| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Paleobotany |
L. H. Bailey Hortorium, Department of Plant Biology, 228 Plant Science Building, Cornell University, Ithaca, New York 14853 USA
Received for publication December 17, 2002. Accepted for publication April 4, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSSYSTEMATICSDISCUSSIONLITERATURE CITED |
|---|
Key Words: angiosperm • Cretaceous • fossil • Itea • Iteaceae • Raritan Formation, New Jersey • Saxifragaceae • Turonian
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSSYSTEMATICSDISCUSSIONLITERATURE CITED |
|---|
), the group has been ill-defined and its relationships poorly understood until recently. Historically, the family Saxifragaceae, as defined by Engler (1930)
, included a broad array of woody and herbaceous taxa that share such features as bisexual flowers, a pentamerous perianth, bicarpellate gynoecium, axile or parietal placentation, and anatropous, bitegmic ovules. This suite of characters, however, was not characteristic of all Engler's Saxifragaceae, and no single character or suite of characters strongly united the family. While various authors postulated that Engler's Saxifragaceae was a highly heterogeneous grouping of taxa based on anatomical, morphological, and developmental evidence (Palmatier, 1942
; Klopfer, 1973
; Bensel and Palser, 1975a
, b
, c
, d
; Krach, 1976
, 1977
), it is only recently, with the advent of cladistic and molecular techniques, that the relationships among saxifragalean taxa have been tested phylogenetically (Morgan and Soltis, 1993
; Soltis and Soltis, 1997
; Fishbein et al., 2001
). As a result, many taxa within Engler's family are now thought to be only distantly related to one another. This is reflected by contemporary classification schemes (e.g., Stevens, 2001
), in which Engler's family Saxifragaceae is distributed among many different families and orders, and Saxifragaceae is reduced to a family of about 30 herbaceous genera roughly equivalent to Engler's (1930)
tribe Saxifrageae (hereafter referred to as Saxifragaceae sensu stricto [s.s.] as in Soltis and Soltis, 1997
). Those families now considered to have closest phylogenetic affinities to Saxifragaceae s.s., but formerly placed therein, include Grossulariaceae de Candolle (Ribes L. and, if recognized separately, Grossularia Miller), Iteaceae J. Agardh (Itea L. and Choristylis Harvey), and the unigeneric Pterostemonaceae Small (Morgan and Soltis, 1993
; Soltis and Soltis, 1997
; Fishbein et al., 2001
). This group of four families includes both woody and herbaceous species, most of which have bisexual flowers with a pentamerous perianth, one or two cycles of stamens, two carpels, and capsular fruit opening apically.
While a wealth of literature exists on fossils placed within Saxifragaceae and Saxifragales, or simply described as "saxifragaceous," the historic definition of the family and the long-running confusion over the relationships among taxa included within it has led to a broad concept of "saxifrage" throughout the paleobotanical literature. In this paper, we abandon that concept in introducing new fossil saxifrages from the Old Crossman Clay Pit Locality of the Raritan Formation, Sayreville, New Jersey, USA, for one that more accurately represents our current understanding of phylogeny as discussed above (Stevens, 2001
). While other saxifrages of Late Cretaceous age have been reported in the literature, the fossils introduced herein may be the oldest with a floral morphology clearly suggesting affinities to extant Saxifragaceae s.s. and Iteaceae, which heretofore had macrofossil records extending back to the Oligocene (Conwentz, 1886
) and Eocene (Wolfe and Wehr, 1987
), respectively.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSSYSTEMATICSDISCUSSIONLITERATURE CITED |
|---|
Extant material for pollen and morphological studies (see Supplemental Data accompanying the online version of this article) was removed from herbarium or preserved specimens, mounted on stubs, sputter coated with Au/Pd, and viewed with a Hitachi 4500 SEM or examined with a dissecting microscope.
Cladistic analyses
Cladistic analyses were performed using a combination of molecular and morphological data. Molecular data for five genes (atpB, matK, rbcL, 18S rDNA, and 26S rDNA; Fishbein et al., 2001
), including 40 taxa and 9237 characters, were downloaded from the Soltis laboratory website on 6 February 2002 (www.wsu.edu.8080/
soltilab/Saxifragales.txt). Data were unmodified save for conversion to a WinClada (Nixon, 2000
) readable format and fusion of the four Paeonia L. species into a single terminal in order to simplify coding of the morphology matrix (the analysis by Fishbein et al. [2001]
indicates that the Paeonia species fused in this analysis form a monophyletic group). A total of 37 morphological characters were scored (Table 1). Because the genus Saxifraga L. is thought to be polyphyletic based on prior molecular analyses (Soltis et al., 1993
, 1996
, 2001
), S. integrifolia W.J. Hooker and S. mertensiana Bongard were coded as for the species. Other extant taxa were coded as for the genus (Table 2). General references for coding of the morphology matrix include Cronquist (1981)
