| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Paleobotany |
Florida Museum of Natural History and Department of Botany, University of Florida, Gainesville, Florida 32611-7800 USA
Received for publication January 29, 2007. Accepted for publication July 11, 2007.
ABSTRACT
Seeds are useful in distinguishing among extant genera of Vitaceae and provide a good basis for interpretation of fossil remains in reconstructing the evolutionary and phytogeographic history of this putatively basal Rosid family. Seeds of Ampelocissus s.l. including Pterisanthes and Nothocissus are distinguished from those of all other vitaceous genera by long, parallel ventral infolds and a centrally positioned oval chalazal scar. Principal component analysis facilitates recognition of four Ampelocissus s.l. seed morphotypes differentiated by dorsiventral thickness, width of ventral infolds, chalazal depth, and degree of dorsal surface rugosity. While these intergrade, their end-member morphologies are distinctive and coincide well with inflorescence morphology, extant geographic distribution, and ecology. Seven fossil morphospecies are recognized. Ampelocissus parvisemina sp. n. (Paleocene of North Dakota; Eocene of Oregon) and A. auriforma Manchester (Eocene of Oregon) resemble extant Central American species; A. bravoi Berry (Eocene of Peru) is similar to one group of Old World extant species; and A. parachandleri sp. n. (Eocene of Oregon) and the three European fossil species A. chandleri (Kirchheimer) comb. n., A. lobatum (Chandler) comb. n., and A. wildei sp. n. (Eocene to Miocene) resemble another group of extant Old World Ampelocissus. All these fossils occur outside the present geographic range of the genus, reflecting warmer climates and former intercontinental links.
Key Words: Ampelocissus • Cayratia • fossils • Nothocissus • Pterisanthes • seeds • Tertiary • Vitaceae
The evolutionary history of the Vitaceae, the grape family, is of interest because of the wide geographic and ecological range of the family today and because of its apparently basal position within the Rosids. This family contains around 700 species, assigned to 
15 genera. They are mostly lianas distributed in tropical and temperate areas throughout the world. This estimate includes Leea, a genus of shrubs that is alternatively treated as a separate but closely allied family (Ingrouille et al., 2002
; Rossetto et al., 2002
; Soejima and Wen, 2006
). The precise phylogenetic position of this family within the Eudicots is not certain (Judd and Olmstead, 2004
; Soltis and Soltis, 2004
; Schonenberger and von Balthazar, 2006
); however, a recent study based on the complete chloroplast genome sequence, although with comparatively limited taxon sampling across the angiosperms, strongly supports its position as the earliest diverging lineage of rosids (Jansen et al., 2006
).
Morphologically, the family is well delimited and easily recognized. Leaf-opposed tendrils and inflorescences and unique seed morphology are the most useful characters to distinguish Vitaceae from other families. The genera are delimited by floral morphology, inflorescence types, and sometimes seed morphology. Seed morphological characters, while commonly applied in recognizing the family and component genera, have not been thoroughly surveyed.
The Vitaceae illustrate interesting geographical distribution patterns. The temperate to warm temperate genera Ampelopsis Michx. (20 spp.), Parthenocissus Planch. (15 spp.), and Vitis L. (60 spp.) include disjunctions between eastern Asia and North America. Ampelocissus Planch. (90 spp.), Cayratia Juss. (50 spp.), Leea Royen ex L. (30 spp.), and Tetrastigma Planch. (90 spp.) are mostly in southern Asia and Malesia; Ampelocissus also occurs in Africa. Cissus L. (300 spp.) is distributed worldwide with greatest diversity in South America and tropical Africa. Cyphostemma (Planch.) Alton (200 spp.) and Rhoicissus Planch. (12 spp.) are mostly in Africa. Nothocissus (Planch.) Latiff (6 sp.) and Pterisanthes Bl. (20 sp.) are endemic to the Malay-Borneo region, Yua C. L. Li (3 sp.) is endemic to China and India, Clematicissus Planch. (1 sp.) is endemic to Australia, and Acareosperma Gagnep. (1 sp.) is a rare species from Laos. Phylogenetic relationships within the family are not fully resolved; certain species may be incorrectly placed and will need to be revised as more information becomes available (Ingrouille et al., 2002
; Rossetto et al., 2002
; Rossetto and Jackes, 2005
; Soejima and Wen, 2006
). The fossil records, along with more phylogenetic data, will help with the understanding of the biogeographical history of this family.
The fossil record of Vitaceae includes leaves, pollen, stems, and seeds. Leaf impressions from the Cretaceous and Tertiary of Europe, North America, and Asia have commonly been assigned to this family, but Kirchheimer (1939)
questioned the validity of many of these assignments. A similar situation exists for the paleopalynological record. Although fossil pollen grains have sometimes been assigned to extant genera of the Vitaceae, (Muller, 1981
; Ferguson et al., 1998
; Nemejc et al., 2002
), the criteria used to distinguish these pollen grains from those of other angiosperm families with similar pollen have not been clearly indicated. Occasional permineralized stems have contributed to our knowledge of the Vitaceae fossil record through careful comparison with the stem anatomy of extant representatives, but these reports are relatively rare (Watari, 1951
; Prakash and Dayal, 1964
; Wheeler and Lapasha, 1994
; Poole and Wilkinson, 2000
; Wheeler and Manchester, 2002
).
Fossil seeds of Vitaceae are easily recognized from a suite of unique and distinctive morphological characters (particularly a pair of ventral infolds and a dorsal chalazal scar) and are relatively common in Tertiary floras (for review, see Kirchheimer, 1939
, 1957
; Tiffney and Barghoorn, 1976
). Because seeds of this family show patterns of morphological variation that often reflect infrafamilial relationships, the fossil seed record is potentially most useful in addressing questions of evolutionary and phytogeographic divergences. Paleobotanists have long considered that vitaceous seeds can be differentiated to the generic level. Most paleobotanical treatments of the family have been based on comparative work with extant vitaceous seeds, but with limited sampling of the extant species. With the exception of some surveys of seed morphology in the extant genus Vitis (Tiffney and Barghoorn, 1976
), intrageneric variation has not been fully explored. Because a thorough morphological survey of modern vitaceous seeds with good representative sampling has been lacking, some of the generic identifications in the paleobotanical literature are dubious. A more comprehensive survey of modern vitaceous seeds and a revision of the fossil records of this family are in progress.
