Floral Development and Morphology of Vochysiaceae. I.

The Structure of the Gynoecium

Author(s): Amy Litt and Dennis W. Stevenson
Source: American Journal of Botany, Vol. 90, No. 11 (Nov., 2003), pp. 1533-1547
Published by: Botanical Society of America
Stable URL: http://www.jstor.org/stable/4123710 .
Accessed: 08/04/2011 17:29

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=botsam. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Botanical Society of America is collaborating with JSTOR to digitize, preserve and extend access to American
Journal of Botany.

http://www.jstor.org
AmericanJournalof Botany90(11): 1533-1547. 2003.

FLORAL DEVELOPMENT AND MORPHOLOGY OF

VOCHYSIACEAE. I. THE STRUCTURE OF
THE GYNOECIUM'

AMY LITT2 AND DENNIS W. STEVENSON

New YorkBotanicalGarden,Bronx,New York10458USA

Vochysiaceaeare dividedinto two tribeson the basisof ovarystructure(superiortrilocularor inferiorunilocular). The superior
trilocularovaryhas been consideredbasalin the family.and the term "pseudomonomerous" was used to indicatethe presumed
evolutionaryderivationof the unilocularconditionfromthe trilocular.However.recentevidencethatVochysiaceaeare Myrtalean
suggeststhatthe superiorovarymay be secondarilyderived.In addition,publishedfigurescast doubton the interpretation of the
putativelyunilocularovaries.To understand these features.floralontogenyand anatomywere examinedusing scanningelectron
microscopy andserialsectioning.Inall taxaexamined,theovarydevelopsin anepigynousfashion,on a concavefloralapex,supporting
thehypothesisthatthesuperiorovaryis secondarily derived.Subsequent growthresultsin ovaries
to initiationof theovary,differential
thatare superior,inferior,or partlyinferiorin differentgenera.Sectionsof floralbuds of the two uniloculargenera.Erismaand
Erismadelphus, show abortedloculesin the latterbut not in the former.The applicationof the term"pseudomonomerous" to both
generaobscuresthis significantdifference.Thepositionof the placentain the trulyuniloculargenusvariesamongspecies,suggesting
a charactertransformationseriesfrommultilocular throughintermediatesto trulyunilocular.

Key words: floraldevelopment;

gynoecium:pseudomonomerous:
Vochysiaceae.

Vochysiaceae, a small family of tropical trees, are known
for their distinctive and beautiful flowers. The flowers are
strongly monosymmetric,with a spurredcalyx, a single fertile
stamen, and most often a reduced number of petals (Fig. 1).
The family is confined to the New World with the exception
of one small west African genus, Erismadelphus.Phylogenetic
analyses based on DNA sequence data indicate that Vochysi-
aceae are most closely relatedto Myrtaceae(e.g., Chase et al.,
1993; Conti et al., 1996; Savolainen et al., 2000; Soltis et al.,
2000), a relationshipsupportedby a suite of anatomical,em-
bryological, and developmentalcharacters(Conti et al., 1997;
Litt, 1999).
Taxonomists have recognized seven or eight genera in
Vochysiaceae (Table 1) depending upon whether or not Eu-
phronia is included in the family. Recent analyses indicatethat
Euphroniais more appropriatelyallied with Chrysobalanaceae
(Litt and Chase, 1999), thereforeit will not be discussed here.
The other seven genera are grouped into two tribes that differ
in charactersof the ovary and fruit (Dumortier,1829; Stafleu,
1952)(Table 1). According to published descriptions, Vochy-
sieae (Vochysia, Salvertia, Qualea, Callisthene, and Ruizter-
ania, a segregate of Qualea) have a superiortrilocularovary
that develops into a loculicidal capsule, whereasErismeae(Er-
isma and Erismadelphus) have an inferior unilocular ovary
that maturesinto an indehiscent winged fruit.
A new genus from west-centralAfrica, Korupodendron(Litt
and Cheek, 2002), was discovered after this work was com-
pleted. It has clear affinities with Erismeae;however,little ma-
terial is available and it was not included in this study.
Vochysiaceae flowers are unusual, but the placement of the
family as sister-groupto Myrtaceaeraises even more questions
I
Manuscriptreceived 28 February2003; revision accepted 1 July 2003.
The authorsthank Scott Mori, Mark Chase, Favio Gonzalez. Louis Ronse
De Craene, and Peter Endress.
2Currentaddress: Yale University, Departmentof Molecular.Cellular,and
Developmental Biology, P.O. Box 208104, New Haven, Connecticut 06520
USA (amy.litt@yale.edu).
about the morphologyand evolution of these flowers. Whereas
flowers of Myrtales are generally polysymmetric, with those
of Myrtaceaeoften having numerousstamens, those of Voch-
ysiaceae are decidedly monosymmetric(or even asymmetrical)
and have a single, frequentlylarge, fertile stamen. Thus, there
is little obvious morphological common ground between the
two families. However, the inclusion of Vochysiaceaein Myr-
tales casts the inferior ovary of Erismeae in a new light, as
epigyny is characteristicof the order. This indicates that the
superiorovary of Vochysieae may be secondarilyderived.Rel-
atively few examples of such independentderivationsof a su-
perior ovary are documented(e.g., Eyde and Tseng, 1969; Ig-
ersheim et al., 1994; Chase et al., 1995; Gustafssonand Albert,
1999; Kuzoff et al., 2001; Soltis and Hufford, 2002).
A review of the literaturealso raised an additionalquestion
about the gynoecium of Erismeae. Whereas the ovary of both
genera, Erisma and Erismadelphus,has a single locule, Kopka
and Weberling (1984) reported additional aborted locules in
transversesections of floral buds of Erismadelphusbut not of
Erisma. Therefore, the unilocular ovaries of the two genera
might not be equivalent and the charactermight not be a syn-
apomorphyof the Erismeae.Their study was limited, however,
and included neither observations of development nor exten-
sive sampling. In light of this informationon the relationships
and the floral structureof Vochysiaceae, it was appropriateto
examine floral development, anatomy,and morphologyin this
family.
In this paper,we reportour observations of floral develop-
ment and vascular anatomy in the Vochysiaceae gynoecium.
In particular,we attemptedto determine:(1) whether there is
evidence that more than one carpel is involved in the devel-
opment of the gynoecium in Erisma and Erismadelphusand
(2) if there is anatomical or morphologicalevidence that the
ovary of Vochysieae is not homologous with the superiorova-
ries of other nonmyrtaleantaxa.
MATERIALSAND METHODS
Taxa-Floralbudsof five generawereused(Table2). Suitablematerialof
Salvertia and Ruizterania was not available; however, phylogenetic analyses

1533
1534 AMERICAN JOURNAL OF BOTANY [Vol. 90

0 AA
s
stigma al xO
Vochysia Salvertia Erismadelphus

spur

fertile
%-- Qualea,Ruizterania,Callisthene Erisma
Erismeae
petalsVochvsieae .
0 inflorescence
axis petal sepal * fertile
stamen
0 spur • O gynoecium
- petal
rudimentary x staminode

Figs. 1-2. Floral structurein Vochysiaceae. 1. Vochysia venezuelana Stafleu. Note zygomorphy. spurredsepal, single fertile stamen, and reduced number
of petals (three). Drawn from photograph(Mori 22871). 2. Floral diagramsof the seven genera of Vochysiaceae. Vochysieae (superiortrilocularovary) are on
the left, Erismeae (inferiorunilocularovary) on the right. In top row, genera have three or five petals. in bottom row, one petal. Callisthenediffers from Qualea
and Ruizteraniain never having rudimentarypetals or staminodes, but some species of the other two genera also lack these structures.

