A Preliminary Molecular Phylogeny of the Abarema Alliance (Leguminosae) and Implications for Taxonomic Rearrangement

Abstract

Premise of research. Here we present the first molecular phylogeny of the Abarema alliance, based on a large species sampling from across its geographical range. Our aim was to test the monophyly of the alliance and to analyze the major biogeographical patterns throughout the Neotropics. Methodology. DNA sequence data were derived from the chloroplast matK region and nuclear external transcribed spacers (ETSs) and were phylogenetically analyzed in order to resolve systematic relationships. Pivotal results. Our results agree in part with Barneby and Grimes's circumscription of the alliance, including the genera Hydrochorea and Balizia and part of Abarema. However, Abarema, the largest genus within the alliance, is polyphyletic. The type species of Abarema, Abarema cochliacarpos, is closer to genera of the Inga alliance. There is also an Andean clade distinct from the remaining species of Abarema sensu lato, Hydrochorea, and Balizia. Conclusions. Two groups of Abarema sensu lato (one restricted to the Andes), but excluding the type species A. cochliacarpos, are here proposed, together with the genera Hydrochorea and Balizia, as a new alliance. The results demonstrate that monophyletic groups have to be recircumscribed and some described as new taxa. We also present preliminary observations about the biogeography of the group in different Neotropical forest formations. The Andean clade appears as sister to all other taxa within the new alliance when combined ETS and matK data are analyzed. Multiple independent events must have occurred during the colonization of the West Indies and Central America. Species in the Atlantic forest and Amazonia are the result of the most recent radiation events within the group.

Full text

SYMPOSIUM PAPER
A PRELIMINARY MOLECULAR PHYLOGENY OF THE ABAREMA ALLIANCE (LEGUMINOSAE)
AND IMPLICATIONS FOR TAXONOMIC REARRANGEMENT
João R. V. Iganci,
1,
* Marcos V. Soares,†Ethiéne Guerra,* and Marli P. Morim‡
*Programa de Pós-Graduação em Botânica, Universidade Federal do Rio Grande do Sul, Avenida Bento Gonçalves 9500, Bloco IV,
Prédio 43433, Porto Alegre, Brazil; †Coordenação de Botânica, Museu Paraense Emílio Goeldi, Avenida Perimetral, 1901,
Montese, Belém, Brazil; and ‡Diretoria de Pesquisas, Instituto de Pesquisas Jardim Botânico do Rio de Janeiro,
Avenida Pacheco Leão, 915, 22460-030, Rio de Janeiro, Brazil
Editors: Gwilym P. Lewis and Patrick S. Herendeen
Premise of research. Here we present the first molecular phylogeny of the Abarema alliance, based on a
large species sampling from across its geographical range. Our aim was to test the monophyly of the alliance
and to analyze the major biogeographical patterns throughout the Neotropics.
Methodology. DNA sequence data were derived from the chloroplast matK region and nuclear external
transcribed spacers (ETSs) and were phylogenetically analyzed in order to resolve systematic relationships.
Pivotal results. Our results agree in part with Barneby and Grimes’s circumscription of the alliance, in-
cluding the genera Hydrochorea and Balizia and part of Abarema. However, Abarema, the largest genus within
the alliance, is polyphyletic. The type species of Abarema,Abarema cochliacarpos, is closer to genera of the Inga
alliance. There is also an Andean clade distinct from the remaining species of Abarema sensu lato, Hydrochorea,
and Balizia.
Conclusions. Two groups of Abarema sensu lato (one restricted to the Andes), but excluding the type species
A. cochliacarpos, are here proposed, together with the genera Hydrochorea and Balizia, as a new alliance. The
results demonstrate that monophyletic groups have to be recircumscribed and some described as new taxa. We
also present preliminary observations about the biogeography of the group in different Neotropical forest for-
mations. The Andean clade appears as sister to all other taxa within the new alliance when combined ETS
and matK data are analyzed. Multiple independent events must have occurred during the colonization of the
West Indies and Central America. Species in the Atlantic forest and Amazonia are the result of the most recent
radiation events within the group.
Keywords: Amazonia, Fabaceae, Ingeae, Mimosoideae, phylogeny, rain forest.
Online enhancements: appendix figures.
Introduction
The largest taxonomic review of Abarema Pittier and allied
genera was prepared by Barneby and Grimes (1996) during their
studies on the synandrous taxa from the Americas. In those stud-
ies, the genera of tribe Ingeae were grouped into seven “alli-
ances,”and the Abarema alliance was circumscribed to com-
prise Abarema,Balizia Barneby & J.W.Grimes, and Hydrochorea
Barneby & J.W.Grimes. Furthermore, Lewis and Rico Arce
(2005) recognized an emended circumscription of the Abarema
alliance that comprised the genera Abarema,Hydrochorea,and
Pararchidendron I.C. Nielsen. These authors considered Bali-
zia a synonym of Albizia Durazz., in accordance with the publi-
cation of Rico Arce (1999).
A comprehensive general overview of the systematics of the
tribe Ingeae was provided by Brown (2008). Recent efforts to
understand the relationships between the genera of the tribe
Ingeae were carried out by Souza et al. (2013), who focused
on the genus Calliandra Benth. The Abarema alliance was rep-
resented by the three genera placed in the alliance by Barneby
and Grimes (1996), and these clustered in a monophyletic group.
