The Rondeletia Complex (Rubiaceae): An Attempt to Use ITS, rps16 , and trnL-F Sequence Data to Delimit Guettardeae, Rondeletieae, and Sections within Rondeletia 1

2009·3 MB·English

by  Johan H E Rova

n
,
THE RONDELETIA COMPLEX
Johan E. Rova 2 Piero DelpreteP and
G.
ti.
,
(RUBIACEAE): AN ATTEMPT TO Birsitta Bremer4
AND
USE TRNL-F
RPS16,
ITS,
SEQUENCE DATA TO DELIMIT
GUETTARDEAE, RONDELETIEAE,
AND
SECTIONS WITHIN
RONDELETIA
1
Abstract
In the present study, a molecular phytogeny of the Rondeietia L. complex (Rubiaceae, Rondeletieae) was constructed with
the following main objectives: to evaluate the sections of Rondeietia proposed by Fernandez Zequeira; to test if Stevemia
(1) (2)
Poit. belongs to the Rondeletieae s. str.; (3) to check if ITS data from Rondeletieae support previous phylogenetic results from
trnL-F data regarding circumscription of Rondeletieae; and to verify Hodgkimonia F. Muell. belongs to Guettardeae or
(4) if
elsewhere. Two analyses were performed, one with ITS sequences from 46 taxa in the Rondeletieae—Guettardeae complex, and
the other mo th combined ITS, rps!6 and trnL-F sequences of 21 taxa. Representatives of nine of the 10 sections of Rondeietia
,
recognized by Fernandez Zequeira were included in the ITS analysis. Five of her sections could be tested for monophyly.
Support was only found for Rondeietia, sect. Leoninae M. Fernandez Zeq., while representatives from section Chamaebuxifoliae
M. Fernandez Zeq., section Hypoleucae Standi., and section Nipemes M. Fernandez Zeq. together form a well -supported clade
that could be distinguished also based on morphology. The latter clade is sister to Stevemia which is thus placed within
,
Rondeietia s. str. In addition, ITS sequence data confirm the separation of Rovaeanthus Borhidi from Rogiera Planch. Support
&
is low for inclusion of Blepharidium Standi., Mazaea Krug L'rb., Phyllomelia Griseb., Rachicallis DC., part of Rogiera and
,
&
Suberanthus Borhidi M. Fernandez Zeq. in Rondeletieae. Rachicallis Mazaea and Phyllomelia form a clade with strong
, ,
support. The tribe Rondeletieae s. str. W'as found to be monophyletic in all trees, although with low support; however, a re-
delimitation of the tribe is proposed here based on this study and previous phylogenetic analyses. The monophyly for the tribe
Guettardeae is weakly supported, with the inclusion of Arachnoihryx Planch, (including Cuatrecasasiodendron Steyerm.),
&
Gonzalagunia Ruiz Pav., Hodgkinsonia and Timonius DC. Although was recently the subject of a molecular phylogenetic
it
,
study, the tribe Guettardeae is still in need of a wide-ranging survey in order to confirm its monophyly and delimit its
taxonomic boundaries. Because Cuatrecasasiodendro was found within the Arachnothryx clade, the twro genera are here
synonvmized as Arachnothryx, and in turn positioned within the tribe Guettardeae. In addition, based on herbarium and field
&
studies, the tw7o species described under Cuatrecasasiodendron C speciabile Steyerm. and C. colombianum Standi.
.
(
&
Steyerm.) are treated as synonyms to the new7 combination Arachnothryx spectabilis (Steyerm.) Rova, Delprete B. Bremer,
which proposed here.
is
Key words: Arachnoihryx Cuatrecasasiodendron Guettardeae, Hodgkinsonia ITS, phylogeny, Rogiera Rondeietia
, , , , ,
Rondeletieae, rpsl6 Rubiaceae, Stevemia trnL-F
, ,
We
1 are grateful to Attila Borhidi (Janos Pannonius University, Pecs) for help with identification of specimens of Rondeietia
and related taxa and to Nahid Heidari and the staff at the Evolutionary Biology Centre (Uppsala University) tor technical
assistance. The National Tropical Botanical Garden, Lawai, Kauai, Hawaii, provided material for Rondeietia inermis and
Rovaeanthus sirigosus This work was supported by a grant from the Bergius Foundation. Part of this research was realized
.
during a Visiting Scientist fellowship to Piero Delprete (PD) from the Conselho Nacional de Desenvolvimento Cientifico e
Tecnologico (CNPq) at the Universidade Federal de Goias, G< Tania, Goias, Brazil. PD’s fieldwork in Cuba was supported by a
grant from the MaeAr thur Foundation through the American Council of Learned Societies/Social Science Research Council
W orking Group on Cuba. W e also thank the Cuban government for giving permission to PD to perform fieldwork, and Ramona
Oviedo, Mayra Fernandez, Jaqueline Salazar, and Lazara Sotolongo Academia de Ciencias, Institute de Ecologia
y
(
HAC) PD
Sistematica, La Habana, Cuba; for accompanying during fieldwork in 2002.
• I Tagal >erg Baskarp, SE-566 92 Habo, Sweden. Author for correspondence: [email protected].
3 CNPq
Visiting Scientist, Institute of Biological Sciences—ICBrl, Department of General Biology/Botany Universidade
,
Federal de Goias, Campus II, 74001-970 Goiania, Goias, Brazil. Current address: Ins titut de Recherche pour le
Developpement, Botanique et Bioinformatique de FArchitecture des Plantes (AMAP), TA-A51/PS2, Blvd. de la Lironde,
34398 Montpellier Cedex France, [email protected].
5,
4 Bergius Foundation at the Royal Swedish Academy of Sciences and Botany Department, Stockholm University, SE-106 91
Stockholm, Sweden.