based on morphological and molecular data indicate that Sahlertia may not
be taxonomically distinct from Vochysia, nor the segregate genus Ruizterania
from Qualea (Litt, 1999). Furthermore,Salvertia and Ruizteraniado not differ
from Vochysia and Qualea, respectively, in features of the gynoecium. Thus.
we assume that in these two genera, the development and the vasculatureof
the ovary do not differ significantly from what is observed in Vochysia and
Qualea, respectively.
General descriptions of the flowers of the genera of Vochysiaceae follow.
All have a complete whorl of five sepals, one of which (the fourth) bears a
spur or enlarged sac at the base. All have a single fertile stamen. The genera
vary in features such as petal number and relative sepal size, as well as in
the features that define the two tribes.
Vochysieae-The flowers of the genera in this tribe (Vochysia, Salvertia,
Qualea, Ruizterania,Callisthene) are described as perigynous (de Sainte-Hi-
laire, 1820; Warming, 1875; Baillon, 1878; Stafleu, 1948), the floral organs
being inserted on a short hypanthiumthat surroundsthe superior,trilocular
ovary. In flowers of Vochysia (100+ species), the fourth sepal is approxi-
mately four times the length of the other sepals, and the corolla is almost
always composed of three petals (Figs. 1, 2), although a very few species
(none included in this study) have one or none. The fertile stamen is in the
plane of symmetry, in front of the middle petal; two staminodes are nearly
always present and are in front of the two lateral petals. The ovary contains
two ovules in each of the three locules. Flowers of Salvertia (one species, not
included in this study) differ in having sepals that are nearly equal in size
and five petals.
The remainingthree genera of Vochysieae (Qualea, Ruizterania,Callisthe-
tie) each have one petal, although some species of Qualea (50 species) also
have two minute rudimentarypetals (Fig. 2). The fertile stamen is not directly
in front of the petal but is slightly offset towards the fifth sepal, and these
flowers are asymmetrical (Figs. 2, 33). Two staminodes in antepetalouspo-
sitions are present in some species of Qualea and Ruizterania.Each of the
three locules of the ovary contains numerousovules. Ruizterania(13 species,
not included in this study) and Callisthene (eight species) differ from Qualea
in features of anther ornamentationand of fruit dehiscence and seed shape,
respectively (Stafleu, 1952. 1953; Marcano-Berti,1969; Martins, 1981)(Table
1).
Erismeae-The flowers of the two genera in this tribe (and Korupoden-
dron) are epigynous, although the ovary is free on the spurred side of the
flower in Erismadelphusand some species of Erisina. The ovary is described
as unilocular (Stafleu, 1954: Kopka and Weberling, 1984; Kawasaki, 1998;
Litt and Cheek, 2002). Erisnma(16 species) flowers have one petal, although
they may have up to four rudimentarypetals (Fig. 2). The fertile stamen is
in front of the petal (in the plane of symmetry). Staminodes (most often two)
are also common and are antepetalousas in the Vochysieae. The single locule

TABLE1. Some features distinguishing the tribes and genera of Vochysiaceae, according to Dumortier (1829), Stafleu (1952), and Kawasaki (1998).
Indehis = indehiscent.

Tribe Ovary Locule Fruit Fourthsepal Petal Ovules/ Seeds/

Genera position number type elongate number locule locule Seed wing

Vochysieae superior 3 capsule

Vochysia present 3 2 1 unilateral
Salvertia absent 5 2 1 unilateral
Qualea absent 1 many many unilateral
Ruizterania present I many many unilateral
Callisthene present/absent 1 many few-many circumferential
Erismeae inferior 1 winged, indehis
Erisma absent 1 2 1
Erismadelphus absent 5 1 1
November 2003] LITT ANDSTEVENSON-VOCHYSIACEAEGYNOECIUM
STRUCTURE 1535

TABLE 2. Taxausedin thisinvestigation

to studyorganinitiationand/orvasculature.
Organogenesis Vasculature Voucher

Vochysia
Section Ciliantha
Subsection Ferrugineae V. densiflora Spruce ex Warm. + JRN 716 (INPA)
Subsection Lutescentes V. tucanorum Mart. + + Litt et al. 69 (NY)
Section Vochysiella
Subsection Calophylloideae V. haenkeana Mart. + Rizzo et al. 11395 (NY)
Subsection Decorticantes V. pumila Pohl + + Litt et al. 75 (NY)
V. elliptica Mart. + Litt et al. 4 (NY)
Qualea
Subgenus Amphilochia Q. dichotoma (Mart.) Warm. + Gomez Klein 3279 (NY)
Subgenus Qualea
Section Costatifolium Q. grandiflora Mart. + unvouchered
+ Litt et al. 54 (NY)
Q. multiflora Mart. + Litt et al. 76 (NY)
Q. parviflora Mart. + unvouchered
+ + Litt et al. 48 (NY)
Section Qualea Q. lineata Stafleu + Daly 6220 (NY)
Q. mori-boomii Marc.-Berti + + Mori et al. 24723 (NY)
Q. rosea Aubl. + Mori et al. 24688 (NY)
Callisthene
Section Cataphyllantha C. fasciculata Mart. + Litt et al. 52 (NY)
Section Callisthene C. major Mart. + unvouchered
Erisma E. bracteosum Ducke + + Litt et al. 64 (NY)
+ + Vicentini 1111 (INPA)
E. floribundum Rudge + + Mori et al. 24721 (NY)
+ + Mori et al. 21630 (NY)
E. japura Spruce ex Warm. + Berry 6844 (NY)
E. uncinatum Warm. + Mori et al. 24725 (NY)
Erismadelphus Erismadelphus exsul Mildbr. + Jean-Louis 2031 (NY)

