Distinguishing angiophytes from the earliest angiosperms: A Lower Cretaceous (Valanginian-Hauterivian) fruit-like reproductive structure^†

Systematics

Order: Corystospermales?

Genus: Doylea gen. nov.

Generic diagnosis: Cupule/carpel-like megasporangiate structures, ellipsoidal to globose, glabrous; bearing one, tightly enclosed, orthotropous, tetrahedral seed on abaxial surface; opening by three valves. Ovule attached by cup-shaped zone of transfusion tissue; either completely enclosed or exposed along edges by apparent valvate dehiscence of enclosing structure. Integument unvascularized with three longitudinal ribs, nucellus free from integument to chalaza, apparent pollen chamber with nucellar beak.

Species: Doylea tetrahedrasperma sp. nov.

Specific diagnosis: Cupule/carpel-like organs 3–4 mm wide, 5 mm long; outer zone parenchymatous; inner zone sclerenchymatous, cells with helical thickenings. Ovules 2–3 mm wide, 3 mm tall; triangular in cross section; micropyle trifid. Integument parenchymatous, three-layered: outer of small cells, middle of isodiametric cells that are thicker in corners, inner with elongate, flattened, rectangular cells; suture line in corners (angles).

Holotype hic designatus: P13394 A (UAPC-ALTA) (Figs. 1, 11, 12, 16, 25, 27).

Paratypes: P 13021 G bot, H; P 13133 F bot, P13243 A, P13249 B bot, P 13373 I bot, P13396 C bot, P13497 D top, C bot; P13955 A, B; P14221 E top; P14726 C bot, P14741 B top, P14784 B top. (UAPC-ALTA).

Locality: Apple Bay, northern Vancouver Island, British Columbia, Canada (50°36′21″N, 127°39′25″W; UTM 9U WG 951068).

Stratigraphic position and age: Longarm Formation equivalent, Valanginian-Hauterivian boundary, Early Cretaceous.

Etymology: The genus is named for James A. Doyle, University of California at Davis, in honor of the contributions he has made to our understanding of angiosperm evolution through his unrelenting search for the origin of the carpel and outer integument. The specific epithet refers to the distinctive tetrahedral seed found in this taxon.

Description

Fourteen specimens of this unique seed-bearing plant have been found (Figs. 1–10), including both empty cupule/carpel-like seed-enclosing structures and dispersed seeds (Figs. 8–10). One specimen has been found in the graywacke matrix (Fig. 4) in which the concretions are embedded, but the other 13 specimens occur inside the concretions where the tissues are better preserved.

Isolated seed-enclosing structures are 3–4 mm in diameter, up to 5.5 mm long, and ellipsoidal to globose (Figs. 1–5). Reconstructions of three of the most complete specimens were made using AMIRA (see Appendix S1–S5 with the online version of this article and Figs. 25–36). Internally, a triangular locule encloses a single tetrahedral seed, which is triangular in cross section (Figs. 1–4). One of the specimens has a fragment of external tissue attached to the enclosing structure at one end, but this is discontinuous and not well preserved (see online Appendix S3 and Fig. 30). Attachments of the reproductive structures to any additional axes are not known.

The seed-enclosing structures are composed mostly of elongate sclerenchyma cells with spiral thickenings that grade externally to a few rows of parenchyma (Figs. 1, 3). Most appear to be somewhat abraded and often the parenchymatous tissues have been removed (Figs. 2, 4, 5, 8, 9). Both the inner and outer surfaces appear to have lacked trichomes. The wall is very thin at the corners (Figs. 2, 4, 7, 9) and appears to have broken open to release the seeds by means of valves. Two specimens appear to have split open into three valves, one valve that contains vascular tissue and two that do not (Figs. 6, 7). The numerous spiral or helical thickenings on the sclerenchymatous cells of the walls probably facilitated the opening of the structures to release seeds (Figs. 14, 15) in a fashion that is similar to the dehiscence of some modern fruits (16).

