Amer. J. Bot. 74(11): 1722-1736. 1987.

DREWRIA POTOMACENSIS GEN. ET SP. NOV., AN EARLY CRETACEOUS MEMBER OF GNETALES FROM THE POTOMAC GROUP OF VIRGINIA1

PETERR. CRANEAND GARLANDR. UPCHURCH,JR.~,~ Department of Geology, Field Museum of Natural History, Roosevelt Road at Lake Shore Dr., Chicago, Illinois 60605, and U.S. Geological Survey, Box 25046, Mail Stop 919, Denver, Colorado 80225

ABSTRACT Drewriapotomacensis gen. et sp. nov. from the Lower Cretaceous Potomac Group of Virginia (Zone I, probably Aptian) provides the first definite Mesozoic megafossil record of Gnetales. The stems are slender, display no evidence of secondary growth, and show axillary monopodial branching. Attached leaves are opposite and decussate, borne at swollen nodes, and have clasping sheathing bases. Each leaf is oblong, up to 20 mm long, and has a dense network of longitudinally aligned subepidermal fibers. Leaf venation consists of two, or possibly three pairs of longitudinal parallel veins that form a reticulum at the apex and higher-order crossveins that form apically oriented chevrons. Reproductive structures consist of short, loose spikes that are borne in dichasially arranged groups of three. The dichasia are either terminal or lateral in the axil of a leaf. Reproductive units of the lateral spikes in a dichasium contain seeds, each surrounded by at least one pair of opposite, broadly elliptical or ovate bracts. Characters of D. potomacensis that suggest a gnetalean relationship include the opposite bracts surrounding the seeds, the network of subepidermal foliar fibers, and the distinctive leaf venation, which is very similar to that seen in the cotyledons of extant Welwitschia. Other features consistent with a gnetalean relationship include opposite and decussate leaves, swollen nodes, and the dichasial arrangement of the reproductive spikes. Masses of polyplicate gnetalean pollen comparable to that of extant Welwitschia occur in the same bed as the megafossils. The morphology of D. potomacensis indicates that it was an herb or possibly a shrub. The growth habit of D. potomacensis and associated plants, combined with the sedimentary occurrence of the fossils, indicate that this species and perhaps related taxa were important components of early successional vegetation during the mid-Cretaceous.

EXTANTSEED PLANTS comprise five major Gnetum, and the monotypic Welwitschia. N6- clades: Coniferales, Cycadales, Ginkgoales, mejc (1967) suggested that Gnetales were never Gnetales, and angiosperms. All ofthese groups, diverse, but as early as 1908, Arber and Parkin with the exception of Gnetales, have an exten- (p. 507) concluded, "Although the Gnetales are sive megafossil record that extends back into unknown to us in the fossil state, we are by no the Mesozoic (e.g., Taylor, 198 1; Stewart, means precluded ipso facto from the conclu- 1983). The Gnetales, in contrast, lack any well- sion that the group may have once flourished documented Mesozoic megafossil record and more vigorously than at the present day." comprise only three extant genera: Ephedra, Arber and Parkin's suggestion has subsequent- ly been corroborated by the recognition of a

I Received for vublication 23 October 1986: revision widespread and diverse group of Mesozoic dis- accepted 2 arch 1987. persed pollen comparable to grains of extant We thank C. B. Beck. J. A. Doyle, M. J. Donoghue, L. Ephedra and Welwitschia in shape, aperture J. Hickey, C. Hotton, and J. A. ~olfefor their assistance configuration, surface sculpture, and exine ul- and discussion of issues raised in this paper. Research was supported in part by National Science Foundation Grant trastructure (e.g., Trevisan, 1980), which are BSR-83 14592 to PRC and a National Research Council especially abundant and diverse in the Early Postdoctoral Research Associateship, U.S. Geological Sur- Cretaceous of low paleolatitudes (e.g., Africa, vey, to GRU. Field work was supported by a Scott Turner South America) (Brenner, 1976). Only one pos- Award in the Earth Sciences, University of Michigan, to GRU. Ms. P. Quasthoff drew the reconstructions, and R. sible kind of fossil gnetalean pollen (Equise- McBride and D. Pocknoll assisted with photography. tosporites chinleanus Daugherty) has been Any use of trade names is for descriptive purposes only found within a pollen-bearing organ (Mascu- and does not imply endorsement by the U.S. Geological lostrobus clathratus Ash, 1972), but the sys- Survey. tematic position of the M. clathratus plant is The listing of authors is alphabetical. Current address: Museum, University of Colorado, uncertain (Ash, 1972) and pollen wall ultra- Campus Box 3 15, Boulder, CO 80309-03 15. structure differs from that of extant Gnetales November 198 71 CRANE AND UPCHURCH -CRETACEOUS GNETALES 1723

