American Journal of Botany 85(6): 876–887. 1998.
IRCUMSCRIPTION OF THE
RELATIONSHIPS TO OTHER
Department of Botany, University of Wisconsin, 430 Lincoln Drive, Madison, Wisconsin 53706;
Harvard University Herbaria, 22 Divinity Avenue, Cambridge, Massachusetts 02138;
Molecular Systematics Section, Jodrell Laboratory, Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3DS, UK; and
De Paul University, Department of Biological Sciences, 2330 W. Kenmore Avenue, Chicago, Illinois 60614
The order Malvales remains poorly circumscribed, despite its seemingly indisputable core constituents: Bombacaceae,
Malvaceae, Sterculiaceae, and Tiliaceae. We conducted a two-step parsimony analysis on 125 rbcL sequences to clarify the
composition of Malvales, to determine the relationships of some controversial families, and to identify the placement of the
Malvales within Rosidae. We sampled taxa that have been previously suggested to be within, or close to, Malvales (83
sequences), plus additional rosids (26 sequences) and nonrosid eudicots (16 sequences) to provide a broader framework for
the analysis. The resulting trees strongly support the monophyly of the core malvalean families, listed above. In addition,
these data serve to identify a broader group of taxa that are closely associated with the core families. This expanded
malvalean clade is composed of four major subclades: (1) the core families (Bombacaceae, Malvaceae, Sterculiaceae,
Tiliaceae); (2) Bixaceae, Cochlospermaceae, and Sphaerosepalaceae (Rhopalocarpaceae); (3) Thymelaeaceae sensu lato(s.l.);
and (4) Cistaceae, Dipterocarpaceae s.l., Sarcolaenaceae (Chlaenaceae), and Muntingia. In addition, Neurada (Neuradaceae
or Rosaceae) falls in the expanded malvalean clade but not clearly within any of the four major subclades. This expanded
malvalean clade is sister to either the expanded capparalean clade of Rodman et al. or the sapindalean clade of Gadek et
al. Members of Elaeocarpaceae, hypothesized by most authors as a sister group to the four core malvalean families, are
shown to not fall close to these taxa. Also excluded as members of, or sister groups to, the expanded malvalean clade were
the families Aextoxicaceae, Barbeyaceae, Cannabinaceae, Cecropiaceae, Dichapetalaceae, Elaeagnaceae, Euphorbiaceae s.l.,
Huaceae, Lecythidaceae, Moraceae s.l., Pandaceae, Plagiopteraceae, Rhamnaceae, Scytopetalaceae, Ulmaceae, and Urtica-
Key words: angiosperm phylogeny; Elaeocarpaceae; Malvales; Muntingia; Neurada; rbcL.
The order Malvales, as traditionally circumscribed, in-
cludes four core families, Bombacaceae (
Malvaceae (1500 spp.), Sterculiaceae (1000 spp.), and
Tiliaceae (400 spp.), plus from one to eight other families
depending upon author (Takhtajan, 1987, 1997; Cron-
quist, 1988; Dahlgren, 1989; Thorne, 1992). Considering
only the core four families, the order comprises predom-
inantly woody and tropical trees, including several eco-
nomically important genera such as cotton (Gossypium
Manuscript received 3 February 1997; revisions accepted 23 Sep-
The authors thank D. Swofford for the use of beta versions of PAUP
4.0; H. Ballard for his translation of Huber (1993) and P. Endress for
calling our attention to that article; H. Ballard, S. Dayanandan, S.Hodg-
es, S. Hoot, V. Savolainen, and K. Wurdack for the use of unpublished
sequences; C. Dunlop, S. Mori, G. Schatz, and M. Zjhra for collecting
plant tissue; and C. Bayer and P. F. Stevens for comments on the manu-
script. This work was supported by National Science Foundation grants
BSF-8800193 to W. Alverson and DEB-9419997 to D. Baum.
Current address: Harvard University Herbaria, 22 Divinity Avenue,
Cambridge, MA 02138.
Current address: Laboratory of Molecular Systematics, MRC 534,
MSC, Smithsonian Institution, Washington, DC. 20560.
Current address: Biology Department, Ithaca College, Ithaca, NY
Current address: Academy of Natural Sciences of Philadelphia,
1900 Benjamin Franklin Parkway, Philadelphia, PA 19103.
Author for correspondence: email: firstname.lastname@example.org.
spp.), chocolate (Theobroma cacao), cola nuts (Cola
spp.), durians (Durio zibethinus), balsa wood (Ochroma
pyramidale), jutes (Corchorus capsularis and C. olito-
rius), and okra (Abelmoschus esculentus). The family
Malvaceae includes herbaceous elements that have un-
dergone extensive radiation in the temperate zones, per-
haps being derived from within tropical Bombacaceae
(Judd, Sanders, and Donoghue, 1994). Bombacaceae,
Sterculiaceae, and Tiliaceae are important components of
many tropical forest ecosystems (Terborgh, 1983; Gentry,
1993) and are often exploited for timber.
The core malvalean families have a good palynological
record stretching back to the Maastrichtian, late Creta-
ceous (Wolfe, 1975; Muller, 1984; Nilsson and Robyns,
1986; Taylor, 1988) and are thought to belong to a lineage
of Gondwanan origin (Raven and Axelrod, 1974; Taylor,
1988). Given this age, it is not surprising that the group
shows a great deal of diversity in vegetative and ﬂoral
morphology, especially in respect to the development of
androphores and gynophores and the splitting and fusion
of the stamens (Edlin, 1935; Venkata Rao, 1952, 1954;
van Heel, 1966; Gibbs, Semir, and da Cruz, 1988). Ad-
ditionally, the group displays a great diversity of chro-
mosome numbers (2n
270; Baker and Baker,
1968; Krapovickas, 1969; Cristo´bal, 1967; Bates and
Blanchard, 1970; Bates, 1976; Ferna´ndez, 1981; Baum
June 1998] 877A
LVERSON ET AL
IRCUMSCRIPTION OF THE
1. Putative members of the Malvales according to four recent authors.
Putative Malvales Takhtajan (1997) Thorne (1992) Dahlgren (1989) Cronquist (1988) Takhtajan (1987) Exem-
The identities of the exemplars included in this study are listed by family in Appendix 1.
Elaeocarpaceae (Elaeocarpales) grouped with Cistales and Malvales in Superorder Malvaneae of Takhtajan (1997).
and Oginuma, 1994), and much variation in pollination
and breeding systems, (e.g., Cope, 1958, 1962; Seavey
and Bawa, 1986; Gibbs and Bianchi, 1993; Baum, 1995).
Nonetheless, the four core families have certain typical
features, such as ﬁbrous bark, palmate venation, and val-
vate calyx aestivation that make them morphologically
coherent (Cronquist, 1981). In recent decades, additional
technical characters that may be diagnostic for the order
have been highlighted, such as mucilage canals or ducts,
tangentially stratiﬁed phloem, tile cells (specialized, up-
right cells) in xylem rays, and cyclopropenoid fatty acids
(Cronquist, 1981, 1988). Indeed, the close relationship
among these families has been recognized since the time
of Linnaeus (e.g., Bentham, 1862). Beyond the core four
families, however, there has been no agreement as to
which other families or taxa should be included in the
order nor has there been agreement as to the closest rel-
atives of the Malvales (Table 1).
The lack of a sound malvalean phylogeny is a major
impediment to systematic and evolutionary studies of this
order. For example, recent attempts to discuss historical
phytogeography (Aubre´ville, 1975; Krutzsch, 1989) and
the evolution of life history traits (Eriksson and Bremer,
1992) at familial and interfamilial levels depended heavi-
ly on tenuous assumptions about phylogenetic relation-
ships inferred from traditional classiﬁcation systems. At
smaller taxonomic scales, attempts to determine intertri-
bal relationships of Sterculiaceae (Kelman, 1991) and to
determine generic limits of Bombacaceae using cladistic
principles (Alverson, 1994) were thwarted by the lack of
a resolved phylogeny at higher levels within this group.
Manchester (1992) also noted that ‘‘. . .the need for mod-
ern systematic revision [of the Malvales] has impeded the
understanding of paleobotanical data.’’ In sum, the de-
termination of phylogeny at all taxonomic scales within
the ‘‘Malvales’’ depends critically on the identiﬁcation of
a monophyletic malvalean clade and determination of its
In addition to the traditional, intuitive systems of clas-
siﬁcation, two recent studies have used cladistic methods
to study the internal phylogeny of Malvales. La Duke and
Doebley (1995) employed chloroplast restriction frag-
ment length polymorphism (RFLP) data to determine a
phylogeny of the Malvaceae, using Tilia as an outgroup.
Based on morphological data, Judd and Manchester
(1997) inferred the relationships of the four core malva-
lean families, using representatives of Dipterocarpaceae
and Elaeocarpaceae as outgroups. The latter study con-
cluded that, of the core malvalean families, only Mal-
vaceae appeared to be monophyletic. However, the anal-
ysis resulted in highly unresolved trees, and those clades
that were obtained had weak internal support.