of the gynoecium contains two ovules. Erismadelphus(two species) differs RESULTS

from Erisma in having five petals (Fig. 2) and only one ovule (Keay and
Stafleu, 1953). The spur in Erismadelphusis particularlybroad and saclike.
so that the ovary is free for perhaps a third of its circumference (Figs. 41.
42).
With the exception of the buds of Erismadelphus,all materialwas fixed in
the field in formalin: propionic acid : 50% ethanol (FPA, II 1 : 18 v/v) and
stored in 70% ethanol. Materialof Erismadelphuswas obtained from herbar-
ium specimens (Table 2), rehydratedin 10% Aerosol OT, fixed in FPA. and
stored in 70% ethanol.
Material for scanning electron microscopy was dissected in 95% ethanol.
dehydratedin an ethanol/acetone series, critical-point dried, sputter coated
with gold/palladium,and viewed at 5 kV with a JEOL 5410LV.
Material for serial sectioning was dehydratedin an ethanol/tolueneseries,
embedded in paraffin,and sectioned with a rotary microtome at 10-15 p.m.
Buds that were particularlydifficult to section were softened by exposing their
tips in the paraffinblock and soaking them for at least I wk in 2 : 5 glycerin
:ethanol (70%) with acetic acid added to 10% of the total volume. Material
that was particularly rich in tannins was treated with Stockwell's bleach
(Schmid, 1977) before staining. All material was stained with Johannson's
safranin and counterstainedwith either chlorazol black E or fast green fol-
lowed by orange G, or counterstained with astra blue only. All available
species of Qualea, Callisthene, Erisma, and Erismadelphuswere used. Spe-
cies of Vochysia were chosen to represent as many of Stafleu's taxonomic
sections and subsections as possible given the available material(Table 2).
The available materialof Erismadelphuswas not suitable for scanningelec-
tron microscopy (SEM); thus, for this genus, we obtained data from serial
sections only. SEM data are also available only for a very few stages of floral
development in Callisthene. Relatively complete series of developmentaland
anatomicaldata were obtained for multiple species of Qualea. VochYsia,and
Erisma.
Interpretationsof the organizationof Vochysiaceae flowers
are complicated by the fusion and (where they are free) ex-
tensive overlap of the sepals, the reductionof the corolla and
androecium, and, in Qualea and Callisthene, the slant of the
receptacle. This last can be seen in longitudinal sections of
buds, in which the ovary and the base of the flower are lower
(relative to an imaginaryline drawn perpendicularto the ped-
icel) on the spurredside than on the opposite side (Fig. 45).
Thus transversesections taken near the base of the flower fre-
quently include oblique sections of the ovary. In dissections,
this tilt of the base of the flower and of the ovary complicates
efforts to locate the horizontalplane and the center of the floral
apex.
The term "hypanthium"is used here to refer to a free floral
cup surroundingan ovary (a perigynous flower). This term is
also used by some authorsto denote the outer layers of tissue
surroundingthe inferior ovary of an epigynous flower (e.g.,
Eames, 1961), but this latterusage leads to confusion and im-
plies homology of these outer tissue layers to the free floral
cup. Although this homology is commonly accepted among
morphologists, it generally cannot be proven. Thus, we will
restrict our use of the term to the floral cup of a perigynous
flower.
Organogenesis-General aspects of floral developmentwill
be presented next, followed by descriptions of the develop-
ment of the gynoecium.
1536 AMERICANJOURNALOF BOTANY
Calyx-In all taxa, the five sepals are initiated in a 2/5 spi-
ral, clockwise or counterclockwise(Figs. 3, 4). The floralapex
becomes noticeably concave with the initiation of the sepals
and remains so throughoutthe remainderof floral develop-
ment, althoughit may be relatively flat in some Qualea species
(Fig. 9). The sepals broaden rapidly. Aestivation is quincun-
cial, although this patternmay become difficult to observe as
the extent of overlap of the sepals increases. In almost all
Vochysiaceae, the fourth sepal is overgrown by the fifth and
completely surroundsthe inner organs. In Vochysia, Ruizter-
ania, and some species of Callisthene, it is also significantly
longer than the other four sepals, which are subequal in size.
In most taxa, the first two sepals are free to a greaterextent
than are the other three, which is related to the fact that the
spur,at the base of the fourth sepal, emerges between the first
two (Figs. 2, 27, 28). There is disagreementin the literature
as to the natureof the spur.Historically it has been described
as an outgrowth of the fourth sepal (e.g., de Candolle, 1828;
Warming, 1875; Stafleu, 1952; Cronquist, 1981), whereas
Kopka and Weberling(1984) interpretedit as an extension of
the floral axis. Recent work has shown that the spur develops
in a position that is ambiguous and cannot be unequivocally
designatedreceptacle as opposed to calyx (R. Carmo-Oliveira,
Universidade Federal do personal communica-
tion). However in species of Uberla.ndia,
Qualea, the first and second se-
pals, which flank the fourth, are inserted below the level at
which the spur emerges. In some species these sepals enclose
the spur until anthesis. Furthermore,the spur abscises with the
calyx (A. Litt, personal observation, and P. Oliveira, Univer-
sidade Federal do Uberlandia, personal communication).
These lines of evidence support the hypothesis that the spur
is part of the calyx and not the floral axis.
The sepals of several species of Qualea and Callisthene
were observed to have small triangular,claw-like structuresat
their tips, which do not enlarge as the sepals grow and are
thus inconspicuous at maturity(Fig. 5). These structures,pos-
sibly glandular,were also observed at the tips of young leaves
and bracts in these species (see also Martins, 1981).
The sepals of the two species of Callisthene bear structures
at the base of their adaxial surfaces that are reminiscent of
colleters (see also Martins, 1981)(Fig. 6). In C. major they are
small and roughly triangular,whereas in C. fasciculata they
are elongate, in some cases reachingthe tip of the sepal. These
flattened structures,also possibly glandular,were irregularin
number and shape and were also found between the petiole
and the axis (Figs. 7, 8).
Corolla and androecium-Petals are initiated when the se-
pals (at least the first three) are alreadyquite large and broadly
overlapping(Figs. 9, 10). Stamens are initiatedwhen the petal
primordiaare still relatively small. In most species, the petals
and stamen/staminodesarise on the vertical flank of the floral
cup, and in early stages, the organs, particularlyin Vochvsia
and Erisma species, grow horizontally (Fig. 11) and perpen-
dicular to the surface on which they insert. As the organs
enlarge and the ovary begins to elongate, the petals and sta-
men/staminodes become more vertically oriented (Fig. 12).
Vochysiaceae anthers are characterizedin general by a rela-
tively massive connective, and the style lies between the two
thecae. In Vochysia, the connective also forms a conspicuous,
large-celled cap over the stigma (Fig. 13).
The position of the thecae on either side of the style is
importantin the secondary pollination presentationof Voch-
[Vol. 90
ysia (Yeo, 1993; Oliveira, 1998). In all species observed so
far, the anthersshed their pollen onto the style before anthesis.
As the bud opens, the stamen abscises or is broken off at the
base and falls out of the flower, leaving the exserted style (but
not the stigma) covered with pollen.
It is common in the development of Vochysiaceae flowers
for packing and twisting to occur within the developing buds.
It is not unusual for the style and/orthe filamentto be curved,
for the antherto become bent at an angle relative to the fila-
ment, or for the single petal of Erisma and Qualea species to
be highly crumpled in bud. The limbs of staminodes and the
lateral or rudimentarypetals are generally displaced, presum-
ably to maximize space-packingefficiency (Fig. 12). In open
flowers of many species the style and filament are strongly
twisted, variably in some species (e.g., Ruizteraniacassiqui-
arensis) and predictablyin others (e.g., Qualea grandiflora).
Gynoecium-In all taxa, the gynoecium is first visible, after
all other floralorganshave been initiated,as a ring primordium
on the floral apex (Figs. 14, 17). Early development is similar
in all species, and differences among the taxa do not become
apparentuntil near maturitywhen the difference in ovary po-
sition between Erismeae and Vochysieae becomes apparent.
Vochysia, Qualea, and Callisthene (Figs. 14-16)-The
opening at the center of the ring primordiumis roughly tri-
angular (Fig. 14). In most species observed, it becomes dis-
tinctly three-lobed as development progresses (Figs. 15, 16).
Eventually the opening closes to a three-prongedslit at the
apex of the style.
Erisma (Figs. 17, 18)-A ring primordiumis formed as
well, but the opening in the center is oval or round (Fig. 17).
As developmentproceeds, it narrowsto a longitudinalslit (Fig.
18). In other features, the development of the inferior ovary
of Erisma is indistinguishablefrom that of the ovary of Voch-
ysia, Qualea, and Callisthene.
Vasculature-The following descriptions are based on se-
rial sections, presented here in acropetal sequence. General
observations of floral vasculature and anatomy will be pre-
sented, followed by descriptionsof the vasculatureof the gy-
noecium.
In all species of Vochysiaceae the floral vasculature,which
has the internalphloem characteristicof the vasculatureof the
vegetative structuresof this family, is derived from a cylinder
of vascular tissue in the pedicel. Medullary bundles were ob-
served in Vochysia, Erisma, and Erismadelphus (Fig. 19);
these appear to anastomose with the vascular cylinder or in
some cases contributeto the vasculatureof the gynoecium. In
the base of the flower, the vascular cylinder separatesentirely
into or splits off discrete bundles that supply the floral organs
(Figs. 20, 21, 28, 34). These bundles extend through the hy-
panthium(in Vochysieae) or the outer layer of the ovary/floral
cup (in Erismeae) to enter the sepals. In Qualea, Callisthene,
and Erisma, all of which have one petal, one bundle branches
to supply the petal and the fertile stamen (Figs. 30-32, 37,
38). Rudimentarypetals and staminodes,when presentin these
species, were not observed to contain vascular tissue. In Vo-
chysia and Erismnadelphus, with two staminodesand three and
five petals respectively, three or five of the "floral cup" bun-
dles branchto supply the petals and stamen/staminodes(Figs.
23, 24, 42, 43).
November 2003] LITT AND STEVENSON-VOCHYSIACEAE GYNOECIUM STRUCTURE 1537

Figs. 3-8. Scanning electron micrographsof floral development in Qualea and Callisthene. Scale bar = 100 ptm.3, 4. Sepal initiationin Qualea grandiflora
(unvouchered). Some sepals removed. 3. Clockwise initiation. 4. Counterclockwise initiation. 5-8. Appendages of unknown function (colleters?). 5. Qualea
grandiflora (unvouchered).Claw-shapedappendage at tip of sepal. 6. Callisthene major (unvouchered).Structures(colleters?) at base of inner surface of sepal.
7. C. fasciculata (Litt et al. 52). Structures(colleters?) between petiole (removed) and stem. flanking bud (with hairs). 8. C. major (unvouchered).Appendages
(colleters?) in between petiole (removed) and stem.
1538 AMERICAN JOURNALOF BOTANY [Vol. 90

Figs. 9-16. Scanning electron micrographsof floral developmentin Vochysiaceae.Sepals removed. Scale bar = 100 p.m. 9, 10. Petal initiation.9. Qualea
grandiflora(Litt et al. 54). Initiationof two petal primordia.Note extensive growth of sepals and relativelyflat floral apex. Only one primordiumwill continueto
develop. 10. Erisma bracteosum(Vicentini 1111). Initiationof two petal primordia.Note slightly less extensive growthof sepals and considerablydeeperconcave
floral apex than in Fig. 9. Three more primordiawill be initiatedbut only one will develop to full size. Early petal growth is nearlyhorizontalover the concavity.
11-13. Developmentof the corolla and androeciumin Vochysia. 11, 12. Vochysiatucanorum(Litt et al. 69). 11. Horizontalearly growth of petals perpendicular
to floralcup. Stamen primordiaare directlyunderneathpetals. 12. Petals, fertile stamen,and staminodesenlarge,become verticallyoriented,and theirfree portions
shift to fill the available space within the floral cup. 13. Vochvsia sp. (Litt et al. 21). Nearly maturebud. Note connective of antherforming cap over stigma.
Staminodesare not visible. 14-16. Gynoecium developmentin Vochysieae. 14. Vochvsiapumnila(Litt et al. 75). Gynoeciumappearsas a ring primordiumwith a
triangular-shaped center.All otherfloralorganshave been removed. 15. V. tucanoruin(Litt et al. 69). Laterdevelopment.Three-partnatureof gynoeciumis apparent.
All other floralorganshave been removed. 16. Qualea parviflora(Litt et al. 48). Three-partopening at apex of gynoecium is visible understamenand petal.Sepals
have been removed. Abbreviations:a, anther;aloc, abortedlocule; as, accessory organ (not identifiableas rudimentarypetal or staminode);br, bract;dcb, dorsal
carpel bundle;f, filamentof fertile stamen;fcv, floralcup vasculature;fs, fertile stamen(or primordium);fst, fertile stamentrace;g, gynoecium;gc, gum canal;gv,
gynoecium vasculature;Icb, lateralcarpel bundle;loc, locule of gynoecium; mb, medullarybundle:mlb, major lateralbundle;ov, ovule; p, petal (or primordium);
pt, petal trace;rp, rudimentarypetal; s, sepal; s2, second sepal (and similar):set, sepal trace; sp, spur: st, staminode:stt, staminodetrace; sty, style; vbr, ventral
bundle ring; vc, vascularcylinder,vcb, ventralcarpel bundle;vcbr,ring of vasculartissue that supplies ovules. Open circles representvascularbundles.
November 2003] STRUCTURE
LITT AND STEVENSON-VOCHYSIACEAEGYNOECIUM 1539

of gynoeciumdevelopmentin Erisma.All floralorgansotherthangynoeciumhave been removed.17. E.