However, since Souza et al.’s main objective was not to study
the Abarema alliance, species within the genera were poorly sam-
pled, including only two species of Abarema,Abarema flori-
bunda (Spruce ex Benth.) Barneby & J.W.Grimes and Abarema
piresii Barneby & J.W.Grimes, plus Hydrochorea corymbosa
(Rich.) Barneby & J.W.Grimes and Balizia pedicellaris (DC.)
Barneby & J.W.Grimes.
All three genera of the Abarema alliance, as circumscribed
by Barneby and Grimes (1996), and also Pararchidendron were
previously recognized as sections of Pithecellobium Benth. (Ben-
tham 1844). The species segregated from Pithecellobium sensu
lato (s.l.) were subordinated to several genera of Ingeae, includ-
1
Author for correspondence; e-mail: joaoiganci@gmail.com.
Manuscript received April 2015; revised manuscript received July 2015;
electronically published November 25, 2015.
Int. J. Plant Sci. 177(1):34–43. 2016.
q2015 by The University of Chicago. All rights reserved.
1058-5893/2016/17701-0004$15.00 DOI: 10.1086/684078
34

ing Abarema,Albizia,Balizia,Chloroleucon (Benth.) Britton &
Rose, Hydrochorea,Leucochloron Barneby & J.W.Grimes,
Macrosamanea Britton & Rose ex Britton & Killip, and Sama-
nea (Benth.) Merr., among others (Burkart 1964; Barneby and
Grimes 1996; Lewis and Rico Arce 2005).
The genus Abarema is the largest within the alliance, com-
prising in its current circumscription approximately 50 species
(Barneby and Grimes 1996; Bässler 1998; Iganci and Morim
2009, 2015).The genus extends from the Atlantic forest in south-
eastern and southern Brazil to Amazonia, Central America, and
the West Indies and to Andean valleys and high-elevation forests
from Colombia to Bolivia (Barneby and Grimes 1996; Iganci
and Morim 2012). The main center of diversity within the genus
is found in the Amazonian rain forests and savannas, followed
by the Atlantic forest in Brazil, occurring in rain forest, coastal
scrub, and seasonal forests, mainly close to sea level, and in a
great variety of habitats in Colombia (Barneby and Grimes 1996;
Iganci and Morim 2012).
The genus Abarema comprises unarmed trees and shrubs with
bipinnate leaves, petiolar nectaries, and axillary synflorescences.
One of the most prominent morphological characters to distin-
guish the genus is its curved to spiral pods, with chartaceous
valves, generally with red endocarp, and seeds bicolored as a re-
sult of the partially translucent testa that displays the color of
the inner embryo. The seeds also have a pleurogram and a per-
sistent funicle (Barneby and Grimes 1996; Iganci and Morim
2012). Even though Barneby and Grimes (1996) implied mono-
phyly of Abarema on the basis of morphology, morphological
characteristics are variable within the genus, in comparison with
other genera within the tribe Ingeae that have a similar but more
consistent morphology.
The other genera traditionally included in the Abarema alli-
ance are species poor and almost restricted to the Amazon
and Central American forests. Both Hydrochorea and Balizia
have three species each, occurring predominantly in the Ama-
zon basin and Central America (Barneby and Grimes 1996);
but note that Balizia was reduced to synonymy in Albizia by
Rico Arce (1999). Both Abarema and Hydrochorea also occur
in the Cerrado (Iganci and Morim 2015; Morim and Soares 2015).
Pararchidendron comprises only one Australasian species (Lewis
and Rico Arce 2005).
The three genera Balizia,Hydrochorea,andPararchidendron
have a compound leaf morphology and leaflet shape similar to
those of Abarema. However, the three genera present only a
capitate or umbelliform inflorescence, whereas Abarema spe-
cies display a more complex inflorescence variation (Iganci and
Morim 2012). The most important morphological character to
distinguish Abarema from Hydrochorea and Balizia is found
in the pods (fig. 1E). While Abarema has typical legumes that,
upon dehiscence, retain the seeds on their funicles, Hydrochorea
has a modified lomentiform, indehiscent legume, with hard ar-
ticles that detach from the pods, each one containing one seed
(Burkart 1964; Barneby and Grimes 1996). Balizia has two types
of fruit: a follicle with a continuous endocarp not forming ar-
ticles (i.e., the fruit not septate) and another fruit type, some-
times tardily dehiscent, with a septate endocarp breaking up to
form articles, each containing one seed, similar to the mimosoid
fruits that Burkart (1943, 1964) called a “crypto-loment.”On
the other hand, Pararchidendron (and also Archidendron F.
Muell.) has fruits very similar to those of Abarema, a typical le-
gume, that, upon dehiscence expose the brownish to reddish
endocarp and colored seeds, which remain attached to the per-
sistent funicles.
To test the various taxonomic hypotheses presented above,
we evaluated the systematic relationships of the genera within
the Abarema alliance and here present a preliminary molecu-
lar phylogenetic analysis of the group that includes a wide spe-
cies sampling of the genera that traditionally comprise the tribe
Ingeae. We addressed the following questions: (1) Is the Aba-
rema alliance as circumscribed by Barneby and Grimes (1996)
or by Lewis and Rico Arce (2005) a monophyletic group? (2) Is
Abarema a monophyletic genus? (3) What are the relationships
of Abarema with other genera within the Abarema alliance?
(4) What can we observe about the geographical and ecological
patterns within the Abarema alliance?