Five previous studies speak to the circumscription and
placement of ‘‘Malvales’’: the broad rbcL analysis of
Chase et al. (1993), the more focused rbcL studies of
Capparales (Rodman et al., 1993), Myrtales (Conti, Litt,
and Sytsma, 1996) and Sapindales (Gadek et al., 1996),
and the broad 18S rDNA studies of Soltis et al. (1997).
In the rbcL studies, the malvalean exemplars formed a
clade within Rosidae that was closely associated with the
expanded Capparales, Sapindales, and Myrtales. The
study of nuclear rDNA resulted in an equivocal place-
ment of core malvalean taxa within rosids. However,
these studies did not include many of the taxa that have
at one time or another been associated with Malvales.
Therefore, we undertook a study using rbcL data that
directly addresses the phylogeny of Malvales.
MATERIALS AND METHODS
Taxon sampling—The total number of rbcL sequences used for phy-
logenetic analyses of Malvales was 125 (see Appendix 1 for a complete
list of sequences and their provenance), of which 18 were generated for
this study. All taxa that have been suggested as being within Malvales
were sampled except for Diegodendron and Dirachma, which are being
studied by Fay et al. (1998), and Thulin et al. (in press), respectively.
In addition to 29 sequences from putative Malvales (see Table 1), 29
sequences were included to represent all the groups that have been
suggested by recent authors (Cronquist, 1968, 1988; Dahlgren, 1983,
1989; Thorne, 1983) to be potential malvalean sister groups: Aextoxi-
caceae, Barbeyaceae, Cannabinaceae, Dichapetalaceae, Elaeagnaceae,
Euphorbiaceae (subfamilies Acalyphoideae, Crotonoideae, Euphorbioi-
deae, Oldﬁeldioideae, and Phyllanthoideae), Lecythidaceae, Pandaceae,
Rhamnaceae, Scytopetalaceae, Simmondsiaceae, Thymelaeaceae (Aqui-
larioideae, Gonystyloideae, and Thymelaeaeoideae), Ulmaceae (Celtoi-
878 [Vol. 85A
Fig. 1. Structure of backbone constraint trees used to test member-
ship in expanded Malvales. The core malvean clade was represented by
exemplars of Bombacaceae, Malvaceae, Sterculiaceae, and Tiliaceae.
Three exemplars of Papaveraceae, Ranunculaceae, and Trochodendra-
ceae (Appendix 1) were used as outgroups.
deae and Ulmoideae), and Urticaceae (Cecropioideae, Moroideae, and
Urticoideae). A further 25 sequences represented taxa that have been
linked to the core four families based on recent rbcL analyses; our
selection of taxa was guided by the published rbcL phylogenies for
these groups: expanded Capparales (Rodman et al., 1993), Sapindales
(Gadek et al., 1996), and Myrtales (Conti, Litt, and Sytsma, 1996).
Finally, an additional 42 sequences were included to represent the major
clades of eudicots shown in trees produced by Search II of Chase et al.
(1993): 13 from the rosid I clade (in which the core Malvales families
fell); four from each of the rosid II and rosid III clades; ﬁve from the
rosid IV clade; four from each of the asterid I and asterid III clades;
three from the asterid II clade; two from the asterid IV clade; two
placeholders from the ranunculid clade; and one from the hamamelid II
clade (Appendix 1). We did not extensively sample the core malvalean
families because our preliminary data showed little rbcL sequence di-
vergence among members of these core families, and because our em-
phasis was on relationships at a broader scale. A concurrent and com-
plementary study employs the more rapidly evolving chloroplast gene
ndhF to examine phylogenetic structure at the tribal level within the
core malvalean families (Alverson et al., unpublished data). Molecular
studies of malvalean chloroplast atpB (M. Chase and C. Bayer) and
single-copy nuclear gene sequences (F. Blattner, Johannes Gutenberg
Universita¨t Mainz, and M. Jenny, Palmengarten, Frankfurt) are also un-
Laboratory methods—DNA extraction, rbcL ampliﬁcation, and se-
quencing followed the general procedure described in Conti, Litt, and
Sytsma (1996). Overlapping sequence fragments were obtained from
both strands of the gene using a total of eight primers. Primer sequences,
of length 26–34 nucleotides, were provided by G. Zurawski (Zurawski
et al., 1981; Zurawski, Whitﬁeld, and Bottomley, 1986). Sequences
were easily aligned by visual inspection and no gaps were needed.
Translation of the nucleotide sequences to the corresponding amino acid
sequences with the TRANSLATE program of the Genetics Computer
Group (GCG) package (http://www.gcg.com/) indicated no internal stop
codons. A total of 1402 nucleotides, from positions 27 to 1428 of the
rbcL exon were used.
Phylogenetic analyses—The analysis was carried out in two separate
steps. The ﬁrst step was designed to identify the membership of a broad
clade, the ‘‘expanded Malvales,’’ comprising taxa closely associated
with the core malvalean families. The second step was a focused study
of the expanded malvalean clade plus its closest relatives. Unless oth-
erwise stated, all parsimony searches used random addition sequences
(RAS) and were run with tree bisection reconnection branch swapping
1, steepest descent off, and zero-length branches col-
Identiﬁcation of the malvalean clade—The ﬁrst step of the analysis
used all 125 taxa. The analysis used PAUP 4.0.0d versions 31 to 52
(Swofford, 1997) on ﬁve boxes: Power Macintosh models 7100/66 and
7100/80AV running Systems 7.5 and 7.5.3, a Macintosh PowerBook
5300/100 running System 7.5.2, a Macintosh Duo 2300/100 with Sys-
tem 7.5.5, and a Sun UltraSparc running Solaris OS 5.5, for a total of
50 complete, unconstrained RAS searches. The shortest trees from all
50 runs were condensed and then summarized by a single strict con-
sensus tree. The phylogenetic signal present in the full rbcL data set
was estimated using PAUP by analyzing the distribution of 1000 000
random trees and calculating the skewness of their distribution (Hillis
and Hulsenbeck, 1992; but see Ka¨llersjo¨ et al., 1992 for shortcomings
of this method). The inordinate amount of time required for these runs
(average 39 h each) made standard bootstrap analysis impractical. In-
stead, the strength of support for 12 clades of particular interest (see
Results) was assessed by decay analysis (‘‘Bremer support’’; Bremer,
1988) using inverse constraints and 100 RAS searches with MULPARS
off, and by a total of 100 bootstrap searches, each consisting of ten
RAS searches without branch swapping.
We wished to avoid excluding any taxa from the second step of the
analysis that are actually members of the expanded Malvales. Therefore,
for each putatively malvalean taxon that was excluded from the ex-
panded Malvales in the consensus tree, we assessed the cost of forcing
it into the expanded Malvales. This was achieved by conducting inverse
backbone-constraint searches (Swofford, 1997). The backbone con-
straint was that the taxon in question was more closely related to the
core Malvales than are the Capparales and Sapindales (Fig. 1). To orient
the constraint tree, Caltha, Papaver, and Trochodendron were used as
outgroups. For each taxon evaluated in this way (see Results), three
RAS searches were conducted.
Structure of the malvalean clade—The second step of the analysis
included 43 taxa representing the expanded malvalean clade plus the
Capparales and Sapindales. This data set was subjected to 1000 uncon-
strained RAS searches with steepest descent on and members of the
Sapindales used as an outgroup. To see the effect of the limited sam-
pling within Capparales and Sapindales, an additional 1000 RAS/steep-
est descent searches were conducted with these two clades constrained
to have the strict consensus topology obtained by Rodman et al. (1993)
and Gadek et al. (1996), respectively. A total of 100 bootstraps, each
consisting of ten RAS searches with TBR branch swapping, were con-
ducted to estimate bootstrap support (Felsenstein, 1985). Decay indices
were estimated for clades on the strict consensus tree. The phylogenetic
signal present in this subset of rbcL data was also assessed by exam-
ining the skewness of 1000000 random trees, as discussed above.
Identiﬁcation of the malvalean clade—The full data
set included 506 potentially informative characters.
Equally weighted parsimony analysis (Fitch, 1971) yield-
ed 4616 trees of 4733 steps (including autapomorphies).
These trees are distributed on one large island (4344
trees), which was found just once, and ﬁve small islands
(16, 16, 40, 48, 48, and 168 trees) found 15 times. The
4616 trees have a consistency index (CI) of 0.240 (in-
cluding all characters) and a retention index (RI) of
0.484. The data are signiﬁcantly skewed as judged by a
0.280 (Hillis and Huelsenbeck, 1992).