and Takhtajan (1997)
. Other references include: Aphanopetalum Endl., Dickison (1975
, 1994)
; Crassulaceae J. St.-Hil., Spongberg (1978)
, Endress and Stumpf (1991)
, Bartel (1993)
; Haloragaceae, Bentham and Mueller (1864)
, Praglowski (1970)
, Orchard (1979)
, Faegri (1982)
; Hamamelideae Takht. (incl. Cercidiphyllum Siebold et Zucc., Daphniphyllum Blume, Hamamelidaceae R. Br., Platanaceae Lestib., Tetracentraceae A. C. Sm., and Trochodendraceae Prantl), Vink (1958)
, Morley and Chao (1977)
, Bogle (1986
, 1989
), Endress (1986
, 1987
, 1989
, 1993a
, b
, c
), Hufford and Endress (1989)
, Huang (1997)
, Endress and Igersheim (1999)
; Iteaceae J. Agardh, Harvey (1859)
, Schneider (1906)
, Spongberg (1972)
, Verdcourt (1973)
, Bensel and Palser (1975b)
, Hideux and Ferguson (1976)
, Grierson (1987b)
, Endress and Stumpf (1991)
, Al-Shammary and Gornall (1994)
, Shuying and Ohba (2001)
; Lambertia Sm., Rao (1969)
; Leea Royen ex L., Ridsdale (1974)
; Paeonia L., Stern (1946)
, Keefe and Moseley (1978)
, Inamdar et al. (1983)
; Penthorum L., Spongberg (1972)
, Haskins and Hayden (1987)
, Endress and Stumpf (1991)
, Jintang and Soltis (2001)
; Pterostemon Schauer, Small (1905)
, Wilkinson (1994)
; Ribes L., Stern et al. (1970)
, Spongberg (1972)
, Bensel and Palser (1975b)
, Hideux and Ferguson (1976)
, Endress and Stumpf (1991)
, Gleason and Cronquist (1991)
, Lingdi and Alexander (2001)
; Saxifragaceae Juss. s.s., Rosendahl (1927)
, Hitchcock et al. (1961)
, Ferguson and Webb (1970)
, Spongberg (1972)
, Bensel and Palser (1975b)
, Hideux and Ferguson (1976)
, Wells (1984)
, Gornall and Bohm (1985)
, Grierson (1987a)
, Gornall (1989)
, Webb and Gornall (1989)
, Endress and Stumpf (1991)
, Jintang and Ohba (2001)
; Tetracarpaea Hook., Hils et al. (1988)
; Vitis L., Brizicky (1965)
. Direct observations were made on the specimens listed in the Supplemental Data accompanying the online version of this article.
|
|
). Before tree searches were conducted, uninformative characters were highlighted and deactivated using the "Mop uninformative characters" and "Deactivate selected characters" options under the "Characters" menu; this left a total of 1475 informative molecular characters and 35 informative morphological characters. Heuristic tree searches were performed under equal weights parsimony using NONA (Goloboff, 1998
) spawned from WinClada. NONA was set so that up to 10 000 trees could be held per analysis. For each analysis including molecular data, 100 replications were performed starting from a random Wagner tree, followed by tree bisection-reconnection (TBR) branch-swapping. Where morphological data alone were analyzed, the same procedure was followed but with 1000 replications. A maximum of 10 shortest trees were held per replicate, so that up to 1000 shortest trees were held for the analyses including molecular data and up to 10 000 shortest trees were held for the analyses using only morphological data during this step. Trees held from the replicates were then swapped to completion, and all unique most parsimonious trees (up to 10 000 unique trees for each analysis) were saved. Trees were viewed and strict consensus trees were constructed in WinClada; all unsupported nodes were collapsed both in most parsimonious trees and in consensus trees. Nine separate tree searches were performed, including the following combinations of taxa and characters: (1) extant taxa and molecular characters; (2) extant taxa and all characters; (3) all taxa (extant and fossil) and all characters; (4) extant taxa, Divisestylus brevistamineus, and all characters; (5) extant taxa, D. longistamineus, and all characters; (6) extant taxa and morphological characters; (7) all taxa and morphological characters; (8) extant taxa, D. brevistamineus, and morphological characters; (9) extant taxa, D. longistamineus, and morphological characters.
|
SYSTEMATICS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSSYSTEMATICSDISCUSSIONLITERATURE CITED |
|---|
Type species
Divisestylus brevistamineus Hermsen, Gandolfo, Nixon, et Crepet, sp. nov. here designated.
Etymology
The name Divisestylus is from the Latin adverb divise, "distinctly" or "separately," and adjective stylus, "styled," referring to the unique feature of this taxon wherein the styles remain free, whereas the carpels and stigmas are fused.
Generic diagnosis
Flowers bisexual or possibly functionally unisexual, actinomorphic, epigynous and pedicellate; sepals five, basally connate into a hypanthium, free above, with valvate aestivation; petals five, free, alternate with the sepals, with imbricate aestivation; stomata anomocytic; androecium haplostemonous, stamens/staminodes five, opposite sepals; anthers dithecal, tetrasporangiate, basifixed; pollen tricolpate, prolate, exine striate; gynoecium bicarpellate; ovary bilocular, inferior, placentation axile, more than 10 ovules per locule; intrastaminal nectary ring smooth; styles two, distinct and free proximally but fused distally, style canal open; stigmas capitate, fused into a bilobed platform; fruit a two-beaked dehiscent capsule, with stigmas remaining connate or separating upon dehiscence. Vegetative morphology unknown.
