ArticlePDF Available

Speciation slowing down in widespread and long-living tree taxa: Insights from the tropical timber tree genus Milicia (Moraceae)

  • asbl Nature+
  • Institut de Recherche pour le Développement (IRD), Montpellier, France

Abstract and Figures

The long generation time and large effective size of widespread forest tree species can result in slow evolutionary rate and incomplete lineage sorting, complicating species delimitation. We addressed this issue with the African timber tree genus Milicia that comprises two morphologically similar and often confounded species: M. excelsa, widespread from West to East Africa, and M. regia, endemic to West Africa. We combined information from nuclear microsatellites (nSSRs), nuclear and plastid DNA sequences, and morphological systematics to identify significant evolutionary units and infer their evolutionary and biogeographical history. We detected five geographically coherent genetic clusters using nSSRs and three levels of genetic differentiation. First, one West African cluster matched perfectly with the morphospecies M. regia that formed a monophyletic clade at both DNA sequences. Second, a West African M. excelsa cluster formed a monophyletic group at plastid DNA and was more related to M. regia than to Central African M. excelsa, but shared many haplotypes with the latter at nuclear DNA. Third, three Central African clusters appeared little differentiated and shared most of their haplotypes. Although gene tree paraphyly could suggest a single species in Milicia following the phylogenetic species concept, the existence of mutual haplotypic exclusivity and nonadmixed genetic clusters in the contact area of the two taxa indicate strong reproductive isolation and, thus, two species following the biological species concept. Molecular dating of the first divergence events showed that speciation in Milicia is ancient (Tertiary), indicating that long-living tree taxa exhibiting genetic speciation may remain similar morphologically.Heredity advance online publication, 19 February 2014; doi:10.1038/hdy.2014.5.
Content may be subject to copyright.
Speciation slowing down in widespread and long-living tree
taxa: insights from the tropical timber tree genus Milicia
3, J Duminil4,5, CW Dick6,7, J-L Doucet1, ASL Donkpe
´gan1, M Pluijgers1,
B Sinsin2, P Lejeune8and OJ Hardy4
The long generation time and large effective size of widespread forest tree species can result in slow evolutionary rate and
incomplete lineage sorting, complicating species delimitation. We addressed this issue with the African timber tree genus
Milicia that comprises two morphologically similar and often confounded species: M. excelsa, widespread from West to East
Africa, and M. regia, endemic to West Africa. We combined information from nuclear microsatellites (nSSRs), nuclear and
plastid DNA sequences, and morphological systematics to identify significant evolutionary units and infer their evolutionary and
biogeographical history. We detected five geographically coherent genetic clusters using nSSRs and three levels of genetic
differentiation. First, one West African cluster matched perfectly with the morphospecies M. regia that formed a monophyletic
clade at both DNA sequences. Second, a West African M. excelsa cluster formed a monophyletic group at plastid DNA and was
more related to M. regia than to Central African M. excelsa, but shared many haplotypes with the latter at nuclear DNA. Third,
three Central African clusters appeared little differentiated and shared most of their haplotypes. Although gene tree paraphyly
could suggest a single species in Milicia following the phylogenetic species concept, the existence of mutual haplotypic
exclusivity and nonadmixed genetic clusters in the contact area of the two taxa indicate strong reproductive isolation and, thus,
two species following the biological species concept. Molecular dating of the first divergence events showed that speciation in
Milicia is ancient (Tertiary), indicating that long-living tree taxa exhibiting genetic speciation may remain similar
Heredity (2014) 113, 74–85; doi:10.1038/hdy.2014.5; published online 19 February 2014
Keywords: Milicia; speciation; phylogeny; phylogeography; Tertiary diversification
Angiosperm diversification was termed ‘abominable mystery’ by
Charles Darwin because this is likely one of the most complex
processes ever investigated by ecologists and biologists. This is
because of the difficulty to understand the biogeographical history
of specific populations as it depends on interactions between infinite
variants of abiotic and biotic phenomena: geological events, climate
changes and characteristics, life traits, gene flow between distant
populations and sister taxa and so on (Davies et al., 2004). Although
most biologists thought that the use of modern tools in phylogenetic
taxonomy and molecular ecology in conjunction with data from
environmental sciences may provide more accurate insights for
inference of biogeographical and evolutionary patterns, the existence
of heterogeneous genomic divergence (differentiation across the
genome can be highly variable) and the fact that differentiation and
speciation are not always synonymous of morphological change
(producing cryptic species) can blur the outcomes (Bickford et al.,
2006; Nosil et al.,2009).
Especially, addressing speciation in sister taxa is a real challenge
complicated by debates on definition of species, as different
approaches can reveal different patterns. Most definitions of species
are derived from two concepts: (1) the biological species concept
(BSC) that defines species as reproductively isolated groups of
living organisms (Mayr, 1942), and (2) the phylogenetic species
concept (PSC) that groups in a given species individuals forming a
monophyletic group (or a clade), according to the monophyly version
(De Queiroz, 2007). Because testing for reproductive isolation is
difficult and PSC requires many (molecular) traits, historical taxo-
nomy has mostly been based on the morphological similarity of
individuals, and ideally the occurrence of multiple diagnostic
qualitative traits to define species (Hey, 2006; Mallet, 2010).
Nowadays, when multilocus genetic markers are available, large-scale
1Laboratory of Tropical and Subtropical Forestry, Unit of Forest and Nature Management, Gembloux Agro-Bio Tech, University of Liege, Gembloux, Belgium; 2Laboratory of
Applied Ecology, Faculty of Agronomic Science, University of Abomey-Calavi, Cotonou, Benin; 3Biodiversity and Landscape Unit, Gembloux Agro-Bio Tech, University of Liege,
Gembloux, Belgium; 4Evolutionary Biology and Ecology, Faculte
´des Sciences, Universite
´Libre de Bruxelles, Brussels, Belgium; 5Bioversity International, Forest Genetic
Resources Programme, Sub-Regional Office for Central Africa, Yaounde
´,Cameroon;6Department of Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, MI,
USA; 7Smithsonian Tropical Research Institute, Gamboa, Republic of Panama and 8Unit of Forest and Nature Management, Gembloux Agro-Bio Tech, University of Liege,
Gembloux, Belgium
Correspondence: Dr K Daı
¨nou, Laboratory of Tropical and Subtropical Forestry, Unit of Forest and Nature Management, Gembloux Agro-Bio Tech, University of Liege, 2 Passage
des De
´s, 5030 Gembloux, Belgium.