10.3417/2006179
doi:
-
Ann. Missouri Bot. Card. 96 182 193 Published on 23 April 2009
: . .
n
Volume Number Rova 183
96, 1 et al.
2009 The Complex
Rondeletia
The tribe Rondeletieae (Rubiaceae, Cinchonoi- are closely related, and this study also aimed to
deae) includes predominantly shrubs and trees and investigate ITS data would suggest a similar
if
New
is mostly distributed in the World tropics circumscription of Rondeletieae as previous studies
(Robbrecht, 1988; Delprete, 1999a), with the main had. Finally, the study was aimed to investigate ITS
if
A
center of diversity in the Greater Antilles. thorough sequence data would support Bremer’s (1992) inclu-
description of the taxonomic and systematic history of sion of the Australian genus Hodgkinsonia F. Muell.
the tribe found in Delprete (1999a). in the Guettardeae or Delprete (1996) transfer of the
is ’s
The largest genus of the tribe, Rondeletia L., is genus from the Guettardeae to the Chiococceae.
mainly Antillean and comprises approximately 120
species. Standley (1918) divided Rondeletia into 15 M M
ATERIA AND ETHOD
LS S
sections based on morphological and distributional
data. Since then, several morphological and molecular For the ITS analyses, material was sampled from as
studies in the Rondeletieae have argued about the many Rondeletia species and subspecies as possible.
An made
statu s of Stand ley's sections and the circumscrip - effort was to include representatives
tion of the genus Rondeletia One opinion is that from all genera in Rondeletieae sensu Rova et al.
.
Rondeletia should be regarded as a narrowly circum- (2002). The outgroup consisted of Luculia Sweet
scribed genus, separated from morphologically similar (basal in Rubiaceae, e.g., Bremer et ah, 1995),
genera such as Arachnothryx Planch., javorhaea Cateshoea fueitesu Urb., Chiococca alba (L.) Hitchc.
&
Borhidi Jarai-KomL, Rogiera Planch., Roigella (Chiococceae and 12 accessions representing 11
s.l.),
&
Borhidi M. Fernandez Zeq., Rovaeanthus Borhidi, species in the following six genera of the tribe
and Suberanthws Borhidi <& M. Fernandez Zeq. Guettardeae (based on available material and the
&
A
(Stevermark, 1967; Borhidi Fernandez Zequeira, results from Rova et ah, 2002): racknothryx
,
&
1981a, b; Borhidi, 1982, 1989, 1994; Borhidi Jarai- Cuatrecasasiodendro Steyerm., Gonzalagunia Ruiz
&
Kornlodi, 1983; Fernandez Zequeira, 1994; Delprete, Pav., Guettarda L., Rogiera and Tirnonius DC.
,
1999a, as Rondeletia complex sensu Delprete; Rova, Authors of species names are given in fable or
1,
1999; Rova et ah, 2002; Borhidi et ah, 2004; Rova, otherwise when first mentioned in the text.
DNA
On
unpublished). the other hand, Lorenee (1991) Fresh or silica gel—dried leaves were used for
recognized Rondeletia as a widely circumscribed extraction when available, but often herbarium
DNA
germs, treating the names applied to Mexican and material had to be used. was extracted using
CTAB &
Central American taxa of the complex as synonyms. the method (Doyle Doyle, 1987) and cleaned
Based on morphological data, Fernandez Zequeira with the QIAquick PC R Purification Kit (QIAGEN
(1994) made an attempt to classify the Greater GmbH, Hi! den, Germany). The cocktail for polymer-
Antillean (especially the Cuban) Rondeletia species ase chain reaction (PCR) amplification w as mixed as
7
X
into 10 sections. Her classification comprised 104 follows (to ca. 25 pi): 2.5 pi 10 buffer, 2.5 pi
DNA
species, most of them endemic to Cuba. This means 25 niM MgCl 2 pi dNTP, 0.125 pi Taq poly-
2
,
pM
that a majority of the species of Rondeletia were merase, 0.625 pi 10 forward primer, 0.625 pi
s. str.
M
pM
included in her study. According to Fernandez 10 reverse primer, 2.5 pi 0.1 EM A Cl, 2 pi
Zequeira (1994), the sections are distinguished by dimethyl sulfoxide (DMSO), 2 pi template, and 10 pi
various combinations of (often multistate) morpholog- water. In some cases, the amount of primer or template
ical characters such as position and shape of was doubled (replacing some of the water). Primers
&
inflorescence, flower merosity, calyx lobe shape, leaf P17 and 26S-82R (Popp Oxelman, 2001) were used
indumentum, and phyllotaxy (leaves opposite for amplification. Sequencing reactions were realized
vs.
DYEnamic ET
However, her focus on Cuban species using the terminator Cycle Sequencing
verticil late).
did little to resolve the problem in the larger kit (Amersham Biosciences, Buckinghamshire, Eng-
Rondeletia complex. The aim of the present land) following the protocol of the manufacturer
first
(DMSO
study was to test Fernandez Zequeira’s sections of was added in the same concentration as in
if
DNA
PCR MegaBACE
Rondeletia are supported by phylogenies obtained the mix) anti run on a 1000
from molecular sequences. The second aim was to test Analysis System (Amersham Biosciences). Eor se-
PCR
if Stevensia Poit. was closely related to Rondeletia or quencing, the same primers were used as in the
not. Stevensia, has not been included previously in amplification.
molecular phylogenies, but morphology suggests a For the ITS study, 50 new ITS sequences were
close affinity between the genera. Earlier studies produced, and five additional sequences were down-
&
(Bremer et ah, 1995; Bremer Thu in, 1998; loaded from Gen Bank and included in the data matrix.