Seeds of Vitaceae have a bipartite seedcoat including a thin, outer sarcotesta composed of several layers of parenchyma cells and an inner, lignified endotesta composed of columnar cells (Periasamy, 1962
; Corner, 1976
). The surface of the endotesta testa varies from smooth to rugose. The endosperm is ruminate because of two main longitudinal infolds of the endotesta on the ventral (adaxial) side of the seed. The vascular strand of the raphe extends from the hilum, passing medially along the ventral side and over the apex, terminating on the dorsal (abaxial) side of the seed as an enlarged lignified chalaza. Although the sarcotesta and the vascular strand usually are not preserved in fossil seeds, the external seed morphology is mirrored in the underlying lignified part of the seed coat that is often preserved in fossils. The combination of paired ventral infolds and the dorsal chalaza (Fig. 1) is unique to the Vitaceae. A pair of cup-shaped indentations on the ventral side of a compressed seed also occurs in some species of Menispermaceae, tribe Tinosporeae; however, they never have a chalaza on the dorsal side. Other features of the vitaceous seed include an apical notch, a beak on the hilum, and a median groove on the dorsal side extending apically and/or basally from the chalaza (chalaza-apex and chalaza-base grooves, or "a median dorsal groove" referring to both) (Fig. 1).
|
|
), mostly distributed in Malesia, southern Asia, and Africa. Only five species occur in Central America. Pterisanthes, a Malesian endemic genus with 20 species, has seeds very similar to those of Ampelocissus. This genus also has tendril-associated inflorescences, but the inflorescence itself is an unusual laminar structure (see description in Materials and Methods). The morphological similarity between Pterisanthes and Ampelocissus has long been recognized (Latiff, 1982a
), and molecular data indicated that Pterisanthes is nested with other Ampelocissus (Soejima and Wen, 2006
). Hence, in this paper we consider Pterisanthes as a subclade of Ampelocissus s.l. According to Soejima and Wen (2006)
, another potential close ally of Ampelocissus is Nothocissus spicifera (Griff.) Latiff. The genus Nothocissus was elevated from one of Planchon's four sections of Ampelocissus by Latiff (1982b)
. Originally, N. spicifera was the only species in this genus, but later five more species were transferred to it (Latiff, 2001
). Three of these new combinations involved Australian species formerly placed in Cissus; however, based on the inflorescence morphology, the new combinations are questionable (B. R. Jackes, James Cook University of North Queensland, personal communication). Nothocissus spicifera is similar in floral structure to Malesian species of Ampelocissus but is distinguished by its unusual whip-like inflorescences (see Materials and Methods). Further, the seeds of N. spicifera are rugose and very different from those of other Malesian Ampelocissus (Latiff, 1982b
). In this paper we use the term Ampelocissus s.l. to include Ampelocissus, Pterisanthes, and Nothocissus.
Besides Nothocissus, the other three sections of Ampelocissus established by Planchon (1887)
are sect. Euampelocissus Planch. (= sect. Ampelocissus), with thyrsoid inflorescences (Africa and Asia); sect. Kalocissus (Miq.) Planch., with elongate panicles of spikes (Malesia); and sect. Eremocissus Planch., with small linear panicles (one species only; Central America). Seed morphology was included among the features used to distinguish the sections, with sect. Ampelocissus described as having boat-shaped seeds with a pair of distinct ventral furrows, sect. Kalocissus having boat-shaped seeds with a ventral keel, and sect. Eremocissus having oval-triangular seeds with a ventral keel (Süssenguth, 1953
). Hence, seed morphology seems to be as informative as inflorescence structure for delimiting subclades of Ampelocissus s.l., each of which occurs in distinct geographical regions. However, a detailed and thorough seed survey of this genus is needed to test whether the seed characters correlate with phylogeny.
In this research, a comprehensive morphological survey of extant Ampelocissus s.l. seeds was conducted and a principal component analysis was used to show the variation of seed characters and associations with inflorescence type, geographical distribution, and habitat. Using the survey of extant seeds as a framework, we review previously published fossil records and recognize some additional fossil representatives of the genus. Possible phylogenetic relationships and implications for biogeography of Ampelocissus s.l. are discussed.
MATERIALS AND METHODS
Survey of extant Ampelocissus seeds
Specimens were obtained though herbarium loans from A, MO, NY, SING and US, or investigated during visits at FLAS, IBSC, L, SING, and SYS. The species determinations were verified and annotated by careful examination of the herbarium specimens in comparison with published descriptions. The herbarium voucher sheets examined, including those sampled for seed morphology (labeled with OTU number), are listed in Appendix 1.
Mature fruits removed from selected sheets were boiled in water for 2 min and soaked until the loosened fruit tissue could be removed by hand to free the seed(s). Sarcotesta was thoroughly removed by scraping gently with a small brush to expose the surface sculpture. This process is extremely important because the sarcotesta tends to attach tightly to the dry seeds and obscures most of the underlying morphological characters. The seeds were transversely sectioned by hand with a razor blade to show the median cross section.
Images of the cross section, dorsal, ventral, lateral, top and bottom view of the seeds were recorded using a Nikon Coolpix 4500 digital camera (Tokyo, Japan). The associated inflorescence/infructescence types were observed from the herbarium sheets; the information of habitat and geographical distribution of these taxa were obtained from the labels of the specimens and the previous taxonomical treatments (Merrill, 1923
, 1929a
, b
; Backer and Bakhuizen van Den Brink, 1965
; Latiff, 1982a
–c
, 1987
, 1991
; Jackes, 1984
; Long and Rae, 1991
; Lombardi, 2000
, 2005
).