Received 23 May 2013; revised 18 December 2013; accepted 19 December 2013; published online 19 February 2014
Heredity (2014) 113, 74–85
2014 Macmillan Publishers Limited All rights reserved 0018-067X/14
population sampling and Bayesian clustering analyses allow identify-
ing groups of individuals with a shared recent common ancestry,
groups that have been reproductively isolated (historically at least)
and can be thought as candidate units corresponding to the BSC.
Moreover, when DNA sequences are available, they constitute ideal
data to identify monophyletic groups of individuals that can be
considered as evolutionary units corresponding to the PSC,
although incomplete lineage sorting can blur the evidence. Whether
historical taxonomic species delimitation converges with the BSC or
the PSC has been rarely examined in African tree species (but see
Koffi, 2010).
Although different approaches may lead to incongruent results
because of reticulated evolution and incomplete lineage sorting, their
integration in a pluralistic species concept can help make a consistent
decision for taxonomic classification and conservation issues
(Freeland et al., 2011). Hence, species delimitation requires to
integrate data from molecular phylogeny, classical taxonomy
(morphology-based), ecology, historical and contemporaneous bio-
geography in a way that permits the reconstruction of the evolu-
tionary history of the target taxon (see, for example, Datwyler and
Weiblen, 2004; Odee et al., 2012; Scotti-Saintagne et al.,2012).We
adopted a pluralistic approach to infer biogeographical history and
delimit significant evolutionary units in the African genus Milicia
Sim. Taxonomists recognize two sister tree species, M. excelsa (Welw.)
C.C. Berg and M. regia (Chev.) C.C. Berg, both called Iroko in the
wood trade. Iroko is one of the most important timber woods of
Africa. At present, M. regia is viewed as a vulnerable taxon by IUCN
(International Union for Conservation of Nature), whereas M. excelsa,
a near-threatened species according to the same source, is considered
threatened or endangered in some West and East African countries
(Ebert, 2004; Ofori and Cobbinah, 2007).
Milicia belongs to the Moreae tribe that also contains Morus,the
closest genus to Milicia (Zerega et al., 2005; Nepal, 2012). The most
recent common ancestor of Milicia and Morus was estimated between
41 and 72 million years ago (mya) (Zerega et al., 2005). The Moreae
have a Eurasian origin and may have primarily diversified during the
Palaeocene and the Eocene. At the genus level, few genetic studies
have been conducted in Milicia. Using the chloroplast intron trnL and
the intergenic spacer trnL-F,Oforiet al. (2001) found no variation in
populations collected from the Upper Guinea, the rainforest zone at
the western part of the West Africa (Supplementary Figure S1). Ofori
et al. (2003) tried to assess genetic differentiation between Milicia
species via chloroplast markers, but their populations were assigned to
species according to vegetation types because of difficulties in
obtaining diagnostic morphological traits of the sampled individuals.
Within M. excelsa, Dainou et al. (2010) demonstrated the existence of
different gene pools in the Lower Guinea (the rainforest from the
Atlantic Ocean until the western side of the Oubangui-Congo River;
Supplementary Figure S1), despite extensive gene dispersal (Bizoux
et al., 2009), and suggested a relatively recent fragmentation during
Pleistocene ice ages. Curiously, Dainou et al. (2010) also showed that
West Af r i c an M. excelsa populations from Benin in the Dahomey gap
(the savannah interval separating the Upper Guinean and Lower
Guinean forest blocks; Supplementary Figure S1) strongly diverged
from Central African samples, despite the fact that M. excelsa presents
a continuous range from West to Central Africa. This pattern
questions whether M. excelsa individuals from Benin are more closely
related to West African M. regia or to Central African M. excelsa
populations. In fact, in contrast to the Lower Guinea, we have
relatively little information about biogeographical patterns in the
Upper Guinea or the Congolia (the humid forest zone eastwards from
the Oubangui-Congo River; Supplementary Figure S1; OJ Hardy,
The present study aims to disentangle phylogenetic relationships
between these sister species using an integrative approach that
combines morphological and molecular data. For the present study,
we collected samples from most of the distribution range of Milicia in
Africa, from Senegal to Tanzania and Democratic Republic of the
Congo. Our study represents the largest scale phylogeography study
focused on tropical moist forests in Africa. Our investigations were
motivated by the following questions: (1) In West Africa, are
morphological and genetic markers congruent in delineating species
boundary in Milicia? (2) At the scale of Africa, what are the significant
evolutionary units within the Milicia species complex and which of
them may represent different species? (3) What are the divergence
times between evolutionary units and can they be related to the
paleodynamics of African rainforests?
Study taxa
Milicia is a tree up to 35–45 m tall and 1.5 m in diameter. The annual
increment in diameter was estimated at 5.7 mm (Durrieu de Madron, 2003),
and regular fruiting was observed on trees of at least 60cm in diameter
(Dainou et al.,2012b).M. regia is only distributed in Upper Guinea mostly in
the evergreen forests of West Africa, westwards of the Dahomey gap
(Supplementary Figure S1). M. excelsa displays a larger geographic distribution
from West to East Africa and can occur in a wide range of forest habitats, but is
more abundant in semi-deciduous forests (Supplementary Figure S1). The
range of the two taxa overlaps from the central part of Liberia to the East of
Ghana. M. excelsa and M. regia are morphologically so similar that they are
usually confused in the wood trade (ATIBT, 2010) and even by researchers and
botanists (Chevalier, 1917; Schnell, 1946; Ofori et al.,2003).Theirwoodsdo
not present significant differences (White, 1966).
Very little information is available regarding the ecology of M. regia (Dainou
et al.,2012a).M. excelsa is a pioneer species with increasing population
densities from evergreen forests to semi-deciduous forests (Nichols et al.,
1998). Both species are dioecious. M. excelsa is wind pollinated and its seeds
are mostly dispersed by birds and the fruit-bat Eidolon helvum (Dainou et al.,
2012b). In West Africa, the two species flower and fruit at the same period
from January to March (Berg, 1977; Nyong’o et al.,1994).