Seed-enclosing structures have one collateral vascular bundle (Figs. 1, 2, 6–9, 17, 18). In specimens where the vascular bundle is well preserved, it can be determined that ovules were borne on the abaxial surface of a bilateral, modified leaf-like structure. This position is demonstrated by the position of the phloem, which is represented by a tissue gap to the inside of the xylem (Fig. 17). Seeds are attached to this inner (i.e., abaxial) surface, and vascular bundles have been traced by serial sections to the broad ovule bases. Seed bases are vascularized by a cup-shaped zone of transfusion tissue (Figs. 11, 12) composed of short, pitted, tracheary elements (Fig. 19). Tracheary elements in the vascular strands between the collateral bundle of the enclosing structure and the vascular cup at the ovule base have helical secondary wall thickenings (Figs. 13, 16).

One tetrahedral, orthotropous ovule is very tightly enclosed inside each cupule/carpel-like structure. Ovules are 2–3 mm in diameter and 3 mm long with an integument of at least three zones of small parenchymatous cells, 4–8 cells thick on the flat sides, that increase to 24–28 cells in the angles or corners (Figs. 20–23). Cell size increases in the corners (angles) of the seed (Figs. 7 at left, 20–22). The innermost cells of the integument are more elongate and flattened (Fig. 21). In the reconstruction of specimen P13243A (Fig. 5), the micropylar end of the ovule shows three arms of integument, two of which are incurved (Fig. 33). Both enclosed and isolated seeds show what appears to be a suture line in the angles (Figs. 6, 7, 10, 20, 21). Thus, like the cupule/carpel-like organ, the seeds probably also split along three lines—the enclosing structures at the time of dispersal, the seeds at the time of germination.

The nucellus is attached only at the base of the seed (Figs. 1, 2, 5, 10–12, 20, 24, 25). The cuticle covering the nucellus is quite prominent and thick, and outlines of former cells are discernable where the cuticle extended between the cells of the outer layer (Figs. 25, 26). So far, no pollen chamber has been identified with certainty, leaving the nature of the nucellar apex in question. However, one seed cut in oblique section shows what may be prominent rectangular nucellar cells at one at one side (Fig. 2, at arrow) that are similar to cells of the nucellar beak in other fossils that have a pollen chamber (e.g., 53, fig. 18; 54, figs. 17, 21, 23). Because more distal levels of the seed were lost in the saw cut, the identification of a pollen chamber in Doylea remains uncertain.

The megaspore membrane is thick and shrunken in the ovules (Figs. 1, 25, 27). In some seeds, there is light-colored material inside the megaspore membrane that may represent megagametophyte, but tissues are not well preserved (Figs. 1, 25, 27). In one of the smaller ovules (Fig. 24), the shrunken nucellus and megagametophyte form a trilobed structure with a hollow center (Fig. 24) that appears similar to gymnosperm ovules with megagametophytes in the free-nuclear phase of development (see e.g., 63, figs. 22, 23, 38). So far, no embryo tissues have been identified.

Closure of the reproductive structures

To determine the morphology and extent of the ovule enclosing structure and the three-dimensional configuration of the seeds, several of the most complete specimens were reconstructed using AMIRA (Figs. 25–32, also see Appendix S1–S5 in online Supplemental Data). The most complete specimen in longitudinal section (Figs. 5, 28, 31) reveals that the ovule is orthotropous and completely enclosed. Although small areas of the enclosing tissue remain open along one edge (Figs. 5, 31, at top), this surface of the enclosing structure has been abraded. Therefore, it is probable that seed enclosure may have been even more complete in life than in the fossils. Other reconstructions of the most complete specimens with enclosed seeds do not preserve the micropylar end because it was lost in the saw cuts (Figs. 30, 31). The only ovule in which the micropylar end is intact occurs in the specimen illustrated in Figs. 5, 28, 30, 33). In this specimen, the micropylar arms are situated directly beneath the small holes in the enclosing tissue (see also online Appendix S1–S5). Because of the tightness of the fit of the enclosing tissue to the ovule, these holes may have provided an entry point for pollination. So far, no pollen grains have been identified in the Apple Bay specimens.