The portion of the consensus tree relevant to the Mal-
vales exhibited robust phylogenetic structure (Fig. 2) in
the form of seven major clades: (a) the four core mal-
valean families, (b) a bixalean clade, (c) a thymelaealean
clade, (d) a dipterocarpalean clade, (e) the expanded Cap-
parales, (f) the Sapindales, and (g) the Myrtales. The core
malvalean, bixalean, and thymelaealean clades together
form a clade to which either the dipterocarpalean clade
or the expanded Capparales clade is a sister group. The
Sapindales clade was sister to clades a–e and the Myrtales
June 1998] 879A
LVERSON ET AL
IRCUMSCRIPTION OF THE
Fig. 2. Portion of the consensus tree, representing the consensus of 4616 trees at 4733 steps, resulting from step 1 of the analysis (including
125 taxa), in which core malvalean taxa and their close relatives were found. Numbers above branches of the major clades of interest represent
decay values derived from 100 TBR/RAS searches without MULPARS. Numbers below these branches are bootstrap values from 100 bootstrap
searches, each consisting of ten RAS searches without branch swapping.
clade was sister to clades a–f. This result provides a spe-
ciﬁc criterion for selecting taxa to be included in the more
focused analysis: namely all taxa falling within clades a–f.
Representative place-holders for 30 putative malvalean
taxa were not located in any of these seven major clades.
Before rejecting a malvalean placement of these taxa, we
tested the strength of this exclusion using backbone con-
straint searches (see Materials and Methods). By using
backbone rather than monophyly constraints, taxa other
than those speciﬁed in the constraint tree were free to
ﬁnd their optimal position. For all 30 taxa, at least nine
additional steps were required to satisfy the backbone
constraint. In none of these cases did the test taxon fall
into one of the clades of the expanded Malvales (clades
a–d, above) but, rather, they were forced into a position
between the expanded Malvales clade and the Capparales
and Sapindales clades. This result justiﬁes the exclusion
of the following families from further consideration as
members of the expanded malvalean clade: Aextoxica-
ceae, Barbeyaceae, Cannabinaceae, Cecropiaceae, Di-
chapetalaceae, Elaeocarpaceae, Elaeagnaceae, Euphorbi-
aceae s.l., Huaceae, Lecythidaceae, Moraceae, Pandaceae,
Plagiopteraceae, Rhamnaceae, Scytopetalaceae, Sim-
mondsiaceae, Ulmaceae, and Urticaceae. Concurrent
work by Thulin, Bremer, and Chase (in press) also ex-
cludes Dirachma as a member of the expanded Malvales.
Structure of the malvalean clade—The reduced data
matrix of 43 taxa included 285 phylogenetically infor-
mative characters. The data are signiﬁcantly skewed, as
judged by a g
0.549 (Hillis and Huelsenbeck, 1992).
An unconstrained, equally weighted search yielded two
equally parsimonious trees of length 1213 (including all
characters), with CI
0.472 and RI
0.609. These two
trees, which only differed by their topologies within Sap-
indales, were located in a single island that was found in
663 of the 1000 RAS replicates.
The strict consensus of the two most-parsimonious
trees is shown in Fig. 3, with branch lengths, and boot-
strap and decay values. The tree was rooted on the branch
880 [Vol. 85A
Fig. 3. Internal structure of the expanded malvalean clade. Shown is the strict consensus of the two most parsimonious trees with the set of 43
taxa used during the second step of the analysis, after 1000 unconstrained TBR/RAS searches with MULPARS. Numbers reported above the
branches indicate branch lengths. Decay and bootstrap values are reported in parentheses below those branches with bootstrap values
names follow Mabberley (1987).
between Sapindales and the remaining taxa, as suggested
by the ﬁrst step of the analysis, and was identical to the
consensus tree from the ﬁrst analysis with two excep-
tions: the bixalean, thymelaealean, and and dipterocar-
palean clades together form a weakly supported clade
sister to the core malvalean clade; and Neurada is re-
moved from the dipterocarpalean clade to become a sister
group to all other members of the expanded Malvales.
A reanalysis of the same data matrix in which the Cap-
parales and Sapindales were constrained to have the to-
pology found in Rodman et al. (1993) and Gadek et al.
(1996) found nine trees (not shown) of length 1218, CI
0.740 (with autapomorphies), and RI
uted in a single island found in 1000 of the 1000 RAS
replicates. The trees resulting from this constrained
search showed a reduced resolution within the core mal-
valean clade and at its base, where there was an unre-
solved polytomy with the bixalean, thymelaealean, and
Identiﬁcation of the expanded Malvales—These anal-
yses suggest the existence of a clade that we refer to as
the expanded Malvales, here deﬁned as all taxa that are
more closely related to the core four families than they
are to either Capparales or Sapindales. The expanded
Malvales clade comprises four major lineages (see
above): (a) a core malvalean clade consisting of Bom-
bacaceae, Malvaceae, Sterculiaceae, and Tiliaceae; (b) a
bixalean clade that includes Bixaceae, Cochlosperma-
ceae, and Sphaerosepalaceae; (c) a thymelaealean clade
that includes Thymelaeaceae s.l. (i.e., including Aquilar-
iaceae and Gonystylaceae); and (d) a dipterocarpalean
clade that includes Cistaceae, Dipterocarpaceae s.l. (i.e.,
including Monotaceae), Sarcolaenaceae, and Muntingia
(which has been variously placed in Elaeocarpaceae, Fla-
courtiaceae, or Tiliaceae). In addition, the monotypic
Neurada (Neuradaceae or Rosaceae) falls within the ex-
panded Malvales, but its exact placement is uncertain (see
Our analyses are equivocal as to the relationships
among these four main clades. The ﬁrst analysis, using
all 125 taxa, indicated that either the bixalean or the thy-
melaealean clade was sister to the core malvalean clade
(Fig. 2), but the second, more focused analysis placed
these two lineages, together with the dipterocarpalean lin-
eage, into a single clade that was sister to the core Mal-
vales (Fig. 3). The reason for this disparity is apparent in
the second analysis as the branch supporting the uniﬁed
sister clade was weak (
50% bootstrap, decay at one
additional step), as was the branch supporting the thyme-
June 1998] 881A
LVERSON ET AL
IRCUMSCRIPTION OF THE
To quantitatively assess the differences between the to-
pologies resulting from the ﬁrst and second analyses, we
used the resulting consensus trees as reciprocal con-
straints. To impose the topology found by the ﬁrst anal-
ysis onto the data set of 43 taxa used in the second anal-
ysis, we pruned 82 taxa from each of the 4616 shortest
trees from the ﬁrst analysis and then condensed the trees.
This procedure yielded 26 trees. Two trees were 1214
steps long, one step longer than the shortest trees origi-
nally found by the second analysis. In these trees, either
the bixalean or bixalean
thymelaealean clades were
sister to the core Malvales. The other 24 trees were 1215
steps in length and showed the bixalean and/or thyme-
laealean clades as sister to the core Malvales. The topol-
ogy found by the second analysis was imposed onto the
full data set by using each of the two shortest trees from
the second search as a strict constraint tree during ﬁve
RAS searches of the full data set. These searches yielded
128 trees of 4736 steps, three steps longer than those
originally found in the ﬁrst analysis. Thus, there is a
slight quantitative basis for choosing the consensus to-
pology of the ﬁrst analysis over that of the second.
The two analyses also differed qualitatively. The ﬁrst
analysis, using all taxa, had the potential advantage of a
better sample of plesiomorphic character states for the
expanded malvalean clade, whereas the second analysis
offered a much more thorough analysis of the data. For
want of a more compelling basis to choose one topology
over the other, however, the relationship among the four
main clades within the expanded Malvales should be con-
sidered as unresolved.
Morphological data are consistent with the results of
our analyses of molecular data. The occurrence of an ex-
otegmic seed coat with a palisade layer, cyclopropenoid
fatty acids in seeds, complex chalazal anatomy, dilated
phloem rays, and stratiﬁed phloem are potential synapo-
morphies for the expanded Malvales clade (Comer, 1976;
Cronquist, 1981). Other characters common within, but
not unique to, the expanded Malvales clade include ly-
sigenous mucilage canals, stellate hairs, peltate scales,
strong phloem ﬁbers (‘‘bast’’), valvate sepals, ellagic
acid, and the presence of an epicalyx.