Figs. 17-18. Scanningelectronmicrographs
floribundum (Mori et al. 21630). Gynoecium appears as a ring primordiumwith an oval thumb-printshaped center. Scale bar = 20 L.m. 18. E. bracteosum
(Vicentini 1111). Later development. Opening at apex of gynoecium is a longitudinalslit. Scale bar = 100 pm.

Several other features that were observed in serial sections rounding tissue (Figs. 21, 22). Thus, the vasculature to the
characterizeVochysiaceae flowers; in general these structures gynoecium becomes organized at the same level at which
have not been included in the figures. These include gum ca- branchesare being supplied to the floral cup. In Vochysia,the
nals, found in the pedicel and often in the sepals (V. tucano- bundles at the inner edge of each septum come together to
rum, V. pumila, all Qualea species except Q. dichotoma, and form a ring in the center of the gynoecium (Fig. 23), which
E. bracteosum)(Fig. 19), and crystals, found in all organs (all then separates into three bundles that lie one in front of each
species except possibly absent from Q. rosea, E. floribundunm, locule. These bundles supply the two ovules of each locule
and E. uncinatum). Simple intra-ovarianhairs were seen in (Fig. 24). The three synlateral bundles and the three smaller
Erismadelphusexsul and all species of Erisma, as well as in dorsal bundles continue into the style (Fig. 25).
Qualea mori-boomii and Q. lineata. In all species, densely 2. Qualea and Callisthene (Figs. 26-33)-In transversesec-
staining cells line the internal surface of the nectary spur and tion, the ovaries of these genera are more or less circularand
are also found in the outer cell layers of the ovary; they are unlobed (Figs. 30, 31, 33). There is no evidence that the loc-
also seen in the outer cell layer of the sepals and other organs ules extend below the level at which the ovary becomes free
in some species. Additionaldensely staining cells surrounding from the surroundingfloral cup, as in Vochysia. In both Qua-
the apparentstylarcanal in all species were identifiedas pollen lea and Callisthene, the base of the flower and ovary are mild-
tube transmitting tissue. An epidermal layer of large, non-
ly to strongly slanted, with the side opposite the spur being
staining cells was seen in the connective of all species. higher (Fig. 26). The ovary becomes free on the spur side at
a lower level, and thus the vasculatureto the ovary also be-
Vochysieae gynoecium-In all species of Vochysieae the comes separatefrom that of the floral cup at a lower level on
vascular cylinder of the pedicel produces branches that enter
that side (Figs. 26, 27). Furthermore,the slant of the ovary
the hypanthium(or the sepals directly), and the remainderof
the cylinder supplies the gynoecium (Figs. 20-22, 28, 29). In means that in most species the vascular cylinder cannot con-
most cases the vasculartissue that enters the gynoecium forms tinue straight into the ovary but must curve towards the hor-
three dorsal carpel bundles and three major synlateralcarpel izontal as the ovary emerges at an angle from the base of the
bundles. These latter are located at the outer edge of each of flower (Fig. 26). Thus sections throughthe ovary often contain
the three septa and may be as large as or largerthan the dorsal vascular tissue that appears to be branching horizontally but
bundles (Figs. 22, 23, 29, 30). Arcs of numeroussmall bundles in fact is bending to enter the ovary (Fig. 27). The species
between the dorsalbundles form an anastomosingnetworkthat described next differ in details such as the numberof bundles
supplies the ovary wall (Fig. 31). In most species the major in the style and the configurationof the ventralcarpelbundles.
synlateralbundles divide, each producinga bundle that travels a. Q. parviflora, Q. dichotoma (Figs. 28-30)-There are no
through the septum to lie at its inner margin. These ventral significant deviations from the pattern described for Vochy-
carpel bundles (Figs. 22, 29) supply the ovules of the two sieae. In the ovary wall of Q. parviflora, the dorsal carpel
adjacentcarpels. bundles are relatively inconspicuous as opposed to the larger
1. Vochysia (Figs. 19-25)-No significant variability was synlateralcarpel bundles (Fig. 30).
observed in the species of Vochysia. In transversesection, the b. Q. lineata, Q. rosea, Q. mnori-boomii (Fig. 31)-The or-
is
ovary three-lobed; a furrow marks the center of each locule, igin of the ventral bundles is slightly different in these species
where the fruit will dehisce (Fig. 24). The ovary is adherent in that numerous lateral bundles contributeto them. There is
to the floral cup at its base: the locules extend below the level no conspicuously larger synlateralbundle at the outer edge of
at which the ovary becomes completely free from the sur- each septum, as in Vochysiaor other Qualea species (Fig. 31).
1540 AMERICAN JOURNAL OF BOTANY [Vol. 90

sp Icb O dcb
loc
br
,

vc vc o
mlb
gcgv vcb

19 0.5mm 20 0.5mm 21 0.5mm 22 0.5mm

dcb s
dcb

vcboo
-0

vbr ov s
oo
•
•pt f0s
2t 0

s••/pt

0.5mm 24 0.5mm 25 0.5mm

23
Figs. 19-25. Drawings of transverse sections of the structureand floral vasculatureof Vochysia tucanorum(Litt et al. 69). 19. Pedicel. Note medullary
bundles (mb), present in flowers only in the pedicel, and gum canals (gc), present in other sections but not drawn. 20. Vascularcylinder begins to separateinto
bundles that will supply the floral organs and the gynoecium. 21. One locule is visible, three segments of gynoecium vasculatureare identifiable. 22. Three
dorsal carpel and three major synlateral bundles are recognizable. Two major synlateral bundles have split to produce ventral carpel bundles; the third has
startedto do so (asterisk). Ventralcarpel bundles extend through septa to center of gynoecium. 23, 24. Three ventral carpel bundles form a ring in the center,
which then divides into three discrete ventral bundles, one per carpel, which supply the ovules. Note three-lobed shape of ovary. 25. Dorsal and synlateral
bundles continue into style.