(Zavada, 1984). In this report we describe gne- volved removing the cuticle from the rock with talean megafossils, including stems with at- a needle, dissolving the acrylic plastic in ace- tached leaves and reproductive structures, from tone, demineralizing overnight in HF, mac- the Lower Cretaceous Potomac Group of Vir- erating in bleach, staining in Safranin 0 dis- ginia. These remains, found in association with solved in a 5050 mixture of ethanol and masses of Welwitschia- type pollen, provide toluene, and mounting in Histoclad. A centri- the first fully documented report of Mesozoic fuge was used for the bleaching, staining, and gnetalean megafossils and have important im- dehydration stages of processing due to the plications for understanding the ecology and extremely fragmentary nature of the cuticle. evolution of this enigmatic group of seed plants. Dispersed cuticles and associated masses of pollen, prepared in the course of an earlier MATERIALSAND METHODS-Thefossils de- study (Upchurch, 1984a, b), also were exam- , scribed in this paper were collected from a ined to assess variation in cuticular structure 4-cm-thick bed of micaceous gray clay exposed and to determine the possible pollen of the at the northern end of Drewrys Bluff, approx- megafossils. Cuticles were photographed with imately 9 m above high tide level. This bed is Kodak Technical Pan Film 24 15 using a Zeiss dated palynologically as upper Zone I of Bren- Photomicroscope I. Pollen grains were pho- ner (1963), or probably Aptian (J. A. Doyle, tographed with Kodak Technical Pan Film oral communication, 1986) and yields the early 24 15 on an Olympus Vanox Photomicroscope angiosperm leaves with cuticles reported by using Nomarski optics. Upchurch (1984a, b). Although fossil vegeta- tive shoots were collected on the outcrop, al- DIAGNOSIS-Generic diagnosis- Drewria most all of the reproductive material was re- gen. nov.: Stems bearing opposite and decus- covered by splitting blocks of matrix in the sate leaves. Branching monopodial, branches laboratory and examining them under a dis- arising in the axils of leaves. Leaves simple, secting microscope. Organically preserved oblong, margin entire. Leaf venation of two specimens were coated with Krylon acrylic distinct vein orders. First-order venation of plastic spray to prevent further fragmentation. two (possibly three) pairs of longitudinal par- Over 100 specimens on approximately 80 allel veins; the innermost pair running to the pieces of matrix were examined. Leaf cuticles lamina apex and branching to form a reticu- were prepared from one organically-preserved lum. Second-order venation of crossveins that specimen. All specimens are deposited in the typically form apically oriented chevrons. University of Michigan Museum of Paleon- Lamina with abundant, longitudinally orient- tology (UMMP). Systematiccomparisons were ed, subepidermal fibers. Reproductive struc- made using the published literature (e.g., Pear- tures either terminal or in the axil of a leaf, son, 1929; Martens, 197 l; Sporne, 197 l), the consisting of short, loose spikes borne in di- U.S. Geological Survey reference collections of chasial groups of three. Reproductive units of cleared leaves and modern pollen, and speci- the lateral spikes in a dichasium containing mens in the herbarium of the Field Museum seeds, each surrounded by at least one pair of of Natural History. broadly elliptical to broadly ovate bracts. Photographic techniques were critical to de- Holotype: UMMP 654 14b (Fig. 22). termining venation and morphological fea- Etymology: From Drewrys Bluff, Virginia, tures in many specimens. Organically pre- the type locality. served remains and reddish-brown impressions Specijic diagnosis: Drewria potomacensis sp. of vegetative parts were photographed with nov.-As for genus. Kodak Technical Pan 241 5 Film developed Syntypes: UMMP 65409-654 14a, 654 15- with HC-110 developer to medium or high 65419 (Fig. 1-16, 19-21, 23-27). contrast levels. Light gray impressions of veg- Etymology: From Potomac Group, the fos- etative parts were photographed either with sil-bearing rock unit. Polaroid Type 5 1 high contrast film or Kodak Locality: 37"25'25"N, 77"25'30"W, Chester- Technical Pan Film 241 5 under fluorescent field County, Virginia, south of Richmond on light, which enhanced many details of venation the James River at the north end of Drewrys not readily visible to the naked eye. Repro- Bluff, approximately 9 m above high tide level. ductive structures were photographed with Po- Stratigraphic position: Potomac Group (un- laroid Type 55 film. Polarizing filters were used differentiated), upper Zone I of Brenner (1 963), to increase contrast and reduce glare from spec- Lower Cretaceous, probably Aptian. imens coated with acrylic plastic. Cuticles of megafossils were prepared using DESCRIPTION-Plantsconsisting of slender a combination of standard methods. This in- branching stems that bear opposite and de-

November 198 71 CRANE AND UPCHURCH -CRETACEOUS GNETALES 1725

cussate leaves. Stems 1-3 mm in diameter, longitudinally striated but showing no obvious internodes up to 30 mm long, stem diameter venation. constant for most of internode length but grad- ually increasing at the nodes. Stems showing DISCUSSION-Morphologyand anatomy- At no evidence of secondary growth, consisting of least twelve specimens show attachment be- an inner core and an outer, more readily flat- tween leaves and reproductive structures; these tened zone. Inner core up to 1.5 mm in di- reproductive shoots are identical to associated ameter, containing 6-8 longitudinally oriented sterile shoots in size, nonvenational features strands, these strands of similar thickness to of foliar morphology, and mode of preserva- the foliar veins. Outer zone up to 0.3 mm wide, tion. The stems of some specimens have a dis- typically one-fourth the radius of the stem. tinct central core and a thinner border (Fig. 2- Branching monopodial, with pairs of opposite 5), interpreted as a stele and surrounding cor- branches arising in the axils of leaves, no evi- tex, respectively. The central core in one or- dence of scale leaves associated with the bases ganically preserved specimen shows 6-8 thick of the branches. longitudinal strands (Fig. 3) and longitudinal Leaves simple, oblong, 10-20 mm long, 2- striations similar in number and size also occur 6 mm wide, midrib absent, apex acute and in some stem impressions (Fig. 4). Although usually with a sharp tip, base narrow-acute to the exact number of strands cannot be deter- decurrent, margin entire, texture thin. Bases of mined, they are the same thickness as the foliar opposite leaves clasping the stem to form a veins and are therefore interpreted as vascular short sheath. Venation of two distinct orders. bundles. One specimen also shows additional, First-order venation consisting of 2 or possibly narrower striations on the stem that possibly 3 pairs of parallel longitudinal veins that ex- represent fibers. The outer zone is always the tend from the base of the leaf, 2 pairs of veins same proportion relative to stem diameter and present in the middle part of the lamina, veins is interpreted as a decayed parenchymatous about 1 mm apart at the widest part of the cortex; the presence of organic material in the lamina. Longitudinal parallel veins forming a outer zone on one specimen shows that this reticulum at the leaf apex. Second-order ve- zone was not produced by drying and shrinkage natibn consisting of crossveins that typically of the specimens. None of the stems are prom- form apically-oriented chevrons. Leaves with inently impressed into the sediment and were abundant, longitudinally oriented, parallel evidently easily flattened during preservation. subepidermal fibers. Fibers much thinner than Evidence of extensive wood, therefore, is ab- veins, occasionally branching and anastamos- sent. Although the possibility of secondary ing, tending to obscure the venation. Leaf mar- growth in some parts of the plant cannot be gin mechanically reinforced (possibly by an excluded, the largest stems observed are 3 mm additional pair of veins). Foliar cuticle very in diameter and appear to be herbaceous. thin, difficult to prepare, with poorly-devel- Branching of the vegetative shoots is op- oped flanges. Cells in rows, probably elongate, posite, axillary and monopodial (Fig. 1, 2, 4). with straight anticlinal walls, about 35-40 pm None of the lateral branches shows evidence wide. External and internal sculpture smooth of scale leaves at the base (e.g., Fig. 4). In one under light microscopy. Stomata unknown. lateral shoot where the first three nodes are Reproductive structures consisting of short preserved, the distance between the base and loose spikes in dichasially arranged groups of the lowest node is much greater than the dis- three. Dichasia either terminal or lateral in the tance between the first and succeeding nodes axil of a leaf. Reproductive units of the lateral (Fig. 4, 7). The leaves of the same shoot show spikes in a dichasium larger than those of the no decrease in size from the lowest to the up- central spike and bearing seeds. Seeds flat- permost node, making it unlikely that the un- tened, narrowly ovate, 1-2.5 mm long, 1-2 mm equal internode length is due to fossilization wide, apex acute, base rounded. Seeds borne of an immature shoot. The absence of scale between at least one opposite pair of broadly leaves at the base of the lateral shoot and the elliptical to broadly ovate inner bracts with presence of an elongate region between its base rounded apices. Each bract up to 1.5 mm long, and the lowest node are characteristic of extant