The core Malvales—It is perhaps no great surprise that
our analyses of rbcL data strongly support the monophyly
of the core four families consisting of exemplars of Bom-
bacaceae, Malvaceae, Sterculiaceae, and Tiliaceae: these
families have been at the heart of every modern classi-
ﬁcation of Malvales. Judd and Manchester (1997) noted
that the best contenders for morphological synapomor-
phies for the four families may be the type of nectaries,
composed of tightly packed, multicellular hairs and nor-
mally found on the adaxial surface of the sepals (Brown,
1938), and the distinctive upright ‘‘tile’’ cells in wood
rays (Manchester and Miller, 1978). C. Bayer (1994; Bay-
er and Kubitaki, 1996) recently determined that inﬂores-
cences of all members of these core families are com-
posed of specialized modules, called ‘‘bicolor units,’’ that
also appear to provide a synapomorphy for the core Mal-
Due to the low bootstrap and decay values generated
by this data set (Fig. 3), little can be said at this time
about the relationships within the core Malvales. How-
ever, two features warrant some comment. The fact that
among the traditional families, only the family Malvaceae
appears to be monophyletic is not surprising. Previous
systematic analyses have usually placed the Malvaceae
as the most ‘‘advanced’’ group derived from Bombaca-
ceae, which in turn is derived from within Sterculiaceae
or Tiliaceae (Edlin, 1935; Venkata Rao, 1952; Cronquist,
1981, 1988). Hence, we expected paraphyly of the latter
three families, a result also obtained in the morphological
cladistic analysis of Judd and Manchester (1997). Sec-
ondly, the separation of Byttneria and Theobroma (sub-
family Byttnerioideae) from the rest of Sterculiaceae and
their sister group position (along with some Tiliaceae) to
the other core Malvales has also been found in our anal-
yses of ndhF sequences (Whitlock, Alverson, and Baum,
1996; Alverson et al., unpublished data). Members of the
Byttnerioideae have long been seen as distinct elements
of Sterculiaceae, and the group has even been recognized
at the familial rank as Byttneriaceae (Edlin, 1935; Cron-
quist, 1981, 1988). Further work is needed, however, to
conﬁrm and elaborate this unexpected but very plausible
The bixalean clade—Bixaceae, Cochlospermaceae,
and Sphaerosepalaceae formed a clade in all analyses
(Figs. 2–3). These three families were linked with Cis-
taceae by Dahlgren (1983, 1989), in part because of the
presence of an exotegmic seed coat with a palisade layer.
Takhtajan (1969, 1997) and Thorne (1992) also grouped
Bixaceae and Cochlospermaceae together with Cistaceae
but excluded Sphaerosepalaceae (Table 1). Notably,
Thorne (1992) associated the monotypic Malagasy family
Diegodendraceae with Sphaerosepalaceae, a placement
supported by morphological data reported elsewhere
(Dickison, 1988; J. Horn, University of North Carolina,
personal communication). A recent molecular study also
indicated the existence of a clade formed by Diegoden-
dron, Bixaceae, and Cochlospermaceae, but was equiv-
ocal on the placement of Sphaerosepalaceae (Fay et al.,
The taxa comprised in the bixalean clade possess many
characteristics traditionally associated with malvalean
taxa, e.g., mucilage cells and canals, wedge-shaped phlo-
em rays, stratiﬁed phloem, ﬁbrous bark, exotegmic seed
coats with palisade layers (at least in Bixa and Cochlo-
spermum; Corner, 1976; Dahlgren, 1983), but theselikely
are synapomorphies for a broader clade rather than the
bixalean clade alone. Cronquist (1981: 394) excluded
Bixaceae s.l. (including Cochlospermaceae) from Mal-
vales because ‘‘Neither [Bixa nor Cochlospermum] have
been reported to produce the characteristic cyclopropen-
oid fatty acids of the Malvales.’’ However, to our knowl-
edge, neither family has been examined for the presence
of these chemical compounds. The homology of the dis-
coid staminal nectaries of Bixa and Cochlospermum (as
well as in Capparaceae, Cistaceae, Sarcolaenaceae, and
Thymelaeaceae; Brown, 1938) with the nectariferous sur-
faces inside the sepals of core Malvales also warrants
The thymelaealean clade—Aquilariaceae, Gonystyla-
ceae, and Thymelaeaceae have been grouped together by
numerous authors and are sometimes combined in a sin-
882 [Vol. 85A
gle family, Thymelaeaceae (e.g., Mabberley,1997). Dahl-
gren and Thorne (1984) and Conti, Litt, and Sytsma
(1996) discussed the placement of this lineage, which has
been often associated with Myrtales because of the oc-
casional presence of internal phloem and vestured pits.
However, neither character unambiguously groups
Thymelaeaceae s.l. with Myrtales, since vestured pits are
also common in the dipterocarpalean clade, and both fea-
tures are absent from Gonystylaceae. The presence of cy-
clopropenoid fatty acids (Vickery, 1980, 1981) and ex-
otegmic seed coats with palisade layers (Corner, 1976;
Dahlgren, 1983), mucilage cells, tough phloem ﬁbers
(Dahlgren and Thome, 1984), broad phloem rays (Cron-
quist, 1981), and stratiﬁed phloem (at least in Dirca; Za-
hur, 1959) make for a comfortable placement of the
Thymelaeaceae clade within the expanded Malvales.
The rbcL trees here, and in Conti, Litt, and Sytsma
(1996), suggest that Gonystylus is sister to the rest of this
clade, as ﬁrst suggested by Dahlgren (1983: 129): ‘‘If
[subfamily Gonystyloideae] can be convincingly shown
to be Thymelaeaceae [then it] probably represents a prim-
itive group, and the internal phloem and vestured pits
would have evolved within the family ....’’ Further
study is needed to identify morphological synapomor-
phies that might support the monophyly of the thymelae-
The dipterocarpalean clade—This clade consists of
representatives of Cistaceae, Dipterocarpaceae, and Sar-
colaenaceae. These three families have been placed sin-
gly or collectively in Malvales by Dahlgren (1983, 1989)
and Thorne (1992). Ashton (1982) also discussed the ap-
parently close relationship between Dipterocarpaceae and
Sarcolaenaceae. Traits linking these families with core
Malvales include the presence of cyclopropenoid fatty ac-
ids (in several genera of Sarcolaenaceae; Gaydou and Ra-
manoelina, 1983), exotegmic seed coats with a palisade
layer (in Cistaceae and Dipterocarpaceae; Corner, 1976),
mucilage cells or canals and stratiﬁed phloem (with un-
certain occurrence in Cistaceae but known from both oth-
er families, including the dipterocarp subfamily Pakarai-
maeoideae; de Zeeuw, 1977), and ellagic acid (with un-
certain status in Sarcolaenaceae but, likewise, reported
from Pakaraimeoideae by Giannasi and Niklas, 1977).
The association of the members of this clade with one
another is consistent with the presence of vestured pits
in all three families (including Pakaraimaea; de Zeeuw,
1977). Interestingly, de Zeeuw noted the presence of in-
ternal phloem in Pakaraimaea, a trait also found in Myr-
taceae and Thymelaeaceae. Flavonoid spectra are report-
edly consistent with a close association of these families
as well (Gornall, Bohm, and Dahlgren, 1979; Thorne,
1981; Young, 1981).
Muntingia—The neotropical genus Muntingia, which
has been historically placed in the Elaeocarpaceae, Fla-
courtiaceae, or Tiliaceae (Cronquist, 1981; Benn and
Lemke, 1992), falls at the base of the dipterocarpalean
clade. The fact that it falls within the expanded Malvales
is not surprising, given that this monotypic genus has
stellate hairs, mucilage cells, and wedge-shaped second-
ary phloem rays (Zahur, 1959). The seed coat is exo-
tegmic but apparently without a palisade layer (Corner,
1976), and vestured pits and stratiﬁed phloem are lacking
(Metcalfe and Chalk, 1950; Zahur, 1959).
In corroboration of the molecular analysis here, Met-
calfe and Chalk (1950: 265) note that, ‘‘Dicraspidia and
Muntingia differ considerably from the other genera [of
Elaeocarpaceae studied] in their wood anatomy. Except
for these two genera, which have many points in common
with Tiliaceae . . . wood anatomy supports the establish-
ment of the Elaeocarpaceae as a separate family.’’ Di-
craspidia and Neotessmannia, genera with comparably
peripatetic taxonomic histories, share many morphologi-
cal and anatomical traits with Muntingia. Investigation of
their molecular, morphological, and chemical traits may
help to clarify the placement of Muntingia. Meanwhile,
Bayer, Chase, and Fay (1998) recognize Muntingiaceae
as a separate family, which includes Dicraspidia and
Neurada—The placement of Neurada on morpholog-
ical and anatomical grounds has been somewhat contro-
versial. All recent authors have included this and two
closely associated genera (Neuradopsis and Grielum)in
the Rosaceae or as a closely related segregate family, the
Neuradaceae. Ronse Decraene and Smets (1995) also
‘‘conﬁdently suggest’’ a relationship with the Rosaceae
based on ﬂoral development. However, a recent review
of the systematic position of Neurada by Huber (1993)
provides convincing evidence that Neurada exhibits two
potential synapomorphies for the expanded Malvales: an
exotegmic seed coat with a palisade layer and the pres-
ence of cyclopropenoid fatty acids in seeds. Neurada also
shares other characteristics common in, but not unique
to, malvalean taxa, including valvate sepals, lysigenous
mucilage canals, stellate hairs, and an epicalyx (Soler-
eder, 1908; Murbeck, 1916, 1941; Willis, 1973; Huber,
The molecular data presented here argue for placement
of Neurada at or near the base of the expanded Malvales
(Figs. 2–3). It takes six additional steps to force the Neu-
rada sequence out of the expanded Malvales (using a
backbone constraint tree and three inverse RAS search-
es), and even then it appeared at the base of the combined
expanded Malvales rather
than in the vicinity of Rosaceae. Three additional inverse
RAS searches indicate that 12 additional steps are needed
to force Neurada outside the combined Myrtales
expanded Malvales clade. Thus,
rbcL supports the idea that the true afﬁnities of Neurada
are with the Malvales rather than Rosales.