1
&
Andersson Rova, 1999; Rova et al., 2002) have Taxon names, authors, vouchers, and Gen Bank
shown that the tribes Guettardeae and Rondeletieae accession numbers are presented in Table
1.
6 6
s
184 Annals
of the
Garden
Missouri Botanical
Manual alignment and gapcoding of the ITS and resulted in high-quality sequences. was not
It
sequences were performed with the following criteria: possible to obtain sequences from section Lindenianae
an effort was made to see if gaps/insertions could be M. Fernandez Zeq., although extractions were at-
(1)
interpreted as caused by repeats or inversions, and tempted from two different specimens. was also not
if It
PGR
so, sequences were aligned to fit these possible events; possible to obtain products from Roigella
&
gaps (he., inferred insertion/deletion events) were eorreifolia (Griseb.) Borhidi M. Fernandez Zeq.,
(2)
introduced into the sequences to keep the number of which trnL-F data showed to be closely related to
substitutions in an aligned region to a minimum; (3) Rondeletia s. str. (Rova et ah, 2002). For four sections
insertions/deletions and substitutions were considered Rondeletia sect. Odoratae Standi., section Pedicel-
(j
equally probable events; and gaps/insertions of lares Standi., section Rigidae M. Fernandez Zeq., and
(4)
equal length shared by two or more taxa were inferred section Chamaebuxifoliae was only possible to
it
),
We
to be homologous and binary coded. Gaps of more than sequence one species from each section. were
one position in length introduced due to multiplication unfortunately not able to sequence the type species of
of single nucleotides, e.g., poly-A, were not coded. Rondeletia R. americana L. This species seems to be
,
Regions where alignment could not be unambiguously very rarely collected, and extractions made from the
interpreted were removed from the analysis. After herbarium material that we found in the Swedish
Museum
alignment, two ITS matrices were produced, one of Natural History Herbarium did not amplify
We
including gap codings and the other without them. despite several attempts. were not able to
Two parsimony analyses, conducted with PA UP* establish contact with anyone who could assist us
version 4. Ob 10 (Swofford, 2000), were performed for with recently collected material from St. Vincent,
each matrix. The ITS analysis was a heuristic where the species endemic, and was not possible
first is it
search (random addition sequence with 1000 repli- to do such fieldwork ourselves. Ten species that
cates, tree hi section-reconnec tion [TBR] branch swap- yielded sequences were not listed under any section in
ping, and MU LTREES option in effect), and the second the work of Fernandez Zequeira, but four of them
analysis was a jackknife search (Iasi step search option, could be assigned to sections based on the key
10,000 replicates, and Jac resample emulation). provided in her paper (Fernandez Zequeira, 1994): R.
&
For the combined analyses, the data matrix from the inermis (Sprung.) Krug Urb. and R. pilosa Sw
r
.
ITS study was combined with the entire matrices from belonging to section Leoninae and R. hameliifolia
,
&
the trnL-F study of Rova et ah (2002) and from an Dwyer M. V. Hayden and R. purdiei Hook.
f.
rpsld analysis (Rova, unpublished), keeping the ind belonging to section Calophyllae M. Fernandez Zeq.
el
codings from each matrix. Previous analyses of each Sectional affinities are indicated in Figure 1.
separate data set resulted in similar trees, which The first ITS matrix, without indel coding, included
implied that the data sets were congruent. Taxa not 699 characters, of which 174 were parsimony
included in the combined analyses were then deleted informative. The second ITS matrix, where indel
using the command DELETE in the PA U P block. The were coded, included 723 characters, of which 198
resulting set of sequences comprised 20 ingroup taxa. were parsimony informative. The combined ITS,
This set included all taxa where all three sequences rpsl and trnL-F matrix included 2751 characters,
,
were available and all Rondeletia species where at 1451 of which were parsimony informative.
least ITS and rps 1 sequences were available. The strict consensus tree obtained from the ITS
Chiococca alba was used as outgroup, because, analyses is presented in Figure 1. Heuristic searches
previously, it had been clearly shown not to be part of both data sets each resulted in 48 most parsimo-
of the ingroup (Rova et ah, 2002). The data were nious trees. Tree lengths were 768 (consistency index
= —
analyzed by a heuristic search (random addition [CI] 0.56, retention index RE 0.76) in the
TBR
sequence with 1000 replicates, branch swap- heuristic search of the data set without indel coding
MU ETHELS — —
ping, and option in effect). and 805 (CI 0.65, RI 0.77) in the heuristic
search where indels were coded. Strict consensus
Results trees were identical for both data sets, jackknife
support was not found for all clades in the strict
DNA
More than 50 extractions were obtained from consensus trees from the heuristic searches, and
Rondeletia representatives, but only 27 of these jackknife support values for a clade could vary up to
PGR
(representing 23 species) were amplified by more than 10 units between the two data Tree
sets.
and yielded sequences. Extractions that did not topologies differed only marginally between the two
produce usable sequences were mostly made from jackknife searches. The jackknife search without
herbarium material more than 50 years old. Material indel codings found one clade that was not found in
PGR
collected in silica gel almost always worked for the other jackknife search (or in the heuristic
,
Volume Number Rova 185
96, 1 et al.
2009 The Complex
Rondeletia
searches), and the jackknife search with hide codings both belong to Rondeletia sect. Leoninae. In our
I
resulted in two c lades not found in the jackknife analysis, they form a clade with strong support.
search without indeis coded (Fig. Support for these Rondeletia sect, leoninae would thus be the only one
1).
c lades was low in all cases. of Fernandez Zequeira’s sections that is supported by
The analysis of the combined matrix resulted in 12 our ITS sequence data.