Generic description
Flowers bisexual or possibly functionally unisexual, actinomorphic, pentamerous, and epigynous, 0.4–1.2 mm in diameter, on pedicels up to 0.8 mm long (Figs. 1, 2, 11, 12, 14, 16). Sepals five, fused below into a hypanthium and free above, lobes triangular 0.2–0.5 mm wide x 0.4–0.9 mm high (Figs. 1, 2, 3, 11, 12, 14), separated by sinuses 0.06–0.1 mm wide (Figs. 2, 11), with acute trichomes on adaxial and abaxial surfaces (Figs. 2, 3, 12, 14), or trichomes lacking; aestivation valvate. Hypanthium 0.2–0.7 mm high, often bilobed and showing the position of the locules (Fig. 14), or sometimes with prominent nerves (Fig. 24), outer surface with acute trichomes (Figs. 2, 12, 14), or trichomes lacking. Stomata anomocytic (Fig. 4). Petals five, alternate with and wider than the sepals (Figs. 1, 7, 15), papillae or acute trichomes on adaxial and abaxial surfaces, on the abaxial surface, or trichomes lacking (Figs. 7, 8, 15); aestivation imbricate (Fig. 1). Androecium haplostemonous, of stamens with short, tapering filaments 0.06–0.1 mm wide at the point of attachment to the hypanthium by 0.03–0.2 mm high and basifixed, sagitate anthers 0.1–0.2 mm wide at their widest point by 0.1–0.2 mm high (Figs. 5, 6) or of ribbonlike structures possibly representing either staminodes or filaments (anthers unknown) 0.07–0.2 mm wide at the point of attachment to the hypanthium by up to 0.9 mm high (Figs. 11, 14, 16); androecial members five, borne opposite the sepals (Fig. 11). Pollen observed on 14 specimens is tricolpate and prolate, radially symmetrical about the polar axis, 4–10 µm wide on the equatorial axis by 6–20 µm high on the polar axis, sculpture striate (Figs. 28, 29). Gynoecium of two fused carpels, ovary inferior (Figs. 12, 14, 24, 25), biloculate, with numerous (more than 10) ovules on axile placentas (Fig. 20), ovules unornamented and sometimes folded over (Figs. 21, 22), indicating they are possibly anatropous (no structure representing a micropyle has been identified). Intrastaminal ring 0.1–0.9 mm wide, of smooth tissue (Fig. 16), interpreted as a nectary, surrounding densely papillate tissue at the base of the styles describing an oblong region 0.2–0.6 x 0.3–0.8 mm in diameter (Figs. 16, 17), papillate tissue grading into a sparse to dense surface of trichomes distally on the styles (Figs. 10, 19). Styles two, 0.1–0.2 mm wide x 0.3–0.9 mm high, free proximally and fused distally, sometimes appressed (Figs. 9, 18), canals open (Figs. 16, 23). Stigmas capitate, fused into a single, bilobed stigmatic platform (Figs. 9, 25, 18). Fruit a dehiscent capsule, splitting along the inner style sutures and opening apically (Figs. 23, 27), the stigmas remaining connate or becoming free (Figs. 25, 26), sepals and ribbonlike androecial structures sometimes persistent on the fruits (Figs. 26, 27). Vegetative morphology unknown.
|
|
|
|
Holotype
L. H. Bailey Hortorium Paleobotanical Collection CUPC 1340 here designated. Figs. 1, 2, 3, 4, 5, 7.
Paratypes
L. H. Bailey Hortorium Paleobotanical Collection CUPC 1341–1348.
Repository
Cornell University Paleobotany Collection, L. H. Bailey Hortorium, Department of Plant Biology, Cornell University, Ithaca, New York, USA.
Type locality
Old Crossman Clay Pit, Sayreville, New Jersey, USA.
Stratigraphic position
South Amboy Fire Clay, Raritan Formation.
Age
Turonian, Late Cretaceous.
Etymology
The epithet brevistamineus is from the Latin brevis, "short," and stamineus, "stamened," referring to the distinctive short stamens of this species.
Specific diagnosis
Petals verrucose, with papillae on adaxial and abaxial surfaces, or petals glabrous; androecium of stamens with short, tapering filaments; anthers basifixed, sagittate, dithecal and tetrasporangiate, with a conspicuous connective extending apically; styles fully appressed, bearing blunt trichomes distal to the region of papillate tissue, and not protruding above the tips of the erect sepal lobes.
Species description
Flowers bisexual, actinomorphic, pentamerous, and epigynous, 0.4–0.7 mm in diameter, on pedicels up to 0.4 mm long (Figs. 1, 2). Sepals five, fused below into a hypanthium and free above, lobes triangular 0.2–0.4 mm wide x 0.4–0.6 mm high, separated by sinuses 0.06–0.1 mm wide (Fig. 2), with acute trichomes on adaxial and abaxial surfaces (Figs. 2, 3, 12, 14), or trichomes lacking. Hypanthium 0.3–0.6 mm high, outer surface with acute trichomes (Fig. 2). Petals five, spade shaped, alternate with the sepals (Figs. 1, 7), papillae on adaxial and abaxial surfaces giving them a warty appearance (Fig. 8), or papillae lacking, 0.15 mm in width at their widest point and 0.2–0.3 mm high. Androecium haplostemonous, stamens five, borne opposite the sepals; filaments short, blunt, and tapering, 0.06–0.1 mm wide at the point of attachment to the hypanthium by 0.03–0.2 mm high (Fig. 6); anthers basifixed, sagittate, dithecal and tetrasporangiate, 0.1–0.2 mm wide at their widest point by 0.1–0.2 mm high from the base of the anther sacs to the tip of the connective (Figs. 5, 6), connective large and conspicuous, extending distally, with a surface covered by polygonal cells (Fig. 5). One pollen grain of the type described for the genus has been found on the external calyx surface of a single specimen. Gynoecium as for the genus. Intrastaminal ring 0.1 mm wide, of smooth tissue, interpreted as a nectary, encircling an oblong region of densely papillate tissue at the base of the styles 0.2–0.4 x 0.3–0.4 mm in diameter, papillate tissue grading into a sparse covering of blunt trichomes distally on the styles (Fig. 10). Styles two, 0.1–0.2 mm wide x 0.3–0.4 mm high, free proximally but fused distally, appressed, not protruding above the distal ends of the erect sepal lobes (Fig. 9). No fruits are definitively assignable to this species, but assumed to be as for the genus.