Families excluded from the expanded malvalean
clade—With the possible exception of the Elaeocarpa-
ceae, none of the rejected families (see Results) should
prove particularly controversial because of the relatively
weak evidence linking each to Malvales. Of these ex-
cluded taxa, the most notable are the ﬁve exemplars for
Elaeocarpaceae, since they have been frequently and con-
sistently associated with the core malvalean families.
Cronquist (1988: 342) noted that ‘‘The Elaeocarpaceae
stand somewhat apart from the rest of the [order Mal-
vales], but even so the relationship is so close that they
have often been included in the Tiliaceae.’’ Yet, Cron-
quist linked Elaeocarpaceae with Malvales on the basis
June 1998] 883A
LVERSON ET AL
IRCUMSCRIPTION OF THE
of a single shared morphological character: valvate sepals
versus the (mostly) imbricate sepals of the putative thea-
lean ancestors. He also considered Elaeocarpaceae to be
the ‘‘most archaic’’ family in the order because of the
absence of virtually all of the diagnostic characters shared
by the core malvalean families. Other modern authors
have agreed with Cronquist’s placement of the Elaeocar-
paceae, with the notable exception of Dahlgren (1989),
who moved the Elaeocarpaceae from Malvales to Rhi-
zophorales but provided no explanation for this change.
The molecular data examined here unequivocally support
a placement of Elaeocarpaceae (Aceratium, Elaeocarpus,
Crinodendron, Sloanea, and Vallea, but not Muntingia)
at a considerable distance from the expanded Malvales,
close to Ceratopetalum (Cunoniaceae).
Molecular results and changes in taxonomic nomen-
clature—Before this new phylogenetic information on
Malvales is formalized taxonomically, our results should
be further corroborated by other morphological and mo-
lecular data. Judd and Manchester (1997) assembled an
extensive morphological data matrix for the core malva-
lean families that can serve as a starting point for further
studies. Such an analysis might provide further resolution
within the expanded Malvales and might serve to rein-
force (or contradict) some of the unanticipated features
of our analyses. Also, we hope that the circumscription
and phylogenetic structure of the expanded Malvales sug-
gested by rbcL will serve to focus attention on some
interesting characters that have not yet been studied in
all the taxa. For example, the presence of cyclopropenoid
fatty acids, until recently thought to serve as a synapo-
morphy for the core four families (Cronquist, 1981,
1988), is now known to occur commonly throughout the
expanded Malvales. This underscores the need for a com-
prehensive and thorough review of the occurrence of
these compounds throughout this part of the Rosidae.
Similarly, other characters often used to diagnose the core
families (e.g., valvate sepals, lysigenous mucilage canals,
stellate hairs, bast ﬁbers, and epicalyces) are scattered
throughout the malvalean, capparalean, and sapindalean
clades, raising questions as to their path of evolution.
Although by no means offering a deﬁnitive phylogeny of
the Malvales, we hope that the rbcL data reported here
will catalyze further phylogenetic studies of the order and
its close relatives.
, V. A., S. E. W
M. W. C
. 1992. Carnivorous
plants: phylogeny and structural evolution. Science 257: 1491–
, W. S. 1994. New species and combinations of Catostemma
and Pachira (Bombacaceae) from the Venezuelan Guayana. Novon
, P. S. 1982. Dipterocarpaceae. In C. G. G. J. van Steenis [ed.],
Flora Malesiana, ser. 1—Spermatophyta, 237–552. Nijhoff, The
, A. 1975. Essais de ge´ophyle´tique des Bombacace´es. Adan-
sonia n. se´r. 15: 57–64.
, H. G.,
. 1968. Chromosome numbers in the Bom-
bacaceae. Botanical Gazette 129: 294–296.
, D. M. 1976. Chromosome numbers in the Malvales III: miscel-
laneous counts from the Byttneriaceae and Malvaceae. Gentes Her-
barum 11: 143–150.
O. J. B
. 1970. Chromosome numbers in the
Malvales. II. New or otherwise noteworthy counts relevant to clas-
siﬁcation in the Malvaceae tribe Malveae. American Journal of
Botany 57: 927–934.
, D. A. 1995. The comparative pollination and ﬂoral biology of
baobabs (Adansonia—Bombacaceae). Annals of the Missouri Bo-
tanical Garden 82: 322–348.
. 1994. A review of chromosome numbers in
Bombacaceae with new counts from Adansonia. Taxon 43: 11–20.
, C. 1994. Zur Inﬂoreszenzmorphologie der Malvales. Disserta-
tiones Botanicae 212. J. Cramer, Berlin.
M. F. F
. 1998. Muntingiaceae, a new
family of dicotyledons with malvalean afﬁnities. Taxon 47: 37–42.
. 1996. Inﬂorescence morphology of some
Australian Lasiopetaleae (Sterculiaceae). Telopea 6: 721–728.
D. E. L
. 1992. Taxonomy of Neotessmannieae
(Tiliaceae). American Journal of Botany 79 (6, supplement): 166–
, G. 1862. Notes on Malvaceae and Sterculiaceae. Journal of
the Linnaean Society 6: 97–123.
, K. 1988. The limits of amino-acid sequence data in angio-
sperm phylogenetic reconstruction. Evolution 42: 795–803.
, W. H. 1938. The bearing of nectaries on the phylogeny of
ﬂowering plants. Proceedings of the American Philosophical So-
ciety 79: 549–595.
. 1993. Phylogenetics of seed plants: an analysis
of nucleotide sequences from the plastic gene rbcL. Annals of the
Missouri Botanical Garden 80: 528–580.
, E., A. F
K. J. S
. 1993. Tribal relationships
in Onagraceae: implications from rbcL sequence data. Annals of
the Missouri Botanical Garden 80: 672–685.
K. J. S
. 1996. Circumscription of Myrtales
and their relationships to other Rosids: evidence from rbcL se-
quence data. American Journal of Botany 83: 221–233.
, F. W. 1958. Incompatibility in Theobroma cacao. Nature 181:
. 1962. The mechanisms of pollen incompatibility in Theobroma
cacao. Heredity 17: 157–182.
, E. J. H. 1976. The seeds of dicotyledons, vols. 1 and 2. Cam-
bridge University Press, Cambridge.
, C. L. 1967. Cromosomas de Malvales. Kurtziana 4: 139–
, A. 1968. The evolution and classiﬁcation of ﬂowering
plants. Houghton Mifﬂin (reprinted 1978 by Allen Press), Boston,
. 1981. An integrated system of classiﬁcation of ﬂowering
plants. Columbia University Press, New York, NY.
. 1988. The evolution and classiﬁcation of ﬂowering plants, 2d
ed. New York Botanical Garden, Bronx, NY.
, G. 1989. The last Dahlgrenogram: system of classiﬁcation
of the Dicotyledons. In K. Tan, R. R. Mill, and T. S. Elias [eds.],
Plant taxonomy, phytogeography and related subjects, 249–260.
Edinburgh University, Edinburgh.
, R. 1983. General aspects of angiosperm evolution and mac-
rosystematics. Nordic Journal of Botany 3: 119–149.
R. F. T
. 1984. The order Myrtales: circumscription,
variation, and relationships. Annals of the Missouri Botanical Gar-
den 71: 633–699.
, C. 1977. Stem anatomy, Part 3 in B. Maguire et al. [eds.],
Pakaraimoideae, Dipterocarpaceae of the western hemisphere. Tax-
on 26: 368–380.
, W. C. 1988. Xylem anatomy of Diegodendron humbertii.
IAWA Bulletin 9: 332–336.
, H. L. 1935. A critical revision of certain taxonomic groups of
the Malvales, parts 1 and 2. New Phytologist 34: 1–20, 122–143.
. 1992. Pollination systems, dispersal
modes, life forms, and diversiﬁcation rates in angiosperm families.
Evolution 46: 258–266.
, M. F., C. B
. 1998. Plastid rbcL sequence data indicate a close afﬁnity
between Diegodendron and Bixa. Taxon 47: 43–50.
, J. 1985. Conﬁdence limits on phylogenies: an approach
using the bootstrap. Evolution 39: 783–791.
884 [Vol. 85A
, A. 1981. Recuentos cromoso´micos en Malvales. Bonplan-
dia 5: 63–71.