of axillary branch in Fig. 2, showing subtending leaf (Lf) borne at swollen node. The longitudinal striations on the main stem are interpreted as the impressions of vascular bundles. UMMP 65410a1, x 5. 5. Close-up of stem in Fig. 2, showing outer zone (arrows) and inner zone. UMMP 65410a1, x 10. 6. Impression of stem showing opposite leaves with sheathing bases (Sh). UMMP 6541 1, x 6. 7. Counterpart of shoot in Fig. 2, showing first node of axillary shoot. N = node, Sh = sheathing leaf base, Lf = free portion of leaf. UMMP 65410a, x 10.

November 19871 CRANE AND UPCHURCH -CRETACEOUS GNETALES 1727

Fig. 14-17. Foliar venation of D. potomacensis and Welwitschia. 14. Line drawing of D. potomacensis leaf in Fig. 8,traced from photograph. Note inner (I)and middle (M) veins. UMMP 654 lob, x 5.15. Line drawing of D. potomacensis leaf in Fig. 9, traced from photograph. UMMP 65410b, x 5. 16. Line drawing of D. potomacensis leaf apex in Fig. 8, drawn from photograph with camera lucida. UMMP 65410b, x 15. 17. Line drawing of cotyledon of Welwitschia mirabilis. Redrawn from Rodin (1953), fig. 7. x 5. plants that lack a period of bud dormancy innermost pair of veins runs to the apex (Fig. (Tomlinson and Gill, 1973; Tomlinson, 1978). 8, 12, 14). Their more zig-zag course at the tip Leaf impressions of D. potomacensis have of the leaf (Fig. 12, 16) suggests that they are little relief in the rock, especially relative to giving off lateral branches; simultaneously,they those of associated conifers; thus, the leaves become less distinct and appear to form a re- were thin-textured. Preservation varies greatly, ticulum with the crossveins at the leaf apex, but a few leaves from vegetative and repro- rather than running to a point of fusion (Fig. ductive shoots show both the vascular tissue 12, 16). and the subepidermal fibers. In most leaves The foliar cuticle of D. potomacensis is very with good organic preservation, the densely thin and fragments readily upon preparation packed fibers obscure the venation (Fig. 1I), (Fig. 13). The cuticular flanges are difficult to but in a few the fibers appear to have mostly see under light microscopy even with phase- decayed and the details of venation can be contrast, but the observed cells are organized observed (Fig. 8-1 0,14-16). None of the leaves into distinct rows. No stomata were observed has a midvein, and the innermost pair of lon- on either in situ or dispersed cuticle, presum- gitudinal parallel veins runs on either side of ably because only adaxial cuticle was pre- the longitudinal plane of symmetry (Fig. 8, 14). served. The exact number of vein pairs is difficult to Figure 18 shows a reconstruction of the veg- determine, but the two inner pairs run at least etative shoots. as far as the middle part of the lamina (Fig. 8, The reproductive structures of D. potomac- 9, 14, 15). The margin of some specimens ap- ensis consist of three spikes arranged in ter- pears to be reinforced by an additional pair of minal dichasia (Fig. 19-24, 26, 32). One spec- longitudinal parallel veins (Fig. 8), but these imen appears to show a single spike borne in are thin and cannot be confidently differen- the axil of the leaf on an otherwise vegetative tiated from marginal bundles of subepidermal axis (Fig. 25), but is too poorly preserved to fibers, as occur in some extant dicotyledons as exlcude the possibility that it is a fragment of well as in Gnetum and Welwitschia.Excluding a dichasium. Seeds have been observed only the possible pair of marginal veins, only the on the lateral spikes of dichasia and never on AMERICAN JOURNAL OF BOTANY [VO~.74

