown in (g,h) (DMNH15683). (g) Epidermal cell pattern on the adaxial side of the leaf; note isodiametric anticlinal walls. (h) Cyclocytic stoma on the abaxial side of the leaf. Scale bars: (a) 1 cm, (b) 5 mm, (c) 50 µm, (d) 50 µm, (e) 1 cm, (f) 5 cm, (g) 50 µm, (h) 50 µm. (Online version in colour.)

Material and methods

Seedling foliage, a slab with numerous fully differentiated leaves of an adult plant, and many detached leaves assigned to cycads were reported among the fossils from the early Palaeocene (Danian; radiometrically dated at 63.84 ± 0.06 Ma) Castle Rock flora in the Denver Basin, CO, USA ([17-19]; electronic supplementary material, figures S1 and S2). The detailed investigation of these fossils is presented here for the first time. The specimens are housed in the palaeobotanical collection of the Denver Museum of Nature and Science, Denver, CO, USA (seedling fossil and its counterpart—DMNH15662, locality DMNH1200; fossil foliage of adult plants—DMNH15683 (leaf flush), DMNH15674, DMNH8993, locality DMNH1200).

Seedling macromorphology of extant cycad genera was examined for comparison (electronic supplementary material, table S1). Epidermal characters of the fossil foliage were studied using epifluorescence and transmitted light microscopy. The geology of the locality, the associated Castle Rock megaflora, detailed descriptions, inventories, storage and additional images of studied fossil cycad material and the methods applied for the investigation of epidermal characters on the fossils are discussed in the electronic supplementary material.

A matrix including 15 leaf morphological and cuticular characters, 10 characters modified from Martinez et al. [20] and five new characters (electronic supplementary material, note S1 and table S2), was scored for all extant genera of cycads and Dioonopis based on new character observations from the Castle Rock fossils and data from previous matrices and observations [21]. The most parsimonious placement of Dioonopsis on a phylogenetic tree based on [4] was tested by moving the taxon by hand in Mesquite [22] on a backbone phylogeny based on the results of the analyses of Salas-Leiva et al. [4], testing all possible placements of the fossil. Moreover, we conducted a maximum-parsimony analysis using PAUP v. 4.10 [23]. Most parsimonious trees were obtained using exhaustive enumeration of the trees compatible with the backbone constraint topology from Salas-Leiva et al. [4], and 1000 bootstrap replicates were run using the same constraint and a heuristic search strategy. A consensus network of the bootstrap trees was generated using SplitsTree [24] applying a 15% cut-off. We also conducted a Bayesian analysis using MrBayes v. 3.2.6 [25]. The Markov-k model with correction for variable characters and a gamma-distributed rate variation was applied. The topology from Salas-Leiva et al. [4] was used as a backbone constraint. Two independent runs with four chains (one cold, three heated) were run for 1 000 000 generations. After discarding 25% of each run as burn-in, the runs were combined and a consensus network was generated using SplitsTree applying a 15% cut-off.

Results and discussion

We assign the fossil foliage of adult specimens, including approximately 25 whole leaves in a single flush and many detached whole and partial leaves, to Dioonopsis praespinulosa described previously from the Palaeogene of Alaska [16] based on both macro- and micromorphology of the leaves (figure 1; electronic supplementary material, figures S3 and S4). The following characters support this taxonomic assignment: pinnate leaves; sub-opposite leaflets inserted with broad, decurrent bases laterally on rachis sides; a single order of parallel veins, which frequently dichotomize and locally form N-shaped anastomoses (i.e. neighbouring parallel veins traversed by a steep vein); isodiametric epidermal cells, cyclocytic stomata randomly arranged on the lower side of the leaves with four to seven radially arranged subsidiary cells. The 25 associated fully differentiated pinnate leaves (figure 1f), which presumably represents one leaf flush, indicate a plant with approximately 1-m-long leaves that grew in a bunch. This dimension is comparable to smaller or mid-sized members of extant cycads.

The fossil leaves are readily distinguished from other previously described Cenozoic cycads with pinnate leaves and parallel venation. The leaves of the extinct genus Pseudodioon from the Miocene of Turkey [26] differ by lacking vein anastomoses and having stomata in bands. Similarly, the extinct cycad Pterostoma from early Cenozoic floras in Australia [27] differs by having leaflets inserted on the rachis with slightly contracted bases and by an epidermal cell pattern with strongly undulate anticlinal walls, which contrasts the broad base of leaflets and basically straight anticlinal walls of Dioonopsis. Finally, the extinct foliage type, Ctenis, typically reported from Mesozoic floras, but also found in early Cenozoic floras of North America [28], is distinguished by having leaves with, in many cases, irregularly segmented leaflets with a high frequency of vein anastomoses. Compared to extant members of Cycadales, Dioonopsis is distinguished by its N-shaped vein anastomoses.