, W. M. 1971. Toward deﬁning the course of evolution: minimal
change for a speciﬁc tree topology. Sytematic Zoology 20: 406–
, P. A., E. S. F
M. C. S
M. W. C
. 1996. Sapindales: molecular
delimitation and infraordinal groups. American Journal of Botany
E. S. F
. 1992. Afﬁnities of the Australian en-
demic Akaniaceae: new evidence from rbcL sequences. Australian
Systematic Botany 5: 717–724.
A. R. P. R
. 1983. A survey of the
Sarcolaenaceae for cyclopropene fatty acids. Phytochemistry 22:
, A. H. 1993. A ﬁeld guide to the families and genera of woody
plants of northwestern South America (Colombia, Ecuador, Peru).
Conservation International, Washington, DC.
K. J. N
. 1977. Phytochemistry, part 4 in B.
Maguire et al. [eds.], Pakaraimoideae, Dipterocarpaceae of the
western hemisphere. Taxon 26: 380–385.
M. T. C
. 1991. Evolutionary
relationships of the Caryophyllidae based on comparative rbcL se-
quences. Systematic Botany 17: 1–15.
. 1993. Post-pollination events of Chorisia
(Bombacaceae) and Tabebuia (Bignoniaceae) with late-acting self-
incompatibility. Botanica Acta 106: 64–71.
. 1988. A proposal to unite the
genera Chorisia Kunth and Ceiba Miller (Bombacaceae). Notes
from the Royal Botanic Garden Edinburgh 45: 125–136.
, R. J., B. A. B
. 1979. The distributions
of ﬂavonoids in the angiosperms. Botaniska Notiser 132: 1–30.
, L. E., G. K
D. E. G
. 1994. Phylogenetic
relationships of the Juglandaceae. Plant Systematics and Evolution
, D. M.,
J. P. H
. 1992. Signal, noise, and reli-
ability in molecular phylogenetic analyses. Journal of Heredity 83:
, H. 1993. Neurada—eine Gattung der Malvales. Sendtnera 1:
S. R. M
. 1997. Circumscription of Malva-
ceae (Malvales) as determined by a preliminary cladistic analysis
of morphological, anatomical, palynological, and chemical char-
acters. Brittonia 49: 384–405.
M. J. D
. 1994. Angiosperm fam-
ily pairs: preliminary phylogenetic analyses. Harvard Papers in
Botany 5: 1–51.
, M., J. S. F
. 1992. Skew-
ness and permutation. Cladistics 8: 275–287.
, W. M. 1991. A revision of Fremontodendron (Sterculiaceae).
Systematic Botany 16: 3–20.
, J., R. K. J
. 1992. Phylogenetic implications of rbcL sequence varia-
tion in the Asteraceae. Annals of the Missouri Botanical Garden
, A. 1969. Notas citotaxono´micas sobre Malva´ceas. Bon-
plandia 3: 9–24.
M. W. C
. 1993. Systematics of the Ericaceae,
Empetraceae, Epacridaceae and related taxa based upon rbcL se-
quence data. Annals of the Missouri Botanical Garden 80: 735–
, W. 1989. Paleogeography and historical phytogeography
(paleochorology) in the Neophyticum. Plant Sytematics and Evo-
lution 162: 5–61.
, J. C.,
. 1995. A chloroplast DNA based phy-
logeny of the Malvaceae. Systematic Botany 20: 259–271.
, D. J. 1987. The plant book. Cambridge University Press,
. 1997. The plant book, 2d ed. Cambridge University Press,
, S. R. 1992. Flowers, fruits, and pollen of Florissantia, an
extinct Malvalean genus from the Eocene and Oligocene of western
North America. American Journal of Botany 79: 996–1008.
. 1978. Tile cells and their occurrence in Mal-
valean fossil woods. IAWA Bulletin 1978: 23–28.
. 1995. Comparison of the evolution of
ribulose 1,5-bisphosphate carboxylase (rbcL) and atpB-rbcL non-
coding spacer sequences in a recent plant group, the tribe Rubieae
(Rubiaceae). Journal of Molecular Evolution 41: 920–927.
J. M. D
. 1993. Using sequences of rbcLto
study phylogeny and biogeography of Nothofagus species. Austra-
lian Systematic Botany 6: 441–447.
. 1950. Anatomy of the dicotyledons,
vol. 1. Clarendon Press, Oxford.
D. E. S
. 1993. Phylogenetic relationships
among members of Saxifragaceae sensu lato based on rbcL se-
quence data. Annals of the Missouri Botanical Garden 80: 631–
K. R. R
. 1994. Systematic and evolu-
tionary implications of rbcL sequence variation in Rosaceae. Amer-
ican Journal of Botany 81: 890–903.
, J. 1984. Signiﬁcance of fosil pollen for Angiosperm history.
Annals of the Missouri Botanical Garden 71: 419–443.
, S. 1916. U
¨ber die Organisation, Biologie und verwandt-
schaftlichen Beziehungen der Neuradoideen. Lunds Universitets
˚rsskrift N. F. Avd. 2, 12(6): 1–29.
. 1941. Untersuchungen u¨ber das Androeceum der Rosaceen.
Lunds Universitets A
˚rsskrift N. F. Avd. 2, 37(7): 1–56.
. 1986. Bombacaceae, no. 14. In S. Nilsson
[ed.], World pollen and spore ﬂora. Almqvist & Wiksell, Stock-
, R. G., B. B
J. D. P
A parsimony analysis of the Asteridae sensu lato based on rbcL
sequences. Annals of the Missouri Botanical Garden 80: 700–722.
P. A. R
. 1995. Evidence for the polyphyly of the
Scrophulariaceae based on chloroplast rbcL and ndhF sequences.
Annals of the Missouri Botanical Garden 82: 176–193.
, R. A.,
J. D. P
. 1993. Phylogenetic relationships of the
Geraniaceae and Geraniales from rbcL sequence comparisons. An-
nals of the Missouri Botanical Garden 80: 661–671.
, Y., M. W. C
C. R. P
. 1993. Molecular
phylogenetics of the Magnoliidae: cladistic analyses of nucleotide
sequences of the plastid gene rbcL. Annals of the Missouri Botan-
ical Garden 80: 587–606.
D. I. A
. 1974. Angiosperm biogeography and
past continental movements. Annals of the Missouri Botanical Gar-
den 61: 539–673.
, J. H., H. D. W
J. R. M
. 1992. Nucleotide
sequences of Amaranthus tricolor rbcL. Taxon 41: 201–209.
, J., R. A. P
. 1993. Nucleotide sequences of the rbcL gene indicate
monophyly of mustard oil plants. Annals of the Missouri Botanical
Garden 80: 686–699.
E. F. S
. 1995. The ﬂoral development
of Neurada procumbens L. (Neuradaceae). Acta Botanica Neerlan-
dica 44: 439–451.
, V. A., J. F. M
Molecular phylogeny of families related to Celastrales based on
ﬂanking sequences. Molecular Phylogenetics and Evolu-
tion 3: 27–37.
, S. R.,
K. S. B
. 1986. Late-acting self-incompatibility.
Botanical Review (Lancaster) 52: 196–217.
, H. 1908. Systematic anatomy of the dicotyledons: a hand-
book for laboratories of pure and applied botany. Clarendon Press,
, D. E., D. R. M
1993. Molecular systematics of Saxifragaceae sensu stricto. Amer-
ican Journal of Botany 80: 1056–1081.
M. L. E
. 1990. rbcL
seqence divergence and phylogenetic relationships in Saxifragaceae
sensu lato. Proceedings of the National Academy of Sciences, USA
. 1997. Angiosperm phylogeny inferred from 18S ribo-
June 1998] 885A
LVERSON ET AL
IRCUMSCRIPTION OF THE
somal DNA sequences. Annals of the Missouri Botanical Garden
, D. 1997. PAUP: Phylogenetic analysis using parsimony, pre-
release versions 4.0.0d31 to 4.0.0d52. Laboratory of Molecular
Systematics, Smithsonian Institution, Washington, DC. and Sinauer,
, A. 1969. Flowering plants, origin and dispersal. Oliver &
. 1987. Systema magnoliophytorum. Editoria Nauka, Leningrad
. 1997. Diversity and classiﬁcation of ﬂowering plants. Colum-
bia University Press, New York, NY.
, D. W. 1988. Paleobiogeographic relationships of thepaleogene
ﬂora from the southeastern U.S.A.: implications for West Gondwa-
naland afﬁnities. Palaeogeography, Palaeoclimatology, Palaeo-
ecology 66: 265–275.
, J. 1983. Five New World primates: a study in comparative
ecology. Princeton University Press, Princeton, NJ.
, R. F. 1981. Phytochemistry and angiosperm phylogeny: a sum-
mary statement. In D. A. Young and D. S. Siegler [eds.], Phyto-
chemistry and angiosperm phylogeny, 233–295. Praeger Scientiﬁc,
New York, NY.