Methods
Taxon Sampling
In total, 224 accessions of the Abarema alliance (consider-
ing both the circumscription of Barneby and Grimes 1996 and
that of Lewis and Rico Arce 2005) were sampled, representing
41 species of Abarema (ca. 90% of the total species in the genus)
plus Balizia,Hydrochorea,andPararchidendron. The sampling
completely covered the geographical range of the alliance. In ad-
dition, further accessions of 34 species of other genera of tribe
Ingeae were included. The analyses also included 21 accessions
representing genera of Mimosoideae belonging to other tribes.
These samples were chosen from across the systematic diver-
sity of the mimosoid legumes. Plant material was sampled from
the wild during fieldwork and from herbarium collections, espe-
cially from the herbaria RB and NY. Additional sequences were
downloaded from GenBank. All collections were deposited in
the herbarium RB, at the Rio de Janeiro Botanical Garden, and
duplicates were sent to collaborative institutions. Fresh leaves
were dried in silica gel for later DNA extraction.
DNA Extraction, Amplifications, and Sequencing
Total DNA was isolated from 0.3 g of silica gel–dried leaf
tissue with an Invisorb Spin Plant Mini Kit (Invitek, Berlin) ac-
cording to the manufacturer’s protocol. The molecular markers
analyzed were the nuclear ribosomal (nrDNA) external tran-
scribed spacers (ETSs; primers:18S-IGS-CAC ATG CAT GGC
TTA ATC TTT G/AcR2-GGG CGT GTG AGT GGT GTT
TGG; Baldwin and Markos 1998; Ariati et al. 2006) and the
chloroplast (cpDNA) trnK intron including matK (primers:
trnK685F-GTA TCG CAC TAT GTA TCA TTT GA/trnK2R-
CCC GGA ACT AGT CGG ATG G; Lavin et al. 2000, 2001;
Wojciechowski et al. 2004; Iganci et al. 2013). Polymerase
chain reaction (PCR) was conducted with a reaction volume
of 25 mL containing 1 mL(∼20) ng of DNA template, 10 mLof
GoTaq Hot Start Green Master Mix (Promega, Madison, WI),
1mL of DMSO, 1 mLofprimer3
0
(10 ng), 1 mLofprimer5
0
(10 ng), and 11 mLofpurified water. PCR conditions were as
follows: ETS: 957C for 5 min and 30 cycles of 1 min at 947C,
1minat557C, and 2 min at 727C, followed by a final extension
of 7 min at 727C; matK:957Cfor5minand30cyclesof30s
at 947C,30sat507C, and 2 min at 727C, followed by a final ex-
IGANCI ET AL.—MOLECULAR PHYLOGENY OF THE ABAREMA ALLIANCE 35

tension of 7 min at 727C. PCR products were purified and se-
quenced by Macrogen (Seoul, South Korea). Alignments of the
sequences were made with the default settings and the L-INS-i al-
gorithm in MAFFT, version 7.017 (Katoh and Toh 2008),
implemented within Geneious, version 6.1.5. Gaps were not
coded.
Phylogenetic Analyses
Bayesian analyses were performed with MrBayes 3.1.2 (Ron-
quist and Huelsenbeck 2003). The best-fit models of sequence
evolution were chosen using the Akaike information criterion
implemented in the program JModeltest (Nylander 2004). The
GTR 1G and GTR 1I1G nucleotide-substitution models
were chosen for matK and ETS analyses, respectively. Bayesian
analyses were initiated from random starting trees in two inde-
pendent runs of 10,000,000 generations and four chains. In each
run, trees were sampled every 100 generations, log-likelihood
scores were compared for convergence, and the first 25% of
trees were discarded as burn-in. Then, 50% majority-rule con-
sensus and Bayesian posterior probabilities were generated for
the resulting trees. The maximum likelihood analyses were con-
ducted with RAxML (Stamatakis 2014) applied in the Windows
command prompt. Substitution models were defined with Jmodel-
test (Nylander 2004). Each analysis included 1000 repetitions,
without partitions, and then bootstrap support values were cal-
culated. Resulting trees were visualized and edited with FigTree,
version 1.4.0 (Rambaut 2009). A combined Bayesian analysis
using both ETS and matK sequences was also implemented. This
analysis considered only those taxa that present sequences for
both DNA regions, including 48 terminal taxa and 1318 bp.
Partitioned analyses of nuclear and plastid data subsets were
carried out, and the most likely genealogical species tree was
reconstructed with a multispecies coalescent approach imple-
mented in MrBayes 3.2.1.
Results
The alignment of ETS sequences included 224 terminal taxa,
and the sequences were 508 bp long. The ETS phylogenetic
analysis (fig. 1) suggests that Abarema is not a monophyletic
genus. Furthermore, the Abarema alliance has to be recircum-
scribed and appropriately renamed. The results show high Bayes-
ian posterior probability and bootstrap support for major clades
(fig. 1). Abarema cochliacarpos (Gomes) Barneby & Grimes clus-
teredwiththegeneraInga Mill., Macrosamanea,Enterolobium
Mart., and Zygia P. Browne within the Inga alliance, with a pos-
terior probability of 1.0 (figs. 1A,1B,2A; clade X). The re-
maining species of Abarema also formed two groups. One small
clade of Andean species (figs. 1A,1C,2A; clade Y) comprises
A. callejasii Barneby & J.W.Grimes, A. centiflora Barneby &
J.W.Grimes, A. killipii (Britton & Rose ex Britton & Killip)
Barneby & J.W.Grimes, and A. lehmannii (Britton & Rose ex
Britton & Killip) Barneby & J.W.Grimes and is sister to all other
species of Abarema plus Hydrochorea and Balizia.Inaddition
to A. cochliacarpos positioned within the separate Inga alliance,
the presence of a clade formed by Hydrochorea plus Balizia
(figs. 1D,2A; clade Z) nested within taxa of Abarema (figs. 1,
2A; clade W) renders the genus Abarema polyphyletic in its cur-
rent circumscription.