B; 27, middle arrows). However, in some specimens additional bracts are associated with each reproductive unit, and typically these bracts are positioned in an anterior-posterior plane with respect to the flattening of the seed (Fig. 27, bottom arrow). The interpretation of these anterior-posterior bracts is unclear. Some may represent a primary bract subtending a reproductive unit while others could represent tertiary bracts. Each ovulate reproductive unit therefore consists of at least three, and perhaps five "bracts" surrounding a single seed. These ,. would be interpreted most straightforwardly, respectively, as a primary bract plus a pair of lateral secondary bracteoles, or as a primary bract with two pairs of opposite and decussate secondary and tertiary bracteoles. Seeds of D. potomacensis were sufficiently hard to resist compression during fossilization, and occasional specimens preserve the seed outline as a cavity in the matrix. The seed is Fig. 18. Reconstruction of vegetative parts of Drewria potomacensis, based on specimens in Fig. 1-12. bilaterally symmetrical, and some seeds appear to have a prominent ridge down the center of both flattened surfaces (Fig. 22, Mc). Whether central spikes. The central spike may have borne the ridge is a feature of the seed or the keel of pollen-producing reproductive units, but poor a pair of anterior-posterior bracts is uncertain. preservation precludes more detailed consid- A single specimen (Fig. 22, Mc) shows a pro- eration of this possibility. jection 0.25 mm long from the apex of a seed The lateral spikes are always larger than the that probably is an elongated micropyle. central spike of a dichasium (Fig. 19-24), but Figure 32 shows a reconstruction of the re- the number and arrangement of reproductive productive structures. units in each spike is difficult to determine. In central spikes, the reproductive units appear Comparison with other fossil material- to be borne in opposite pairs (Fig. 19, 20). Drewria potomacensis cannot be assigned to These pairs probably are arranged decussately, any previously described genus of fossil plant considering the phyllotaxy of the rest of the but shares significant similarities with four fos- plant. If this interpretation is correct, the cen- sil taxa. One isolated leaf of Conospermites tral spike may have contained about 8-12 re- hakeaefolius Ettingshausen described from the productive units. Each reproductive unit in the mid-Cretaceous (Cenomanian) of Czechoslo- central spike clearly includes a bract, but other vakia (Vidovle locality) by Velenovsk? and Vi- details are not preserved. niklh? (1926, pl. 3, fig. 5, left-hand specimen) None of the lateral spikes is complete, and has venation almost identical to that of D. po- the arrangement of the reproductive units is tomacensis.The lamina of the specimen is larg- unclear. Although the bracts are larger, the lat- er (up to 2 cm wide) than that of D. potoma- eral spikes probably contained about the same censis and has a more rounded apex and more number of reproductive units as the central gradually tapering base. However, there are spikes. Several specimens show that each re- two prominent veins that arise at the base of productive unit consists of a seed surrounded the leaf, and these run parallel toward the apex by at least one pair of opposite bracts and that where they become less distinct and form a these bracts are apparently lateral with respect reticulum. The longitudinal veins are linked to the plane of flattening of the seed (Fig. 22 by crossveins that frequently form apically ori-

Fig. 19-24. Reproductive structures of D. potomacensis gen. et sp. nov. 19. Dichasium showing central spike with bracts and two poorly preserved lateral spikes. UMMP 65413, x 13. 20. Dichasium showing subtending leaves, central spike, one well-preserved lateral spike, and one poorly preserved lateral spike. UMMP 65414a, x 9. 21. Dichasium showing the remains of two leaves, poorly preserved central spike, and two lateral spikes. UMMP 65415, x 16. 22. Holotype showing stem with a swollen node bearing a pair of opposite leaves and the terminal dichasium. Dichasium

November 19 871 CRANE AND UPCHURCH-CRETACEOUS GNETALES 1731

izona was described as a peculiar conifer of uncertain relationship (Ash, 1972) but shares several similarities with D. potomacensis. The leaves of Decheyllia are attached in opposite and decussate pairs, and the linear vegetative leaves are flattened into a single plane. The linear vegetative leaves have a thickened mid- rib, while the wider sporophylls lack a midrib and have venation composed of two broad lon- gitudinal "ribs" and two narrow marginal "strands." (A pair of longitudinal strands is illustrated for the midrib of linear vegetative Fig. 32. Reconstruction of dichasium of D. potomac- i leaves, along with a pair of marginal strands ensis. [Ash, 1972, Text-fig. 6A-C]; however, vena- tion on the type specimens is difficult to in- terpret.) Despite these similarities to D. po- tomacensis, crossveins appear to be absent in Systematics-The habit of Drewria poto- Decheyllia, and other features ofvegetative and macensis, combined with the morphology of reproductive morphology differ from those of the reproductive organs and leaves, excludes D.potomacensis. It is ofinterest, however, that a close relationship to any known group of the pollen organs associated with Decheyllia living or fossil seed plants except Gnetales and (Masculostrobus clathratus Ash) contain pollen angiosperms. Within flowering plants, a close grains of Equisetosporites chinleanus Daugh- relationship to the monocotyledons is excluded erty (Ash, 1972). The longitudinal striations of by the organization of the foliar venation, es- these grains are similar to those in pollen in pecially the absence of a midvein and the thin- extant Ephedra (Scott, 1960), but the details ning and branching of the longitudinal veins of pollen wall ultrastructure are different (Za- at the leaf apex to form a reticulum. In both vada, 1984). extant monocotyledons that have few pairs of The single specimen of Cyperacites sp. de- longitudinal parallel veins and in putative fos- scribed by Krassilov (1982) from the Lower sil relatives such as the Early Cretaceous Aca- Cretaceous (Valanginian-Barremian) Manlaj ciaephyllum, each leaf has a midvein and lon- locality, Mongolia, consists of a stem with a gitudinal veins that persist to the leaf apex, terminal dichasial arrangement of three repro- where they fuse (Doyle, 1973). Additionally, ductive "spikes" arising from a swollen node. almost all extant monocotyledons have alter- Each spike apparently consists of bracts nate, rather than opposite, phyllotaxy. The ("glumes") and long bristles. Although addi- group of extant dicotyledons most similar to tional information is not available, the size of D. potomacensis is the Chloranthaceae. This plant, the swollen node, and the dichasial ar- family and closely related extinct forms are of rangement of three spikes are very similar to particular interest because of their well-docu- what is seen in D. potomacensis. mented Early Cretaceous fossil record based Eoanthus zherikhinii Krassilov, an ovulate on pollen, leaves, and reproductive structures reproductive structure from the Lower Cre- (Upchurch, Hickey, and Niklas, 1983; Walker taceous of the Lake Baikal area (Krassilov, and Walker, 1984; Upchurch, 1984a, b; Friis, 1986), resembles D. potomacensis in several Crane, and Pedersen, 1986). Similarities shared respects. In E. zherikhinii, each ovule has an by D. potomacensis, Chloranthaceae, and Gne- elongate micropyle and is associated with a tales include opposite and decussate leaves flattened, presumably foliar structure (com- borne at swollen nodes, dichasially arranged pared with an opened follicle by Krassilov); spike-like inflorescences, and reproductive additionally, polyplicate gnetalean pollen is structures surrounded by pairs of bracts. How- found in the micropyles of the ovules. How- ever, leaves of extant Chloranthaceae are ever, E. zherikhinii differs from D. potoma- toothed, possess a midrib with pinnately ar- censis in having a whorled arrangement of ranged secondary veins, and lack a network of ovules and transverse ridges on the structures subepidermal fibers. In contrast, the leaves of subtending the ovules. In addition, the type of both Gnetum and Welwitschia lack teeth and gnetalean pollen found in the micropyles of E. have a dense network of subepidermal fibers, zherikhinii differs from that found in associ- and the venation of Drewria leaves is strikingly ation with D. potomacensis (cf. below) in hav- similar to that ofthe cotyledons of Welwitschia ing fewer and wider plications that are spirally (Takeda, 19 13; Rodin, 1953; Fig. 17). These arranged. cotyledons are supplied by two vascular bun- 1732 AMERICAN JOURNAL OF BOTANY [VO~.74