Fossil Ampelocissus seeds
Fossil seeds were analyzed to assess the same characters as in the extant seeds and were photographed in the same orientations and with the same apparatus. Sometimes it was necessary to remove adhering sediment to expose the seed surface morphology. Permineralized seeds were sectioned using a paper-thin diamond saw blade mounted on a Microslice II annular saw (Malvern, England).
Tertiary fossils were studied from museum collections including the Florida Museum of Natural History at University of Florida (locality and specimen numbers cited with the prefix UF), the Smithsonian Institution, Washington, D.C. (cited by the prefix USNM), the Forschungsinstitut und Naturmuseum Senckenberg, Frankfurt, Germany (cited by the prefix SM.B.), and Denver Museum of Nature and Science (cited by the prefix DMNH). For other Ampelocissus fossils, the pictures and descriptions in original publications were consulted. Images of Cayratia jungii (Gregor) comb. n. were scanned from the original publication (Gregor, 1977
) with permission of the copyright holder, E. Schweizerbart'sche Verlagsbuchhandlung OHG, Stuttgart, Germany (http://www.schweizerbart.de/j/pal-z/).
The geologic ages of the fossil localities cited in this paper were taken from the original articles and revised where necessitated by new data. The age of the Belen fruit and seed assemblage in western Peru, from which Ampelocissus bravoi was collected, was considered Eocene by Berry (1929a)
; although he did not provide direct evidence for the assignment. Olsson (1931
; pp. 15, 16) determined the age as Early Oligocene based on correlations of mollusks from nearby marine strata; however, subsequent revisions to the geologic time scale have resulted in repositioning of the Eocene-Oligocene boundary (Gradstein et al., 2005
) such that these beds are now considered Eocene (Burnham and Johnson, 2004
). Additional fieldwork is needed to determine more precisely the placement of the Belen fruit and seed assemblage within the Eocene epoch. The vitaceous specimens cited here from Early Tertiary of southern England include those from the London Clay Formation (Reid and Chandler, 1933
; Chandler, 1962
; Collinson, 1981
), which is considered to be Early Eocene based on stratigraphic correlations with mammals (Collinson and Cleal, 2001b
), and from younger sites including Hordle, which is considered Late Eocene (Collinson and Cleal, 2001a
). The oldest of the mainland European sites we considered is Messel, near Darmstadt, which is dated as Middle Eocene (Mertz and Renne, 2005
). Many of the younger German brown coal deposits that yielded seeds studied by Kirchheimer (1938)
have been revised in their ages since the time of his publications. The Wiesa locality that he considered Oligocene is now treated as Miocene (Mai, 2000
, 2001
).
The North American sites from which fossil Ampelocissus seeds were identified include the Beicegal Creek locality in North Dakota and the Nut Beds locality in Oregon. The Beicegal Creek locality is considered Late Paleocene based on stratigraphic position and floristic composition (Manchester et al., 2004
). The Nut Beds locality was previously considered to be mid-Middle Eocene based on fission track and 40Ar/39Ar dates of about 44 million years ago (mya) (Manchester, 1994
; Wheeler and Manchester, 2002
). However, Hanson (1996)
considers the age of these beds to be Early Middle Eocene, based on mammals from the Nut Beds site (which are interpreted to represent the Bridgerian land mammal age) and on an additional unpublished 40Ar/39Ar date by Swisher of 48.32 ± 0.11 mya (cited in Hanson, 1996
). We currently accept this Early Middle Eocene age for the Nut Beds.
Measurements and morphometric analysis
Forty-one extant species or subspecies of Ampelocissus s.l. and five fossil seeds were selected for a principal component analysis (PCA). Twenty-two seed characters (labeled C1–C22) were measured as described in Fig. 2 and Table 2 using digital images and the program ImageJ (Rasband, 1997–2006
). In a few instances, where the characters could not be measured directly from the fossil specimens (i.e., testa thickness, C21, of A. bravoi and A. parachandleri) a predicted value was deduced from the most common condition of all the seeds examined. PCA analyses were performed by the program Minitab 14.0 (Minitab Inc., Pennsylvania, USA).
|
|
and Tiffney (1979)
. Inflorescences of Ampelocissus s.l. were differentiated into six types (Fig. 3b): (1) bifurcated spikes—a pair of long whip like "spikes," up to 30 cm long, joined at the base, as occurs in Nothocissus; (2) compact/cymose—the inflorescence is round in outline, possessing several primary axes rather than a continuous main axis; terminal axes tending to be closely clustered; (3) laminar—the specialized inflorescence of Pterisanthes having small sessile flowers dotted on both sides of a more or less fleshy, bright red laminar structure; (4) lax/elongate—a thyrsoid inflorescence, elongate and triangular in outline, with a continuous main axis and loosely arranged lateral axes; (5) linear panicle—the inflorescence is linear in outline, with cymose units spirally arranged on a main axis; (6) panicle of spikes—the inflorescence of sect. Kalocissus is elongate or triangular in outline, with a main axis and spirally arranged lateral axes that are spikelike because the flowers are usually sessile. With the exception of the bifurcated spike inflorescence, all of these inflorescence types include a tendril attached on the basal part of the main axis—a characteristic feature of the genus.
|
RESULTS
Seed morphology of extant Ampelocissus s.l.