The eophylls are paripinnate having two pairs of leaflets, which bear teeth on the apical margin and frequently dichotomizing parallel veins (figure 1; electronic supplementary material, figure S3). Veins sporadically anastomose as well, forming N-shape cross-connections similar to those of fully differentiated pinnate leaves of adult plants. Based on shared venation characteristics (anastomoses) and epidermal features (figure 1; electronic supplementary material, figure S5), the eophylls represent the juvenile stage of D. praespinulosa. Detailed descriptions of the fossil specimens and additional photo documentation are given in the electronic supplementary material.

Nomenclature

Dioonopsis praespinulosa (Hollick) Erdei, Manchester et Kvaček, emend. nov. Erdei

Genus: Dioonopsis Horiuchi et Kimura, Review of Palaeobotany and Palynology, 51:217, 1987. Species: Dioonopsis praespinulosa (Hollick) Erdei, Manchester et Kvaček, International Journal of Plant Sciences 173(1):83, figs 1–3a, 4 and 5, 2012, lectotype USNM38688b, figs 2d–e, 4a,d.

The emended diagnosis of Dioonopsis praespinulosa, given by Erdei, Manchester et Kvaček, International Journal of Plant Sciences 173(1):83, is completed by adding the description of the eophylls (DMNH15662, Denver Museum of Nature and Science, Denver, CO, USA; figure 1a–d, electronic supplementary material, figures S3d,e and S5a–e).

Emendation of the diagnosis: Eophyll paripinnate, with few pairs of leaflets, leaflets obovate, apex slightly rounded, base decurrent, acute teeth on the apical one-third of lamina; 8–10 parallel veins entering leaflets, veins frequently dichotomize in the apical half of lamina, some veins sporadically anastomose forming N-shape; lamina hypostomatic, stomata cyclocytic, scattered, guard cells sunken, surrounded by four to seven subsidiaries, coronal rim formed around stomata, anticlinal walls isodiametric.

We surveyed the macromorphology of extant cycad seedlings and compared them to the seedling of Dioonopsis praespinulosa (figure 2; electronic supplementary material, table S1). The basic morphology of the D. praespinulosa eophylls closely resembles the eophylls of several extant cycads (e.g. Encephalartos), which is consistent with the cycadalean affinity of Dioonopsis. The eophylls of D. praespinulosa differ most strongly from those of Cycas and Stangeria, which have prominent midribs. In the majority of the extant members of Zamiaceae, which includes all extant cycad genera except Cycas, the eophylls and the fully differentiated leaves of adult plants of the same species have similar gross morphology. For instance, the eophyll and the fully differentiated leaves of adult specimens of most Dioon species are nearly indistinguishable. When differences do occur, they usually involve the shape of the leaflet (e.g. the eophylls of some Zamia and Encephalartos species, i.e. E. ferox, Z. furfuracea, are more ovate) or the frequency of teeth or spines on the leaflet margins as in some Dioon and Encephalartos species (e.g. D. merolae, E. kisambo). Finally, as seen in some Ceratozamia, Encephalartos and Zamia species, eophylls have considerably fewer leaflet pairs than the fully differentiated foliage of adult specimens has.

Figure 2.

Comparison of eophylls of extant cycads and extinct Dioonopsis (a–h). (a) Dioon merolae. (b) Encephalartos ferox. (c) Encephalartos transvenosus. (d) Zamia furfuracea. Line drawings of cleared eophyll leaflets of extant cycads and extinct Dioonopsis showing venation details. (e) Zamia furfuracea. (f) Dioon spinulosum. (g) Encephalartos hildebrandtii. (h) Dioonosis praespinulosa. Scale bars: (a–h) 1 cm. (i) Number of steps needed for placing Dioonopsis on the molecular backbone topology. MP indicates the most parsimonious placements. (j) Consensus network showing support for the placement of Dioonopsis from the maximum-parsimony bootstrap and the Bayesian inference posterior probability. The two stronger splits are shown. (Online version in colour.)

In contrast to many species of Zamiaceae, the seedling of D. praespinulosa has eophylls that differ markedly in leaflet shape, margin and number of leaflets from the fully differentiated leaves of adult plants. The morphological disparity of the seedling and the adult foliage of Dioonopsis may be compared to that of some Encephalartos species (figure 2).