Divisestylus longistamineus
Hermsen, Gandolfo, Nixon, et Crepet sp. nov.
Holotype
L. H. Bailey Hortorium Paleobotanical Collection CUPC 1349 here designated. Figs. 11, 12.
Paratypes
L. H. Bailey Hortorium Paleobotanical Collection CUPC 1350–1355, 1357, 1361–1399.
Repository
Cornell University Paleobotanical Collection, L. H. Bailey Hortorium, Department of Plant Biology, Cornell University, Ithaca, New York, USA.
Type locality
Old Crossman Clay Pit, Sayreville, New Jersey, USA.
Stratigraphic position
South Amboy Fire Clay, Raritan Formation.
Age
Turonian, Late Cretaceous.
Etymology
The epithet longistamineus is from the Latin longus, "long," and stamineus, "stamened," referring to the long filaments or staminodes that distinguish this species.
Specific diagnosis
Petals with acute trichomes on their abaxial surfaces; androecium of ribbonlike structures possibly representing stamens or staminodes, anthers unknown; styles appressed to widely spaced, bearing flat trichomes distal to the region of papillate tissue and protruding above the tips of the erect sepal lobes.
Species description
Flowers bisexual or possibly functionally unisexual (anthers unknown), actinomorphic, pentamerous, and epigynous, 0.6–1.2 mm in diameter, on pedicels up to 0.5 mm long (Figs. 11, 14, 16, 26). Sepals five, fused below to form a hypanthium and free above, lobes triangular 0.2–0.5 mm wide x 0.5–0.9 mm high (Figs. 11, 12), separated by sinuses 0.06–0.1 mm wide (Fig. 12), with acute trichomes on adaxial and abaxial surfaces (Fig. 13), or trichomes lacking. Hypanthium 0.2–0.7 mm high, often bilobed and clearly showing the position of the locules (Fig. 14), outer surface with acute trichomes (Figs. 12, 14), or trichomes lacking. Petals five, shape unknown, alternate with the sepals (as inferred from the positions of partially preserved petals opposite sinuses between sepals; Fig. 15), with acute trichomes on the abaxial surface in a longitudinally aligned band, outer margins glabrous (Fig. 15), up to 0.8 mm wide and at least 0.5 mm long (full length unknown). Androecium haplostemonous, ribbonlike structures representing staminodes or stamens with ribbonlike filaments five, borne opposite the sepals, 0.07–0.2 mm wide at the point of attachment to the hypanthium by up to 0.9 mm long (Figs. 11, 14, 16); anthers unknown. Pollen of the type described for the genus known from eight specimens (Fig. 28). Gynoecium as for the genus. Intrastaminal ring 0.1–0.3 mm wide of smooth tissue, interpreted as a nectary ring, encircling an oblong region of densely papillate tissue at the base of the styles 0.3–0.6 x 0.4–0.7 mm in diameter (Figs. 16, 17), papillate tissue grading into a dense covering of flat trichomes distally on the styles (Fig. 19), trichomes becoming sparse toward the stigmas. Styles two, 0.1–0.2 mm wide x 0.5–0.9 mm high, free proximally but fused distally, appressed to widely spaced, protruding above the distal ends of the erect sepal lobes (Figs. 12, 18). Fruit a two-beaked capsule opening apically, sometimes with persistent sepals and androecial structures (Figs. 26, 27).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSSYSTEMATICSDISCUSSIONLITERATURE CITED |
|---|
). Characteristics common among all members of Iteaceae, some Saxifragaceae s.s., and Divisestylus include a pentamerous calyx united below to form a hypanthium, a pentamerous corolla, a pentamerous androecium with stamens (or staminodes) situated opposite the calyx lobes, a bicarpellate gynoecium, axile placentation, presence of style canals, and capsulate fruit opening apically along the style sutures (Spongberg, 1972
; Bensel and Palser, 1975b
; Takhtajan, 1997
). Ovary position in Iteaceae and Saxifragaceae s.s. varies (Spongberg, 1972
; Bensel and Palser, 1975b
; Takhtajan, 1997
), and both families reportedly have species that can produce unisexual flowers, though hermaphroditic flowers represent the predominant condition (for exceptions, see Small and Rydberg, 1905
; Knuth, 1908
; Ohwi, 1965
; Verdcourt, 1973
; Elvander, 1982
; Webb and Gornall, 1989
; Shuying and Ohba, 2001
). The primary differences between members of Iteaceae and Saxifragaceae s.s. with similar floral morphology include habit, pollen type, degree of fusion of the carpels, and in the stamens, anther attachment position. Other families within the Iteaceae-Saxifragaceae s.s. clade, including Pterostemonaceae and Grossulariaceae, differ markedly from the fossils, the former in having a pentamerous gynoecium, filaments with appendages, and dicyclic androecium, the latter in having parietal placentation and fleshy fruit (Takhtajan, 1997
).
Iteaceae and Saxifragaceae s.s. are similar in perianth structure. In both families the calyx is usually formed by five sepals united below to form a hypanthium adnate (at least proximally) to the carpels and free above; the corolla is composed of five petals alternating with the sepal lobes and inserted on the rim of the hypanthium (Hitchcock et al., 1961
; Spongberg, 1972
; Wells, 1984
). Symmetry of the flowers in each family is usually actinomorphic, though there is a tendency toward zygomorphy in some Saxifragaceae s.s. (Wells, 1984
; Gornall and Bohm, 1985
; Takhtajan, 1997
). Both sepal and petal aestivation in Iteaceae can be valvate, whereas in Saxifragaceae s.s. they can be imbricate or valvate (Britton, 1905
; Spongberg, 1972
; Gornall and Bohm, 1985
; Takhtajan, 1997
). Fossils of Divisestylus clearly show imbricate petal aestivation (Fig. 1), and sepal aestivation is interpreted as valvate, given the shape, alignment, and spacing of the sepals (Figs. 2, 12).