Seeds of Ampelocissus s.l. are pyriform, oval, or round in dorsal or ventral view; oval is the most common condition. They are more or less dorsiventrally compressed regardless of the number of the seeds per fruit. The number of seeds per fruit ranges from one to four, but two- and three-seeded fruits are most common. In the same plant seeds from a two-seeded fruit would be more compressed dorsiventrally than the seeds from a one-seeded fruit, and seeds from three- or four-seeded fruits would be more triangular in top view. Nevertheless, all other features of the seeds remain the same. The apical notch ranges from shallow to deep. The ventral infolds are almost as long as the full length of the seed and are parallel to each other. They may be narrow furrows or wider concavities occupying most of the ventral surface. The chalaza is round to oval, located in the center of the dorsal side. Chalaza-apex and chalaza-base grooves, when present, are linear furrows which are V-shaped in cross section. The dorsal surface ranges from smooth, with faint marks radiating from the chalaza, to highly rugose. The testa is usually thinner in the ventral infolds and slightly to strongly thickened on the lateral margins. Seed length ranges from 4 to 12 mm (Fig. 4). In general, young seeds have the same morphology as mature ones, differing only in that the endotesta is not fully lignified. Based on the specimens that we examined (Appendix 1), the shape and the size of mature Ampelocissus s.l. seeds are not highly variable among individuals of the same species.
|
|
1. Smooth, flattened, with broad ventral infolds
Seeds oval, highly compressed dorsiventrally, apical notch shallow or deep, beak usually acute, smooth angled and not sharply pointed. Ventral infolds broad, dish-like, smooth, occupying almost the whole ventral surface. Dorsal surface smooth or very finely rugose with protrusions scattered on the surface and without a special pattern. Chalaza not sunken, dorsal median groove absent (Fig. 5a). Seeds of this type are relatively large, mostly over 7 mm and up to 12 mm long (Fig. 4, open triangle). The majority of the species from Malesia have this type of seed (Fig. 3c). These species are rainforest dwellers (Fig. 3d), and the type 1 seed is tightly associated with the "panicle of spikes" inflorescence type (Fig. 3b).
|
3. Rugose, flattened with linear ventral infolds
The shape of these seeds is similar to that of type 2, except that the ventral infolds appear furrowlike (Fig. 5d) or linear (Fig. 5e) on the surface. The edge of the linear infolds and the margin of the ventral surface can have irregular ridges or furrows, repeating the rugose pattern of the dorsal side. In cross section the ventral infold cavities have an irregular outline (Fig. 5d–f cross section). The chalaza is sunken, and the median dorsal groove is present. The degree of rugosity varies as in the type 2 seed. Some extremely rugose seeds have multiple deep furrows cutting into the endosperm (Fig. 5f; Fig. 3, OTU28). Width of ventral infolds is a continuous quantitative character; therefore there is no clear boundary between type 2 and type 3 seeds. Type 3 seeds are mainly distributed in Africa, South Asia, and Australia (Fig. 3c), in dry, open or dry, cool forest (Fig. 3d) and are associated with more compact, cymose inflorescences (Fig. 3b). Ampelocissus latifolia, the type species of Ampelocissus, has this type of seed.
4. Rotund, with wide infolds
Seed dorsiventrally thicker than the previous three types, and only slightly longer than wide. In transverse section, the broad ventral infold cavities have a circular and smooth outline. Two subgroups can be further differentiated: (a) Rotund with cup-shaped ventral infolds—Seeds round, less compressed, apical notch shallow or absent, beak acute. Ventral infolds round and cup-shaped, occupying most of the ventral surface. Dorsal surface smooth, chalaza not sunken, dorsal median groove absent (Fig. 6a). Pterisanthes, the Malesian taxon with laminar inflorescences (Fig. 3) has this type of seed. (b) Cordiform with broad ventral infolds—Seed oval, cordiform or pyriform, less compressed, with prominent apical notch and mucronate or stipitate beak. Dorsal surface with faint radiating rugose marks and a prominent median dorsal groove, and with the chalaza sunken to varying degrees (Fig. 6b–e). Some Central American species have this type of seed, but they are not associated with particular inflorescence types or habitat (Fig. 3).
|
, seeds of N. spicifera are similar in shape to seeds of some species of Tetrastigma.
|
Dorsiventrally compressed and rugose seeds also occur in some species of Ampelopsis and Vitis, but they have much smaller seeds (<5 mm) and comparatively short ventral infolds. Seeds of Vitis rotundifolia, however, do have the features of the type 3 Ampelocissus s.l. seed and are distinguishable only because they have faint radiate markings and a chalaza that is not as much sunken (Fig. 7e). Yua austro-orientalis, a Southeast Asian species with floral structures similar to Parthenocissus, has pyriform, dorsiventrally compressed seeds; a pair of parallel, long, and linear ventral infolds; and an oval chalaza with rugose marks radiating out from the edge (Fig. 7d). These features correspond to those of the type 3 Ampelocissus s.l. seed. The differences are that Y. austro-orientalis seeds are more pyriform and triangular (vs. oval with small sharp beak) and have slightly shorter ventral infolds.
Broad, cup-shaped ventral infolds, while useful in recognizing Ampelocissus seed types 1, 2, and 4, are not limited to Ampelocissus; they also occur in some species of Ampelopsis (e.g., A. denudata, A. grossedentata) and Cayratia (e.g., C. oligocarpa, C. formosana). However, Ampelopsis has a droplet-shaped or triangular chalaza and Cayratia has an elongate, linear chalaza. Seeds of Central American Ampelopsis denudata resemble those of Central American Ampelocissus erdvendbergiana and A. robinsonii because the seed shape is cordiform and the chalaza is almost oval (Fig. 7f).
Our seed survey of extant Vitaceae shows that species of the same genus share certain characters (Table 1); however, all the characters are variable, with indistinct boundaries between genera. There are suites of characters that can be considered typical of each genus, yet some seeds may fall within zones of overlap between two or more genera. The aforementioned Ampelopsis denudata is an example. Similar conditions also apply to the four seed types of Ampelocissus s.l. Although the boundaries among them are not always clear, a fossil seed can still be recognized as one particular seed type if it possesses characters uequivocally confined to that type. To determine whether a fossil seed is within a particular morphological category, one can conduct a morphometric analysis to indicate the fossil seed's position relative to extant seeds.