Comparison of seed-enclosing structure of Doylea to the carpel of angiosperms

Several specimens of the most extensively enclosed seeds are nearly completely isolated from the outside environment by the enclosing tissue (e.g., Figs. 1–4, 29), while other specimens have openings along the corners of the seed (e.g., Fig. 6). The enclosing structure tightly covers the seed, but it apparently dehisced at the corners at maturity by means of three distinct valves, thus releasing seeds into the environment. This interpretation is supported by the occurrence of isolated seeds (e.g., Fig. 10) and also by two specimens that have partially open enclosing structures with seeds still inside (Figs. 6, 7). The existence of such a dispersal mechanism would also account for the range of variation that is present in the degree of seed enclosure from specimen to specimen. The enclosing structure appears to have nearly completely surrounded the seed in at least some stages of development, with only small openings near the micropylar end of the cupule being potentially available for the possible entry of pollen (see Appendix S1–S5 with online Supplemental Data). Unlike carpels of most living angiosperms, in which tissue completely isolates seeds from the external environment at the time of pollination, several living “basal” angiosperms have carpels that are partly closed by secretions, rather than by postgenital tissue fusion (13). So far, we have no way of determining whether such a secretion occurred in the fossil specimens.

The prominent nucellus attached only at the base, an apparently single integument, a megaspore membrane with enclosed thin-walled tissue, a surrounding cupule-like structure that was probably at least partially open near the micropylar end, and a trifid micropyle indicate that these plants were probably gymnospermous rather than angiospermous. A developmental sequence for these plants and attachment to some sort of branching system would be of great help in determining the exact affinities of the Apple Bay seed plant. It should be noted that while no pollen has been found in or associated with these reproductive structures, the small openings at the micropylar end of the enclosing structure and the trifid micropyle could indicate that these plants had a pollination droplet. In living gymnosperms such as pinaceous conifers (e.g., 50, 15), micropylar arms are common in ovules that produce pollination droplets, and these taxa usually have saccate pollen (e.g., 77). Thus, it appears likely that these reproductive structures were ovulate cupules borne by an extinct gymnosperm.

Comparison of Doylea to the cupules of fossil seed ferns

The cupule of the Apple Bay specimens is an abaxially enrolled leaf-like structure as evidenced by the single collateral vascular bundle with the phloem zone oriented toward the ovule. In the Paleozoic seed ferns such as Elkinsia 56, evidence shows that cupules probably evolved by fusion of terete distal segments of foliar branching systems (62; 70). In “Mesozoic” or post-Carboniferous pteridosperms, however, cupules are usually interpreted as modified leaves or leaflets that enclose the ovules (4, 5; 48; 57; 9, 10; 29). In addition, it has been suggested that cupules in the various “Mesozoic” or post-Carboniferous seed fern groups probably had several separate origins (e.g., 48; 57; 29; 10).

The post-Carboniferous seed ferns are usually treated in three or four orders (minus the very distinctive glossopterids): Peltaspermales, Caytoniales, Petriellales and Corystospermales, which are known mostly from compression/impression fossils (62; 70; 73; 39). All these groups are reviewed by 70, pp. 237–251 in this issue). Of these so-called “Mesozoic” seed ferns, the peltasperms have seeds borne abaxially on peltate, umbrella-like discs or fan-shaped megasporophylls (24; 78; 35, 5; 51; 44; 37; 43; 34) arranged in cones that appear distinct from the seed-enclosing structures from Apple Bay.

While the seed-bearing reproductive structure of the Caytoniales, Caytonia, first described from the Jurassic flora of Yorkshire (75; 23, 25, 26, 27) has received substantial attention, it also represents one of the more poorly known seed ferns (71). Although several species have been described, all are known from compression/impression specimens, and internal detail is generally lacking. The cupules of Caytonia have been reported to have 8–30 seeds (Table 1). Seeds of Caytonia are ovoid with an outer epidermis and an inner integumentary layer of thick-walled cells (26), unlike the tetrahedral seeds from Apple Bay, which have only thin-walled integumentary layers (Table 1). Evidence provided by 52 further suggests that Caytonia cupules enclosed a series of elongate canals from the outside, leading to the seed micropyles. This type of structure is completely lacking in Doylea, where the single seed nearly fills the cupule and is tightly enclosed over all surfaces except the micropylar area (see reconstructions in Appendix S1–S5 with online Supplemental Data and Figs. 28–31).