Sampling and DNA extraction
Samples were obtained from 849 individuals in 17 countries. These included
550 individuals previously studied in Dainou et al. (2010) and sampled from
the Lower Guinea and Benin (Dahomey gap) and 46 collected from herbarium
collections at the National Botanical Garden of Belgium. During the sampling
work, collaborators did not pay attention to the taxonomic identity of
individuals. Whenever possible, samplers were asked to collect one entire leaf
per individual especially for West African specimens in order to characterize
leaf traits, but this was not always possible and many samples consist of a
fragment of limb or cambium tissue. The morphological characterization of
herbarium vouchers was conducted directly on specimens, as we could not
collect entire leaf samples from herbarium collections. Fresh materials were
silica-dried. DNA was extracted using the DNeasy Plant Mini kit (QIAGEN,
Venlo, The Netherlands).
Genotyping and sequencing
Nuclear microsatellites (nSSRs) were genotyped on the whole sample through
seven loci: Mex51, Mex63, Mex81, Mex95, Mex137, Mex163a and Mex202
(Ouinsavi et al., 2006). Amplification conditions were as reported in Bizoux
et al. (2009). Genotyping was repeated until each individual is scored for at
least six loci.
The amplification and polymorphism of nuclear DNA sequences (nDNA)
were tested at 12 gene regions: ADH,GScp,ADHc,RnI1,TPI 6rn/4rn,LF4
Cl1R,LF4 Cl4R,Agt1,Apg1,At103 (Li et al.,2008),andPEPC E1/E2 and PEPC
E2/E3 (J Duminil et al., unpublished). Three of them successfully amplified,
Evolutionary history of the genus Milicia
¨nou et al
Agt1,Apg1 and At103, but only the latter was polymorphic and then selected
for further analyses. For the plastid genome (pDNA), two intergenic regions,
psbA-trnH and trnC-ycf6, were selected. The three nDNA and pDNA sequences
were successfully amplified on fresh material but not on the herbarium
vouchers in most cases. To overcome this problem, we defined new internal
primers that amplified a significant proportion of herbarium samples. The new
primer sequences were as follows: 50-TTCTTGTTCTATCACGAAGTTTGG and
30-AATCCACTGCCTTGATCCAC (411bp) for psbA-trnH,5
GAG (342 bp) for At 103. PCRs were performed in a thermal cycler PTC 200
(MJ-Research, Waltham, MA, USA). psbA-trnH was amplified in a total
volume of 25 ml with the Phusion polymerase (Finnzymes Espoo, Finland)
following the manufacturer’s protocol. PCR conditions for trnC-ycf6 were as
described in Dainou et al. (2010) and were also applied for At103.Fromthe
subsample of 260 individuals selected for sequencing, 195, 225 and 202 were
successfully amplified for psbA-trnH,trnC-ycf6 and At103, respectively, with
172 individuals common to the three fragments. In order to test whether
Milicia forms a monophyletic group, we also included a sample from Morus
indica in the sequencing processes.
Morphological characterization
In an exhaustive literature review of morphological traits that differentiate
M. regia and M. excelsa,Dainouet al. (2012a) reported that there could be a
slight difference in female flowers according to Aubre
´ville (1959), whereas
most botanists agree that ‘in adult specimens the only differences can be found in
the leaf venation and the indument’ (Berg, 1977; pg 349). One qualitative and
three quantitative leaf characters distinguish the adult specimens of the two
species (Hawthorne and Jongkind, 2006). (1) Leaf lower surface is soft to touch
(variable Soft) because of the presence of rounded areoles containing
microscopic hairs in M. excelsa (Soft ¼1), whereas this feature is absent in
M. regia (Soft ¼0). (2) The number of pairs of secondary veins (variable
N_veins) should not exceed 11 in M. regia, whereas it ranges from 12 to 22 in
M. excelsa. However, Berg (1977) and Hawthorne and Jongkind (2006)
reported a few cases of M. regia leaves displaying up to 14 veins and
M. excelsa leaves with only 10 veins. In such cases, the authors implicitly used
only the character Soft to identify each species. (3) The length/width ratio of
limb (variable L/w_limb) equals B1.4 in M. regia and B1.7 in M. excelsa.
(4) The mean distance between two contiguous veins (variable D_veins)in
M. excelsa is expected to be about half the value found in M. regia specimens
We characterized these four traits on a subsample of 114 individuals with
entire leaves that came from 11 countries of West and Central Africa. We
performed a principal component analysis using STATISTICA 6.0 (StatSoft, 2004)
considering the quantitative variables L/w_limb,N_veins and D_veins.The
qualitative variable Soft was integrated in the principal component ordination
using two different symbols to represent individuals with Soft ¼0or1.
Identification of genetic clusters
Identification of the number of genetic clusters Kwas based on variation at
nSSRs, using the Bayesian clustering algorithm implemented in TESS 2.3.1
(Chen et al., 2007) without prior information of the morphological descrip-
tion. We applied the method as described in Dainou et al. (2010) with some
slight differences. Only the admixture model was run with an interaction
parameter c¼0, which means that no spatial information was included in the
analysis. For each fixed value of the number of clusters (Kmax) ranging from
2 to 10, we performed 10 runs with a burn-in of 20000 for a total number of
sweeps of 100 000. The best value of Kwas determined after plotting values of
log-likelihood and deviance information criterion against Kmax: for both
approaches, the plateau starting point indicated the number of clusters K
(Pritchard et al.,2000andFranc¸ois and Durand, 2010, respectively). We also
applied the method in Evanno et al. (2005) to confirm the previous methods.
The program CLUMPP (Jakobsson and Rosenberg, 2007) was then used to
summarize the five most reliable simulations for the best K, and to estimate
averaged cluster membership coefficients qfor each individual. At this step, an
individual was assigned to a given cluster whenever q40.5.
Phylogeography and phylogeny at pDNA and nDNA
The following analyses were performed on the subsample of 172 individuals
sequenced at all pDNA and nDNA fragments. Two sets of haplotypes were
defined based on variation at (1) At103, and (2) over the two plastid sequences
psbA-trnH/trnC-ycf6 (Supplementary Tables S1 and S2). Median joining
networks were constructed with NETWORK 4.6 (Bandelt et al., 1999).
Bayesian inferences of phylogenetic relationships were performed using
BEAST 1.7.3 (Drummond and Rambaut, 2007). M. indica was used as outgroup.