dles that dichotomize at the base of the lamina individual "ephedroid" pollen grains, prepa- to produce a pair of inner bundles, a pair of rations of dispersed cuticles from the same bed lateral bundles and a pair of marginal bundles as the megafossils yielded three large clumps that run longitudinally on either side of the of polyplicate pollen similar to that of extant midline. Toward the apex of the cotyledon, the Ephedra (Steeves and Barghoorn, 1959) and six longitudinal veins become less distinct and Welwitschia (Wodehouse, 1935; Erdtman, form a reticulum (Rodin, 1953, 1958a). Sec- 1965). These clumps are all of similar size and ondary crossveins form apically oriented chev- ovoid shape (Fig. 28) and are interpreted as rons between the longitudinal veins, as in the pollen sacs. One clump is enclosed by a cuticle mature foliage of Welwitschia (Rodin, 1958b). (Fig. 29) and shows numerous Ubisch bodies As far as we are aware, this distinctive pattern associated with the pollen grains (Fig. 3 1). In- of venation is unique to Welwitschia and dividual pollen grains are boat-shaped, ellip- Drewria, and in combination with previously soidal, 18-32 pm long, 8-1 8 pm in diameter, cited features provides strong evidence for a with a L/W ratio ranging from 1.7-2.9 (Fig. close relationship between the fossil and extant 29, 30). Some grains in each clump are weakly Gnetales. monosulcate, with a long narrow sulcus (Fig. Recent cladistic analyses of seed plants re- 30). Many other grains show no sulcus under solve the Gnetales as a monophyletic group in light microscopy, but some of these are folded which Gnetum and Welwitschia are more longitudinally along an apparent line of weak- closely related than either is to Ephedra (Hill ness. The surface of each grain has over 20 and Crane, 1982; Crane, 1985; Doyle and Don- ridges (plications) parallel to the long axis of oghue, 1986). Coulter and Chamberlain (1 9 17) the grain, each less than 0.5 pm wide. The exine reached a similar conclusion. Hypothesized is 0.5-1.0 pm thick and appears homogeneous synapomorphies of Gnetum and Welwitschia under light microscopy. These pollen grains do include the presence of reticulate venation not belong to any species described from the composed of two or more vein orders, syn- Potomac Group by Brenner (1963), but clearly detocheilic stomata, a well-developed system relate to other types of Early Cretaceous pollen of hypodermal fibers, tetrasporic megagame- allied with extant Gnetales by Trevisan (1 980). tophyte development, absence of archegonia, Among extant Gnetales they are most similar exclusively cellular embryogenesis, and the to the pollen of Welwitschia in having a sulcus presence of a "feeder" in the embryo (see Crane and a large number of narrow longitudinal [I9851 and literature cited therein). Of these ridges. Although there is no direct evidence features, only two are preserved in Drewria. that these grains were produced by Drewria, The presence of reticulate venation with more the presence of gnetalean pollen clumps in the than one vein order and a well-developed sys- megafossil-bearing bed supports our interpre- tem of hypodermal fibers suggests that Drewria tation that gnetalean plants were present in the is more closely related to Gnetum and Wel- local flora. witschia than to Ephedra. However, the valid- ity of this hypothesis is largely dependent on Paleoecology-The functional morphology the arrangement of outgroups accepted for the of D. potomacensis, coupled with its floristic Gnetales. Cladistic analyses of seed plants sug- and sedimentary associations, indicate that it gest that the most appropriate outgroups are was an herbaceous or shrubby, early succes- angiosperms, Bennettitales, or Pentoxylon, in sional plant of mesic environments. The stems one or another arrangement (Crane, 1985; of D. potomacensis appear to have a well-de- Doyle and Donoghue, 1986). Subepidermal fi- veloped cortex and show no evidence of ex- bers are known in some Bennettitales, and the tensive secondary tissues. These features in- combination of subepidermal fibers and retic- dicate that the plant was either herbaceous or, ulate venation composed of two or more vein like many suffrutescent shrubs, had young orders is present in certain extant angiosperms. shoots that lacked secondary growth. The fo- Thus, the possibility exists that these features liage is thin-textured, with a thinner cuticle have been lost in the small simple leaves of than almost all other gymnosperms seen at Ephedra. In the absence of a clearer under- Drewrys Bluff and other Potomac Group lo- standing of the relationships of the Gnetales calities. This indicates that the plant probably to other seed plants, the phylogenetic position grew under conditions of relatively low mois- of Drewria within the group cannot be confi- ture stress, perhaps resulting from its low stat- dently established. ure (cf. Salisbury, 1927) and streamside habit Indirect evidence supporting the assignment (cf. below). Like extant Gnetum and many ex- of Drewria to Gnetales is derived from pollen tant tropical angiosperms (Tomlinson and Gill, associated with the megafossils. In addition to 1973; Tomlinson, 1978), D. potomacensis ap- November 19871 CRANE AND UPCHURCH -CRETACEOUS GNETALES 1733