The most recently described pteridosperm order, the Petriellales, is based on permineralized cupules with enclosed seeds from the Triassic of Antarctica (73). Similarities of Petriellaea Taylor, del Fueyo & Taylor to the Apple Bay reproductive structures include cupulate orthotropous seeds with a vascular disc that are triangular in cross section with a parenchymatous integument (Table 1). However, Petriellaea cupules are bilateral with a dorsal groove, contain 5–6 seeds about half the size of those from Apple Bay, and appear elliptical in longitudinal view rather than tetrahedral (Table 1). The integumentary cells of the outer layer in Petriellaea have dark contents and a vascular disc is reported at the base of the nucellus (73). The Apple Bay seeds lack dark contents and the vascular disc- or cup-shaped zone of tracheids is in a completely different position, outside of the seed integument where the seed is attached to the cupule. The cupule in Petriellaea contains many vascular strands to the seeds that are attached on the adaxial surface, while in the Apple Bay material there is a single vascular strand and the seed is attached on the abaxial surface. The nucellus is described as completely attached to the integument in Petriellaea, while the nucellus in the Apple Bay seeds is attached to the integument only at its base. Cupules of Petriellaea may be borne in groups of up to seven (73). Attachment of the cupules described here is thus far unknown.

The most recent occurrence of pteridosperms is that of Ktalenia described by 2 from the Cretaceous (early Aptian) of Argentina (assigned to Caytoniales, Archangelsky and Taylor, 1985). Cupules, now known to be attached to Ruflorinia Archangelsky foliage, contain one or two orthotropous seeds with a prominent nucellar beak (Table 1; 72). Although these cupules are compression/impressions, cuticles are well preserved and the outer surfaces are papillate, unlike the structures from Apple Bay (Table 1). Like Caytonia, Petriellaea, and the structures from Apple Bay, it appears that seed micropyles in Ktalenia may have been enclosed inside the cupules. These cupules appear to have been fleshy and unlobed like those of Caytonia (Table 1).

Among the Mesozoic seed ferns, the cupules from Apple Bay most closely resemble taxa in the Corystospermales. Corystosperms typically have only one or two ovules per cupule (Table 1), and in those where anatomy is known, the seeds are attached to the abaxial surface of the cupule as in the material described here (39). Unlike the cupules described here, however, they usually have seeds with exposed bifid micropyles (Table 1). Only a few specimens of Spermatocodon Thomas are known; however, its cupules are spherical to campanulate (76), unlike the globose structures from Apple Bay. Pilophorosperma Thomas has more distinctly helmet-shaped cupules and conical seeds with curved exposed micropyles that differ from the enclosed tetrahedral seeds of the Apple Bay material. It has long been speculated that the corystosperm cupule was at least somewhat fleshy, as evidenced by the wrinkled surfaces on many compressions of Umkomasia Thomas, Pilophorosperma, and Spermatocodon (Table 1; 76; 80). The cupules described here have a thin parenchymatous outer zone that is underlain by a very sclerotic inner zone.

Umkomasia resinosa 39 from the Triassic of Antarctica is, thus far, the only known anatomically preserved corystosperm cupulate structure. Cupules are recurved and attached to a branching system. In U. resinosa, there are usually two orthotropous ovules (occasionally one) per cupule (Table 1). Umkomasia resinosa also had a rugose, glabrous surface like that in the Apple Bay structures, but U. resinosa cupules are typically bilobed rather than trilobed and have prominent secretory cavities (39) that are not found in the material from Apple Bay.