The style of Q. lineata contains a ring of discretebundles.The
three dorsal bundles are conspicuously the largest, with nu-
merous small bundles between. In the style of Q. rosea, which
is significantly twisted in the bud, there is a continuous six-
pointed ring of vascular tissue. The style of Q. mori-boomii
contains only three bundles, the dorsal carpel bundles.
c. C. major-The available material of C. major was not
adequateto determinethe details of the vascular supply to the
gynoecium, however the structureof the flower and the pattern
of vasculaturein this species is similar to that seen in Qualea.
d. C. fasciculata (Figs. 32-33)-The ovary of this species
is more elongate and slender than that of other species of
Vochysieae that were examined in this study. Although the
base of the ovary is slanted, the vascular tissue does not have
to curve appreciablyto enter the ovary as it does in Qualea
species. After the dorsal and synlateralcarpel bundles branch
off from the vascular tissue in the base of the gynoecium, a
triangularmass of vascular tissue remains in the center. This
coalesces into a ring, which appears to supply the numerous
ovules of each of the three carpels (Figs. 32, 33). The synla-
teral bundles and the dorsal bundles extend into the style.
Erismeae gynoecium-In these epigynous flowers the vas-
cular cylinder separates into discrete bundles, which extend
the length of the ovary and supply the three outer whorls of
floral organs. These bundles also produce numerous small
branchesthat form an anastomosing network surroundingthe
single locule of the ovary (Figs. 37, 41). There is no clearly
November 2003] LITTAND STEVENSON-VOCHYSIACEAE STRUCTURE
GYNOECIUM
a
et
et
9 ~~pt
fst
sp gv
26 27
Figs. 26-27. Drawings depicting the slant of the ovary of Qualea and Callisthene. Qualea panriflora (unvouchered).26. Median longitudinalsection. Base
of ovary is higher on the side of the flower opposite the spur;ovary inserts on receptacleat an angle. Vascularsupply from pedicel must bend to entergynoecium
as at the asterisk. 27. Cross-section throughbase of ovary. In serial sections the bend in the vasculatureas it enters the gynoecium gives the appearanceof the
vasculaturebranchingas at the asterisk (sections include horizontalsegments of bundles).
identifiable dorsal carpel bundle, although in some cases a
bundle can be interpretedas such.
1. Erisma (Figs. 34-40)-Four species of this genus had no
evidence of abortedcarpels. The shape and orientationof the
single locule and details of the vasculaturevaried among the
species examined.
a. E. bracteosum, E. floribundum (Figs. 34-39)-In these
species, the spur is inserted at the top of the ovary (Figs. 36-
38). The locule is oriented somewhat diagonally in the bud,
although it may be more symmetrically oriented in E. flori-
bundum.The placenta is more or less in front of the position
of the fifth sepal or between the second and fifth in E. flori-
bundum.This position is either clockwise or counterclockwise
from the spur (used as a point of reference) depending on the
direction of sepal initiation.
The vascular bundles on the adaxial side of the floral bud
produce small branches that anastomose in the center with
some of the medullary bundles. An arc of bundles is thus
formed, which coalesces into one bundle that supplies the two
ovules (Figs. 34-36). The bundles that form the network sur-
roundingthe locule anastomose into a ring of several bundles
that enter the style and further anastomose to form three
(sometimes four or five) bundles that surroundthe stylarcanal
(Figs. 38, 39). In E. floribundum,the bundles are conspicu-
ously off-center, and this is also variably the case in E. brac-
teosum (Fig. 39).
b. E. uncinatumand E. japura (Fig. 40)-In these species
the spur is inserted at the base of the ovary, thus the ovary is
free on that side of the flower (Fig. 40). The locule is sym-
metrical, lying on the midline of the bud with the placenta in
1541
00000
n
vcs
0.5mm
front of the petal (between the third and fifth sepals)(Fig. 40).
In E. japura, the bundle that supplies the two ovules is formed
from an arc of bundles in the same way as in E. bracteosum
and E. floribundum. In E. uncinatum, one vascular bundle
branchesfrom the bundle that will eventually become the me-
dian bundle of the fifth sepal; this bundle supplies the two
ovules. The small bundles surroundingthe locule anastomose
to form several bundles that enter the style, but the exact num-
ber of bundles in the style could not be determinedaccurately.
One bundle is larger than the others, which form an arc sur-
roundingthe transmittingtissue.
2. Erismadelphus(Figs. 41-44)-One species, Erismadel-
phus exsul, was available for this study. Examinationof serial
sections revealed one aborted locule in two of the four spec-
imens available (Figs. 41, 42). The fertile locule is irregularly
shaped, although it should be noted that these observations
were made on rehydratedherbariummaterial. In contrast to
Erisma, in which the placenta is in front of the position of the
fifth sepal or thereabouts,the placentain Erismadelphusexsul
is between the positions of sepals one and three (Figs. 42, 44).
As with Erisma, this position may be eitherclockwise or coun-
terclockwise from the spur.
As the vascularcylinder separatesinto discrete bundles, one
bundle remains separateand supplies the ovule (Figs. 41, 42).
No bundles occupy the position of a ventral carpel bundle of
the abortedcarpel.
There is no bundle associated with the fertile locule thatcan
be identified as a dorsal carpel bundle. There is a bundle that
can be followed into the style in the position where one might
look for the dorsal bundle of the abortedcarpel. However,the
1542 AMERICAN JOURNALOF BOTANY [Vol. 90

ss o mlb

dcb dcb
to * vcb
00

o u o 0
,

vc p

28 gv 0.5 mm 0.5mm 30 0.5 mm

2---m
ov Icb
*0*st
. loc dcb 0
dc-vcb
c~L o c
gvV pt p
3128 0.5mmo 29 0.5mm 30 0.5mm
TO
31 0
.

Figs. 28-33. Drawings of transversesections of floral structureand vasculatureof Qualea and Callisthene. 28-30. Qualea parviflora (unvouchered).28.
Vascular supply to gynoecium separates at a lower level on the spurredside of the flower in keeping with the slant of the base of the ovary. Labeled sepals
are the first and second; spur emerges between them. 29. Locules are not visible until the ovary is free from the surroundingtissue. Major synlateralcarpel
bundles split to produce ventral bundles (asterisk). 30. Vasculatureof the ovary. 31. Qualea mori-boomii (Mori et al. 24723). Note absence of large major
lateral carpel bundles. Ventralcarpel bundles are formed by coalescence of several small bundles from the carpel wall supply. 32-33. Callisthenefasciculata
(Litt et al. 52). 32. Dorsal, lateral, and ventral carpel bundles all form concurrentlyout of mass of vascular tissue in base of gynoecium. 33. Ring of vascular
tissue in center of gynoecium supplies ovules.

curve of the style makes tracingthis bundle difficult (Figs. 42,
43). This appearsto be the only vascular bundle in the style,
although in some specimens this could not be determined.
DISCUSSION
This study was undertakento provide information on the
development and vasculatureof the flowers of Vochysiaceae,
as well as to answer two questions about the gynoecium: (1)
is there evidence of more than one carpel in the gynoecium
of Erismeae and (2) is there evidence that the superiorovary
of Vochysieae is secondarily derived? With regardto the first
question, our results show that there is at least one additional
carpel in Erismadelphusbut not in Erisma. Regardingthe sec-
ond question, development and morphology do provide some
confirmationthat the superiorovary of Vochysieae is not ho-
mologous with the superiorovaries of other angiosperms.
Ontogeny of the flowers of Vochysiaceae is basically sim-
ilar, despite their morphological differences at maturity.The
mannerin which the organs of all whorls are initiatedand the
broad outlines of their developmentare fundamentallysimilar,
whether the ovary is superioror inferior,trilocularor uniloc-
ular,or the corolla includes three petals or one. Vasculatureis
also similar among the genera. The biggest difference is in the
epigynous Erismeae, which have no discrete dorsal or other
regularly organized carpel-wall bundles as in Vochysieae. In-
stead, a networkof anastomosingbundles surroundsthe locule
and forms the bundle(s) that will enter the style (Figs. 37, 38,
41, 42)(as in Hufford, 1989). At first glance, the vasculature
of Qualea and of Callisthene major appearto differ substan-
tially from that of C. fasciculata and Vochysia.The observed
differences, however, reflect the slant of the ovary in the for-
mer species as well as the more slender shape of the ovary of
C. fasciculata. If the tilt is taken into account, it can be seen
that in all species of Vochysieae the vascular cylinder of the
pedicel enters the gynoecium more or less directly, after split-
ting off branchesthat will supply the other floral organs.
In a few cases, the observed patternsfollow establishedsub-
November 2003] STRUCTURE
AND STEVENSON-VOCHYSIACEAEGYNOECIUM
LITTrr 1543

s p
0
0 0

gv ov o

4 0m00

00
spOgos
00 b
*c 0a0 p

sp

ot 00 a
'P.0
*0n
0 ,O 00 0O sty
45 loc o
aXO o

61C a 0 0

37 CO O-O 0.mm38 .
0_5m_ 363 O
__ O
oC
0-0
. . *010 0
o
. 0

.
T

a o
gcv o'op,

pt of s
1370. m 18 .5mm390. m

"•fst fst
-
pt pt__s_
".