The androecium is haplostemonous in Iteaceae and Divisestylus and haplostemonous or obdiplostemonous in Saxifragaceae s.s. (Spongberg, 1972
; Takhtajan, 1997
). In Iteaceae and most haplostemonous Saxifragaceae s.s., a single whorl of five stamens is situated opposite the calyx lobes (Hitchcock et al., 1961
; Spongberg, 1972
; Wells, 1984
; Gornall and Bohm, 1985
; Takhtajan, 1997
), as it is in Divisestylus (Fig. 11). Anthers in Iteaceae, Saxifragaceae s.s., and D. brevistamineus are tetrasporangiate (Cronquist, 1981
; Fig. 7). Anthers of Iteaceae are dorsifixed, whereas those of Saxifragaceae s.s. are basifixed to slightly dorsifixed with the distal end of the filament inserted in a pit at the base of the anther, a condition characteristic of Saxifragaceae s.s. and Crassulaceae (Endress and Stumpf, 1991
; sometimes referred to as centrifixed, Baum and Leinfellner, 1953
). Anthers of D. brevistamineus are clearly basifixed (Fig. 8), though they show no evidence of a basal pit (Figs. 7, 8). Anthers vary from sagittate to X-shaped in Saxifragaceae s.s., with or without shallow ventral and dorsal furrows, whereas anthers are reported as X-shaped in Iteaceae with prominent ventral furrows (Endress and Stumpf, 1991
). Anthers in both Saxifragaceae s.s. and Iteaceae have thick connectives (Endress and Stumpf, 1991
). An inconspicuous apical connective protrusion occurs sometimes in Saxifragaceae s.s. (Endress and Stumpf, 1991
) and Itea virginica L. (Schneider, 1906
; E. J. Hermsen, personal observation) and a conspicuous apical connective protrusion has been reported for the monospecific genus Choristylis (Verdcourt, 1973
). Divisestylus brevistamineus anthers are sagittate with a thick connective that protrudes apically and have no dorsal or ventral furrows (Fig. 7). The mode of dehiscence in Iteaceae and Saxifragaceae s.s. is by longitudinal slits (Endress and Stumpf, 1991
; Takhtajan, 1997
), though it has not been observed for D. brevistamineus. Filament length relative to other floral structures varies in both Saxifragaceae s.s. and Iteaceae (Harvey, 1859
; Franchet, 1896
; Gagnepain, 1916
; Rosendahl, 1927
; Chun, 1934
; Yamamoto, 1937
; Wu, 1940
; Hitchcock et al., 1961
; Verdcourt, 1973
; Roxburgh and Wallich, 1975
; Wells, 1984
; Gornall and Bohm, 1985
; Rehder et al., 1988
), encompassing the variation displayed between D. longistamineus and D. brevistamineus specimens (Figs. 8, 9).
Species of Iteaceae are not reported as having staminodes, though they do occur within some Saxifragaceae s.s. (Saxifragodes D. M. Moore, Heuchera L., and Mitella L.), usually along with functional stamens (Hitchcock et al., 1961
; Spongberg, 1972
; Moore, 1983
). Within Saxifragaceae s.s., Tanakaea radicans Fr. et Sav. and some Saxifraga section Ciliatae Haworth are dioeceous (Ohwi, 1965
; Webb and Gornall, 1989
), polygamodioecy occurs in Astilbe Buch.-Ham. ex D. Don (Small and Rydberg, 1905
), gynodioecy is exhibited in other Saxifraga species (though not consistently; Knuth, 1908
; Elvander, 1982
; Stevens and Richards, 1985
; Webb and Gornall, 1989
), and some species are protandrous with deciduous anthers (Spongberg, 1972
). Within Iteaceae, flowers of Choristylis can achieve unisexuality through abortion of the stamens (Harvey, 1859
). The androecium of D. longistamineus may by composed of staminodes, representing a dioecious, monoecious, gynodioecious, or polygamous syndrome, or stamens with caducous anthers, or the anthers may simply have been lost during fossilization.
Although no pollen has been found inside anthers of D. brevistamineus, pollen of the type described for the genus has been found variously on the calices, nectaries, papillate tissue at the base of the styles, ribbonlike filaments/staminodes, and stigmas of 14 individual specimens of Divisestylus. These include five specimens assigned to Divisestylus sp. (CUPC 1359, 1421, 1430, 1456, 1472; Fig. 29), eight specimens assigned to D. longistamineus (CUPC 1354, 1357, 1365–1367, 1376–1378; Fig. 28), and one specimen assigned to D. brevistamineus (CUPC 1340). Over 80 individual Divisestylus-type pollen grains are known from these specimens, between one and ca. 30 per specimen (Figs. 28, 29). Colpate pollen is also found on three addition specimens of Divisestylus (CUPC 1358, Divisestylus sp., one grain; CUPC 1351, D. longistamineus, one grain; CUPC 1441, Divisestylus sp., ca. two grains), though on these specimens a definitive interpretation of exine ornamentation cannot be made. No other pollen is known. Because tricolpate, striate pollen is found on numerous Divisestylus specimens, it is reasonable to infer that the pollen produced by Divisestylus sp. flowers was of this type. The pollen morphology of Saxifragaceae s.s. includes types similar to that found on Divisestylus specimens in being tricolpate or tricolporate with a striate surface sculpture (Ferguson and Webb, 1970
; Hideux and Ferguson, 1976
). Pollen of Iteaceae is of a distinctive diporate, psilate form (Fig. 30; Erdtman, 1955
; Wakabayashi, 1970
; Hideux and Ferguson, 1976
; Takhtajan, 1997
), a morphological syndrome characteristic of all Itea pollen examined for this paper (save for that of one I. macrophylla Wall. specimen [A. D. E. Elmer 18023], which had some surface sculpture). This pollen type appears to unite extant Itea and Choristylis (Erdtman, 1955
; Pastre and Pons, 1973
) and, in the fossil record, is first reported from the Eocene of Europe (Graus-Cavagnetto, 1976
; Muller, 1981
).