First, JMODELTEST 0.1.1 (Posada, 2008) indicated F81 and HKY as the likely
substitution models for psbA-trnH /trnC-ycf6 and At103, respectively, after
likelihood ratio tests. Then, for each type of genomic sequence, BEAST was run
and set up as follows: (1) for the model F81, the appropriate code was
imported from BEAST webpage (
code) and we modified the input file accordingly; (2) we assumed an
uncorrelated lognormal relaxed molecular clock (in general, the uncorrelated
relaxed clock models perform well even if the analyzing data are clocklike;
Drummond et al., 2006) and a Yule tree prior; (3) the tree root height was
constrained with the age of the MRCA of Milicia and Morus (Zerega et al.,
2005), assuming a normal distribution with a mean of 56 mya and the s.d. of
9 mya (corresponding to 95% of the distribution lying between 41 and
71 mya); (4) the Markov Chain Monte Carlo was run independently five times
for 50 million generations each, sampling 1 tree every 2000 generations. The
output files of BEAST were checked with TRACER considering the effective sample
size of 4100 and combined using LOGCOMBINER, both programs distributed
with BEAST.TREEANNOTATOR 1.7.3 (also distributed with BEAST) was then used to
select the maximum clade credibility tree (the tree with the highest sum of
posterior probabilities on its internal nodes) with the height of nodes
calculated as the mean height of all sampled trees for that clade. FIGTREE
1.3. (A Rambaut; was finally used to
visualize the phylogenetic trees.
Diversity and differentiation parameters: detecting
phylogeographic signal
The genetic clusters determined from nSSRs data were used as basis for
computing diversity and differentiation parameters for each type of marker.
Relative to the degree of genetic diversity, the following genetic diversity
parameters were computed for each genetic cluster: A0,totalnumberof
haplotypes (pDNA and nDNA) or alleles (nSSRs); RS, haplotypic (pDNA and
nDNA) or allelic richness (nSSRs); %Rpriv, percentage of private haplotypes or
alleles; and HTand VT
, haplotypic diversity based on unordered and ordered
haplotypes, respectively (after Pons and Petit, 1996). Differentiation at nSSRs
between pairs of genetic clusters was computed using FSTand tested with FSTAT (Goudet, 1995). To check for a phylogeographic signal at nSSRs, we also
computed RST that is based on allele size and expected to be larger than FST if
mutations (under stepwise model) have contributed to differentiation (Hardy
et al., 2003). We also estimated differentiation between clusters and tested for a
phylogeographic signal at pDNA and nDNA using GST (for unordered
haplotypes) and NST (for ordered haplotypes) (Pons and Petit, 1996). Note
that a genetic distance matrix between haplotypes was first inferred following
the number of polymorphic characters that differed in state between each pair
of haplotypes for pDNA and nDNA. In both cases, 10000 permutations were
performed. Unless specified, computations described above were carried out
using SPAGEDI 1.3 (Hardy and Vekemans, 2002) except haplotypic and allelic
richness computed using the rarefaction procedure implemented in HP-RARE 1.1
(Kalinowski, 2005). RSand %Rpriv were computed for subsamples of seven
genes and 55 individuals at pDNA and nDNA sequences and nSSRs,
respectively, taking into account the minimum size of the clusters for each
type of markers.
Morphology-based species delimitation
The two first axes of the principal component analysis explained
93.05% of the total variance (Figure 1). The first axis, which was
mainly determined by the variable N_veins (45.3% of contribution),
split individuals in two groups, one containing samples from Central
Africa, Benin and Nigeria presenting soft leaves (Soft ¼1), and the
Evolutionary history of the genus Milicia
¨nou et al
other one containing samples from Senegal to Sierra Leone and Ivory
Coast, without hairs underneath the leaves (Soft ¼0), although there
is a small overlap on the principal component analysis plane. As a
consequence, the characters N_veins,D_veins and L/w_limb display
significantly different mean values between individuals with Soft ¼0
and those with Soft ¼1 but their distributions overlap (Table 1). For
example, four trees with Soft ¼0 presented 14–15 pairs of secondary
veins that is unexpected in that group.
Thereafter, as there are no clear species-specific limits for the three
quantitative traits, we will assign M. excelsa to individuals with
Soft ¼1andM. regia to individuals with Soft ¼0, as Berg (1977) and
Hawthorne and Jongkind (2006) suggested. The character Soft
allowed us to identify 335 specimens (135 from West Africa and
200 from Central Africa) because any small piece of leaf was enough
for that. All individuals assigned to M. regia were distributed from
Senegal to Ghana and never occurred in Central Africa. In West
Africa, M. excelsa was found from Guinea-Conakry to Nigeria,
making the overlap zone of the two species from Guinea-Conakry
to Ghana (Supplementary Figure S1). M. regia mostly occurred in the
evergreen forest zone, whereas M. excelsa was mostly present in semi-
deciduous forests.
The nSSR genetic clusters and relationships with morphological
Log-likelihood and deviance information criterion methods
(Pritchard et al.,2000andFranc¸ois and Durand, 2010, respectively)
suggested K¼5 as the number of clusters, whereas variation of DK
(Evanno et al., 2005) suggested two solutions: K¼2andK¼5
(Supplementary Figure S2). The case of K¼2clustersmaybe
explained by the existence of two sister species in our data set, as
the method of Evanno et al. (2005) is known to detect the uppermost
structure in situations of strong differentiation. It performs less well
on situations of weak and moderate genetic differentiations (Waples
and Gaggiotti, 2006). Therefore, we have to consider K¼5clustersas
the true number of clusters (K) in our data set. Cluster K1 grouped
together all morphology-based M. regia individuals (Figure 2). All
West African M. excelsa individuals were grouped into K2. Clusters
K3,K4 and K5 were constituted only of M. excelsa specimens from
Central Africa. A proportion of 5.4% of the whole sample was not
assigned to any cluster, with qnever exceeding 0.50 in these cases.
When assessing the degree of purity of each cluster through the
proportion of admixed individuals for q40.8, for example, K1 was
the purest (82%), followed by K2 and K3 (70%), K4 (35%) and K5
(22%) (Supplementary Figure S3). Clearly, most individuals assigned
to K4 or K5 present a significant portion of their genome also
assigned to another of the Central African clusters K3,K4 and K5
(Supplementary Figure S4).