Figs. 34-39. Drawings of transversesections of floral structureand vasculatureof Erisma bracteosum (Vicentini 1111). 34, 35. Several peripheralbundles
supply branches to center that anastomose to form the ventral carpel bundle. 36, 37. Note asymmetricalorientationof locule and placenta. Peripheralbundles
all supply branches that form a network around locule. 38, 39. Vascularbundles in network (asterisk) anastomose into a few bundles that enter the style and
resolve themselves as 3-5 bundles. Note that the spur emerges at the top of the ovary. Regions dotted with black squares are pollen sacs.

generic taxonomic groupings,althougha phylogeneticanalysis
testing the monophyly of these groups is not yet complete.
Qualea rosea, Q. lineata, and Q. mori-boomiiall have similar
patternsof gynoecial vasculature,differing only in the number
of bundles in the style. Qualea mori-boomii is a segregate of
Q. coerulea (Marcano-Berti,1989), and Q. coerulea was in-
cluded with Q. lineata and Q. rosea by Stafleu (1953) in his
series Calophylloideae (section Qualea, subgenus Qua-
lea)(Table 2). Based on his taxonomy, Q. mori-boomiiwould
also be placed here. However, Stafleu (1953) placed fully 27
of his 59 species in this series (27 of 46 when the segregate
genus Ruizterania is excluded). Furthermore,the buds and
flowers of these three species are larger--considerably larger
in the cases of Q. rosea and Q. lineata-than those of Q.
parviflora and Q. dichotoma.This size differencemay produce
differences in vasculature.Thus, the observed similarities in
vasculature may reflect an evolutionary relationship or may
reflect independentlyderived similaritiesin floraldevelopment
and structure.
Stafleu (1954) recognized two sections in Erisma, each in-
cluding half the species. Erismajapura and E. uncinatumboth
belong to section Rixa, and E. floribundumand E. bracteosum
to section Erisma. The similaritiesand differences seen in loc-
ule and placenta orientationin these species thus conform to
Stafleu'snotions of relationswithin the genus. However,a cla-
distic analysis of the 16 species, based on 21 morphological
features, did not show these sections to be monophyletic and
placed E. uncinatum and E. japura in different clades (Ka-
wasaki, 1992). Erisma bracteosum and E. floribundum are
placed in the same clade. Additional work on gynoecium
structurein Erisma is needed to determinethe patternof char-
acter evolution within the genus and tribe Erismeae.
The data collected also highlight the similarities between
Qualea and Callisthene and accentuate the differences be-
tween these two genera and Vochysia, with which they have
been allied in tribe Vochysieae. All three have a trilocularova-
ry, but it is partlyinferiorin Vochysiaand completely superior
in the other two (Figs. 21, 22). Vochysia also does not have
the strongly slanted receptacle seen in Qualea and Callisthene
(Fig. 45). In fact, it is not entirely clear that there is a hypan-
1544 AMERICAN JOURNAL OF BOTANY [Vol. 90

sp . .
0.5mm sp
sp
o 00
*
O p ,. 6
09D000

4.31.0 00 aloc
0 vcb ov aloc
ov t
0o. m0vcbm

41 0.5 mm 42 0.5 mm

y S

9 .sts
0 t
S

000000 000o
/oo
O o0 o
00 stt*

40

Figs. 40-44. Drawings of transversesetions of floral structureand vasculatureof Erismeae.40. Floral vasculatureof Erisma uncinatum(Mori et al. 24725).
Note symmetrical position of locule and placenta and position of spur at the level of the ovules. Asterisk, vascular bundle that will divide to supply fertile
stamen and petal. 41-44. Floral vasculatureof Erismadelphusexsul (Jean-Louis2041). 41, 42. Aborted locule and asymmetricalposition of fertile locule. 42.
Bundle in position of dorsal carpel bundle of aborted locule (boxed double asterisk) will enter the style. Single asterisk indicates bundle that will divide to
supply fertile stamen and petal. 43. Vasculatureanastomoses into bundles that enter the style but do not appear to go far. 44. Vasculatureof all floral organs.
The two thecae (regions dotted with black squares) and the filament of the antherare all separatein this section.