The gynoecium of Divisestylus bears a closer resemblance to those of modern Iteaceae species than to those of Saxifragaceae s.s. species primarily due to its unique fusion. While Iteaceae often have carpels, styles, and stigmas that are fully fused, sometimes the styles are free, whereas carpels and stigmas, respectively, are connate (Harvey, 1859
; Franchet, 1896
; Gagnepain, 1916
; Verdcourt, 1973
; Bensel and Palser, 1975b
; Shuying and Ohba, 2001
). This characteristic is only reported elsewhere in angiosperms within extant Apocynaceae Juss. (including Asclepiadoideae), to which Divisestylus clearly bears little resemblance, as Apocynaceae are marked by a sympetalous corolla with adnate stamens and frequently by connate anthers (Takhtajan, 1997
). While it is true that most species of Iteaceae are reported to have fully fused styles (Hu, 1925
; Chun, 1934
; Wu, 1940
; Spongberg, 1972
; Roxburgh and Wallich, 1975
; Grierson, 1987b
; Rehder et al., 1988
; Shuying and Ohba, 2001
), even among these, splitting of the capsule takes place along the inner style sutures, and the styles diverge so that the capsule opens apically, often without a separation of the stigmas (Schneider, 1906
; Spongberg, 1972
; Verdcourt, 1973
; Takhtajan, 1997
; E. J. Hermsen, personal observation). This results in a fruit that is nearly indistinguishable from that of some known Divisestylus specimens (Figs. 25, 34). In contrast to carpels of Iteaceae and Divisestylus, the carpels in members of Saxifragaceae s.s. are often only fused below with no fusion of the styles or stigmas, and the fruit is a two-beaked capsule or sometimes two follicles (Hitchcock et al., 1961
; Wells, 1984
; Gornall and Bohm, 1985
; Webb and Gornall, 1989
; Takhtajan, 1997
). Both Saxifragaceae s.s. and Iteaceae can have open style canals as observed in Divisestylus (Bensel and Palser, 1975b
; Figs. 16, 23). Axile placentation occurs in all Iteaceae and some Saxifragaceae s.s., and ovules in both families are anatropous, often numerous (more than 10) and usually bitegmic, though they can be unitegmic (Spongberg, 1972
; Takhtajan, 1997
). In Divisestylus, placentation is clearly axile and ovules are numerous (Fig. 20); however, the position of the micropyle and number of ovule integuments have not been observed.
Conspicuous ringlike intrastaminal nectaries as observed in Divisestylus are common in both Saxifragaceae s.s. and Iteaceae, though not ubiquitous (Britton, 1905
; Chun, 1934
; Hitchcock et al., 1961
; Spongberg, 1972
; Bensel and Palser, 1975b
; Roxburgh and Wallich, 1975
; Huang and Huang, 1977
; Wells, 1984
; Gornall and Bohm, 1985
; Rehder et al., 1988
; Shuying and Ohba, 2001
). On Itea yunnanensis Franchet, one of two extant Itea flowers examined with the SEM (see Supplemental Data accompanying the online version of this article), the nectary ring surrounds a region of papillate tissue at the base of the styles, as in Divisestylus (Figs. 16, 17, 31, 33). Notably, petals of I. yunnanensis display a knobby surface similar to that observed in some specimens of D. brevistamineus (Figs. 6, 32).
Trichomes can occur on any floral surface in Saxifragaceae s.s. and Iteaceae, including the calyx, corolla, filaments, anthers, carpels, and styles. These trichomes may be uniseriate, multiseriate, or multiseriate with globular heads (glandular hairs), and may be useful in taxonomy (Bensel and Palser, 1975b
; Wells, 1984
; Gornall and Bohm, 1985
; Gornall, 1986
, 1989
; Al-Shammary and Gornall, 1994
). Divisestylus appears to lack glandular trichomes, but has simple trichomes that have been observed variously on sepals, petals, and styles (Figs. 3, 6, 10, 13, 15, 19). Stomata in Divisestylus are anomocytic (Fig. 4), also the common condition in Saxifragaceae s.s. (Takhtajan, 1997
). In Iteaceae, stomata are paracytic (Takhtajan, 1997
).