The data matrix of the plastid matK region (partial) included
119 terminal taxa and 810 characters. The resulting trees
agree with results from the ETS analysis for the larger clades,
and the Abarema alliance again was not supported as mono-
phyletic (fig. A1; figs. A1, A2 available online). Abarema coch-
liacarpos grouped with genera of the Inga alliance with 0.99
posterior probability. In the phylogeny based on the partial
matK sequences, A. centiflora, the representative species of the
Andean clade, is closer to the Inga alliance, but with low poste-
rior probability. However, a separate matK analysis including
a smaller species sample (51 terminal taxa) but with full sequence
length of the region generated results similar to the phylogeny
based on ETS, where the Andean lineage is sister to all remain-
ing Abarema species plus Hydrochorea and Balizia, with 0.93 pos-
terior probability (fig. A2).
The combined analyses using both ETS and matK (partial)
sequences also showed the Abarema alliance to be polyphy-
letic, and A. cochliacarpos appeared as sister to Macrosamanea,
with 0.98 posterior probability (fig. 2A). The Andean clade, rep-
resented by A. centiflora, is also positioned as sister to the re-
maining lineages of Abarema plus Hydrochorea and Balizia,
with 1.0 posterior probability (fig. 2A). We consider this topol-
ogy to be the best current hypothesis to explain the phylogenetic
relationships within the group. An incongruence observed be-
tween the ETS and matK analyses is related to the position of
A. cochliacarpos within the Inga alliance. While in both the ETS
and the combined analyses A. cochliacarpos appears as sister
to Macrosamanea,inthematK-only analysis this species is closer
to Inga. The ETS and matK sequences are from different ge-
nomes, nuclear and chloroplast, respectively, and possible incon-
gruences between independent phylogenetic analyses have been
widely reported. Such conflicts could result from different pro-
cesses, including paralogy, hybridization, incomplete lineage sort-
ing, and horizontal gene transfer (Zhang et al. 2015).
Discussion
Phylogenetic Relationships
This is the first molecular phylogenetic analysis specifically
focused on the Abarema alliance. The analysis includes a large
sampling of almost all species within the genera currently rec-
ognized in the alliance, plus a broad sampling of outgroup taxa.
Our results closely agree with the Barneby and Grimes (1996)
circumscription of the Abarema alliance, in that this group in-
cludes most of the genus Abarema, plus Hydrochorea and
Balizia. However, the results require an emended circumscrip-
tion of the alliance, primarily because of the nonmonophyly
of the genus Abarema. The type species of the genus, Abarema
cochliacarpos, does not group with the remaining species of
Abarema, and one or more new generic circumscriptions must
be proposed in the future. During our earlier taxonomic and
morphological studies on Abarema (Iganci and Morim 2009,
2012), it was already possible to recognize A. cochliacarpos by
its very unique morphology. Abarema cochliacarpos is one of
the best-known species of the genus and is also one of the most
economically important species, being used as a popular anti-
inflammatory medicine in northeastern Brazil (Oliveira et al.
2013). On the basis of the morphological variation observed be-
tween populations of A. cochliacarpos, which relate to its geo-
36 INTERNATIONAL JOURNAL OF PLANT SCIENCES

graphical distribution in different biomes (Lewis 1987; Iganci
and Morim 2009, 2012), it is possible that the species as cur-
rently circumscribed may include two or more distinct species.
The results of our phylogenetic analysis suggest that taxo-
nomic realignments and descriptions of new taxa will be needed.
However, resolution and support are not yet adequate to deter-
mine the appropriate taxonomic changes. Hydrochorea plus
Balizia are merged into a clade that also includes the remaining
(and the majority of) species of Abarema s.l. in both the ETS and
combined analyses (figs. 1, 2). However, when analyzing com-
plete matK sequences for a smaller group of samples, the clade
formed by Balizia plus Hydrochorea appears as sister to the re-
maining species of Abarema s.l. (fig. A2). Our results show that
the species of Balizia are distinct from the species of Albizia in-
cluded in this study (except for Albizia subdimidiata) and closer
to Hydrochorea, as suggested by Barneby and Grimes (1996)
but in disagreement with Rico Arce (1999) and Lewis and Rico
Arce (2005).
The Andean species A. killipii,A. lehmannii,A. callejasii, and
A. centiflora form a clade sister to a clade composed of the re-
maining species of Abarema s.l. plus the genera Balizia and
Hydrochorea. The position of A. cochliacarpos within the Inga
alliance, together with this Andean clade and the nesting of the
genera Balizia and Hydrochorea with the remaining species of
Abarema, renders the latter polyphyletic. Considering the mor-
phological (especially related to the fruit morphology and seed
dispersal) and ecological differences that also allow us to recog-
nize Balizia and Hydrochorea as taxa distinct from Abarema
s.l., the Andean clade of Abarema s.l. will be recognized as a
new genus in a forthcoming publication. The Andean species
clade can be diagnosed by its particular morphology, especially
the flowers that are clustered in long, spike-like inflorescences.