Many corystosperm cupules have been described as “lobed” (Table 1; see 73, for summary). In Umkomasia resinosa, these appear to be naturally occurring lobes in cupules that contain two seeds, and there is some indication of a possible abscission layer for the seeds, which could allow them to just drop out of the cupules at maturity without the need for the cupule to open. It has also been suggested that some seed fern cupules may have had valvate dehiscence or that the ribs in the cupule may indicate that they dehisced (e.g., 31; 1). Karibacarpon 40; 31; 30) has been described with 5–9 lobes that opened to leave large star-shaped structures (Table 1). The cupules from Apple Bay like those of Kannaskoppia 1, a Triassic gymnosperm (treated by these authors as a possible ginkgophyte) from the Molteno Formation of South Africa, had three valves. However, in these compression specimens the anatomy of the cupules is not clear, and seeds are unknown. This type of cupule opening mimics the dispersal of angiosperm seeds from a dehiscent fruit. Thus, there appears to be a variety of different types of seed dispersal in corystosperms and related taxa. It is unclear at the present time whether all of these taxa, should be included in the corystospermalean seed ferns. It was reported that ovules of corystosperms probably contained no fibrous cells (76), based on macerations of compression fossils. However, some tabular, thick-walled cells have been reported now in the permineralized Umkomasia from Antarctica (39). Most of the seeds known are flattened (both in compressed and permineralized material). These characters contrast with the Apple Bay seeds, which are distinctly tetrahedral and do not contain any thick-walled integumentary cells. Therefore, we have described the Apple Bay cupules in a new taxon, Doylea tetrahedrasperma gen. et sp. nov., in recognition of its distinctive characters.

A description of Umkomasia from the Triassic of northern China (80) recently brought attention to the occurrence of corystosperms, formerly thought to be mostly Southern Hemisphere taxa, in the Northern Hemisphere at this time. Further occurrences in the Northern Hemisphere include pollen organs of Pteroma Harris and foliage of Pachypteris Brongniart from the Yorkshire Jurassic flora (27), Umkomasia and Pteruchus from the Jurassic of Germany (38), and foliage types from the Triassic of Eurasia (Dobruskina, 1994; 80). The fossils from Apple Bay also show that not only were corystosperm-like plants more common in the Northern Hemisphere in the past, but also that they may have extended up into the Lower Cretaceous of North America. This group, which has had relatively little attention compared to other Mesozoic pteridosperms, is now beginning to be better known and clearly has much more diversity than was previously recognized.

Comparison of Doylea to the outer integument of angiosperms

The cupule vasculature in the Apple Bay material tracks the course of a raphe in an anatropous double-integumented angiosperm ovule (see Fig. 30), and there is sometimes a slight bulge to the cupule on that side (see Figs. 1, 3). Recently, 10 discussed the origin of the second integument in angiosperms as potentially arising from a seed fern cupule like that of Caytonia. This idea originated with 19 and later with 61 and 8. Caytonia, however, has from 8 to 30 seeds per cupule (Table 1), so evolution of the angiospermous outer integument would require the reduction in the number of ovules per cupule to one (10). Assuming that the seeds of Caytonia are borne on the adaxial surface of the cupule wall (a character that has not been documented), evolution of an angiosperm carpel through the cupule of Doylea from that in a Caytonia-like ancestor would require not only the reduction of the number of ovules but a complete turning inside-out of the cupule. Unlike the carpels of extant angiosperms (e.g., 14, 15; 42), seeds of Doylea were borne on the abaxial (rather than the adaxial) surface of the enclosing structure.

Doylea had single seeds that were tightly enclosed in tissue of the cupule like the outer integument of angiosperms, showing that this level of organization was present in pteridosperms in the Lower Cretaceous. The inner integument of angiosperms is attached to the adaxial surface of the outer integument (14; 10; 42), while the seed inside the tight cupule of Doylea is attached abaxially, making the enclosing tissue analogous rather than homologous with the second integument of flowering plants. Doylea seeds were dispersed from the cupules by three valves like those of some dehiscent carpels (fruits) of angiosperms. However, again, these are analogous, rather than homologous to the carpels of flowering plants.

While we know that seed ferns are very diverse in the late Paleozoic and early Mesozoic, Doylea provides evidence that far from dying out, corystosperm-like plants may have become increasingly specialized through time to a point of near convergence with angiosperms in terms of their reproductive morphology. Doylea tetrahedrasperma was probably one of several angiophytes that combined gymnospermous pollination with angiospermous protection of the seeds and flowering plant-type dehiscence for seed dispersal. The biotic pressures for specializations that we usually consider to be limited to angiosperms have led to convergent evolution and provide a cautionary note about the need to distinguish analogy from homology when probing the fossil record for answers to the mystery of angiosperm origins. As more well-preserved material is described and gaps in the fossil record filled, we look forward to coming closer to understanding the evolutionary pathways that led to angiospermy.