Cladistic analyses
Analysis of extant taxa (Table 2) using molecular data alone resulted in two shortest trees of length 4944 steps (consistency index [CI] = 44, retention index [RI] = 58); the same data analyzed by Fishbein et al. (2001)
resulted in two most parsimonious trees of length 5954 steps (CI = 53, RI = 63; however, the Paeonia terminal was unfused in the Fishbein et al. analysis). The strict consensus of the two most parsimonious trees found in each analysis under equal-weights parsimony has nearly the same topology. The only difference is in the placement of Lambertia Smith, which groups ([Platanus L., Lambertia] [all other taxa]) in the Fishbein et al. analysis (Fig. 3 in Fishbein et al., 2001
), but (Platanus [Lambertia, all other taxa]) in our analysis. Addition of morphological data did not change the topology of the trees found using the molecular data; the two trees found including all extant taxa, molecular data, and morphological data are of length 5095 steps (CI = 44, RI = 58).
Analysis of all taxa (fossil and modern) and all characters (molecular and morphology) together resulted in six most parsimonious trees 5095 steps long (CI = 44, RI = 58). In these trees, the branching pattern of the extant taxa is unchanged from the prior analyses, save for within the Pterostemonaceae-Iteaceae clade. Without inclusion of the fossil taxa, this clade has the pattern (Pterostemon [Itea, Choristylis]). With inclusion of the fossil taxa, the clade is resolved in one of three ways, each represented in two of the six most parsimonious trees: (1) (Divisestylus brevistamineus, D. longistamineus, Pterostemon [Itea, Choristylis]); (2) (D. brevistamineus, D. longistamineus [Pterostemon {Itea, Chorstylis}]); or (3) (D. brevistamineus [Pterostemon, D. longistamineus {Itea, Choristylis}]). In the strict consensus of all most parsimonious trees, the fossils form a polytomy with Pterostemon and Iteaceae (Fig. 35).
|
] for a discussion of fossils and missing data in parsimony analyses). When D. brevistamineus was analyzed with the extant taxa, two most parsimonious trees 5095 steps long (CI = 44, RI = 58) were found. In both of these trees, branching order of the Pterostemonaceae-Iteaceae clade is (D. brevistamineus [Pterostemon {Itea, Choristylis}]). When D. longistamineus was analyzed with the extant taxa, two most parsimonious trees 5095 steps long (CI = 44, RI = 58) were also found, but in these trees, the position of Pterostemon relative to the fossil taxon is unresolved and the branching order of the clade is (D. longistamineus, Pterostemon [Itea, Choristylis]). Thus, while the larger numbers of missing data present for the fossil taxa when analyzed together do not alter their placement relative to the extant taxa significantly, analyzing the two fossils together results in greater resolution in the position of D. longistamineus relative to the extant taxa in two out of the six most parsimonious trees found, and less resolution in the position of D. brevistamineus in two out of the six most parsimonious trees found when compared to the single fossil analyses. The results of all analyses indicate, however, that the fossil taxa group within the Pterostemonaceae-Iteaceae clade, but not within extant Iteaceae. Analysis of the morphology matrix without molecular data produced trees that are largely unresolved. When the extant taxa were analyzed alone, 138 trees 121 steps long were found (CI = 37, RI = 64); in the strict consensus of these trees, Platanus and Daphniphyllum Blume branch below a large basal polytomy in which several clades are resolved, one including Chrysosplenium L. and Heuchera L., one containing all Crassulaceae J. St.-Hil., one consisting of Exbucklandia R. W. Br., Rhodoleia Champ. ex Hook., Altingia Noronha, and Liquidambar L., one of Tetracentron Oliver and Trochodendron Siebold et Zucc., and one of Iteaceae and Ribes. Analysis of all taxa and morphological characters produced six trees 121 steps long (CI = 37, RI = 65) and resulted in greater resolution of the strict consensus tree, where the fossils form a polytomy with the clade containing Itea, Choristylis, and Ribes (Fig. 36). The fossil taxa were again analyzed alone to determine whether their interaction was affecting their placement. When D. brevistamineus was analyzed together with the extant taxa, six trees 121 steps long were found (CI = 37, RI = 65); when D. longistamineus was analyzed with the extant taxa, 16 trees 121 steps long were found (CI = 37, RI = 64). In all trees, placement of the fossil is sister to the clade containing Itea, Choristylis, and Ribes. The strict consensus of trees including only D. brevistamineus is identical in the branching order of extant taxa to that found using both fossils (Fig. 36); the strict consensus of trees analyzed with only D. longistamineus is less resolved, with Haloragis, Myriophyllum, and Pterostemon collapsing into the large basal polytomy.
|
; Crepet et al., 2001
). Other elements of the flora include a moss (Polytrichaceae), two species of ferns (Gleicheniaceae R. Br. and Schizaeaceae Kaulf.), representatives of four families of gymnosperms (Cheirolepidiaceae Nakai, Pinaceae Sprengel ex Rudolphi, Cupressaceae Gray, Taxodiaceae Saporta), and numerous angiosperms, including fossils with affinities to modern Chloranthaceae R. Br. ex Sims, Lauraceae Juss., Clusiaceae Lindley, Triuridaceae Gardner, Hydrangeaceae Dumort., and Ericaceae Juss., among others (Crepet et al., 2001
). Modern Itea is an eastern North American–eastern Asian disjunct taxon with one species in the southeastern United States and the remainder in Asia, ranging from Japan to tropical regions in India, Java, and the Philippines (Spongberg, 1972
). In the United States, I. virginica can be found in wet habitats such as swamp forests, moist woodlands, and river banks (Spongberg, 1972
); in Asia, Itea species occur in a variety of forested and streamside habitats (Shuying and Ohba, 2001
). Choristylis is distributed in tropical east Africa to South Africa (Spongberg, 1972
; Verdcourt, 1973
). Most Saxifragaceae s.s. are found in the Northern Hemisphere in temperate, cold-temperate, and alpine zones (Spongberg, 1972
).