Some of these species (A. killipii and A. lehmannii) were placed
by Britton and Rose (1928) in their genus Punjuba Britton &
Rose, along with Punjuba racemiflora (Donn. Sm.) Britton &
Rose (pAbarema) and Punjuba dependens (Rusby) Killip
(pZygia). However, resurrecting the genus Punjuba is not ap-
propriate because the type species of the genus, P. racemiflora,
does not group with the other species of the clade, and further
studies are required before making any taxonomic decisions.
An emended circumscription of the clade that comprises the
Abarema alliance is necessary. In its new circumscription, the
Abarema alliance might be renamed because the type species,
A. cochliacarpos, now belongs to the Inga alliance.
Our results suggest that fruit morphology should be used
carefully in determining relationships and circumscribing taxa,
as pointed out by Nielsen (1981). For example, the fruit mor-
phology in Pararchidendron is similar to that of Abarema, but
Pararchidendron is not part of the alliance, contrary to the sug-
gestion of Lewis and Rico Arce (2005). The similar fruit mor-
phologies do not reflect phylogenetic relationships (figs. 1, 2,
A1). Pararchidendron is sister to Archidendron, in a larger clade
comprising Australian Ingeae (Brown 2008; Brown et al. 2008).
The genus Archidendron comprises 94 species from Asia and
Australia, and many species in this genus were first transferred
from Pithecellobium to Abarema. Although Archidendron has
been well established for a long time, taxonomic misinterpreta-
tions are still common because of the similarities between the
fruit and leaves of Archidendron and those of Abarema. Recent
published floras have cited two species of Abarema for Asia
(Chen and Sun 2006; Zhu et al. 2007). Even though Abarema
cordifolia (T.L.Wu) C.Chen. & H.Sun (combined from Archi-
dendron cordifolium T.L.Wu) and Abarema multifoliolata (H.Q.-
Wen) X.Y.Zhu (combined from Archidendron multifoliolatum)
appear similar in pod morphology to Abarema species, the genus
Archidendron is phylogenetically closer to Cojoba Britton &
Rose and Zygia P. Browne (Lewis and Rico Arce 2005). Thus,
both curved to spiral pods with a reddish endocarp and seeds
with a partly translucent testa appear to have evolved indepen-
dently a number of times within the tribe Ingeae.
Another possible taxonomic circumscription could be to rec-
ognize all the species within the alliance as a single genus, syn-
onymizing Balizia and Hydrochorea into Abarema s.l. (except
A. cochliacarpos). Such a treatment would have to circum-
scribe a large and morphologically heterogeneous genus with
no clear synapomorphies. The more conservative results from
the matK analysis support the presence of a clade composed
of Hydrochorea plus Balizia nested within Abarema s.l. and
dividing Abarema into twoclades. Even though the results from
the ETS analysis do not completely agree with those ofthe matK
analysis, its more variable nature as a nuclear marker better re-
flects the geographical patterns observed in the alliance. Most
samples of Hydrochorea and Balizia cluster together with high
support values but are nested within a larger polytomy, to-
gether with the remaining species of Abarema s.l. The separation
of Balizia and Hydrochorea from Abarema seems to be recent,
as reflected by the genetic heterogeneity observed when analyz-
ing ETS sequences. Adding more molecular markers in future
analyses would likely increase the phylogenetic resolution, per-
mit a better understanding of the internal relationships within
the Abarema alliance, and guide decisions on generic delimi-
tation.
Preliminary Observations of the Biogeographic Patterns
within a Newly Circumscribed Abarema Alliance
Considering the results presented here, it is too early to make
meaningful conclusions about the evolutionary history of the
lineages within the alliance and the processes that drove spe-
cies diversification within Neotropical forests. Such conclu-
sions are hindered by the lack of complete congruence between
the plastid and nuclear markers in the molecular analyses. How-
ever, it is possible to make a few inferences based on the major
diversification events observed. The trees based on ETSs show
strong geographic structure (fig. 2C). It is also clear in our com-
bined phylogenetic analyses that an Andean lineage is sister to
all the remaining Abarema lineages (clade Y on fig. 2A). The
monophyletic Andean lineage is here represented by the blue
dots in figure 2B, clade Y in figures 1 and 2A, and the blue clade
in figure 2C. Other lineages diversified in the lowlands of the
Atlantic coastal rain forest, in the river basins of the Amazon,
or in the rain forests of Central America and the West Indies.
It is possible that an ancestral distribution was more widespread
throughout northern South America, i.e., a pan-Amazonian dis-
tribution (Hoorn et al. 2011) before the uplift of the Andes
mountain range (fig. 3A). A vicariance event during the Andes
uplift could have isolated the Andean lineage from the rest of
Abarema (fig. 3B). In addition to lineage isolation, Andean up-
lift also caused lots of abiotic changes within western South
America, promoting the extinction of some habitats and the ap-
IGANCI ET AL.—MOLECULAR PHYLOGENY OF THE ABAREMA ALLIANCE 37

pearance of other new ones and consequently leading to species
diversification and radiation (Antonelli and Sanmartín 2011;
Hoorn et al. 2011).