. 1983. Proposed new realignments in the angiosperms. Nordic
Journal of Botany 3: 85–117.
. 1992. An updated phylogenetic classiﬁcation of the ﬂowering
plants. Aliso 13: 365–389.
, M., B. B
M. W. C
. In press. Family relationships of the enigmatic rosid
genera Barbeya and Dirachma from the horn of Africa region.
Plant Systematics and Evolution.
, W. A. 1966. Morphology of the androecium in Malvales.
Blumea 13: 177–394.
, C. 1952. Floral anatomy of some Malvales and its bear-
ing on the afﬁnities of families included in the order. Journal of
the Indian Botanical Society 31: 171–203.
. 1954. A contribution to the embryology of Bombacaceae. Pro-
ceedings of the Indian National Science Academy, Part B (Biolog-
ical Sciences) 39: 51–75.
, J. R. 1980. The fatty acid composition of seed oils from ten
plant families with particular reference to cyclopropene and dihy-
drosterculic acids. Journal of the American Oil Chemists Society
. 1981. The occurrence of dihydromalvalic acid in some seed
oils. Journal of the American Oil Chemists Society 58: 731.
, B. A., W. S. A
D. A. B
. 1996. Phyloge-
netic relationships in the tribe Byttnerieae (Sterculiaceae) based on
ndhF sequence data. American Journal of Botany 83 (6, supple-
ment): 166–167 (Abstract).
, J. C. 1973. A dictionary of the ﬂowering plants and ferns, 8th
ed. Cambridge University Press, Cambridge.
, J. A. 1975. Some aspects of plant geography of the northern
hemisphere during the late Cretaceous and Tertiary. Annals of the
Missouri Botanical Garden 62: 264–279.
, D. A. 1981. The usefulness of ﬂavonoids in angiosperm phy-
logeny: some selected examples. In D. A. Young and D. S. Siegler
[eds.], Phytochemistry and angiosperm phylogeny, 205–232. Prae-
ger Scientiﬁc, New York, NY.
, M. S. 1959. Comparative study of secondary phloem of 423
species of woody dicotyledons belonging to 85 families. Cornell
University Agricultural Experiment Station Memoir 358: 1–160.
, G., B. P
P. R. W
The structure of the gene for the large subunit of ribulose-1,5-
bisphosphate carboxylase from spinach chloroplast DNA. Nucleic
Acids Research 14: 3251–3270.
. 1986. Sequence of the
gene for the large subunit of ribulose-1,5-bisphosphate carboxylase
from pea chloroplasts. Nucleic Acids Research 14: 3975.
886 [Vol. 85A
1. The 125 species included in the study, listed alphabetically by family names, which follow Mabberley (1987). Also given are
memberships of these genera in trees found by Chase et al. (1993), GenBank numbers, literature citations for previously published sequences,
and source and voucher information for unpublished sequences.
Species FamilyaChase 93bGenBank Citation, or source and voucher
Acer saccharum L. Aceraceae R2 L01881 Albert, Williams, and Chase, 1992
Aextoxicon punctatum Ruiz & Pavon Aextoxicaceae — X83986 V. Savolainen; voucher unknown
Amaranthus tricolor L. Amaranthaceae R3 (Ca) X53980 Rettig, Wilson, and Manhart, 1992
Schinus molle L. Anacardiaceae R2 U39270 Gadek et al., 1996
Ilex vomitoria Aiton Aquifoliaceae A2 M88583 Chase et al., 1993
Achillea millefolium L. Asteraceae A2 L13641 Kim et al., 1992
Impatiens capensis Meerb. Balsaminaceae A3 — Chase et al., 1993
Barbeya oleoides Schweinf. Barbeyaceae Ca (R3) — M. Chase; Collenette 1/93, K
Tabebuia heterophylla (DC.) Britton Bignoniaceae — L36451 Olmstead and Reeves, 1995
Bixa orellana L. Bixaceae — AF022128 Alverson/Karol; Alverson s.n., WIS
Cochlospermum vitifolium L. Bixaceae — AF022129 Alverson/Karol; Fairchild Bot. Gard. FG X-19-3
Bombax buonopozense P. Beauv. Bombacaceae R2 AF022118 Alverson/Karol; Alverson s.n., WIS
Camptostemon schultzii Mast. Bombacaceae — AF022120 Alverson/Karol; Dunlap s.n., WIS
Durio zibethinus Murr. Bombacaceae — AF022119 Alverson/Karol; Alverson 2180, WIS
Ochroma pyramidale (Cav. ex Lam.)
Urb. Bombacaceae — AF022122 Alverson/Karol; Alverson & Rubio 2246, WIS
Quararibea gomeziana W.S. Alverson Bombacaceae — AF022121 Alverson/Karol; Alverson 2136, WIS
Brassica oleracea L. Brassicaceae R2 M88342 Rodman et al., 1993
Bretschneidera sinensis Hemsl. Bretschneideraceae R2 M95753 Rodman et al., 1993
Bursera inaguensis Britton Burseraceae R2 L01890 Albert, Williams, and Chase, 1992
Humulus lupulus L. Cannabidaceae – (R1) U02729 Chase et al., 1993
Capparis hastata Jacq. Capparidaceae R2 M95754 Rodman et al., 1993
Viburnum acerifolium L. Caprifoliaceae A2 L01959 Olmstead et al., 1993
Carica papaya L. Caricaceae R2 M95671 Rodman et al., 1993
Cecropia palmata Willd. Cecropiaceae — — Conti and Wiegrefe; Fairchild Bot. Gard. 3181
Chrysobalanus icacao L. Chrysobalanaceae R1 L11178 Morgan and Soltis, 1993
Cistus revolii Coste & Soulie Cistaceae — — M. Chase; Chase 524, K
Helianthemum grandiﬂorum DC. Cistaceae — — M. Chase; Chase 525, K
Quisqualis indica L. Combretaceae R2 (R1) L01948 Albert, Williams, and Chase, 1992
Cornus canadensis L. Cornaceae A4 L11213 Qiu et al., 1993
Crossosoma californicum Nutt. Crossosomataceae R2 L11179 Morgan and Soltis, 1993
Alzatea verticillata Ruiz & Pavon Crypteroniaceae — U26316 Conti, Litt, and Sytsma, 1996
Ceratopetalum gummiferum Small Cunoniaceae R1 L01895 Soltis et al., 1990
Daphniphyllum sp. Daphniphyllaceae R4 (R3) L01901 Albert, Williams, and Chase, 1992
Datisca cannabina L. Datiscaceae R1 L21939 Chase et al., 1993
Dichapetalum crassifolium Chodat Dichapetalaceae — X69733 Savolainen et al., 1994
Dillenia indica L. Dilleniaceae R3 (A5) L01903 Albert, Williams, and Chase, 1992
Monotes sp. Dipterocarpaceae — — S. Dayanandan; Harter 3134, MO
Shorea zeylanica (Thwaites) Ashton Dipterocarpaceae – (R2) — Chase et al., 1993
Diospyros virginiana L. Ebenaceae A3 L12613 Kron and Chase, 1993
Elaeagnus angustifolia L. Elaeagnaceae — U17038 Morgan and Soltis, 1993
Aceratium ferrugineum C.T. White Elaeocarpaceae — L28947 P.G. Martin and J. Dowd; voucher unknown
Crinodendron hookerianum Gay Elaeocarpaceae — — M. Chase; Chase 909, K
Elaeocarpus grandis F. Muell. Elaeocarpaceae — L28951 P.G. Martin and J. Dowd; voucher unknown
Muntingia calabura L. Elaeocarpaceae
— — M. Chase; Chase 346, NCU
Sloanea latifolia (Rich.) Schumann Elaeocarpaceae — AF022131 Alverson/Karol; Alverson 2211, WIS
Vallea stipularis L. Elaeocarpaceae — — M. Chase; Chase 654, K
Erica australis L. Ericaceae A3 L12617 Kron and Chase, 1993
Acalypha rhomboidea Raf. Euphorbiaceae — U00435 Gunter, Kochert, and Giannasi, 1994
Croton alabamensis Chapman Euphorbiaceae — — K. Wurdack; Wurdack s.n., NCU
Drypetes roxburghii (Wall.) Hurus. Euphorbiaceae R1 M95757 Rodman et al., 1993
Euphorbia polychroma A. Kerner Euphorbiaceae R1 L13185 Chase et al., 1993
Phyllanthus epiphyllanthus L. Euphorbiaceae — — K. Wurdack; Wurdack s.n., NCU