Modern Saxifragaceae s.s. and Iteaceae are thought to be insect-pollinated, though as Webb and Gornall (1989)
point out, Saxifraga-type flowers do not appear to be specialized to any particular type of insect, and plants may at times reproduce by selfing or by bulbils, runners, or offsets. Among the animals observed on flowers of Saxifragaceae s.s. are bees, wasps, beetles, flies, butterflies, moths, and snails (Knuth, 1908
; Taylor, 1965
; Webb and Gornall, 1989
). Itea virginica is thought to be entomophilous and has been described as "honeybee-associated" (Lieux, 1982
). Given the similarity in morphology between Divisestylus flowers and the flowers of Saxifragaceae s.s. and Iteaceae, as well as the presence of what is interpreted as a nectar-secreting ring, it is likely that Divisestylus flowers were also entomophilous. Grimaldi et al. (2000)
have identified a number of potential insect pollinators in Turonian New Jersey amber, including wasps, primitive Lepidoptera, beetles, and brachyceran flies.
Conclusions
The combination of comparative morphological study and cladistic analyses suggests that Divisestylus has close affinities to extant Iteaceae. The floral morphology characteristic of Divisestylus is nearly indistinguishable from that of Iteaceae, save for the anther attachment, which is basifixed in D. brevistamineus and dorsifixed in Iteaceae, and the type of stomatal apparatus. Divisestylus also shares significant similarities in floral morphology with Saxifragaceae s.s., and its inferred pollen morphology is more similar to that of taxa within Saxifragaceae s.s. than Iteaceae, which is characterized by its unique diporate pollen. Because both Divisestylus species group closer to Iteaceae than to Saxifragaceae s.s. in cladistic analyses (Figs. 35, 36), however, the fossil taxa appear to share a more recent common ancestor with the former family than the latter. Notably, a phenomenon wherein the floral morphologies of a fossil taxon and the extant taxon(a) to which it has closest affinities are remarkably similar, but the pollen morphology of the fossil taxon is tricolpate and that of the extant taxon(a) more derived is also observed in at least one other element of the Raritan Flora, Microaltingia apocarpela (Zhou et al., 2001
). Whether this phenomenon is generalized and represents a trend wherein adaptation of pollen morphology lags floral morphology remains to be confirmed.
Although other Late Cretaceous (Santonian, ca. 86–83.5 mya) fossils, including Scandianthus Friis et Skarby, Silvianthemum Friis, and an unnamed flower, all of Sweden, have been compared to Iteaceae (Friis and Skarby, 1982
; Friis, 1990
), Divisestylus is the oldest fossil taxon with clear affinities to the family. The precise affinities of Scandianthus and the unnamed flower remain undetermined (Friis and Skarby, 1982
), while a cladistic analysis by Backlund (1996)
shows that Silvianthemum is related to the asterid genus Quintinia A. DC. Thus, the occurrence of Divisestylus in the Turonian sediments of New Jersey pushes the record of the Iteaceae lineage, previously known only from the Eocene to the Recent, back more than 35 my. The next oldest Itea-type fossil is a leaf taxon from the Eocene Republic Flora, Washington, USA, assigned to the genus Itea (Wolfe and Wehr, 1987
, 1991
; Wehr and Hopkins, 1994
). Other macrofossils are younger and known only from Eurasia. Flower and fruit remains include Ademanthemum iteoides Conwentz, an iteoid flower preserved in Oligocene amber (Conwentz, 1886
), and the Miocene Itea europaea Mai, represented by multiple compressed fruits and seeds (Mai, 1985
; Pingen, 1987
, as Ericaceae sp. Taf. 6, Figs. 1–3; Dorofeev and Velichkevich, 1988
, as Carpolithus sp. 1–4, Abb. 43 (1–6) and Taf. XL, Figs. 1–8; Pingen, 1996
; Mai and Palamarev, 1997
). Two leaf taxa, I. Faujasii (Unger) Meschinelli et Squinabol (originally Cedrela Faujasii Unger) and I. transsilvanica Petrescu et Givulescu, are reported from the Tertiary of Italy and Romania, respectively (Unger, 1845
, 1850
; Meschinelli and Squinabol, 1892
; Petrescu and Givulescu, 1986
; Bertoldi et al., 1994
). There are more reports from the microfossil record for Itea than from the macrofossil record, probably because dispersed diporate, psilate pollen grains are often easily ascribed to Iteaceae. Pollen reports of Itea, Iteapollis Ziemb.-Tworz., and Iteapollenites (evidently a misspelling of Iteapollis; Petrescu and Givulescu, 1986
) span the Eocene to Pliocene in Europe, with several reports from the Oligocene and Miocene of North America (Traverse, 1955
as Corylus?, Fig. 9 [35]; Stuchlik, 1964
as Corylus americana, Plate XI [7–8]; Schneider, 1965
as Psilodiporites angustiporatus, Taf. 1 [8–10]; Ziembi
ska and Niklewski, 1966
; Wolfe, 1970
; Petrov and Drazheva-Stamatova, 1972
; Sadowska, 1973
, 1977
; Ziembiska-Tworzyd
o, 1974
; Graus-Cavagnetto, 1976
; Menke, 1976
; Muller, 1981
; Ziembiska-Tworzydo and Wa
yska, 1981
; Petrescu and Kolovos, 1982
; Dyjor and Sadowska, 1986
; Petrescu and Givulescu, 1986
; Rylova, 1989
; Kohlman-Adamska, 1993
; Bertoldi et al., 1994
).
|
FOOTNOTES |
|---|