Fossil representatives
Based on our understanding of the morphology of the extant representatives as presented above, we recognize seven fossil species of Ampelocissus seeds, including three that are new and four previously described. They are treated in alphabetical order in this section. In addition, three of the fossil species previously attributed to this genus are now rejected; we discuss other more likely affinities for these taxa. The comparisons of the fossils to the extant seeds are summarized in Table 4.
|
Fossil Ampelocissus
Ampelocissus auriformaManchester (1994)
emend. n. (Fig. 8e–g, i, j)
|
Our revised concept of A. auriforma excludes the smaller seeds from the same locality that we have transferred to A. parvisemina for reasons indicated in the description of this new species. We emend the diagnosis of the species to accommodate revised dimensions: Internal mold of seed coat is 4.7–6.7 mm long, 4.4–6.3 mm wide, 2.5–4.0 mm thick; seed length is always greater than width. Seed outline is sub-oval to triangular in apical view, indicating a two- to four-seeded fruit. Molds of the ventral infold cavities have a thick rim, about 0.5–0.7mm thick (Fig. 8g, i), which can be seen also in transversely fractured specimen in which the whole seed has remained intact within the matrix (Fig. 8f). We interpret it as the extra thickening of the testa along the lateral margin.
This seed is cordiform, with cup-shaped ventral infolds, faint ruminations radiating from the chalaza (preserved in some specimens), and a deeply incised median dorsal groove. It is similar to Central American A. acapulcensis (i.e., to those populations of this species having seeds having wide infolds; Fig. 6b) and A. erdvendbergiana (Fig. 6c) (see also Fig. 3). Seeds of A. auriforma are smaller than A. acapulcensis and are about the same size as A. erdvendbergiana (Fig. 4). The extra thickening of the testa at the lateral margins is not obvious in these two extant Central American species; however, it is observed in some Malesian species, which do not possess a prominent dorsal groove (for example, Pterisanthes cissoides, Fig. 6a, transverse section).
In addition to the specimens from the Clarno Formation, rare specimens from the Early or Middle Eocene Green River Formation in Colorado and Utah might belong to this species, e.g., DMNH 14109 (Fig. 8j) from Douglas Pass, Colorado, and UF 15755-8207 (also cited in Manchester, 1994
) from Evacuation Creek Canyon, Utah, USA. The specimens are preserved in shale as external molds of the ventral infold cavities, which appear as a pair of mounds in an impression with an outline similar to that of the Clarno specimens (e.g., Fig. 8i). However, the counterpart shale specimens, which would have shown the dorsal surface features of these fossils, were unfortunately lost or discarded in the field. In the characters preserved, the Green River specimens are consistent with A. auriforma; however, without knowing the shape of the chalazal knot we cannot confirm their assignment.
Ampelocissus bravoi (Berry) Berry (Fig. 8h)
Basionym: Carpolithus bravoiBerry (1927)
, p. 130, pl. xix, figs. 3, 3c
Ampelocissus bravoi (Berry) Berry (1929a)
, p. 161, pl. III, figs. 1, 1a
The holotype and only known specimen of A. bravoiBerry (1927)
from the Eocene of the Mancora Formation, at Belen, northwestern Peru, is housed at the Smithsonian Institution (USNM 318011). This fossil was correctly assigned to the genus by Berry (1929a)
. This seed is 8 mm long, 6.4 mm wide, and only 2.3 mm in dorsiventral thickness. Its ventral infolds are broad, like shallow dishes. The dorsal side of A. bravoi has an oval chalaza with radiating rugose marks. The chalaza-apex and chalaza-base grooves are shallow. This fossil is similar to type 2 Ampelocissus seed, for example, A. cavicaulis (Fig. 5b).
Ampelocissus chandleri (Kirchheimer) Chen & Manchester comb. n. (Fig. 8l)
Basionym Tetrastigma chandleriKirchheimer (1938)
, p. 593, figs. 17, 18
This species occurs in the Lower Miocene of Wiesa (Kirchheimer, 1938
), Hartau (Mai, 1964
), Upper Lusatica (Mai, 2000
; Czaja, 2003
), Germany, and Turow, Poland (Czeczott and Skirgiello, 1959
); in the Middle Miocene of theLusatica region (Mai, 2001
); and in the Miocene of other sites in central Europe (Mai, 2000
, 2001
). The specimens housed at UF (15794–28873, Lower Miocene of Wiesa) were observed (Fig. 8l). The seeds are oval, maximum length 7 mm, maximum width 5 mm, 3 mm thick, with a small stipitate beak. The ventral infolds are long, linear, irregular in outline, and parallel to each other. The margin of the ventral surface is rugose. The dorsal surface has a deep chalaza-apex groove, a shallower chalaza-base groove and an oval sunken chalaza with radiating ridges. The size and overall features of the seed correspond to Ampelocissus s.l. type 3 seeds. Kirchheimer (1939)
indicated that this seed is very similar to the older fossil species Tetrastigma lobata Chandler (discussed later) but is oval in shape, and he compared this fossil to Tetrastigma lanceolarium (Roxb.) Planch. (Fig. 7h), an extant species from southeastern Asia and Malesia. However, T. lanceolarium has a typical Tetrastigma seed with ventral infolds diverging in a V-shape, and with horizontal furrows flanking an elongate and linear chalaza (a detailed description of extant Tetrastigma seeds can be found in the preceding section). These characters are not present in A. chandleri.
Ampelocissus lobatum (Chandler) Chen & Manchester comb. n. (Fig. 8m)
Basionym Tetrastigma lobataChandler (1925–1926
), p. 32, pl. V, figs. 3a–c
This fossil occurs in the Lower Eocene of the London Clay and Upper Eocene of Hordle, England (Chandler, 1925–1926
, 1961a
, 1962
); in the Lower Miocene of Wiesa, Turow, and Hartau (Mai, 2000
); and in the Miocene of the Lusatica region and other sites in central Europe (Mai, 2000
, 2001
). According to Chandler's description, it is round, 5.5–9 mm long, 4–10 mm wide, 3 mm thick, usually with a stipitate beak and long, parallel ventral infolds. The dorsal surface has an oval sunken chalaza in the center, with radiating lobes or furrows (rugose marks). It is very similar to extant type 3 Ampelocissus s.l. seeds, for example, A. tomentosa (Fig. 2). As Kirchheimer (1938
, 1939
) mentioned, A. lobatum is very similar to A. chandleri, differing only in that A. lobatum is round rather than oval. Chandler (1925–1926
) compared this fossil to extant T. lanceolarium (Roxburgh) Planchon; the differences between extant Tetrastigma and Ampelocissus have been described in previous paragraphs.