—
equally parsimonious trees (length 2046, Cl 0,91 RI In all analyses, there is moderate support for a
,
=
A
0.97). strict consensus of these trees is shown in clade with Rondeletia alaternoides R. odorata jacq.,
v.
,
&
U
Figure 2, and branches from the consensus tree that also and R. pachyphylla Krug rb (Fig. 1, clade F),
.
occur in the ITS analysis are marked in bold in Figure which represent sections Calophyilae Qdoratae and
1.
, ,
Pedicellar respectively. Following the diagnostic
es,
table of sections in Fernandez Zequeira (1994), we
Discussion
were unable to find any morphological characters that
support this group.
SECTIONAL CLASSIFICATION OF RONDELETIA
Rondeletia intermixta Britton and R. ochraeea Urb.
The main aim of our study was to test Fernandez form a clade with strong support. While R. intermixta
M
Zequeira s (1994) classification with 10 sections of belongs to section Rondeletia Fernandez Zeq.,
FI.
Rondeletia using molecular phylogenetic analyses. ochraeea has not been previously classified to any
This goal was hard to reach satisfactorily; despite an section. It is thus possible to argue that R. ochraeea
extensive search, it was difficult to find herbarium or should also belong to section Rondeletia The only
.
silica gel—dried material that would work for PC R and other known representative of section Rondeletia
&
sequencing. For five of the sections, only one included in our analysis, R. portoricensis Krug
representative of each could be sequenced. Further- Urb., is placed in an unresolved relationship to the R.
more, ITS data are obviously not variable enough to intermixta—R. ochraeea clade, although jackknife
provide resolution among sections Hypoleucae and support for this is below 50.
we
J\lipenses Nevertheless, obtained several interest- Our study does not show any support for a
.
ing results with regard to the circumscription of separation of sections Hypoleucae and Nipenses in
Rondeletia and some of Fernandez Zequeira's sec- sequenced
Rondeletia. All representatives of these
tions. sections, except R. berteroana DC., are found in a
There is strong support for the Rondeletia s. str. strongly supported but unresolved clade (Fig. clade
1,
clade (Fig. clade E). This clade consists of No morphological character combination seems
1, PI). to
predominantly Antillean species. The only exceptions be unique for these two sections as one group,
to this distribution are R. hameliifolia from Central according to the character list in Fernandez Zequeira
America (Panama) and R. purdiei from South America (1994). Rondeletia berteroana differs from the other
(Ecuador). Neither R. hameliifolia nor R. purdiei were sequenced species of section Hypoleucae (and section
included in Fernandez Zequeira’s (1994) treatment, Nipenses) in being from Hispaniola. Phis species is
H
but according to her identification key, both species found as sister to clade but with very low support
would belong to section Calophyilae In our study, the
(Fig.
. 1).
A
two species form a clade with strong support. third
Rondeletia chamaebuxifolia Griseb., the only se-
representative of this section is R. alaternoides A. quenced representative of section Chamaebuxifoliae
,
Rich, from Cuba, which found in clade F
is (Fig. 1). is found closely related to the species from sections
Thus, ITS sequence data do not support a monophy-
Hypoleucae and Nipenses Following the diagnostic
.
letic section Calophyilae. characters provided in Fernandez Zequeira (1994) for
Rondeletia deamii (Donn. Sm.) Standi, is found just sections Chamaebuxifoliae Hypoleucae and Nipenses,
, ,
The
outside the Rondeletia clade. generic
s. str. this clade (Fig. clade G) could be distinguished
1,
position of this Central American species has recently from other sections by having - to 3-flowered
1
been under debate. Lorence (1999) supported its inflorescences and retrorse -pilose flowers.
position in Rondeletia but Borhidi (2001a) positioned
in Arachnothryx. Our ITS sequence data suggest that
it
STEVEN A
SI
this species should be treated as a Rondeletia
,
although support for this hypothesis is less than 50. The second aim of our study was to investigate the
Rondeletia inemiis and R. pilosa Sw. were not relationships between Stevensia and Rondeletia
.
included in Fernandez Zequeira’s (1994) treatment of Stevensia is here for the first time included in a
Cuban Rondeletia as these species occur in Puerto molecular phylogenetic study. According to ITS data,
,
Rico and the U.S. Virgin Islands, respectively. there is strong support for an inclusion of at least S.
However, according to her key to sections, they would minutifolia Alain in Rondeletia, s. str. The genus
H H H
HH
i ( 1f 1 i H« i
186 Annals of the
Garden
Missouri Botanical
!M kO o k o
•a vH0£)^0 LO o &S tto)- tLo)O tCotOo
CO to* Lk0 r- ?o— CO k
02
r- 1> f> t- l>
EM
€
(M 04 CoM CM CM 04 COM 04 04 CM
fe LO LO LO L0 LO LO >D
•LO
j— j—— r-H i— r-H s^H i—l p—H f—
fe fe fe tn fe fe fe fe fe
|>H
< < < c
to to to to to to
o H w o o
1 Hf
fe sO CO f— cko o Ov CoO VOO
3 r— CO CO vO CO Ov
& j^H o O o o
so
Os Ov O' o. O' Cv
l""-H CM CM *os CM 04 CM CM o co CM mco co
a
"to
Tf o Tf ii o t?