Tetracoccus dioicus Parry Euphorbiaceae — — K. Wurdack; Levin 2202, SD
Pisum sativum L. Fabaceae R1 X03853 Zurawski, Whitﬁeld, and Bottomley, 1986
Nothofagus balansae (Baill.) Steenis Fagaceae – (R1) L13344 Martin and Dowd, 1993
Plagiopteron suaveolens Griff. Flacourtiaceae — — M. Chase; Chase 1335, K
Geranium cinereum Cav. Geraniaceae R2 L14695 Price and Palmer, 1993
Ribes aureum Pursh Grossulariaceae R4 (R3) L11204 Morgan and Soltis, 1993
Hamamelis mollis Oliv. Hamamelidaceae R4 (R3) L01922 Albert, Williams, and Chase, 1992
Afrostyrax sp. Huaceae — — M. Chase; Cheek 5007, K
Hua gabonii Pierre ex De Wilde Huaceae — — M. Chase, Weiringa 3177, WAG
Hydrangea macrophylla Torr. Hydrangeaceae A4 L11187 Morgan and Soltis, 1993
Krameria lanceolata Torr. Krameriaceae R1 — Chase et al., 1993
Asteranthos brasiliensis Desf. Lecythidaceae — AF022133 Alverson/Karol; Mori 21856, NY
Couroupita guianensis Aubl. Lecythidaceae — AF022134 Alverson/Karol; Alverson s.n., WIS
Eschweilera odora (Poepp.) Miers Lecythidaceae — AF022135 Alverson/Karol; de Granville 5086, CAY
June 1998] 887A
LVERSON ET AL
IRCUMSCRIPTION OF THE
Species FamilyaChase 93bGenBank Citation, or source and voucher
Limnanthes douglasii R. Br. Limnanthaceae R2 L14700 Rodman et al., 1993
Lythrum hyssopifolia L. Lythraceae R2 (R1) L10218 Conti, Fischbach, and Sytsma, 1993
Byrsonima crassifolia (L.) Kunth. Malpighiaceae R1 L01892 Albert, Williams, and Chase, 1992
Gossypium hirsutum L. Malvaceae R2 M77700 Giannasi et al., 1991
Thespesia populnea (L.) Sol. ex Correa Malvaceae – (R2) L01961 Albert, Williams, and Chase, 1992
Rhexia virginiana L. Melastomataceae — U26334 Conti, Litt, and Sytsma, 1996
Guarea glabra Vahl. Meliaceae — U39085 Gadek et al., 1996
Morus alba L. Moraceae R1 L01933 Soltis et al., 1990
Myrica cerifera L. Myricaceae R1 L01934 Albert, Williams, and Chase, 1992
Baeckea ramosissima A. Cunn. Myrtaceae — U26319 Conti, Litt, and Sytsma, 1996
Heteropyxis natalensis Harv. Myrtaceae – (R1) U26326 Conti, Litt, and Sytsma, 1996
Neurada procumbens L. Neuradaceae — U06814 Morgan, Soltis, and Robertson, 1994
Ochna serrulata Walp. Ochnaceae R1 Z75273 Chase et al., 1993
Ludwigia peruviana (L.) Hara Onagraceae R2 (R1) L10221 Conti, Fischbach, and Sytsma, 1993
Hypseocharis sp. Oxalidaceae – (R2) L14699 Price and Palmer, 1993
Oxalis dillenii Jacq. Oxalidaceae R1 L01938 Albert, Williams, and Chase, 1992
Microdesmis puberula Hook. f. Pandaceae — — M. Chase; Cheek 5986, K
Papaver orientale L. Papaveraceae Ra L08764 Chase et al., 1993
Passiﬂora quadrangularis L. Passiﬂoraceae R1 L01940 Albert, Williams, and Chase, 1992
Stegnosperma halimifolium Benth. Phytolaccaceae s.l. R3 (Ca) M62571 Rettig, Wilson, and Manhart, 1992
Polygala cruciata L. Polygalaceae R1 L01945 Albert, Williams, and Chase, 1992
Plumbago capensis Thunb. Plumbaginaceae R3 (Ca) M77702
Giannasi et al., 1991
Caltha palustris L. Ranunculaceae Ra L02431 Albert, Williams, and Chase, 1992
Rhamnus catharticus L. Rhamnaceae R1 L13189 Chase et al., 1993
Rosa woodsii Lindl. Rosaceae — U06824 Soltis et al., 1993
Rubia tinctorum L. Rubiaceae — X81104 Manen and Natali, 1995
Ruta graveolens L. Rutaceae — U39281 Gadek et al., 1996
Cupaniopsis anacardioides Raklk. Sapindaceae R2 L13182 Chase et al., 1993
Sarcolaena oblongifolia F. Ge´rard Sarcolaenaceae — U26337 Conti, Litt, and Sytsma, 1996
Francoa sonchifolia Cav. Saxifragaceae R2 L11184 Soltis et al., 1990
Penthorum sedoides L. Saxifragaceae R4 (R3) L11197 Soltis et al., 1990
Saxifraga integrifolia Hook. Saxifragaceae R4 (R3) L01953 Albert, Williams, and Chase, 1992
Oubanguia alata Baker f. Scytopetalaceae — — M. Chase; Gereau et al. 5202, MO
Ailanthus altissima (Miller) Swingle Simaroubaceae R2 L12566 Gadek et al., 1992
Kirkia wilmsii Engl. Simaroubaceae s.l. — — Gadek et al., 1996
Simmondsia chinensis (Link). S.K. Sch. Simmondsiaceae — — S. Hoot; Hoot s.n., F
Lycopersicon esculentum Mill. Solanaceae A1 L14403 Olmstead et al., 1993
Rhopalocarpus lucidus Bojer Sphaerosepalaceae
— AF022130 Alverson/Karol; Miller and Schatz 6275, MO
Byttneria aculeata (Jacq.) Jacq. Sterculiaceae — AF022123 Alverson/Karol; Alverson s.n., WIS
Fremontodendron mexicanum Davids. Sterculiaceae — AF022124 Alverson/Karol; Thorne 54717, RSA
Sterculia tragacantha Lindl. Sterculiaceae — AF022126 Alverson/Karol; FBG FG X.12-17
Theobroma cacao L. Sterculiaceae – (R2) AF022125 Alverson/Karol; Solheim BUF296, WIS
Camelia japonica L. Theaceae A3 L12602 Kron and Chase, 1993
Daphne mezereum L. Thymelaeaceae — AF022132 Conti/Karol; Conti and Giordano 46, WIS
Aquilaria beccariana Van Tiegh. Thymelaeaceae — — M. Chase; Chase 1380, K
Dirca palustris L. Thymelaeaceae — U26322 Conti, Litt, and Sytsma, 1996
Gonystylus macrophyllus (Miq.) A.
Shaw Thymelaeaceae s.l. — — M. Chase; Chase 1382, K
Phaleria chermsideana (Bailley) C.
White Thymelaeaceae — U26332 Conti, Litt, and Sytsma, 1996
Thymelaea hirsuta Endl. Thymelaeaceae — — M. Chase; Chase 1883, K
Tilia americana L. Tiliaceae – (R2) AF022127 Alverson/Karol; Alverson s.n., WIS
Trochodendron aralioides Sieb. & Zucc. Trochodendraceae H2 L01958 Albert, Williams, and Chase, 1992
Tropaeolum majus L. Tropaeolaceae R2 L14706 Price and Palmer, 1993
Celtis yunnanensis C.K. Schneid. Ulmaceae R1 L12638 Qiu et al., 1993
Ulmus alata Michx. Ulmaceae — U00441 Gunter, Kochert, and Giannasi, 1994
Boehmeria nivea (L.) Gaudich. Urticaceae – (R1) — Chase et al., 1993
Rinorea crenata S.F. Blake Violaceae — — S. Hodges; Ballard 94-006, WIS
Phoradendron serotinum (Raf.) M.C.
Johnst. Viscaceae A1/Ca L11199 Morgan and Soltis, 1993
Vochysia hondurensis Sprague Vochysiaceae — U26340 Conti, Litt, and Sytsma, 1996
Nitraria retusa (Forsskal) Asch. Zygophyllaceae s.l. — U39278 Gadek et al., 1996
With the exception of Muntingia and Rhopalocarpus (see below), all family assignments are the same in Mabberley (1987) and Mabberley (1997).
Clades occupied on trees found by Chase et al. (1993) are indicated by the following abbreviations: A1, A2, A3, A4
asterid clades I, II, III, and
caryophyllid clade; H2
hamamelid clade II; Ra
ranunculid clade; and R1, R2, R3, R4
rosid clades I, II, III, and IV. Citation of a
second clade in parentheses indicates the clade membership of the genus in Search II of Chase et al. (1993) only when different from that of Search I.
Muntingia was later moved to Tiliaceae by Mabberley (1997).
Accession numbers M77701 (listed in GenBank as Plumbago capensis) and M77702 (listed as Rheum
cultorum) are reversed in GenBank, as
evidenced by additional sequences for both genera (Lledo´ et al., unpublished, ﬁde M. Chase).
Rhopalocarpus was later moved to Ochnaceae (Diegodendraceae) by Mabberley (1997).