pears to have lacked a period ofbud dormancy. overlying, plant-bearing gray clay has a sharp Such an interpretation is consistent with recent basal contact and little coarse material except paleobotanical analyses of Early Cretaceous for mica, indicating an abrupt decrease in the climates, which infer relatively mesic condi- energy of the depositional environment. The tions for the Potomac Group (Brenner, 1976; gray color, absence of bedding, preservation of Vakhrameev, 1978). some megafossils at an oblique angle to the dip The megaflora at Drewrys Bluff consists of of the bed, and almost complete absence of D. potomacensis, ferns, angiosperms, conifers, bioturbation suggest relatively rapid deposi- and cycadophytes. Fern fronds, angiosperm tion in a poorly aerated environment, probably leaves, and remains of D. potomacensis are the during floods. The overlying argillaceous sand- commonest fossils. Most species of ferns are stone records a brief resumption of bed-load known from at least one specimen in which deposition, but finer grain size indicates much . one or two orders of frond branching are pre- lower-energy conditions than for the conglom- served, and the fragmentary nature of many eratic sandstone. The thick, gray, sandy mud- specimens appears to be due to the tendency stone at the top of the sequence represents a of the matrix to break irregularly, rather than return to the deposition of predominantly sus- along well-defined bedding planes. The com- pended sediment in a poorly aerated environ- mon occurrence of angiosperm megafossils ment. Extensive bioturbation and presence of contrasts markedly with the absence of angio- numerous carbonized roots indicate that this sperms from most Zone I Potomac Group lo- unit formed a substrate for plant growth and calities (cf. Fontaine, 1889; Doyle and Hickey, probably represents shallower water condi- 1976). Three of the five angiosperm species tions than the fossiliferous gray clay. Features reported by Upchurch (1984a, b) are known. especially indicative of an abandoned channel from at least one complete leaf, and two of fill deposit are the abrupt decrease in the energy these are known either attached or closely as- of the depositional environment seen at the sociated with cutinized stems that show little conglomeratic sandstone/claystone contact, the evidence of secondary growth (Upchurch, lateral discontinuity of the fossiliferous gray Hickey, and Niklas, 1983; Upchurch, 1984a, claystone and overlying units, and the devel- fig. '5). Conifers and cycadophytes are rare and opment of a rooted mudstone horizon in the fragmentary, in contrast to their abundant and upper part of the sequence (cf. Allen, 1965). relatively complete preservation at many other This is consistent with other evidence for the Zone I megafossil localities. Taken together, fluviatile (i.e., stream) origin of Potomac Group these data imply that 1)D. potomacensis, ferns, sediments in this region and other parts of the and angiosperms were the major elements of outcrop belt (Glaser, 1969). The occurrence of an herbaceous to shrubby local vegetation, and abundant fern, angiosperm and Drewria re- 2) conifers and cycadophytes were either rare mains in a stream channel environment, at an local elements or transported to the site of de- abrupt shift from high-energy to low-energy position from a more distant source. depositional conditions, indicates that these Sedimentological interpretations of the plants were early successional colonizers of dis- Drewrys Bluff locality indicate that the plant- turbed habitats, an ecology inferred for early bearing bed is the basal unit of an abandoned angiosperms as a whole by Doyle and Hickey channel fill. The rocks exposed at the locality (1976) and Hickey and Doyle (1977). show a wide range of grain sizes and are, in These conclusions on the probable growth- ascending stratigraphic order, a >1-m-thick habit and habitat of Drewria potomacensis are basal conglomerate; a 4-cm-thick megafossil- of considerable interest in light of recent dis- bearing gray claystone; a 15-cm-thick, fine- cussions on the ecology of Cretaceous Gne- grained, massive, argillaceous sandstone; and tales. Brenner (1 976) and Vakhrameev (1970) a >2-m-thick, extensively root-bioturbated, hypothesize that Cretaceous Gnetales were pri- gray sandy mudstone. Based on sections ex- marily adapted to hot and dry climates and posed 5-15 m to the north and south of the had an ecology similar to that of Classopollis- megafossil locality, the conglomerate is later- producing conifers (Cheirolepidiaceae). Both ally persistent, but the overlying units are re- authors compare the pollen of Cretaceous Gne- placed laterally by trough-crossbedded sand- tales to extant Ephedra, which lives in dry cli- stone with gravel lag. (The lateral contacts mates, and assume comparable ecology for the between lithological units could not be ob- fossil plants that produced this pollen. In ad- served due to soil cover.) The conglomeratic dition, they note that gnetalean pollen is most sandstone is massive and poorly sorted, in- abundant at low paleolatitudes and correlated dicating rapid deposition of bed-load material on a regional scale with possible indicators of under turbulent, high-energy conditions. The aridity such as low percentages of fern spores 1734 AMERICAN JOURNAL