CM CM CM CM CM CM CM CM CM CM
rM
A to to to to to to to to to to to to
d
to to to to to to to to to to to to
PQ
d
€
O
k M o M o o
o
00 ON O} co ^t-y-HO, ' l •l'_ CoM O coo fo— LO X j— VoO
O
00 ©v co 0V CO CO Ov Ov <—
O m 'O. 'MO'. © Ov
£o<i eo^4 o CoM CoM £©* CoO- CoO CoO CoO COM CoM COM COM COM coo coo
H CO
co
fen VO
C>O CO CO cko cko CO CM cko cko cko cko CkO cko cko CkO cko cko cko
t> l> CM CM co
I to co
>H >H >H
to to feH < to to tCo1 fin <1 to to to to to to to to to to
to to to •< to <1 to to to to to to to to to to to to to
p
t B5
to
-p
lsO PQ
o
—
.p^
VJ tP >PH to
O
’O w
^
o o CM Pi
^
O
N
D
E £ Ifl c
^ to £C£
S ©
^ ^ O foi ^§ to ^
*- cm —
00 '00— a f€'l ^a V V r<j P e 'tu;' OX CD
^ M Opa 33 CPDh' ^ 'NO * Pi <o ' Oh
w — 5 P
E£|
d is i—
i—3 '^ ce O CO o p<s to> t0o\ to
03
' ^ k — CD ^ £ P Co a
CM CD Ph
Co p CO to
t^o p i—
Vi -H pi ?a to >-h t2o5' *o* P ow 0> u«a OM f" os¥ tPtoo 0S3 H 0«
S
PL & 00 iOrj Co
=° l to § "3 25; cq e to>
5 O
,J
f-k; r^^O CM CO 9. 2 o
p Of -o, k ^ ^ *5 vi o kto 02
.g £* ® E toK to k
ski "WS -3
Oj ^33 a fN cm a co ^3 to n3 53 J £p
tH/} SL3 5e^0 £ CcMm -•c- “tgo 33 3*Vt*j%.i dQ ^%P f a<kD.nD $ ^
? a 05 to? £ « r 3 rto CQ £ e
23 a
1
k nk s Cto -S E © ^
C3 «
o « » — E
•a to
fe C» k5 # StS <3 o (3S Mh -s to a
(g ~s S , - a
E-h pq
EM a Ck O
<s X
fe to> § eg
03
3p
Mtf- Hn tf- 13 2 tf-- O £ o O "3
"T<3D to cd O OEm a to DOEM eg ~i "3 S' cp Os Pd £
02 ,a-«j T5 r r^H S E £ i CN] E c
d d d -<MQ WrM d d X C^M « .& CD rkDjQa Ja
IS Co d tdj .-W
„^o d 3 fe 3 3 fe S sP QS CD CD 03
U3 d o 3 a 3 Qo Mo P3 (H>i s 3 H H Cs dI I sI a a
_ k E
Ph 5 CJ O CJ CJ
to to
cr
a
pn
V3
a
as
HO
.
& * ki
o
S
p
3
C O.
fe
fe
S? 5 I
0 0
£
2
r-O
S s 4
S
fe
5 3
S € S &D
02
" 3 © fe -p
Q K^-J [>-, IS
fe 3
aa 23 'ML g -<>-D' *<0D j=ol Pi '“O£ O0D 02 spj
o a CD k
5 CD
5 s PQ "S
-a -C,
a 9 £ Si ^3
* >o
PQ « CD r-O <D ^ ' 3 Cto
V ^ P 02 ^3 Cm
€
p^p s ' to3
f^S w
CO 02 <
ffee toe J V.o ’to cb o- pH 3—t3o Pvk 5
, vy
' ork £PP Jr p ffpiH P CD 1-1 A O ^ ' _$P' -ottqoo V55v a^VasI
<^i § dp ^^ ^ £ e tei o * o P>: ktPao i*^
P 4 os
NR
^ S 0. 31 e :C/b.. ^iD ^
fe 5’T r-w P cL 3 io 5C&; p a 'C
^0 e S
1-0 sp 02 12 *?@ * >o ik. HHJ rP
<D »S ^ i-O o a a k. :>} >j
02 K) ^ c e V> a a
IS ii| | b -&p -Ip- Ta^i ppp pp 'KaS §w^> £ 3 ^pp^ ok. ok -ap -Vppj. a a V
"S .si 33
p ^0 to to t** ‘H = Tk 'Ik
e "ws *w* ' r-ii SP P P a a Hk -k j* HdH
Pk*. I g 53 k a a to to>
a
sI£ I k r-O £ a a r-P -S *3 S k.
s£r- Sa' sa -BP >"tjo- "8 ai.j ssQ *wSj> 4° **P*a p^O3: PP(Sj * otcos - Mtt^o: a Si a a
OP O oa ^ 5 s£ ^ o o
^ . S>
t§ C§ c§ C§
-t to to to to to to
H
1 1 i 1
Volume Number Rova 187
96, 1 et al.
2009 The Complex
Rondeletia
number
accession
GenBank
p o m
- r- Ol co LoO VO LO co LpO VO o !——
(M O] c\ Ol r— (—1 j— co CO
F 1
CoO CoO CoO CoO CoO CoO CoO CoO COO CoO CoO CoO CoO CoO CoO CoO C©O
co co co co co co co co co co co co co co co co co
p p p p p P p p p p p p p p p p p
P p p M p p p
Im 8H Ip
f<eH < < < < < < < c < < <
<1 <! <! *#
H
,
188 Annals of the
Garden
Missouri Botanical
Luculia grandifiora
OUTGROUP
*
Chiococca alba
Catesbaea
fuertesii
Acrosynanthus revolutus
Rondeletia pitreana
*
Rogiera cordata
Rogiera cordata
*
amoena
Rogiera
Arachnoth ryx chimboracensis
Arachnothryx sp. indet.