It is also clear that both Balizia and Hydrochorea have un-
dergone morphological responses and ecological trait changes
leading to their diversification. Both genera diversified within
the Amazon basin, Balizia also reaching the Central American
forests. Species of Hydrochorea are usually found near rivers or
in periodically flooded vegetation and are specialized for this
humid environment. Their adaptation to those habitats is seen
in their water-dispersed seed strategies. Hydrochorea is charac-
terized mostly by its lomentiform fruit; Balizia has ligneous, in-
dehiscent, or tardily dehiscent fruits (fig. 1E), sometimes (e.g.,
Balizia pedicellaris,Balizia elegans) occurring in nonflooded
Fig. 1 Phylogeny of the Abarema alliance, based on external transcribed spacer data. A, Phylogenetic tree including all terminals sampled.
B, Detail showing the Inga alliance (clade X), highlighting the position of Abarema cochliacarpos.C, Detail showing the Andean clade (clade Y)
and its relationship with the other taxa of the emended alliance. D, Summary of the phylogenetic tree shown in A, highlighting the main clades.
E, Fruit morphology, comparing Abarema,Balizia, and Hydrochorea. Node labels indicate Bayesian posterior probabilities/bootstrap values.
38 INTERNATIONAL JOURNAL OF PLANT SCIENCES

“terra firme”forest (Barneby and Grimes 1996). These genera
are resolved as closely related in both the ETS and matK anal-
yses, and most samples (but not all) form a clade with high sup-
port values. This clade is nested within Abarema s.l. in the com-
bined analysis. Nevertheless, the species relationships within
each genus and their relationships to the Amazonian and Atlan-
tic forest species of Abarema s.l. are still unclear. The remaining
species of Abarema s.l. (green, yellow, and red clades in fig. 2C)
are part of a more diversified lineage within the alliance, when
compared to the Andean clade.
Some general biogeographical patterns are worthy of men-
tion. After divergence of the Andean lineage, the distribution
Fig. 1 (Continued)
IGANCI ET AL.—MOLECULAR PHYLOGENY OF THE ABAREMA ALLIANCE 39

Fig. 2 Biogeography of the Abarema alliance (emended alliance). A, Phylogeny of the Abarema alliance, based on the combined analysis of matK and external transcribed spacer (ETS)
sequences, indicating the major clades. Node labels indicate Bayesian posterior probabilities. B, Geographic distribution of species from the Abarema alliance (emended alliance: clades Y, Z,
and W in A), highlighting the most important distribution areas. C, Circular phylogenetic tree based on ETS data, corresponding to the large tree in figure 1A(colors correspond to geo-
graphic distribution in B).

of Abarema over eastern South America was probably inter-
rupted by the expansion of drier vegetation formations that to-
gether formed a dry corridor in central Brazil (fig. 3C), includ-
ing the Cerrado and Caatinga biomes (Prado and Gibbs 1993).
Within the Atlantic domain (green dots in fig. 2B, green clade
in fig. 2C), most species of Abarema s.l. are found in coastal
scrub (restinga vegetation) on sandy soils (Iganci and Morim
2009, 2012, 2015). The term “domain”refers to large phyto-
geographic regions that include heterogeneous physiognomies
recognized by their floristic composition under specific ecolog-
ical conditions (Joly et al. 1999; Fiaschi and Pirani 2009). In the
Amazonian forest, Abarema species (red dots in fig. 2B, red
clade in fig. 2C) grow mostly near the rivers, also on sandy soils,
and the most species-rich areas are found near the black-water
courses of the Rio Negro basin. Thus, although covering a rel-
atively large distribution range, the niche is still conserved for
those more recent lineages. Multiple independent events must
have occurred to colonize the West Indies and Central America
(yellow dots in fig. 2B, yellow clade in figs. 2C,3C). The history
of those lineages could reflect a more complex geological his-
tory of those regions. Abarema s.l. has a wide species diver-
sity and many endemic species, exclusive of the West Indian
islands and Central American forests. Immigrating into these
relatively young environments must have involved adaptation
to new habitats and resulted in stochastic mutations over time
to generate new species. These dispersal events probably oc-
curred as the islands of the Antilles emerged, facilitating a mi-
gration path for tropical species. Noteworthy is the occurrence
of some taxa of the Central American lineages on the Pacific
coast of northwestern South America, which could represent
a secondary colonization back into the Andes, specifically into
the Chocó forests of Colombia (see yellow dots in fig. 2Band
the yellow area in fig. 3D). This secondary migration back into
South America might have been promoted by the uplift of the
Panama Isthmus and the connection of Central to South America.
We also highlight a recent and rapid species diversification
within the Amazonian forest, as discussed for the closely related
genus Inga by Richardson et al. (2001), where most speciation
events occurred during the past 10 Myr, the majority during the
past 2 Myr. This pattern could be related to the rapid coloni-
zation of an emerging habitat after the drainage of large areas
of Amazonian wetlands, evidenced by both palynological and
phylogenetic studies (Hoorn et al. 2011). The rapid diversifica-
tion could also have been driven by the climatic shifts during
the Pleistocene that caused forest fragmentation and population
isolation in refugia during cooler periods, even though the exis-
tence of Amazonian refuges is not a universally accepted theory
(Antonelli and Sanmartín 2011; Hoorn et al. 2011). Thus, the
relatively greater number of Abarema species in Amazonia re-
flects this recent and complex diversification history. Abarema
jupunba is the most widespread species in the genus, with most
populations occurring in the Amazon basin, but others occur in
the Atlantic forest, the West Indies, Central America, and the
Andean valleys, covering almost the whole distribution range
of the genus (most of the red dots outside Amazonia in fig. 2B).