One can also assess the hierarchy of clustering when considering
the solutions of the clustering algorithm at lower K(Supplementary
Figures S4 and S5). For K¼2, West African samples (from Senegal to
Benin, including all M. regia but also part of M. excelsa) were
Be Be
Gh Gh
Si Si
Si Si
Si Si
Ca Ca Ca
Ca Ca
Axis 1: 64.94%
Axis 2: 28.11%
Figure 1 Principal component analysis (PCA) projection of Milicia individuals based on three leaf traits: L/w_limb (length/width ratio of limb), N_veins
(number of secondary veins) and D_veins (distance between two contiguous veins). Black triangles represent samples with Soft ¼1, a character expected in
M. excelsa, whereas red triangles are samples for which Soft ¼0, a characteristic of M. regia. Country codes: Be, Benin; Ca, Cameroon; Co, Republic of the
Congo; Ga, Gambia; Gb, Gabon; Gh, Ghana; Gu, Guinea Bissau et Guinea-Conakry; IC, Ivory Coast; Li, Liberia; Ni, Nigeria; Se, Senegal; Si, Sierra Leone.
Table 1 Comparison of leaf traits between Milicia excelsa and
M. regia specimens, identified from the presence (Soft¼1) or
absence (Soft¼0) of hairs at the lower surface of leaf
Taxa Mean value
of N_veins
Mean value
of D_veins (cm)
Mean ratio
of L/w_limb
M. regia (West Africa) 10.6 (8–15) 1.1 (0.7–1.6) 1.5 (1.0–2.4)
M. excelsa
(West and Central Africa)
15.7 (12–22) 0.6 (0.3–1.2) 1.7 (1.2–2.4)
Abbreviations: D_veins, distance (cm) between two contiguous veins; L/w_limb, length/width
ratio of limb; N_veins, number of secondary veins.
All differences are statistically significant (Student’s t-test; Po0.001).
Evolutionary history of the genus Milicia
¨nou et al
separated from the rest of the sampling (from Benin to Tanzania).
The latter group was further subdivided at K¼3andK¼4
(Supplementary Figures S4 and S5), but it was only when K¼5 that
all M. regia samples form a pure genetic cluster.
Degree of differentiation between the West African M. excelsa
cluster K2 and the three clusters from Central Africa (K3 to K5;FST
ranged from 0.121 to 0.163, RST from 0.103 to 0.201) were close or in
the same range to values found at interspecific level (FST ranged from
0.147 to 0.255, RST from 0.171 to 0.308 between M. excelsa-K1 and
any M. excelsa cluster K2 to K5), whereas differentiation was lower
among Central African clusters (FST ranged from 0.087 to 0.158, RST
from 0.027 to 0.099 between K3,K4 and K5; Table 2). Allele size
permutation tests did not detect any phylogeographical signal at nSSR
markers (Table 2).
Phylogeographic structure at pDNA and nDNA sequence variation
Whatever genome is considered, M. regia is well differentiated from
M. excelsa populations as there is no shared haplotype and, when the
trees are rooted, M. regia appears monophyletic (Figures 3a and b). At
pDNA fragments, seven haplotypes were restricted to West Africa and,
surprisingly, M. excelsa haplotypes in Central Africa belonged to a
distinct lineage to the one of West African M. excelsa. When we
excluded repetitive sequences (microsatellites) from the analysis, it
turned out that only one mutation separates the unique haplotype of
M. regia from the most frequent haplotype of M. excelsa in West
Africa (Supplementary Figure S6A). Furthermore, at psbA-trnH only,
excluding microsatellite sites led to a unique and common haplotype
within West African populations, regardless of the taxon
(Supplementary Figure S6B). A similar pattern of differentiation
was observed at nDNA sequence (M. regia haplotypes forming one
clade) except that West and Central African M. excelsa samples shared
several haplotypes (Figure 3b). Nevertheless, differentiation indices
between the West African cluster M. excelsa-K2 and the Central
African clusters M. excelsa-K3 to K5 (GST ranged from 0.479 to 0.716
at pDNA and from 0.201 to 0.319 at nDNA) was nearly as high as
between M. excelsa-K2 and M. regia-K1 (GST ¼0.685 at pDNA and
0.379 at nDNA), and much higher than between Central African
clusters (GST ranged from 0.105 to 0.393 at pDNA and from 0.033 to
0.097 at nDNA) (Table 2). There was a significant phylogeographical
signal (NST4GST) in the whole data set at both pDNA and nDNA,
resulting from the interspecies comparison (Table 2). Within M.
excelsa, a significant phylogeographic signal also occurred between the
West Af r i c a n clust e r K2 and the Central African M. excelsa clusters at
pDNA but not at nDNA, whereas no signal of phylogeographical
structure was found among Central African populations (Table 2).
When haplotypes at pDNA and nDNA, nSSR-based clusters and
morphological assignments (based only on the character Soft)are
compared, there is a strong congruence between all types of markers,
for the subsample of 172 individuals common to them all (Table 3).
All morphology-based M. regia individuals belonged to the lineage
formed by pDNA haplotypes H9-H10-H11 and to the cluster K1.
Similarly, M. excelsa in West Africa (K2 and far west of K3) was only
represented in the lineage formed by pDNA H1-H2-H7-H8. At the
cluster level, the pDNA haplotypes H1 and H8 were shared between
clusters K2 and K3. However, the three individuals in cluster K3 with
these haplotypes were located at the West of Nigeria, closer to Benin
than to Cameroon. Therefore at the regional scale, there was no
shared haplotype between West Africa and Central Africa for pDNA
sequence. In contrast, nDNA sequences displayed shared haplotypes
between West and Central Africa.
Variation of genetic diversity among regions and species
West African clusters displayed higher genetic diversity than those of
Central Africa at nSSRs, with HTvarying from 0.600 to 0.683, whereas
it was always o0.490 in Central Africa (Table 2). However, that trend
was reversed at nDNA and pDNA sequences. Similarly, haplotypic
richness looked higher in Central Africa, but the percentage of private
alleles or haplotypes was threefold higher in West African clusters
than in Central Africa, with the exception of cluster K5 for pDNA that
presented an important proportion of endemic haplotypes (Table 2).
Phylogenetic relationships and timing of divergence
According to both pDNA and nDNA analyses, M. regia is mono-
phyletic. In contrast, M. excelsa appears paraphyletic (Figure 4), at
least at pDNA where West African populations of M. excelsa (cluster
K2) form a well-supported clade (posterior probability of clade
Cluster K1
Cluster K2
Cluster K3
Cluster K4
Cluster K5
Figure 2 Spatial genetic structure of Milicia populations derived from TESS clustering algorithm for an optimal number K¼5 clusters. Black crosses indicate
morphology-based M. regia (group of) individuals, whereas the white circles stand for M. excelsa. Codes of the 17 sampled countries, from West to East:
Se, Senegal; Ga, Gambia; GBi,GuineaBissau;GCo, Guinea-Conakry; Si, Sierra Leo ne; IC, Ivory Coast; Gh, Ghana; Be,Benin;Ni, Nigeria; Ca,Cameroon;
Gb,Gabon;Co, Republic of the Congo; CK, Democratic Republic of the Congo; Bu, Burundi; Ta, Tanzania.