OF BOTANY [VO~.74 and high percentages of Classopollis. Doyle, the preservation of gnetalean megafossils. The Jardink, and Doerenkamp (1982), in contrast, vegetative organs of herbs have a low fossil- hypothesize that Cretaceous Gnetales were ization potential relative to those of trees (e.g., adapted to warm, but not necessarily dry, cli- Spicer, 1980); therefore, if an herbaceous to mates and had an ecology more similar to that shrubby growth habit was widespread in ex- of angiosperms than that of Classopollis-pro- tinct Gnetales it may partially explain the dis- ducers. They argue that palynological diversity crepancy between their extensive microfossil, in Cretaceous Gnetales is too great to justify but sparse megafossil, record. ecological interpretations based on the climatic Notwithstanding these possible biases of tolerances of two relictual extant genera, representation and recognition, the gnetalean Ephedra and Welwitschia, especially in a world megafossil record seems certain to be expanded where angiosperms had yet to become domi- by the discovery of new material and perhaps nant. In addition, Doyle et al. (1982) note that by the reinterpretation of previously described in the Lower Cretaceous Cocobeach sequence Mesozoic fossils. In addition to previously of central Africa, located at a paleolatitude of mentioned taxa, described forms that share 15" S (Smith, Hurley, and Briden, 198 I), the some similarities with extant Gnetales include percentages of gnetalean pollen and Classo- Cyperacitespotomacensis Berry and Casuarina pollis show an inverse relationship typical of covilliI.lrard from the Potomac Group (Ward, ecological segregation. The highest percentages 1895; Berry, 19 1I), Montsechia vidali (Zeiller) of Classopollis are found in the middle Aptian, Teixeira from the Early Cretaceous (Berria- during an interval of basinal flooding by ma- sian) of Spain (Blanc-Louvel and Barale, 1983), rine waters and salt deposition, suggesting ad- Hexagonocaulon minutum Lacey and Lucas aptation to dry upland habitats and possibly (1 98 1) from the Triassic of the South Shetland saline conditions. In contrast, the highest per- Islands, and Gurvanella dictyoptera Krassilov centages of Gnetales are found in an earlier and the "Potamogeton-like spike" from the interval of fluvial-deltaic deposition and are Early Cretaceous of Mongolia (Krassilov, coincident with the highest percentages of an- 1982). Careful reevalution of these and other giosperms, suggesting ecological co-occurrence problematic Mesozoic megafossils may con- and adaptation to floodplain habitats. The data siderably enhance our knowledge of the early from our study provide direct support for the diversity of Gnetales, and this will be crucial view that Gnetales and early angiosperms had to further clarifying how Gnetales are related similar ecological tolerances, probably arising to other groups of seed plants. from fundamental similarities in their repro- ductive biology (Crane, 1987) and vegetative LITERATURE CITED morphology. Both in terms of branching and leaf morphology D. potomacensis is more me- ALLEN,J. R. L. 1965. A review of the origin and char- somorphic than any extant Gnetales, and this acteristics of Recent alluvial sediments. Sedimentol- indicates that these extant genera provide in- ogy 55: 89-191. ARBER,E. A. N., AND J. PARKIN. 1908. Studies on the sufficient basis from which to generalize about evolution of angiosperms. The relationship of the an- the ecology of fossil Gnetales. giosperms to the Gnetales. Ann. Bot. London22: 489- 515. Fossil history of the Gnetales-Despite the ASH, S. R. 1972. Late Triassic plants from the Chinle abundance and diversity of gnetalean pollen in Formation in northeastern Arizona. Palaeontology 15: southern Laurasia and especially northern 598-6 18. BLANC-LOWEL,C., AND G. BARALE.1983. Montsechia Gondwana during the Early Cretaceous (Bren- vidali (Zeiller) Teixeira 1954. Nouvelles Observations ner, 1976; Doyle et al., 1982), gnetalean mega- et Reflexions sur son attribution systematique. Ann. fossils have not been described prior to this Paleontol. 69: 15 1-1 74. study. Two hypotheses have been advanced to BERRY,E. W. 19 1 1. Systematicpaleontology, Lower Cre- account for the apparent scarcity of gnetalean taceous: fossil plants. In W. B. Clark [ed.], Lower Cretaceous, 214-508. Maryland Geological Survey, megafossils. Arber and Parkin (1 908) and Sew- Baltimore. ard (1 9 19) emphasized the possibility of con- BRENNER,G. J. 1963. The spores and pollen of the Po- fusing Gnetum-like foliage with that of dicot- tomac Group of Maryland. Maryland Dept. Geol. yledons, while extrapolations based on extant Mines and Water Res. Bull. 27: 1-215. Ephedra and Welwitschia led Wilson (1959) to -. 1976. Middle Cretaceous floral provinces and suggest that the group had a low fossilization early migrations of angiosperms. In C. B. Beck [ed.], Origin and early evolution of angiosperms, 23- potential because of restriction to arid envi- 47. Columbia University Press, New York. ronments. Drewria demonstrates that not all COULTER,J. M., AND C. J. CHAMBERLAIN.19 17. Mor- Mesozoic Gnetales were restricted to such hab- phology of gymnosperms. University of Chicago Press, itats and indicates a further possible bias against Chicago. November 19871 CRANE AND UPCHURCH-CRETACEOUS GNETALES 1735