thium in Qualea and Callisthene. If one changes the frame of
reference and views the base of the ovary as horizontal, the
putative hypanthiumbecomes a broad receptacle (Figs. 45-
47).
Symmetry in Vochysiaceae-In all species, sepals are ini-
tiated in a spiral sequence, either clockwise or counterclock-
wise (Figs. 3, 4). Subsequent organs are initiated with refer-
ence to the sepals, thus the single petal of Qualea, Erisma,
and Callisthene is always between the third and fifth sepals
(Fig. 2) and opposite the spur, as is the center petal of Voch-
ysia. The stamen in Vochysia and Erisma (and Salvertia and
Erismadelphus)is located in front of this petal, in the plane
of symmetry.In Qualea and Callisthene(and Ruizterania),the
stamen is displaced towardsthe fifth sepal; thus, these flowers
are asymmetrical(Fig. 2). Because the sepals are arrangedin
left-handedand right-handedspirals, the fifth sepal is either to
the left or the right of the petal, and thus the stamen is also
either to the left or the right of the petal. There has been
speculation that these left-handed and right-handedmorphol-
ogies may be significantfor pollination,but both are found on
the same plant and therefore can provide no special opportu-
nity for cross-pollination.However if there were not right- and
left-handedflowers there would in fact be no effective transfer
of pollen from stamen to stigma.
Flowers of Erisma and Erismadelphusare superficiallybi-
laterally symmetrical, with the stamen in the median plane,
but in two of four species of Erisma, and in Erismadelphus,
the single locule of the ovary is not symmetrically oriented
(Figs. 36, 42). In Erismadelphus,Erisma bracteosum,and Er-
isma floribundum,the placenta may be positioned to the left
or the right of the spur, as with the stamen of Qualea. This
again presumably reflects the spiral initiation of the sepals.
Any structurethat is off the plane of symmetry will be found
clockwise from the spurin some flowers and counterclockwise
in others.
Gynoecium of Erismeae-Locule number-The function-
ally unilocular ovaries of the two genera of Erismeae differ
structurally.Erismadelphushas at least one abortedcarpel in
the gynoecium (Figs. 41, 42), and the vasculatureof the style
is derived from a bundle that is associated with the aborted
carpel ratherthan with the fertile carpel (Fig. 42). Two aborted
carpels were reportedby Kopka and Weberling(1984). In the
limited material available for this study, one aborted carpel
was seen, but inconsistently.In contrast,abortedcarpels were
never seen in Erisma, and there is no evidence from vascu-
latureto indicate that any might exist. Developmentaldata are
not available for Erismadelphus,but developmentof the ovary
of Erisma is consistent with the notion that the gynoecium
consists of a single carpel. In all Vochysiaceae,the gynoecium
is initiatedas a ring primordium;in Vochysieae, the depression
in the center of the ring, and subsequentlythe opening at the
apex of the developing ovary, is triangularor three-parted.
This is not the case in Erisma, in which the depressionis oval
and the opening is longitudinal.The presence of abortedcar-
November 2003] LITT AND STEVENSON-VOCHYSIACEAEGYNOECIUMSTRUCTURE
fs
p
g,
fs
p variable because they incorporatedinformation from the lit-
__ b Authors since Eckardt(1937) have followed his lead in de-
Figs. 45-47. Interpretationof the structureof base of the Qualea flower.
Median longitudinal sections. 45, 46. Qualea parviflora (unvouchered).45.
In this orientation,the base of the ovary is strongly slanted and the stamen
and petal are inserted on a concave hypanthium(asterisk). 46. In this orien-
tation, the base of the ovary is flat and there is no apparenthypanthium.but
there is a broad receptacle (asterisk).47. Vochysiatucanorum(Litt et al. 69).
Base of ovary is not significantly slanted.
pels in Erismadelphusflowers makes this an importantgenus
from which to obtain developmentaldata.
The only suggestion in Erisma of a multi-carpellatecondi-
tion is the asymmetricalposition of the locule and the placenta
in E. bracteosumand E. floribundum.This position is similar
to what one might expect if the carpel were one of several in
the gynoecium. Nonetheless, it does not of itself providedirect
evidence of other carpels. Phylogenetic analyses have not yet
resolved relationships within Erismeae, but it is difficult to
refrainfrom speculatingthat the differentmorphologieswithin
the family representstages in a reduction series. The starting
point would be a tricarpellategynoecium such as that of Vo-
chysieae, with Erismadelphus (aborted carpels) and species
such as E. bracteosumand E. floribundumrepresentinginter-
mediate stages. Erisma uncinatumandE. japura, with a single,
symmetricallocule, would representthe final stage.
Pseudomonomery-Ronse De Craeneand Smets (1998) not-
ed that in 95% of unicarpellatetaxa, the carpel is oriented
toward the inflorescence axis. Away from the axis is uncom-
mon, and taxa in which the position is variableare rare.Clarke
(1859), in a comment on the Vochysiaceae, stated that "one
of the most remarkablecharactersin this family is that the
carpel when single is posterior."However,this descriptionwas
based only on Erisma, as Erismadelphus had not yet been
described (Mildbraed, 1913). Clarke was following common
1545
usage in applying the term "posterior"to the spur-sideof the
flower in Vochysiaceae, although it is the second, not the
spurredfourth sepal, that is in front of the axis (Fig. 2). Based
on this usage, Clarke'sdescriptionis correct for E. uncinatum
and E. japura because in these species the locule is oriented
away from the spur (Fig. 40). However, in E. floribundumand
E. bracteosum,the locule is oriented obliquely, more towards
the axis (Fig. 36). Thus, Erisma in fact is one of the raretaxa
in which there is variabilityin the position of the locule.
Eckardt (1937) used Clarke's description, along with the
fact that other Vochysiaceae were trilocular,to conclude that
the gynoecium of Erisma is pseudomonomerous.Eckardtused
this term to describe a unicarpellategynoecium that is derived
by reduction from a syncarpous one, in contrast to a "true"
unicarpellategynoecium derived by reduction from an apo-
carpous one. Pseudomonomery sensu Eckardt (also Eames,
1961; Hufford, 1989) therefore refers to an evolutionaryder-
ivation, not a morphologicalcondition, although Eckardtdoes
present morphologicalevidence of the derivationfor the taxa
that he observed. Ronse de Craene and Smets (1998) defined
pseudomonomeryin morphologicalterms, using it to refer to
an ovary in which several carpels are initiated but all abort
except one. However, their actual applicationof the term was
erature,in which usage is inconsistent.
scribing the ovary of Erismeae, including Erismadelphus,as
pseudomonomerous(e.g., Cronquist, 1981; Takhtajan,1997).
However, there are two problems with using the term in this
historical sense. First, it is applied to gynoecia that differ in
structure.There may be physical evidence of reductionin car-
pel number (as in Erismadelphus),or there may be no such
evidence (as in Erisma). When the same term is used to de-
scribe both, informationis lost, and a reader might conclude
that the ovaries are the same. Second, it is inappropriateto
use assumed evolutionary derivation in describing characters.
To avoid circularity,the patternof characterstate changes is
best deduced from the resultsof a phylogenetic analysis, rather
than assumed a priori.
We recommend restricting the use of the term "pseudo-
monomery" to the morphological condition described by
Ronse De Craene and Smets (1998) and Weberling(1981), in
which there is only one fertile locule in the matureovary, but
in which there is direct evidence that more than one carpel is
initiated. This requiresexamining the ontogeny and morphol-
ogy of the ovary (Weberling, 1981; Ronse De Craene and
Smets, 1998). Using this criterion,the gynoecium of Erisma-
delphus is pseudomonomerous,whereas that of Erisma is not.
We suggest describing the gynoecium of Erisma as monom-
erous rather than unicarpellate.Monomerous refers to a gy-
noecium composed of a single unit, without indicatingthe spe-
cific natureof the unit. In Erisma the ovary is never distinct
from the surroundingfloral tissue and no discrete carpel can
be identified,only a single locule. To date there is no evidence
for more than one structurein the ovary, thus monomerousis
an accuratedescriptionof what can be observed. Because eight
species of Erisma were not examined in this study,futurework
may demonstrateevidence in some species of more than one
carpel.
Ovaryposition in Vochysieae-Mature morphology-Serial
sections of Vochysia buds show that in this genus the base of
the gynoecium is not free from the surroundingfloral periph-
1546 AMERICAN JOURNAL OF BOTANY
ery, that is, the ovary is not wholly superior.The locules of
the three carpels can be seen in sections in which the ovary
is still continuous with the surroundingfloral tissue (Figs. 21,
22), thus matureflowers of Vochysia are epigynous to a small
degree.
In contrast, the ovary of Qualea and Callisthene does not
appear to be inferior to any degree. In sections of flowers of
these genera, the locules are not apparentuntil the ovary is
completely free, and there is no clear-cut morphologicalevi-
dence to distinguishthe superiorposition of the ovary of Qua-
lea and Callisthenefrom that of othertaxa. However,the slant-
ed base of these flowers and of the ovary as well as the pres-
ence of the spur on one side make it difficult to determinethe
exact position of the base of the ovary (Fig. 45). Likewise, it
is difficult to identify what should be considered hypanthium
as opposed to receptacle, if in fact there is any difference be-
tween them. Thus it is possible to interpretthe ovary as being
slightly adherenton one side to the receptacle.
Our results indicate that there are three distinct floral plans
within Vochysiaceae: Erisma and Erismadelphusclearly are
epigynous; Vochysiais partlyepigynous and partlyperigynous
(the ovary is adherentat its base, but free from the hypanthium
above); Qualea, Callisthene, and Ruizteraniaare perigynous.
There are qualificationsto these basic descriptions:the ovary
in flowers of Erismadelphus,E. uncinatumand E. japura, is
free on the spurredside of the flower,and Qualea, Callisthene,
and Ruizterania flowers can perhaps be considered hypogy-
nous if they are tilted (Fig. 46).
Ontogeny:floral apex shape and ovary position-In spite
of these differences in the structureof the matureflowers, the
early stages of development of the gynoecium are the same in