Evolutionary history of the genus Milicia
¨nou et al
D¼0.99; Figure 4a) more closely related to M. regia than to Central
African populations of M. excelsa (posterior probability support of
the clade B ¼0.94; Figure 4a). The nDNA phylogenetic tree does not
display a M. excelsa clade restricted to West Africa, and although
M. excelsa seems paraphyletic in the most likely tree (Figure 4b), there
is no strong support (posterior probability of clade J o0.50).
Molecular dating of the divergence of lineages in the genus Milicia
was quite congruent in the two genomes used. The MRCA of Milicia
was estimated at 31 and 32mya (Middle Oligocene) with a wide
confidence interval (12–55 mya; clades A and F; Figures 4a and b).
Age should correspond to the maximum age of speciation in the
genus Milicia. Most of the well-supported major clades within Milicia
are dated between 16 and 24 mya (Late Oligocene to Early Miocene;
clades B and E in Figure 4a, and clades G and K in Figure 4b). This
includes the nDNA well-supported M. regia clade (K) dated at
16 mya. These results should be considered with caution as the
confidence intervals were relatively wide because of the fact that we
used a single calibration point provided with an important s.d.
The present work examined the pattern of genetic differentiation in
the African timber tree genus Milicia that is threatened in several
African countries because of overexploitation and habitat degrada-
tion. Our divergence date estimates showed that Milicia has a Tertiary
origin. Despite the tenuous morphological differences between the
two species classically recognized by taxonomists, all markers con-
gruently supported the recognition of M. regia (forming one genetic
cluster at nSSRs, and monophyletic groups at nDNA and pDNA
sequences) that can be distinguished from M. excelsa by the absence of
rounded areoles containing microscopic hairs. Differentiation within
M. excelsa is more complex because (1) four genetic nSSR clusters
were detected: one in West Africa well differentiated from three more
related Central African clusters, (2) the West African pDNA clade of
M. excelsa is more related to M. regia than to Central African
M. excelsa haplotypes, whereas many nDNA haplotypes are shared
between West and Central Africa. Hence, West African populations
of M. excelsa constitute a questionable group. Hereafter, we first
discuss the implications of these results for species delimitation
according to the species concepts, and then we question the origin
of the differentiation pattern in relation to the history of African
vegetation changes.
On phylogenetic relationships and interspecific mating
Under the PSC, a species must be a monophyletic group of
organisms. Although DNA sequences show it is the case of M. regia,
this does not apply for M. excelsa. The latter could be subdivided in
two monophyletic groups according to pDNA but these groups are
not supported by nDNA sequences. Hence, under strictly a PSC view,
only one species could be recognized in Milicia, with two subspecies
or varieties. Whereas phylogenetic analyses support only one species
of Milicia, the other findings suggested a more cautious attitude.
Three findings tended to confirm an important reproductive barrier
between Milicia taxa, a key criterion for the BSC.
First, in West Africa where the two species cooccur, we detected a
morphological differentiation for leaf traits. This argument could
seem negligible as just leaf traits are used to distinguish the two
species (see the next section of the discussion for the importance
of using vegetative traits in species delimitation). We think that
these slight leaf-trait differences are not insignificant in the case of
Milicia for the following reasons. Morphological differentiation was
congruent with interspecific genetic divergence. Moreover, assuming
that M. regia leaf characters are simply signs of adaptation to
evergreen forests would lead to expect such characters in Gabonese
evergreen forests too, for example. This genus has likely a Tertiary
origin and we should reasonably observe some individuals similar
to the morphospecies M. regia in Central African evergreen forests.
Table 2 Genetic diversity and differentiation of nSSR clusters detected in Milicia for three types of markers: nSSRs, pDNA sequences and
nDNA sequences
Type of marker Region Cluster at nSSRs Taxon Differentiation parameters Diversity parameters
K1 K2 K3 K4 K5 Ninds A0RS%Rpriv HTVT
nSSRs West Africa K1 M. regia 0.195NS 0.308 NS 0.231 NS 0.171 NS 78 44 6.05 13.39 0.683 /
K2 M. excelsa 0.147 0.185 NS 0.103 NS 0.201 NS 58 43 6.11 9.17 0.600 /
Central Africa K3 M. excelsa 0.196 0.121 0.027 NS 0.099 NS 416 45 5.12 1.76 0.486 /
K4 M. excelsa 0.255 0.163 0.087 0.058 NS 149 37 4.45 0.22 0.477 /
K5 M. excelsa 0.178 0.139 0.110 0.158 102 34 4.48 0.45 0.472 /
M. excelsa samples (K2 to K5)FST ¼0.115; RST¼0.085 NS 725 54 / / 0.585 /
All samples (both M. excelsa and M. regia)FST ¼0.139; RST¼0.140 NS 803 58 / / 0.643 /
pDNA West Africa K1 M. regia 0.900 NS 0.832* 0.881 NS 0.869** 33 3 1.91 100.00 0.277 0.018
K2 M. excelsa 0.685 0.901** 0.933* 0.916** 25 3 1.56 88.46 0.157 0.010
Central Africa K3 M. excelsa 0.696 0.716 0.075 NS 0.448 NS 75 7 2.33 15.88 0.404 0.052
K4 M. excelsa 0.563 0.583 0.215 0.328 NS 31 3 2.21 10.41 0.529 0.036
K5 M. excelsa 0.459 0.479 0.393 0.105 8 4 3.75 70.93 0.750 0.058
M. excelsa samples (K2 to K5)GST ¼0.470; NST¼0.817*** 139 12 / / 0.868 0.213
All samples (both M. excelsa and M. regia)GST ¼0.540; NST¼0.838*** 172 15 / / 0.920 0.214
nDNA West Africa K1 M. regia 0.763** 0.646** 0.668** 0.667** 33 4 2.86 100.00 0.652 0.060
K2 M. excelsa 0.378 0.264 NS 0.449 NS 0.284 NS 25 7 2.86 30.77 0.593 0.090
Central Africa K3 M. excelsa 0.276 0.201 0.067 NS 0.011 NS 75 7 4.00 10.50 0.796 0.124
K4 M. excelsa 0.316 0.319 0.035 0.109 NS 31 7 3.50 20.00 0.715 0.105
K5 M. excelsa 0.291 0.237 0.033 0.097 8 6 3.94 7.36 0.767 0.103
M. excelsa samples (K2 to K5)GST ¼0.160; NST¼0.220 NS 139 12 / / 0.855 0.136
All samples (both M. excelsa and M. regia)GST ¼0.228; NST¼0.478*** 172 16 / / 0.913 0.189
Abbreviations: A0, number of alleles or haplotypes; HTand VT
, gene diversity based on unordered and ordered alleles; nDNA, nuclear DNA; NS, not significant; nSSR, nuclear microsatellite; pDNA,
plastid DNA; %Rpriv, percentage of private haplotypes or alleles; RS, haplotypic or allelic richness.