Besides A. lobatum, at least seven more species were assigned to Tetrastigma from lower Tertiary beds of England (Chandler, 1925–1926
; 1961a
, b
, 1962
, 1978
; Reid and Chandler, 1933
). All of them have long, parallel ventral infolds (except T. davisii), ridges or rugose marks radiating from the oval and more or less sunken chalaza. These are the characters of type 3 Ampelocissus s.l. seeds. However, more investigation is needed to confirm their affinities because some of these fossils are very small and resemble extant Ampelopsis seeds.
We examined an additional specimen similar to A. lobatum from the Eocene of Messel, Germany (SM.B.Me 2352, fig. 8m). The seed is pyriform, 6.6 mm long, and 5 mm wide, with an inconspicuous apical notch and rounded beak. Its ventral infolds are long and linear, with rugose ventral margin. The chalaza on this specimen is round, positioned at the center of dorsal side. The chalaza-apex groove is inconspicuous and chalaza-base groove linear. The dorsal side is highly rugose, with seven pairs of ridges radiating from the edge of chalaza (Fig. 8m). The single specimen is highly compressed by the fossilization process, but its general features closely resemble the Miocene specimens of A. lobatum. This Eocene specimen is significantly older than the type material of the species and seems to be distinguished by the greater prominence and sharpness of the rugose marks on both surfaces.
Ampelocissus parachandleri sp. n. (Fig. 8k)
Description: seed oval, slightly dorsiventrally compressed, 6.7 mm long, 5 mm wide, 4 mm thick, apical notch present, beak small and triangular. Ventral infolds deep, furrow-like, outline irregular, almost as long as the full length of the seed. Margin of ventral side with faint rugose marks. Chalaza oval, large, 3.5 mm long, 1.7 mm wide, deeply sunken in the center of the dorsal side. Dorsal surface rugose, rugose marks radiate from the raised edge of the chalaza. Chalaza-apex groove present as a deep furrow, chalaza-base groove a shallower furrow.
Holotype: USNM 355132. Clarno Nut Beds, Oregon, 44°56.36 N, 120°25.34 W (UF locality 225); Early Middle Eocene.
Paratype: UF 50000, 50001; USNM 355133, 358096. Same locality as for holotype.
Etymology: para-L = aside, plus chandleri referring to the similarity with European fossil seeds assigned to Tetrastigma chandleri Kirchheimer.
Remarks: Nine seeds were identified as this species. All specimens were preserved as endosperm casts missing the testa; however, the testa is judged to be 0.1 mm thick from the space between matrix and the endosperm cast. The description is based on the cast of the endosperm and assumes that the testa was evenly thickened. The depth of the apical notch varies.
Comparative morphology: The most unusual feature of this fossil is the large and deeply sunken chalaza. The margin of chalaza is elevated high with rugose marks radiating outward. The linear infolds with irregular outlines indicate that the edges of the ventral infolds were probably ridged and that the margin of the ventral side was rugose. These features indicate a close resemblance to type 3 Ampelocissus seeds, for example A. grantii from Africa (Fig. 5e). In its oval shape, Ampelocissus parachandleri resembles seeds of the European fossil species, A. chandleri (Kirchheimer) Chen & Manchester, but the former differs by its larger deeply sunken chalaza, and less prominent rugose marks.
Ampelocissus parvisemina sp. n. (Fig. 8a–d)
Description: seed round to pyriform, round-triangular in transverse view; 3–4 mm wide, 3–4 mm high, and 2–3 mm thick; apical notch inconspicuous; beak triangular, sharply pointed. Ventral infolds broad, deep, cup-shaped, and smooth, occupying most of the area of the ventral surface. Dorsal surface smooth, some specimens have faint radiating marks; chalazal knot oval, positioned centrally on the dorsal side. A pair of suture lines running along the lateral margins of the dorsal side. Endosperm in cross section m-shaped, chalazal knot not sunken. Chalazal-apex and -base grooves present but very shallow. Testa 90 µm thick, testa on the lateral margins slightly thicker than on the dorsal and ventral faces.
Holotype: UF 34705. Silica Summit Butte at Beicegal Creek, North Dakota, Bullion Creek Formation of North Dakota, 47°22.37 N, 103°26.29 W (UF locality 18972); Paleocene.
Paratypes: UF 9622, 50003, 5004, 50005; USNM 434993, 434994, 434996, 434997 (these four specimens in the Smithsonian collection were cited by Manchester [1994]
among the paratypes of A. auriforma, but are now excluded from that species). Clarno Nut Beds, Oregon, 44°56.36 N, 120°25.34 W (UF locality 225); Early Middle Eocene.
Etymology: parvi-, L = small, semina, L = seed.
Remarks: Ampelocissus parvisemina is known both from the Late Paleocene of North Dakota and from the Early Middle Eocene of Oregon. Only one specimen was recovered from the type locality, but it is well preserved and is noteworthy as one of the oldest known fossil vitaceous seeds. The same seed type is present in larger numbers in the Early Middle Eocene of the Clarno Formation, Oregon. Originally these small fossil seeds were misidentified as A. auriforma (Manchester, 1994
), a common species with larger seeds and similar morphology from the same locality. When the width was plotted against the length, a clear gap was shown between these two (Fig. 9). They also can be distinguished by the depth of the chalaza and the prominence of median dorsal grooves. Ampelocissus parvisemina has shallow chalaza-apex and chalaza-base grooves, whereas A. auriforma has a prominent median groove running the full length of the dorsal side. Most of the specimens from the Clarno Formation preserve as the cast of the endosperm (or the internal mold of the seed coat); sometimes the cast of the endosperm is loosened and lost, leaving a mold impression in the sediment with a pair of small mounds corresponding to the ventral infold cavities (Fig. 8d).
|
Ampelocissus wildei sp. n. (Fig. 8n–p)
Description: seed oval to pyriform in ventral or dorsal view, large, 11–14 mm long, 7–9 mm wide, dorsiventral thickness uncertain because of compression of the fossils. Apical notch and beak not prominent. Ventral surface smooth; ventral infolds linear, parallel to slightly divergent from each other, and about one-half to two-thirds of the full seed length. Dorsal surface rugose, with 7–8 pairs of ridges radiating out from the edge of the elongate oval chalaza. Chalaza located at the center of the dorsal surface and about one-third of the full seed length. Chalaza-apex and base grooves present, linear.