Abarema jupunba appears as a nonmonophyletic species in the
ETS phylogeny. Multiple copies of ETSs and a possible noncon-
certed evolution in this DNA region can be problematic, al-
though considering the large sampling and the multiple acces-
sions grouping together for many species (e.g., A. filamentosa,
A. langsdorffii,A. microcalyx,A. cochleata,A. adenophora,
and A. leucophylla), ETS seems to be a reliable source of nu-
cleotide data that have been used to reconstruct phylogenetic
relationships with good results in many other taxa. It is possible
that A. jupunba is not a monophyletic species. Within this spe-
cies two varieties are recognized on the basis of morphological
differences. Future studies regarding recent diversification in sis-
ter species, including population-level genetic studies, are re-
quired to elucidate these species-level taxonomic issues.
Conclusion
Our results agree in part with Barneby and Grimes’s (1996)
circumscription of the Abarema alliance, accepting Balizia as
closer to Hydrochorea and part of Abarema. However, both
ETS and matK analyses show that Abarema is a polyphyletic
genus requiring an emended taxonomic circumscription. The
type species of Abarema,Abarema cochliacarpos, clustered with
the genera Macrosamanea,Enterolobium,Inga,andZygia from
the Inga alliance in our analyses. In the combined analysis,
an Andean clade within the Abarema alliance appeared as sister
to a clade that includes the remaining species of Abarema and a
clade formed by Hydrochorea and Balizia. The Andean clade
should be described as a new genus. The remaining species of
Abarema are widespread within the Atlantic forest in Brazil,
the Amazon basin, Central America, and the West Indies. Those
species also require a new generic circumscription and name,
Fig. 3 Hypothesis of the evolutionary history of the distribution range of the alliance (emended alliance: clades Y, Z, and W in fig. 2A).
IGANCI ET AL.—MOLECULAR PHYLOGENY OF THE ABAREMA ALLIANCE 41

although resolution is inadequate at present to take this ac-
tion. The alliance comprising the Andean clade, Hydrochorea,
Balizia, and the bulk of Abarema as traditionally circumscribed
will also require a new informal name because the type species
of Abarema is now placed in the Inga alliance.
This work represents a first phylogenetic study of the Aba-
rema alliance and allows some considerations about the bio-
geography of the alliance. Many outstanding questions will be
addressed in future studies.
Acknowledgments
We thank CNPq/Universal (480530/2012–2) and CNPq/
Sisbiota for the financial support; all the herbaria consulted,
especially the herbaria NY and RB, for allowing us to sam-
ple leaflets from herbarium specimens for molecular analysis;
and the legume researchers H. C. Lima, F. Bonadeu, M. Simon,
F. Garcia, and L.P. Queiroz, the two anonymous reviewers,
and the editors for their valuable suggestions.
Literature Cited
Antonelli A, I Sanmartín 2011 Why are there so many plant species
in the Neotropics? Taxon 60:403–414.
Ariati SR, DJ Murphy, F Udovicic, PY Ladiges 2006 Molecular phy-
logeny of three groups of acacias (Acacia subgenus Phyllodineae)in
arid Australia based on the internal and external transcribed spacer
regions of nrDNA. Syst Biodivers 4:417–426. doi:10.1017/S14772
00006001952.
Baldwin BG, S Markos 1998 Phylogenetic utility of the external tran-
scribed spacer (ETS) of 18S–26S rDNA: congruence of ETS and ITS
trees of Calycadenia (Compositae). Mol Phylogenet Evol 10:449–463.
doi:10.1006/mpev.1998.0545.
Barneby RC, JW Grimes 1996 Silk tree, Guanacaste, monkey’s earring: a
generic system for the synandrous Mimosaceae of the Americas. Part I.
Abarema,Albizia, and allies. Mem NY Bot Gard 74. New York Botan-
ical Garden, Bronx.
Bässler M 1998 Abarema maestrens Bässler. Page 121 in Flora de la
República de Cuba. Ser A. Plantas vasculares. Fasc 2. Mimosaceae.
Koeltz, Königsberg.
Bentham G 1844 Notes on Mimoseae, with a short synopsis of spe-
cies. Lond J Bot 3:195–226.
Britton NL, JN Rose 1928 Punjuba Britton & Rose, gen. nov. Page
28 in North American flora. Vol 23. Pt 1. (Rosales) Mimosaceae.
New York Botanical Garden, Bronx.
Brown GK 2008 Systematics of the tribe Ingeae (Leguminosae-
Mimosoideae) over the past 25 years. Muelleria 26:27–42.
Brown GK, DJ Murphy, JT Miller, PY Ladies 2008 Acacia s. s. and
its relationship among tropical legumes, tribe Ingeae (Leguminosae:
Mimosoideae). Syst Bot 33:739–751.
Burkart A 1943 Las leguminosas Argentinas. Acme Agency, Buenos
Aires.
——— 1964 Leguminosas nuevas o criticas. VI. Darwiniana 13:428–
448.
Chen C, H Sun 2006 Abarema cordifolia (T.L. Wu) C. Chen & H. Sun.
Page 324 in Flora Yunnanica. Vol 10. Science Press, Beijing.