*
Arachnothryx leucophylla
Cuatrecasasiodendron spectabile
Arachnothryx buddleioides
Gonzalagunia
affinis
*
Guettarda scabra
Guettarda uruguensis
Hodgkinsonia
ovatiflora
Timonius
nitidus
*
Biepharidium guatemalense
Suberanthus brachycarpus
*
Rovaeanthus suffrutescens
Rovaeanthus
strigosus
*
Rachicallis americana
*
Mazaea
phialanthoides
Phyllomelia coronata
*
Mazaea
shaferi
*
Rondeletia deamii unknown
Rondeletia hameliifblia Calophyllae
Rondeletia purdiei Calophyllae
Rondeletia alatemoides Calophyllae
*
Rondeletia alatemoides Calophyllae
*
Rondeletia odorata Odoratae
*
Rondeletia pachyphylla Pedicellares
Rondeletia pachyphylla Pedicellares
*
Rondeletia portoricensis Rondeletia
*
Rondeletia intermixta Rondeletia
Rondeletia ochracea unknown
*
Rondeletia inermis Leoninae
*
Rondeletia pilosa Leoninae
Rondeletia stipularis unknown
Rondeletia sp. Jamaica unknown
Rondeletia barahonensis unknown
Rondeletia cincta Rigidae
Stevensia minutifolia
Steven
sia minutifolia
Rondeletia chamaebuxifoiia Chamaebuxifoliae
Rondeletia berteroana Hypoleucae
Rondeletia subcanescens Nipenses
Rondeletia subcanescens Nipenses
Rondeletia nipensis Nipenses
Rondeletia nipensis Nipenses
Rondeletia lomensis Nipenses
*
Rondeletia mirafiorensis Hypoleucae
*
Rondeletia apiculata Hypoleucae
Rondeletia plicatula Hypoleucae
Figure 1. Tree compiled from the strict consensus trees from the two heuristic searches, without and with indels coded,
respectively. The one dotted branch was not found in the heuristic searches, only in the jackknife search without indels coded.
Numbers indicate jackknife support: numbers above branches are support values without indels coded, and numbers below
A—
branches are support values with indels code—d. Jackknife support of 50 and lower is indicated by dashes (—). L—etters
indicate the clades discussed in the text: A. Glade comprising representatives of the tribe Guettardeae. B. Glade
—
corresponding to Rondeletieae sensu Rova et al. (2002). C. Glade comparable to the one in which Acrosynanthus was found
— —
—
in Rova et al. (2002). D. Glade comprising Mazaea Phyllomelia and Rachicallis. E. The Rondeletia s. str. clade. F.
—
,
Glade including representatives sections Odoratae Pedicellares, and the paraphyletie section Calophyllae. G. Glade
of
,
distinguished by having 1- to S-flowered inflorescences and retrorse-pilose flowers (corresponding to sections
—
Chamaebuxifoliae, Hypoleucae, and Nipenses). H. Clade comprising Cuban representatives of sections Hypoleucae and
Nipenses. Taxa. marked by an asterisk are included in the combined ITS, rps!6, and trnL-F analysis presented in Figure 2.
(*)
The circled Roman numerals I—XI on bold branches refer to clades in the combined ITS, rpsl6, and trnL-F analysis (Fig. 2).
,
Volume Number Rova 189
96, et al.
1
2009 The Rondeletia Complex
Chiococca
alba
Rogiera cordata
amoena
Rogiera
Blepharidium guatemalense
Guettarda scabra
Rovaeanthus
suffrutescens
Arachnothryx leucophylla
Cuatrecasasiodendron
spectabile
americana
Rachicallis
Mazaea
phialanthoides
Mazaea
shaferi
unknown
Rondeletia deamii
Rondeletia alatemoides Calophyllae
Rondeletia odorata Odoratae
Rondeletia pachyphylla Pedicellares
Rondeletia intermixta Rondeletia
Rondeletia inermis Leoninae
Rondeletia pilosa Leoninae
Rondeletia portoricensis Rondeletia
Rondeletia miraflorensis Hypoleucae
Hypoleucae
Rondeletia apiculata
Figure 2. Strict consensus tree of the 12 equally parsimonious trees from the analysis of the combined ITS, rpsl6 and
trnL-F data matrix. Guettarda scabra, Cuatrecasasiodendron spectabile, Rondeletia deamii, R. pachyphylla R. miraflorensis,
,
and R. apiculata were only represented by ITS and rpsl 6 data in the data set. The circled Roman numerals 1—XI on branches
refer to clades in the tree in Figure 1. Sectional assignment of Rondeletia species, according to Fernandez Zequeira (1994), is
listed in the rightmost column. Rondeletia inermis and R. pilosa were not listed in Fernandez Zequeira (1994), but have been
assigned to section based on her key. Rondeletia deamii is not assigned to a section (ef. unknown among the
sectional assignments).
Stevensia comprises 11 species endemic to Hispa- Rova et ah, 2002). The fourth aim was to see if ITS
(
niola. recognized by triangular stipules connected sequence data would place Hodgkinsonia in Guettar-
It is
to a sheath, solitary and axillary flowers, two to three deae or elsewhere.
calyx lobes, five to seven stamens attached in the Acrosynanthus revolutus Urb. and Rondeletia pi-
&
Ekman
corolla throat, glabrous style, and ovoid to oblong treana Urb. (not classified to section)
seeds (Borhidi, 2001b). However, several these appear as early diversified lineages in the cladogram.
of
R
One
character states are also found within Rondeletia possible explanation for the position of
s.
.
We
we
according to Fernandez Zequeira (1994). pitreana could be that were not able to read the
Sir.,
therefore suggest that Stevensia should be included sequence in its entirety. Because of this, it is about 40
within Rondeletia pending future studies with bases shorter than the other Rondeletia sequences.
s. str.,
an extended sampling. Another possibility is that R. pitreana does not belong
to Rondeletia. In any case, further studies are needed
R
The
to solve the position of pitreana. position of
GUETTARDEAE AND RONDELETIEAE
.