Holotype: SM.B.Me 7271. Messel, Germany; Middle Eocene.
Paratypes: SM.B.Me 7371, 7552, 8205, 8786. Same locality.
Etymology: This species is named after Dr. Volker Wilde, recognizing his prior and continuing investigations of the Messel flora.
Remarks: This is a relatively common seed type at Messel. Fossils from these oil shales are highly compressed and fragile. Usually only the ventral or dorsal surface is exposed on the rock surface, and careful preparation is required to reveal the other surface. Me7271 was completely removed from the matrix, revealing both ventral and dorsal sides. The actual size of the seed could have been smaller and the actual length of the ventral infolds could be longer if the original shape of the seed was thicker, because compression from the fossilization process could have distorted the relative dimensions.
Comparative morphology: This seed has linear ventral infolds shorter than those of the majority of the extant Ampelocissus seeds, but this is characteristic of the type 3 Central American A. acapulcensis (Fig. 6b). The original thickness of the fossil seeds are unknown; nevertheless the size, shape and surface features of this specimen are similar to those of A. acapulcensis, except that the dorsal side of A. acapulcensis is not strongly rugose like the fossil.
Fossils excluded from Ampelocissus
Ampelocissus scottiiManchester (1994)
This fossil seed species from the Early Middle Eocene Nut Beds of Oregon is known from about eight specimens. It has very short ventral infolds, a feature not seen in extant Ampelocissus seeds. Hence we believe the identification to this genus to be in error but do not reassign it here. Short ventral infolds are more common in seeds of Ampelopsis and Vitis.
Ampelocissus similkameenensisCevallos-Ferriz and Stockey (1990)
This species from the Middle Eocene Princeton chert of British Columbia, Canada was assigned to Ampelocissus based on the anatomical features of the seed coat (Cevallos-Ferriz and Stockey, 1990
). However, seeds of most temperate genera of Vitaceae have testa anatomy similar to that of A. similkameenensis. Morphological features on the surface of the seeds are far more important for generic-level diagnosis. Based on the published pictures of the cross section of the seeds, this small fossil (2.5–3.5 mm wide, 2.5–2.6 mm thick; smaller than most extant Ampelocissus seeds) has linear ventral infolds and is not highly rugose. These features can be found in type 3 Central American A. acapulcensis; however, A. acapulcensis has relatively large seeds (>9 mm) and a much more sunken chalaza. The limited diagnostic features so far demonstrated for this fossil suggest that it is more similar to seeds of the extant temperate genera—Ampelopsis, Parthenocissus, and Vitis subg. Vitis.
Ampelocissites lytlensisBerry (1929b)
(Fig. 10)
|
Cayratia jungii (Gregor) Chen & Manchester comb. n. (Fig. 11a)
|
, p. 214, abb. 8, 9, taf. 20, figs. 1–4
Ampelocissus jungii (Gregor) Gregor (1984)
, p. 16, taf. 2, figs. 1, 2
This fossil seed from the Middle Miocene of Schwandorf, Germany has a pair of cup-shaped infolds, a linear chalazal knot, and several pairs of horizontal dorsal ridges (Fig. 11a). These are the diagnostic seed characters of extant Cayratia (Table 1). The sharp, irregular ridges on the lateral rim of the seed and the dorsal horizontal ridges are similar to the extant seed of Cayratia acris (F. Muell.) Domin from Australia and New Guinea (Fig. 11b). The raphal ridge of the fossil is shallow (see abb. 9 in Gregor, 1977
), which is comparable to another extant Australian species C. cardiophylla B. R. Jackes (see Fig. 3K in Jackes, 1987
). Süssenguth (1953)
set up two sections in Cayratia based on the shape of the cross section of the seed; T-shaped seeds belong to sect. Discypharia, and n-shaped seeds belong to sect. Koilosperma. These Cayratia spp. all have seeds characteristic of sect. Koilosperma; however, the raphal ridge is more or less raised. We are not aware of any species of Ampelocissus having a linear chalaza. Because it conforms in all the observed characters with extant Cayratia, we reassign it accordingly.
DISCUSSION
Extant seeds and implication of phylogeny
Seed morphology, inflorescence type, geographical distribution, and habitat are tightly linked in the extant species of Ampelocissus s.l. that we examined (Fig. 3). Type 1 seeds—flat and smooth with dish-shaped infolds—correspond with the panicles of spikes inflorescence type, and they belong to species from Malesian rainforest. Type 2 seeds—rugose with wide infolds—are from lax, elongate, thyrsoid inflorescences. Type 3 seeds—rugose with linear infolds—are correlated with more compact and cymose inflorescences. The type 2 and 3 seeds belong to species occurring in Africa, Asia, and Australia, usually in dry or cool habitat. Type 4a seeds—rotund and with round cup-shaped infolds—are associated with fleshy, laminar inflorescences of the Malesian rainforest Pterisanthes, whereas type 4b seeds—cordiform in shape—are presently confined to Central America and are not associated with a particular inflorescence type or habitat. Compared to floral structure, which is relatively constant throughout the genus, and characters of leaf type and indument, which are highly variable among species, features of both seed and inflorescence morphology are more useful for the delimitation of subclades in Ampelocissus s.l.