all species of Vochysiaceae and conform to a patterngenerally
but not exclusively associated with an inferior ovary (Boke,
1963, 1964; Kaplan, 1967; Gustafsson and Albert, 1999; Ku-
zoff et al., 2001). The gynoecial primordium,in this case a
ring primordium,arises on a concave floral apex; it is therefore
initiated as an inferior ovary, beneath the level of the other
floral organs. Subsequentdifferentialgrowth results in the ul-
timately superior(or nearly superior)position of the ovary in
Vochysieae. This is in contrast with the patternseen in most
species with superiorovaries, in which the floral apex is con-
vex. In these species, the ovary is initiatedand remainsabove
the level of the other floral organs (Gustafsson and Albert,
1999).
The term "pseudosuperior"has been suggested to designate
ovaries that appear superficiallyto be superiorbut in fact are
inferior up to one-fourthof their length and that develop on a
concave floral apex. At maturity,these ovaries have a small
inferiorregion that can be detected only when examined care-
fully (e.g., with scanning electron microscopy)(Kuzoffet al.,
2001; D. Soltis, University of Florida, personal communica-
tion). However, the shape of the apex in the developing flower
and the ovary position in the mature flower are not always
correlated(Soltis and Hufford, 2002). In addition, although a
concave floral apex is considered characteristicof epigynous
flowers, it is also seen in perigynous flowers (e.g., Evans and
Dickinson, 1999a, b). This may indicatea relationshipbetween
epigyny and perigyny, or it may suggest that the significance
of the correlation of a concave floral apex with epigyny has
been overemphasized.
In Vochysiaceae, all taxa share the same state for the early
development character(concave floral apex) but differ in the
[Vol. 90
mature ovary position character:in Erisma and Erismadel-
phus, it is inferior;in Vochysia, it is semi-inferior;in Qualea
and Callisthene, it is superior.Thus, the genera of Erismeae
fit the common patternfor an epigynous flower, and the ovary
of Vochvsiafits the definitionof pseudosuperior.However,the
situation in Qualea is not as clear: the ovary is initiated on a
concave apex, but the mature position is fully superior.To
avoid confusion we have chosen to treat early and late onto-
genetic stages as two separate characters.Thus, all Vochysi-
aceae share the early ontogenetic characterof concave floral
apex, but they differ in the late characterin having inferior,
semi-inferior,or superiormatureovaries.
By examining development in a phylogenetic context, we
can better understandthe different ovary positions and locule
numbers found in Vochysiaceae flowers. First, developmental
data indicate that the superioror nearly superiorovary seen in
Vochvsia, Qualea, and Callisthene is not homologous with the
superior ovary in other families. Second, the status of Vo-
chysiaceae as a monophyletic family indicates that the differ-
ent ontogenetic trajectoriesthat produce the different anthetic
morphologies share a common evolutionary history. Finally,
the inclusion of Vochysiaceae in Myrtalesand the sister-group
relationshipwith Myrtaceae suggest that this history involves
derivation from a flower with an inferior,multilocularovary.
LITERATURECITED
BAILLON. H. 1878. The naturalhistory of plants. L. Reeve, London,England.
BOKE,N. H. 1963. Anatomy and development of the flower and fruit of
Pereskia pititache. AmericanJournal of Botany 50: 843-858.
BOKE. N. H. 1964. The cactus gynoecium: a new interpretation.American
Journal of Botany 51: 598-610.
CHASE,M. W., ET AL. 1993. Phylogenetics of seed plants: an analysis of
nucleotide sequences from the plastid gene rbcL. Annals of the Missouri
Botanical Garden 80: 528-580.
CHASE. M. W., D. W. STEVENSON, P. WILKIN, AND P J. RUDALL. 1995.
Monocot systematics: a combined analysis. In P J. Rudall, P J. Cribb,
D. E Cutler. and C. J. Humphries [eds.],Monocotyledons: systematics
and evolution, vol. 2, 685-730. Royal Botanic Gardens, Kew, UK.
CLARKE.B. 1859. Notes on the structureand affinities of Batideae, Callitri-
chaceae. Vochysiaceae and Cassytheae. Transactions of the Linnaean
Society of Lotndon 22: 411-414.
CONTI, E.. A. LITT, ANDK. J. SYTSMA. 1996. Circumscriptionof the Myrtales
and their relationshipsto other rosids:evidence from rbcL sequence data.
American Journal of Botany 83: 221-233.
CONTI,E., A. LIrr. P G. WILSON, S. GRAHAM,B. G. BRIGGS,L. A. S.
JOHNSON,AND K. J. SYTSMA. 1997. Interfamilialrelationshipsin Myr-
tales: molecularphylogeny and patternsof morphologicalevolution. Sys-
tematic Botany 22: 629-647.
CRONQUIST,A. 1981. An integrated system of classification of flowering
plants. Columbia University Press, New York, New York, USA.
DECANDOLLE. A. P 1828. Vochysiaceae. Prodromussystematisnaturalisreg-
ni vegetabilis, vol. 3, 25-30. Treutteland Wtirtz, Paris, France.
DE A. 1820. M nmoiresur la nouvelle famille des Vochisides.
SAINT-HILAIRE,
Mimoires du Museumnd'Histoire Naturelle 6: 253-270.
DUMORTIER.B. C. 1829. Analyse des families des plantes avec l'indication
des principaux genres qui s'y rattachent.Vochysiaceae. J. Casterman,
Tournay.France.
EAMES.A. J. 1961. Morphology of the angiosperms. McGraw-Hill, New
York, New York, USA.
ECKARDT,T 1937. Untersuchungen fiber Morphologie, Entwicklungsgeschi-
chte und systematische Bedeutung des pseudomonomerenGynoeceums.
Nova Acta Leopoldina, N.F. 5:1-112.
1999a. Floralontogeny and morphology
EVANS.R. C., ANDT A. DICKINSON.
in subfamily Amygdaloideae T. & G. (Rosaceae). International Journal
of Phlat Sciences 160: 955-979.
1999b. Floralontogeny and morphology
EVANS.R. C.. ANDT. A. DICKINSON.
in subfamily Spiraeoideae Endl. (Rosaceae). International Journal of
Plant Sciences 160: 981-1012.
November 2003] LIrr AND STEVENSON-VOCHYSIACEAE
EYDE,R. H., AND C. C. TSENG.1969. Flower of Tetraplasandragymnocarpa:
hypogyny with epigynous ancestry. Science 166: 506-508.
GUSTAFSSON, M. H. G., ANDV. A. ALBERT.1999. Inferiorovaries and an-
giosperm diversification.In P. M. Hollingsworth,R. Bateman, and R. J.
Gornall[eds.], MolecularSystematicsand Plant Evolution,403-431.Tay-
lor and Francis, London, UK.
HUFFORD, L. D. 1989. Structureof the inflorescenceand flower of Petalonvx
linearis (Loasaceae). Plant Systematicsand Evolution 163: 211-226.
IGERSHEIM, A., C. PUFF,P. LEINS,ANDC. ERBAR.1994. Gynoecial devel-
opment of GaertneraLam. and of presumablyallied taxa of the Psycho-
trieae (Rubiaceae):secondarily "superior" vs. inferior ovaries. Botanis-
che Jahrbiicherfir Systematik 116: 401-414.
KAPLAN,D. R. 1967. Floral morphology,organogenesis and interpretationof
the inferiorovary in Downingia bacigalupii. AmericanJournal of Botany
54: 1274-1289.
KAWASAKI, M. L. 1992. Systematics of Erisma (Vochysiaceae). Ph.D. dis-
sertation,City University of New York, New York, New York, USA.
KAWASAKI,M. L. 1998. Systematics of Erisma. Memoirs of the New York
Botanical Garden 81: 1-40.
KEAY,R. W. J., ANDEFA. STAFLEU. 1953. Erismadelphus.Mededelingenvan
het Botanisch Museumen Herbariumvan de Rijksuniversiteitte Utrecht
114: 594-599.
KOPKA,S., ANDE WEBERLING. 1984. Zur Morphologie und Morphogenese
der Blite von Vochvsia acuminata Bong. subsp. laurifolia (Warm.)Sta-
fleu (Vochysiaceae). Beitriige zur Biologie der Pflanzen 59: 273-302.
KUZOFF, R. K., L. HUFFORD, ANDD. E. SOLTIS. 2001. Structuralhomology
and developmentaltransformationsassociated with ovary diversification
in Lithophragma(Saxifragaceae).AmericanJournal of Botany 88: 196-
205.
A. 1999. Floral morphology and phylogeny of Vochysiaceae. Ph.D.
LrrITTr,
dissertation,City University of New York, New York, New York, USA.
Lrrr, A., ANDM. W. CHASE.1999. The systematic position of Euphronia,
with comments on the position of Balanops: an analysis based on rbcL
sequence data. SystematicBotany 24: 401-409.
LITT, A., AND M. CHEEK. 2002. Korupodendronsongweanumn, a new genus
and species of Vochysiaceae from West-CentralAfrica. Brittonia54: 13-
17.
MARCANO-BERTI, L. 1969. Un nuevo genero de las Vochysiaceae. Pittieria
2: 3-28.
MARCANO-BERTI, L. 1989. Vochysiaceae: novidades y correciones. Pittieria
18: 5-14.
MARTINS, H. E 1981. O genero Callisthene Martius (Vochysiaceae): ensaio
GYNOECIUM STRUCTURE 1547
para uma revisdo taxin6mica. Ph.D. dissertation, Universidade Federal
do Rio de Janeiro, Rio de Rio Janeiro, Brazil.
MILDBRAED.J. 1913. Erismadelphusexsul Mildbr.n. gen. et spec. Eine Vo-
chysiacee aus Kamerun.Botanische Jahrbiicherfiir Svstematik49: 547-
551.
OLIVEIRA. P. E. 1998. Reproductive biology, evolution and taxonomy of
Vochysiaceae in central Brazil. In S. J. Owens and P J. Rudall [eds.],
Reproductivebiology in systematics, conservation,and economic botany,
381-393.Royal Botanic Gardens, Kew, UK.
RONSEDE CRAENE, L. P., ANDE. SMETS.1998. Meristic changes in gynoe-
cium morphology, exemplified by floral ontogeny and anatomy.In S. J.
Owens and P. J. Rudall [eds.], Reproductivebiology in systematics,con-
servation, and economic botany, 85-112.Royal Botanic Gardens, Kew,
UK.
SAVOLAINEN.V., M. W. CHASE,S. B. HOOT,C. M. MORTON, D. E. SOLTIS,
C. BAYER, M. E FAY, A. Y. DE BRUIJN, S. SULLIVAN,AND Y.-L. QIU.
2000. Phylogenetics of flowering plants based on combined analysis of
plastid atpB and rbcL gene sequences. SystematicBotany 49: 306-362.
SCHMID,R. 1977. Stockwell's bleach, an effective remover of tannin from
plant tissues. Botanische Jahrbiicherfiir Systematik98: 278-287.
SOLTIs.D. E., AND L. HUFFORD. 2002. Ovary position in Saxifragaceae:
clarifying homology of epigyny. InternationalJournal of Plant Sciences
163: 277-293.
SOLTIS.D. E., ET AL. 2000. Angiosperm phylogeny inferred from a combined
data set of 18S rDNA. rbcL, and atpB sequences. Botanical Journal of
the Linnaean Society 133: 381-461.
STAFLEU. F A. 1948. A monograph of the Vochysiaceae. I. Salvertia and
E
Vochysia. Recueil des TravauxBotaniques Nierlandais 41: 398-540.
STAFLEU. E A. 1952. A monographof the Vochysiaceae. II. Callisthene.Acta
Botanica Neerlandica 1: 222-242.
STAFLEU, E A. 1953. A monographof the Vochysiaceae. III. Qualea. Acta
Botanica Neerlandica 2: 144-217.
STAFLEU, E A. 1954. A monographof the Vochysiaceae. IV. Erisma. Acta
Botanica Neerlandica 3: 459-480.
TAKHTAJAN.A. 1997. Diversity and classification of flowering plants. Co-
lumbia University Press, New York, New York, USA.
WARMING,E. 1875. Vochysiaceae.hi C. E P Martius[ed.], Florabrasiliensis,
Vol. 13, 17-116. R. Oldenburg,Munich,Germany.
WEBERLING.E 1981. Morphology of flowers and inflorescences. Cambridge
University Press. Cambridge,England.
YEO,.P E 1993. Secondarypollen presentation:form, function,and evolution.
SpringerVerlag, Vienna, Austria.