Acrosynanthus Urb. in the present analysis differs
Our third aim was to compare a nuclear ITS markedly from the results of the tmL-F study from
phylogeny of the Rondeletieae with the results from a Rova et al. (2002), where Acrosynanthus was found in
DNA A
previous trnL-F chloroplast (cpl)NA) study a position equivalent to basal in clade C (Fig.
1).
Sectional assignment of Rondeletia, species, according to Fernandez Zequeira 1994), is listed in the rightmost column.
(
Rondeletia hameliijolia, R. purdiei, R. inermis, and R. pilosa were not listed by Fernandez Zequeira, but have been assigned to
section based on her key (1994: 106).
190 Annals of the
Garden
Missouri Botanical
possible explanation would be that Acrosynanthus is Phyllomelia Griseb. (D in Fig. 1). Rachicallis and
not monophy letic: A. latifolius Stand L was included in Phyllomelia are monotypic genera, and Mazaea only
the trnL-F study, but A. revolutus was sequenced in comprises two species. Based on ITS data, one could
the ITS analysis. However, the possible paraphyly of argue that three genera should be merged together.
all
Acrosynanthus must be left to another study when However, both Mazaea and Phyllomelia are easily
more material of this genus is available. distinguished by the peculiar fruit (pseudosamara,
We
found that the well- supported Guettardeae s.L sensu Delprete, 1999b) and calyx morphology (Del-
and Rondeletieae s. str. c lades in the trnL-F study prete, 1999b), and for this reason we prefer to regard
from Rova et al. (2002) have only weak support from them as separate genera.
ITS sequence
data.
With regard to the tribe Guettardeae, there is
HODGKINSONIA
moderate support for a clade including Arachno-
thryx Gonzalagunia Cuatrecasasiodendron Guet- The fourth aim was to see if ITS sequence data
, , ,
tarda Hodgkinsonia and Timonius (Fig. clade A). would place Hodgkinsonia in Guettardeae or else-
1,
, ,
where. Our study undoubtedly places Hodgkinsonia
According our Cuatrecasasiodendron should
to results,
be synonym zed with Arachnothryx and this is also close to Timonius which means within Guettardeae.
i %
,
morphologically supported (see taxonomic treatment Ihis position is in accordance with the view of
below). Wliile trnL-F data (Rova et al., 2002} showed Mueller (1861) in the original description and Bremer
Rogiera amoena Planch, and R, cordata (Benth.) (1992), but contradicts the supposition ol Delprete
who
Planch, as members of the Guettardeae, the inclusion (1996), tentatively included the genus in tribe
of Rogiera in the Guettardeae clade not Chiococceae in agreement with Robbrecht (1988).
s. str. is
supported by ITS data alone. In the combined analysis
(Fig. Rogiera is found within Guettardeae, while Conclusion
2),
Arachnothryx is found to be more closely related to
Rondeletieae. The ITS sequence data support only one of
In a recent molecular phylogenetic study, A chi e et Fernandez Zequeira's (1994) Rondeletia sections as
1 1
ah (2006) supported the monophy of the Guettar- monophyletic: section Leoninae. Rondeletia sections
ly
deae as recognized here, although they showed that Calophyllae and Rondeletia are paraphyletic accord-
Guettarda Antirhea Comm, ex Juss., and Stenostomum ing to our analysis. However, one should bear in mind
,
C. F. Gaertn. are polyphyletic. However, more genera that we were not able to sequence more than one
and more species need to be included in the study in species from several sections. When we compare our
order to test the monophy ly and delimitation of this ITS phylogeny with the character lists in Fernandez
tribe. Zequeira’s treatment of Rondeletia we were unable to
,
Although Rondeletieae sense Rova et al. (2002) is find morphological characters that correspond with
recognized by ITS data in the consensus trees (Fig. I, our phytogenies. The sections described by Fernandez
clade B), there is no jackknife support for this clade. Zequeira are often defined by various combinations of
In both heuristic ITS searches, the genera Blephar- overlapping character states, which makes compari-
The
idium Standi, and Suheranthus were found basal in the sons difficult. only exception is a clade including
Rondeletieae clade, but again, there no representatives from Rondeletia sections Chamaebux-
s. str. is
jackknife support for this. However, this position ifoliae Hypoleucae and Nipenses. which could be
, ,
corresponds to the results from the tmL-F study ol distinguished by having 1- to 3-flowered inflorescenc-
Rova et ah (2002). es and retrorse -pilose flowers. This clade could
he ITS data place Rovaeanthus strigosus (Benth.) potentially be recognized as one section.
t
Borhidi in the Rondeletieae, as sister taxon to R. Rondeletia s. sir. (be., excluding Arachnothryx
,
suffrutesceiis (Brandegee) Borhidi. just as in the trnL-F Javorkaea Rogiera Roigella Rovaeanthus and Suh-
, , , ,
study (Rova et al., 2002), the ITS data indicate that R. eranthus} has strong support, although some species
suffrutescens belong to the Rondeletieae (although need to be further investigated for their generic
s. str.
this is contradicted in the combined analysis where R. affinity (e.g., R. pitreana and R. deamii).
An
suffrutescens is found as sister to Guettarda). In any important result from our study is that Stevensia
A
case, R. suffrutescens always appears in a separate minutifolia is included within Rondeletia s. str.
position from Rogiera and our study thus supports the reevaluation of the morphological characters in
,
new
transfer of these species from Rogiera into a genus Rondeletia (including Stevensia) based on the results
as proposed by Borhidi et al. (2004). from ITS and other sequence data is certainly needed.
There strong support for a close relationship The present analysis clearly suggests that Stevensia
is
&
between Rachicallis DC., Mazaea Krug rb and should be recognized at most as a section of Rondeletia
l. .