In matrices of pairwise differentiation, GST (FST for nSSRs) and NST (RST for nSSRs) are given below and above the diagonals, respectively. Tests of phylogeographic signal (NST4NSTperm or
RST4RSTperm for nSSRs) are indicated by NS (P40.05), *Po0.05, **Po0.001 and ***Po0.001.
Evolutionary history of the genus Milicia
¨nou et al
That was not the case: Central African individuals in our sample,
regardless of the vegetation type, had the same M. excelsa leaf traits as
individuals from drier forest zones of West Africa (Dahomey gap).
A second argument that rejects the assumption of only one species
in Milicia comes from the existence of mutual allelic exclusivity
within this genus, in the sense of Doyle (1995) and Flot et al. (2010).
This concept states that speciation can be assumed in paraphyletic
groups if they have no common allele (mutual exclusivity). Mutual
exclusivity is reached in diverging groups generally before reciprocal
allelic monophyly, and therefore it should be more suitable than
monophyly for species delineation. Under that concept, we should
accept two species in Milicia considering the cluster K1 with respect to
the others (Figure 5). Rosenberg (2003) showed that the time for a
species (for example, M. excelsa) to become reciprocally monophyletic
Figure 3 Geographical distribution of psbA-trnH /trnC-ycf6 (a)andAt103 (b) haplotypes in Milicia and median joining networks. Repetitive sequences were
included in these analyses.
Evolutionary history of the genus Milicia
¨nou et al
(99% of its loci) is B5.3Ne generations (Ne being the population
size), assuming that its sister species is already monophyletic at all loci
(for example, M. regia). If we assume that NeE100 000 individuals
(a minimum figure considering the distribution range and population
densities; see, for example, Nichols et al.,1998;Fe
´et al.,2004)and
agenerationtimeof100yearsinM. excelsa, then it will take B53
millions of years before monophyly would be expected at almost all
loci. That estimate is much higher than the mean age of maximum
divergence time in Milicia, so that the incomplete lineage sorting may
be justified. Rejection of reciprocal monophyly for species identifica-
tion is in line with the unified species concept of De Queiroz (2007)
that defines hypothetical species as separately evolving metapopula-
tion lineages, a lineage being a branch and several lineages forming a
clade or a monophyletic group. From this point of view, any of the
properties derived from other species concepts can be used as a line of
evidence to confirm existence of different species, although it is
necessary to have several lines of evidence to corroborate the
hypothesis of distinct species.
Third, interspecific admixture seemed very scarce. Only four
individuals presented jointly leaf characters of both species. In
addition, using Bayesian approaches for identifying hybrid individuals
as implemented in STRUCTURE (Pritchard et al., 2000), TESS (Chen
et al.,2007)andN
EWHYBRIDS (Anderson and Thompson, 2002), we
found only three to six putative hybrids among 46 individuals
sampled in the interspecific contact zone, from Guinea-Conakry to
Ghana (K Daı
¨nou, unpublished results). However, this low hybrid
proportion (6.5–13%) may be biased because the best hybrid zone for
Milicia species should be located in the semi-deciduous forest region
of Ivory Coast and Ghana that was not sampled enough for the
present study. Clearly, additional investigations are required to assess
contemporaneous hybridization patterns in Milicia populations.
On the importance of morphological characters for species
Despite the limited number of distinctive characters in Milicia,itis
worth remembering that scarcity of morphological differences does
not mean absence of speciation. Of the animal studies, 23% revealed
paraphyly or polyphyly, suggesting a much common phenomenon
than generally assumed (Funk and Omland, 2003). Investigating
monophyly and species boundaries in higher plants is less common
because of a more limited number of related studies and fewer
attempts to combine morphological and genetic characterizations.
There are very few morphological characters that delimit M. excelsa
from M. regia. Past assessment by taxonomists of the existence of two
species on the basis of a few vegetative characters may look surprising.
The evolution of vegetative characters is often thought to be related to
adaptation to the growth environment, so that these characters are
viewed as less reliable than floral traits for systematic studies
(Ingrouille and Chase, 2004; Das et al., 2007). However, floral traits
may be less commonly used than assumed. Grant (1949) observed
among 416 genera that although differences in floral traits were
employed for classifying 37–40% of bird-, bee- and fly-pollinated
plants, this proportion drops up to 4% for wind- and water-
pollinated taxa. Plant species with generalist pollination systems are
less subject to strong directional selection of floral traits (Johnson and
Steiner, 2000), which may explain the absence of floral differentiation
in Milicia species. In addition, Berg’s ‘correlation pleiades hypothesis
has received notable support from various studies (see, for example,
Armbruster et al.,1999;Menget al., 2008). This hypothesis predicted
a stronger relationship between floral and vegetative traits in wind-
pollinated and generalized flowers than in specialized flowers, because
selective pressures affect jointly both categories of characters in the
former (covariation is expected as flower components derived from
leaves) whereas they are decoupled in the latter (in order to preserve
the flower-pollinator fit, regardless of the environment characteris-
tics). Thus, referring exclusively to vegetative traits for species
delimitation in Milicia, a wind-pollinated genus, is quite justified.
Phylogeographical structuring: potential paleoenvironment
Diversification in gene genealogy can occur in a stable population,
independently of environmental perturbations. However, time of
coalescence is readily affected by fluctuating population sizes, natural
selection and immigration (Freeland et al.,2011).