CRANE,P. R. 1985. Phylogenetic analysis of seed plants -. 1958b. Leaf anatomy of Welwitschia.11. A study and the origin of angiosperms. Ann. Missouri Bot. of mature leaves. Amer. J. Bot. 45: 96-103. Gard. 72: 716-793. SALISBURY,E. J. 1927. On the causes and ecological sig- . 1987. Vegetational consequences of the angio- nificance of stomata1frequency, with special reference sperm diversification. In E. M. Friis, W. G. Chaloner, to the woodland flora. Phil. Trans. R. Soc. Lond., Ser. and P. R. Crane [eds.], The origin of angiosperms and B 216: 1-65. their biological consequences, 107- 144. Cambridge SCOTT,R. A. 1960. Pollen of Ephedra from the Chinle University Press, Cambridge. Formation (Upper Triassic) and the genus Equiseto- DOYLE,J. A. 1973. Fossil evidence on the early evolution sporites. Micropaleontology 6: 27 1-276. of the monocotyledons. Quart. Rev. Biol. 48: 399- SE~ARD,A. C. 19 19. Fossil plants, Vol. 4. Cambridge 413. University Press, London. , AND M. J. DONOGHUE.1986. Relationships of SMITH,A. G., A. M. HURLEY,AND J. C. BRIDEN. 198 1. angiosperms and Gnetales: a numerical cladistic anal- Phanerozoic palaeocontinental world maps. Cam- ysis. In B. A. Thomas and R. A. Spicer [eds.], Sys- bridge University Press, Cambridge. tematic and taxonomic approaches in palaeobotany, SPICER,R. A. 1980. The importance of depositional sort- 177-1 98. Oxford University Press, Oxford. ing to the biostratigraphy of plant megafossils. In D. -, AND L. J. HICKEY. 1976. Pollen and leaves from L. Dilcher and T. N. Taylor [eds.], Biostratigrpahy of the mid-Cretaceous Potomac Group and their bearing fossil plants: successional and paleoecological analy- on early angiosperm evolution. In C. B. Beck [ed.], ses, 17 1-1 83. Dowden, Hutchinson & Ross, Strouds- Origin and early evolution of angiosperms, 139-206. burg, PA. Columbia University Press, New York. SPORN& K. R. 197 1. The morphology of gymnosperms. , S. JAR DIN^, AAN S. DOERENKAMP.1982. Afro- Hutchinson, London. pollis, a new genus of early angiosperm pollen, with STEEVES,M. W., AND E. S. BARGHOORN.1959. The pollen notes on the Cretaceous palynostratigraphy and pa- of Ephedra. J. Arnold Arb. 40: 221-255. leoenvironments of northern Gondwana. Bull. Centres STEWART,W. N. 1983. Paleobotany and the evolution Rech. Exp1or.-Prod. Elf-Aquitane 6: 39-1 17. of plants. Cambridge University Press, New York. ERDTMAN,G. 1965. Pollen and spore morphology/plant TAKEDA,H. 19 13. Some points in the anatomy of the taxonomy. Gymnospermae, Bryophyta (text). Alm- leaf of Welwitschia mirabilis. Ann. Bot. London 27: quist and Wiksell, New York. 347-357. ETTINGSHAUSEN,C. V. 1867. Die Kreideflora von Nied- TAYLOR,T. N. 198 1. Paleobotany. An introduction to erschona. Sitzungsb. Ak. Wiss. Wien, Math C1. 55: fossil plant biology. McGraw-Hill, New York. 235-264. TOMLINSON,P. B. 1978. Branching and axis differentia- FONTAINE,W. M. 1889. The Potomac or younger Me- tion in tropical trees. In P. B. Tomlinson and M. H. sozoic flora. U.S. Geol. SUN. Monogr. 15: 1-377. Zimmerman [eds.], Tropical trees as living systems, FRIIS,E. M., P. R. CRANE,AND K. PEDERSEN.1986. Floral 187-207. Cambridge University Press, Cambridge. evidence for Lower Cretaceous chloranthoid angio- ,AND A. M. GILL. 1973. Growth habits of tropical sperms. Nature 320: 163-164. trees: some guiding principles. In B. J. Meggers, E. S. GLASER,J. D. 1969. Petrology and origin of Potomac Ayensu and W. D. Duckworth [eds.], Tropical forest and Magothy (Cretaceous) sediments, Middle Atlantic ecosystems in Africa and South America: a compar- Coastal Plain. Maryland Geol. Surv. Rep. Invest. 1 1. ative review, 129-143. Smithsonian Institution Press, HICKEY,L. J., AND J. A. DOYLE. 1977. Early Cretaceous Washington, DC. fossil evidence for angiosperm evolution. Bot. Rev. TREVISAN,L. 1980. Ultrastructural notes and consider- 43: 3-104. ations on Ephedripites, Eucommiidites and Mono- HILL,C. R., AND P. R. CRANE. 1982. Evolutionary cla- sulcites pollen grains from Lower Cretaceous sedi- distics and the origin of angiosperms. In K. A. Joysey ments of southern Tuscany (Italy). Pollen Spores 22: and A. E. Friday [eds.], Problems of phylogenetic re- 85-132. construction, 269-36 1. Academic Press, New York. UPCHURCH,G. R., JR. 1984a. Cuticular anatomy of an- KNOBLOCH,E. 1978. On some primitive angiosperm giosperm leaves from the Lower Cretaceous Potomac leaves from the Upper Cretaceous of the Bohemian Group. I. Zone I leaves. Amer. J. Bot. 71: 192-202. Massif. Palaeontographica Abt. B. 166: 83-98. . 1984b. Cuticle evolution in early Cretaceous an- KRASSILOV,V. A. 1982. Early Cretaceous flora of Mon- giosperms from the Potomac Group of Virginia and golia. Palaeontographica Abt. B 18 1: 1-43. Maryland. Ann. Missouri Bot. Gard. 71: 522-550. . 1986. New floral structure from the Lower Cre- , L. J. HICKEY,AND K. J. NIKLAS. 1983. Leaves taceous of Lake Baikal area. Rev. Palaeobot. Palynol. with chloranthoid characters from the Lower Creta- 47: 9-16. ceous Potomac Group of Virginia. Amer. J. Bot. 70, LACEY,W. S., AND R. C. Luc~s.198 1. The Triassic flora Part 2: 81. of Livingston Island, South Shetland Islands. Br. Ant- VAKHRAMEEV,V. A. 1970. Range and paleoecology of arct. SUN. Bull. 53: 157-173. Mesozoic conifers, the Cheirolepidiaceae. Paleontol. MARTENS,P. 197 1. Les Gnetophytes. Handbuch der J. 1: 12-25. Pflanzenanatomie 12(2): 1-295. . 1978. The climates of the Northern Hemisphere NEMEJC,F. 1967. Notes on the origin and past of the in the Cretaceous in light of paleobotanical data. Pa- Chlamydospermophyta. Preslia 39: 345-35 1. leontol. J. 12: 143-154. PEARSON,H. H. W. 1929. Gnetales. Cambridge Uni- VELENOVSK*,J. 1884. Die Flora der bohmischen Krei- versity Press, Cambridge. deformation 11. Beitr. pal. 0sterr.-Ung. 3: 26-47. RODIN,R. J. 1953. Seedling morphology of Welwitschia. , AAN L. VINIKLA~.1926. Flora Cretacea Bohe- Amer. J. Bot. 40: 371-378. miae I. Rozpravy Statniho Geologickeho Ustavu Ces- . 1958a. Leaf anatomy of Welwitschia I. Early koslovenske Republicky 1: 1-57. development of the leaf. Amer. J. Bot. 45: 90-95. WALKER,J. W., AND A. G. WALKER.1984. Ultrastructure 1736 AMERICAN JOURNAL OF BOTANY [VO~.74

of Lower Cretaceous angiosperm pollen and the origin WODEHOUSE,R. P. 1935. Pollen grains. McGraw-Hill, and early evolution of flowering plants. Ann. Missouri New York. Bot. Gard. 7 1: 464-52 1. ZAVADA,M. W. 1984. Angiosperm origins and evolution WARD,L. F. 1895. The Potomac Formation. U.S. Geol. based on dispersed fossil pollen ultrastructure. Ann. SUN. 15th Annual Report, 307-397. Missouri Bot. Gard. 7 1: 444-463. WILSON,L. R. 1959. Geological history of the Gnetales. Okla. Geol. Notes 19(2): 35-40.