Fiaschi P, JR Pirani 2009 Review of plant biogeographic studies in
Brazil. J Syst Evol 47:477–496.
Hoorn C, FP Wesselingh, H ter Steege, MA Bermudez, A Mora,
J Sevink, I Sanmartín, et al 2011 Origins of biodiversity—response.
Science 331:399–400.
Iganci JRV, STS Miotto, TT Souza-Chies, TE Särkinen, BB Simpson,
MF Simon, RT Pennington 2013 Diversification history of Ades-
mia ser. psoraleoides (Leguminosae): evolutionary processes and
the colonization of the southern Brazilian highland grasslands. S Afr
J Bot 89:257–264. doi:10.1016/j.sajb.2013.06.016.
Iganci JRV, MP Morim 2009 Abarema (Leguminosae, Mimosoideae)
no estado do Rio de Janeiro, Brasil. Rodriguesia 60:581–594.
——— 2012 Abarema (Leguminosae, Mimosoideae) in the Atlantic
Domain, Brazil. Bot J Linn Soc 168:473–486.
——— 2015 Abarema.In Lista de espécies da flora do Brasil. Jar-
dim Botânico do Rio de Janeiro. http://floradobrasil.jbrj.gov.br/jabot
/floradobrasil/FB22756.
Joly CA, MPM Aidar, CA Klink, DG McGrath, AG Moreira, P Moutinho,
DC Nepstad, et al 1999 Evolution of the Brazilian phytogeography
classification systems: implications for biodiversity conservation. Cien
Cult 51:331–348.
Katoh K, H Toh 2008 Recent developments in the MAFFT multiple
sequence alignment program. Briefings Bioinf 9:286–298.
Lavin M, M Thulin, JN Labat, RT Pennington 2000 Africa, the odd
man out: molecular biogeography of dalbergioid legumes (Fabaceae)
suggests otherwise. Syst Bot 25:449–467.
Lavin M, MF Wojciechowski, A Richman, J Rotella, MJ Sanderson,
A Beyra Matos 2001 Identifying Tertiary radiations of Fabaceae
in the Greater Antilles: alternatives to cladistic vicariance analysis.
Int J Plant Sci 162:S53–S76.
Lewis GP 1987 Legumes of Bahia. Royal Botanic Garden, Kew. 369 p.
Lewis GP, MLRico Arce 2005 Tribe Ingeae. Pages 193–213in G Lewis,
B Schire, B Mackinder, M Lock. Legumesof the world. Royal Botanic
Gardens, Kew.
Morim MP, MVB Soares 2015 Hydrochorea.In Lista de espécies da flora
do Brasil. Jardim Botânico do Rio de Janeiro. http://floradobrasil.jbrj
.gov.br/jabot/floradobrasil/FB22758.
Nielsen I 1981 Tribe 5. Ingeae. Pages 173–190 in RM Polhill and PH
Raven, eds. Advances in legume systematics. Pt 1. Royal Botanic
Gardens, Kew.
Nylander JAA 2004 Mr. Modeltest v2. Program distributed by the
author. Evolutionary Biology Centre, Uppsala University.
Oliveira RF, PR Ribeiro, GKM Santos, CS Oliveira, PRC Silva, HA
Oliveira, RC Trinidade, LG Fernandez 2013 Evaluation of the hep-
atotoxicity of Abarema cochliacarpos extracts in mice Mus musculus.
Rev Bras Farmacogn 23:674–679.
Prado DF, PE Gibbs 1993 Patterns of species distributions in the dry
seasonal forests of South America. Ann Mo Bot Gard 80:902–927.
Rambaut A 2009 FigTree, ver. 1.3.1. http://tree.bio.ed.ac.uk/software
/figtree.
Richardson JE, RT Pennington, TD Pennington, PM Hollingsworth
2001 Rapid diversification of a species-rich genus of Neotropical
rain forest trees. Science 293:2242–2245.
Rico Arce ML 1999 New combinations in Mimosaceae. Novon 9:
554–556.
Ronquist F, JP Huelsenbeck 2003 MrBayes 3: Bayesian phylogenetic
inference under mixed models. Bioinformatics 19:1572–1574.
Souza ER, GP Lewis, F Forest, AS Schnadelbach, C van den Berg, LP
Queiroz 2013 Phylogeny of Calliandra (Leguminosae-Mimosoideae)
based on nuclear and plastid molecular markers. Taxon 62:1201–1220.
Stamatakis A 2014 RAxML, version 8: A tool for phylogenetic anal-
ysis and post-analysis of large phylogenies. Bioinformatics 30:1312–1313.
42 INTERNATIONAL JOURNAL OF PLANT SCIENCES

Wojciechowski MF, M Lavin, MJ Sanderson 2004 A phylogeny of
legumes (Leguminosae) based on analysis of the plastid matK gene
resolves many well-supported subclades within the family. Am J
Bot 91:1846–1862.
Zhang Q, TS Feild, A Antonelli 2015 Assessing the impact of
phylogenetic incongruence on taxonomy, floral evolution, biogeo-
graphical history, and phylogenetic diversity. Am J Bot 102:566–
580.
Zhu X, Y Du, J Wen, BJ Bao 2007 Legumes of China: a checklist.
International Legume Database and Information Service (ILDIS),
School of Biological Sciences, University of Reading, Reading,
UK.
IGANCI ET AL.—MOLECULAR PHYLOGENY OF THE ABAREMA ALLIANCE 43