Morphological characters shared by the Malesia-centered Ampelocissus section Kalocissus (Miq.) Planchon, Nothocissus, and Pterisanthes suggest close relationships of these taxa: they all have tetramerous floral parts and shortly pedicellate or sessile flowers. Although Kalocissus has type 1 seeds and Pterisanthes has type 4 seeds, both types are smooth, with broad ventral infolds, and lack a median dorsal groove. Nothocissus spicifera, nevertheless, is a morphologically distinct species with type 3 Ampelocissus seeds. With the exception of A. helferi, which occurs on the Andaman Island of India and Myanmar (not sampled in this research), all of these taxa are restricted to Malesian rainforest. Their unusual inflorescences (panicle of spikes, dichotomous spikes, and laminar panicles) suggest specialized pollination phenomena or seed dispersal mechanism; however, no information of this kind was found in the literature.
The species from Africa, southern Asia, and Australia, i.e., sect. Ampelocissus, have either type 2 or type 3 seeds. This section has been divided into two subsections by the shape of inflorescences (Gilg and Brandt, 1911
): subsect. Paniculatae has elongate, lax, and thyrsoid inflorescences; and subsect. Cymosae has more compact and cymose inflorescences. Inflorescence shape is correlated with width of ventral infolds, as shown in Fig. 3b. However, intermediate forms of both characters exist, and there is no clear distinction between these two subsections. They are mostly from dry environments (Fig. 3d). In general, the species from south Asian high altitude deciduous forest tend to have smaller seeds with narrower ventral infolds and a less sunken chalaza, whereas those from Africa tend to have larger seeds with a deeply sunken chalaza. Although inflorescence and seed morphology correlate well in this group of Ampelocissus, exceptions occur: Seeds of A. barbata (OTU7), A. borneensis (OTU9), and A. martinii (OTU24) are finely rugose and have broad infolds apparently corresponding to type 1 Ampelocissus s.l. seeds. Ampelocissus pauciflora (OTU30) from dry coastal forest of Philippines has cymose inflorescences, but type 4a seeds that are similar to those of Pterisanthes (see Fig. 3). These deviations from the type 2 and type 3 seeds typical for sect. Ampelocissus may reflect the difficulty in clearly defining seed types—for example, OTU7 and OTU24 are positioned on the boundary of type 1 and type 2 seeds (Fig. 3a). Indeed, each of the Ampelocissus seed types overlaps to some extent, and the precise placement of the boundaries between them remains somewhat arbitrary. An alternative explanation is that these outliers truly represent the divergent morphology of that particular taxon in an otherwise morphologically coherent group, as seems to apply in the cases of OTU9 and OTU30.
The five species of Ampelocissus from Central America all have a prominent median dorsal groove in their seeds, but their seed size varies greatly (Fig. 4). Ampelocissus javalensis (Fig. 6d) has large seeds with a thick lignified testa and strong beak. Ampelocissus acapulcensis also has large seeds. It is the only species examined with such great variation in the width of the ventral infolds that its seeds were separated into two types, 3 and 4b (Figs. 6b and 7c). It is also the only Central American Ampelocissus species with seeds of type 3. Ampelocissus erdvendbergiana (Fig. 6c) has distinct cordiform seeds with round chalaza, and A. robinsonii (Fig. 6e) has the smallest seeds among all the seeds examined. Ampelocissus mesoamericana was not examined; however, based on the description of Lombardi (2005)
, it is very similar to A. erdvendbergiana, and it has similar small-sized, heart-shaped seeds. All the Central American Ampelocissus have more or less thyrsoid inflorescences; those of A. javanensis can be up to 30 cm long, while those of A. acapulcensis are usually smaller and more compact. Ampelocissus erdvendbergiana has a lax inflorescence with long lateral axes. Ampelocissus robinsonii, a species endemic to West Indies, however, has unusual inflorescences that led Planchon (1887)
to place this species in a separate section Eremocissus.
The small, linear, thyrse inflorescences of A. robinsonii suggest its close relationship to Vitis subg. Vitis. The tendril-associated inflorescences, one of the main features used to define the genus Ampelocissus, are also frequently observed in Vitis. Seeds of A. robinsonii are smaller than those of other Ampelocissus s.l. examined; in size and shape they resemble those of Vitis subg. Vitis. However, no Vitis seeds with wide ventral infolds were observed. Vitis rotundifolia of southeastern North American, on the other hand, has seeds that resemble type 3 Ampelocissus s.l. seeds, differing only because they are not as rugose (Fig. 7e). Vitis rotundifolia belongs to Vitis subg. Muscadinia, which is separated from other Vitis by chromosome number, inflorescence, and seed morphology (Brizicky, 1965
).
Other members of Vitaceae with seeds similar to those of Ampelocissus s.l. include Yua austro-orientalis and Ampelopsis denudata. Yua was separated from Parthenocissus because its tendrils are dichotomous and do not have suction pads, and its seed morphology differs from that of Parthenocissus. Our survey shows that seeds of Y. austro-orientalis and Y. thomsonii (Laws.) C. L. Li do not possess a deep apical notch like those of Parthenocissus. Yua thomsoni has smooth seeds; however, Y. austro-oreintalis has large rugose seeds similar to type 3 Ampelocissus s.l. seeds. Seeds of Y. chinensis C. L. Li were not observed, but based on the description by Li (1998)
, they are similar to those of Y. thomsoni. Central American Ampelopsis denudata has seeds similar to those of Central American A. erdvendbergiana and A. robinsonii; however, the chalaza is pyriform and close to the apical notch, a characteristic feature of Ampelopsis seeds (Table 1).
To summarize our morphological investigation on the extant species, we emphasize four main points: (1) those Malesian species with tetramerous flowers are probably closely related to each other; (2) the species from Africa, southern Asia, and Australia are morphologically similar; (3) the Central American species of Ampelocissus, especially A. robinsonii, seem to be related to Vitis subg. Vitis; and (4) Vitis subg. Muscadinia may be closer to Ampelocissus sect. Ampelocissus. A close relationship among Ampelocissus, Ampelopsis, Parthenocissus, and Vitis has been assumed by various authors based on morphology (Brizicky, 1965