. ,
,
Volume Number Rova
98, et al. 191
1
2009 The Complex
Rondeietia
There is strong support for a division of the a rotate pterophyJ after an the sis and present in all
l
m
Rondeletieae—Guettardeae complex into the tribes flowers in Phyllomelia; corolla hypocraterifor or
Rondeletieae and Guettardeae from trnL-F narrowly infundibuliform, orifice with annular thick-
s. str. s.l.
data (Rova el ah, 2002), but only weak support from ening, white, cream-white, red, green, or yellow,
ITS data. While part of the Guettardeae has moderate membranous to fleshy; aestivation valvate, contorted,
support, support for Rondeletieae in the sense of Rova or imbricate; stamens mostly as many as corolla lobes,
et ah (2002) is weak in the ITS study, although the inserted near the base or at the medial zone or near
Rondeletieae s. str. clade is found in the strict the orifice of corolla tube; anthers included or
consensus trees of all of our analyses, both including exserted, oblong to narrowdy elliptic to button-shaped,
and excluding indel codings. It was not possible to 2 -locular, opening by longitudinal slits, dorsiiixed
compile a sufficiently large data set in order to test the near the base or around the middle, introrse; pollen
delimitations of Guettardeae and Rondeletieae using a released as monads, col pate or colporate, exine
combined ITS, rpsl6 and trnL-F sequence data reticulate or foveolate (not echinate); style branches
,
we
matrix; however, consider that the phvlogenies present, with stigmatic surface smooth vermcate;
to
available up to this point (Rova, 1990; Rova et al., ovary inferior (half-inferior in Rachicallis bilocular,
),
&
2002; Delprele Cortes-B., 2004; the present study) with a few to many ovules (1 to 2 in Mazaea) per
provide sufficient support for a re-delimitation of the locule attached to a central placenta, or exceptionally
tribe Rondeletieae. one ovule per locule basal ly attached Phyllomelia
).
(
Based on the results from ITS sequence data, we Fruits woody capsules, loculicidal or septicidal, or
also reconsider Delprele s (1996) tentative inclusion septicidal and loculicidal contemporaneously (Ble-
(based on morphology) of Hodgkinsonia in the Chio- pharidium Mazaea commonly dehiscing basipetally,
, ),
cocceae, since the present ITS sequence data support or exceptionally pseudosamaras, indehiscent (Phyllo-
Bremer’s (1992) conclusion (also based on morpho- melia); placenta central, rarely apieally incomplete, or
logical data) that Hodgkinsonia part of the tribe shortly stalked; seeds horizontal, imbricate, peltate,
is
Guettardeae. and vertical, minute, 3- to 5-angular or dorsoventrally
convex, not winged, wing concentric or bipolar
Mazaea
(Blepharidiurn or basally inserted, ellip-
Taxonomic Treatment
, ),
seid- ovoid and fleshy Phyllomelia
( ).
Based on the results from the present and other Genera included: Acrosynanthus Acunaeanlhus
,
&
recent studies (Delprete, 1999b; Rova, 1999; Rova et Borhidi, Komlodi Moncada, Blepharidium Glion-
,
&
ah, 2002; Delprete Cortes-B., 2004; Borhidi et al., netia Tirveng., Habroneuron Standi., Mazaea Phyllo-
,
2004), we propose the following taxonomic descrip- melia Rachicallis Rogiera Roigella Rondeietia
|, , , , ,
tions and rearrangements. Rovaeanthus Spathichlamys R. Parker, Stevensia
, ,
SuberanthuSi
Tribe Rondeletieae (DC.) Miq., Flora Nederi. Indie
The description and delimitation of the Rondele-
2: 130, 1 56. 1856. Rondeletiinae DC., Prodr. 4:
tieae here proposed are based on the results of the
342, 401. 1830, as subtribe “Rondeletieae,”
present study in combination with those of Rova
tribe Hedyotideae. Rondeletieae DC. ex Rchb.,
and Rova The
(1999) et al. (2002). description is
Der Deutsche Botaniker 77. 1841, non
1: stat.
basically a reduction of that proposed by Delprete
TYPE:
indie. Rondeietia L.
(1999a), based on his w ide circumscription of the
7
Shrubs or trees; wood whitish or yellowish; raphides tribe to include the Condainineeae and the Sipaneeae,
absent; axillary thorns absent. Stipules free or connate which was produced primarily for the floristic
at base, mostly entire, rarely bifid, mostly mterpetio- treatment and not based on a comprehensive phylo-
lar, frequently with coll etcrs on the adaxial side genetic analysis.
secreting resinous compounds, persistent to readily Rova (1999) and Rova et al. (2002) demonstrated
Condamineeae
caducous; leaves opposite or verticillate, decussate, that the (except the subtribe Portland-
petiolate to sessile, blades chartaceous to thick- iinae, which belongs to the Chiococceae s.l.) should
coriaceous; domatia variably present or absent. be transferred to the Subfamily Ixoroideae, in a
Inflorescences terminal or axillary, cymose, panicu- complex also including the C al y cophy eae and the
1 1
late or thyrsoid, multiflorous or pauciflorous, or Flippotideae (more studies are needed to re -delimit
&
uniflorous. Flowers hermaphroditic, mostly actino- these groups; Kainulainen Bremer, unpublished).
morphic, (3- to)4- to 6-merous; calyx persistent or Delprete and Cortes-B. (2004) and Rova et al. (2002)
caducous; lobes often minute, sometimes foliose; also demonstrated that the Sipaneeae belongs to the
ealyeophylls commonly absent or pterophyllous (green subfamily Ixoroideae and a monophyletic group that
is
to greenish white), with all calyx lobes expanding into was positioned in the same clade as the tribes