The venation of both eophylls and the fully differentiated leaves of Dioonopsis differs significantly from all extant cycads. In particular, the veins anastomose to form an N-configuration—a consistent and distinctive character of both eophylls and fully differentiated foliage of adult plants of Dioonopsis [16,28]. Although this venation pattern does not appear in extant cycads, it is shared by fossil foliage assigned to Cycadales (e.g. Ctenis [28] and Pterostoma [27]) implying that an extinct lineage or lineages of cycads with similar venation patterns lived during the Mesozoic (Ctenis) and persisted into the Palaeogene (Dioonopsis, Pterostoma).

Among extant cycads, Dioon has been inferred to be related to the extinct Dioonopsis based on their shared gross morphology of leaves (parallel-veined leaflets with broad bases inserted laterally to the rachis) [15]. However, these genera are easily distinguished by leaf venation and epidermal characters (electronic supplementary material, table S2; [29]) and by the morphology of their seedlings. In fact, the contrasting morphological characters of the seedlings and adult specimens of Dioonopsis and Dioon suggest a distant relationship between these genera.

When considered in a phylogenetic context, our analysis indicates the placement of Dioonopsis outside Zamiaceae with both MP bootstrap and Bayesian posterior probability supporting this placement (figure 2). Similar results are obtained looking at the most parsimonious placement of the fossil, with only a placement close to Stangeria, being one step less parsimonious, which is due mostly to a reversal of typical zamiaceaous characters in the lineage leading to Stangeria (loss of elongate cells, irregular orientation of the stomata). A close relationship between Dioonopsis and Dioon retrieved in some previous phylogenetic analyses [20,21,30] is due mostly to the coding of Dioonopsis as having two layers of encircling cells (an autapomorphy of Dioon [29]), longitudinally oriented stomata (a potential synapomorphy of Zamiaceae [21]) and anticlinal plugs. These characters are clearly absent from all species of Dioonopsis and were probably originally miscoded based on some ambiguity in the original description [15].

The fossil genus and species, Dioonopsis nipponica, were established to accommodate cycad leaves with well-preserved epidermal details from the Palaeocene Noda Group of northeast Honshu, Japan [15]. Following the description of this material, Eocene cycad fossils from Alaska and California that were previously argued to be related to Ceratozamia and Dioon were found to belong to Dioonopsis [16]. The disjunct biogeographic distribution of Dioonopsis was interpreted as the consequence of spreading via the Beringia phytogeographic pathway [16]. The occurrence of D. praespinulosa in the Castle Rock flora (ca 63.8 Ma) further expands the North American distribution of Dioonopsis during the Palaeogene. Nevertheless, it does not offer additional clues about the place of origin of the genus because Dioonopsis appears roughly simultaneously, during the Palaeocene, in the fossil record of Japan and North America. However, the megafloras accompanying Dioonopsis do provide information on the ecological tolerance of the genus. In particular, the Castle Rock flora has been argued to be a high diversity, subtropical rainforest [31], whereas the Alaskan floras, although the Yakutat tectonic block was probably located further south [32], suggest more temperate conditions [16]. Assuming that, during the Palaeocene, Dioonopsis achieved higher palaeolatitudes (the Beringial passage) estimated at 75°–80° N, its adaptation to extreme light variations [16] was probably established, similarly to some extinct Southern Hemisphere cycads [33]. Considering these ecological factors, Dioonopsis had an ecological tolerance greater than any extant cycad genus.

Although there are rare examples of fossil cycads for which organs of the same genus or species were reported as associated (e.g. Antarcticycas [33]), the early and late ontogenetic stages of Dioonopsis praespinulosa that are preserved in the early Palaeocene Castle Rock flora are unprecedented. Even though there is an extensive record of extinct cycads [15,16,26-28,34-39], fossil data are commonly limited to adult leaves making a comparison to extant genera challenging. The co-occurring eophylls and fully developed adult foliage of D. praespinulosa allow for a much more complete analysis of Dioonopsis and its phylogenetic placement in Cycadales. Our work confirms that Dioonopsis belongs to an extinct lineage of cycads outside Zamiaceae. This pattern corroborates that today unknown cycad lineages flourished during the early Cenozoic (Palaeogene) [5]. The apparent scarcity of modern forms in both the Palaeogene and Mesozoic fossil record of cycads may be interpreted by a younger evolutionary radiation of modern cycads [3]. Nevertheless, the Cenozoic history of cycads is far from being resolved and requires additional fossil data.