Zerega et al. (2005) tracked back the biogeographical history and
migration routes of most of genera known for Moraceae. They
showed that Moraceae may have originated in Eurasia during the
Mid-Cretaceous. The two collisions of the landmasses formed by
Africa and India with Eurasia at B60 and B45 mya have created
colonization routes for at least four tribes, including Moreae. That
period coincided with the first main diversification phase of West
African flora, from the Palaeocene to the Mid-Eocene (65–46 mya;
Jacobs, 2004; Plana, 2004). All the northern part of the continent
including the Sahara was covered by a mosaic of savannahs and
forests. This period may be likely the timing of arrival of Milicia
ancestor on the continent.
It is interesting to note that the mean ages of the two first
divergence events in Milicia took place at the two first major
environmental disturbances reported for the continent, after
Table 3 Concordance between the three types of genetic markers
used and morphological identity in a subsample of 172 individuals of
Milicia excelsa and M. regia
at nSSRs
nDNA haplotype s
M. excelsa M. regia
K1 3 H10 H11—H12—H13—H14
28 H9
K2 23 H1 H3—H4—H5—H6—H9
K3 2 H1 H3—H4—H5—H6—H7—H8—H9
57 H3
11 H4
K4 1 H13 H1—H3—H4—H5—H9—
18 H3
12 H4
K5 1 H14 H3—H4—H5—H6—H7—H9
Abbreviations: nDNA, nuclear DNA; nSSR, nuclear microsatellite; pDNA, plastid DNA.
Evolutionary history of the genus Milicia
¨nou et al
establishment of West African flora at the Mid-Eocene. The start of
speciation in Milicia was estimated atB31–32 mya (mean ages of
nodes A and F; Figure 4) that coincided with a sharp cooling period
at the Late Eocene to the Early Oligocene (35–31mya), leading to a
major extinction in the hygrophilous African flora (Maley, 1996;
Plana, 2004). The second major environmental perturbation occurred
from the Early to the Mid-Miocene (23–15 mya): humid vegetation
disappeared in the Sahara, the continent moved northwards, down
positioning the Equator. The rainforest belt shifted southwards, and
the Tethys Sea was closed (Maley, 1996; Jacobs, 2004). The highly
supported nodes E, B, G and K in Figure 4 had mean ages included in
that range. Obviously, these correlations are just hypotheses as our
estimates of divergence times were not accurate enough.
At a regional scale, three clusters were detected in Central Africa.
Unlike West Africa, there was no phylogeographical structure at any
type of marker in Central Africa. In addition, the degree of
differentiation is far lower within Central African clusters compara-
tively to values noted in West Africa or between West African clusters
and those of Central Africa. Dainou et al. (2010) proposed that
genetic clusters within Lower Guinea diverged more than 100 000
years ago, possibly during the ice age period lying from 160 000 to
130 000 years ago (Maley, 1996). The spatial organization of
M. excelsa clusters in central Africa may result from the past forest
fragmentation, possibly at ice ages. Clusters K3 and K4 centered on
Cameroon and Gabon, respectively, may be expansions of the Atlantic
coastal refuge zones that stretched from the southwestern part of
Figure 4 Phylogenetic trees in Milicia based on (a)pDNA(psbA-trnH and trnC-ycf6 intergenic regions) and (b)nDNA(At103 intron) regions. The tree root
height was constrained with the age of the MRCA of Milicia and Morus: mean of 56 mya and s.d. of 9mya (in millions of years ago (mya)) (Zerega et al.,
2005). Numbers at the right part of nodes are posterior probabilities (p1). The scale bar at the bottom of each figure indicates time period. The estimated
mean ages of selected clades are provided in circles along with their 95% highest posterior density (HPD) interval in brackets (empty brackets: no available
HPD information as posterior probability was p0.50). Letters A to K are names used in the text for selected nodes.
Evolutionary history of the genus Milicia
¨nou et al
Nigeria to the South of Democratic Republic of the Congo. The
North–South divide nearby the Equator between these two clusters
has been documented for other forest tree species (Duminil et al.,
2010; Debout et al., 2011) and a significant floristic shift has recently
been demonstrated at that latitude (Gonmadje, 2012). Cluster K5
centered on the Democratic Republic of the Congo may represent
populations spreading from inland forest refuges surrounding the
Congo River. The larger extent of K3 and K5 comparatively to the
Gabonese cluster K4 may be related to the fact that they are located
on more suitable zones for M. excelsa population dynamics (semi-
deciduous forests).
In West Africa, making such link between genetic clusters and
refuges is possible if we refer to the map in Anhuf et al. (2006) rather
than the one proposed by Maley (1996). These authors diverged
substantially in mapping the closest forest refuge of the East of the
Dahomey gap. According to Maley (1996), a small, well-delimited
refuge was established at the southern frontier of Ivory Coast and
Ghana, whereas that refuge should have extended up to the extreme
eastern boundaries of Ghana, thus partially located in the Dahomey
gap. Under the latter scenario, cluster K2 could have extensively been
developed eastwards from that refuge. However, phylogeographical
signals within West African clusters denoted older isolations than
those observed in Central Africa. Then, we should assume that the
same places have acted as forest refuges since at least the Late
Oligocene-Miocene after tropical forests were established on the
continent (Jacobs, 2004; Plana, 2004). There is no proof of such an
assumption for that period in West Africa, but elsewhere in East
Africa the role of upland areas as refuges many millions years ago was
demonstrated (Yemane et al., 1987).
As Milicia populations in West Africa experienced severe exploita-
tion pressures during past centuries, the lower degree of haplotypic
richness in this region may be caused by rapid decrease of population
density. The higher degree of allelic/haplotypic endemism in West
Africa argues for the hypothesis assuming that West African flora may
be substantially older than Central African vegetation (suggested in
Vande Weghe (2003) and Jacobs (2004)). In the only two other
genetic studies of rainforest tree species (Irvingia gabonensis and
Terminalia superba) that have included both West and Central African
Starting point. One set of allele
frequencies from any representative
sampleOne gene cluster.
Phase 1. Differentiation of allele
frequencies Different gene clusters.
K3vs. K4vs. K5
Phase 2. Extinction of
haplotypes + mutations
Polyphyly or monophyly
depending on the sequence
K2 vs.K3, K4 and K5
Phase 3. Extinction of
lineages + mutations
One monophyletic
group with possibly
mutual allelic exclusivity.
K1 vs. K2, K3, K4, K5
Phase 4. Extinction of
lineages +
Reciprocal monophyly
(no such case for
Figure 5 A schematic representation of lineage divergence and speciation that reflects the case of Milicia. The circles with cross inside represent a lineage,
whereas the empty circles stand for another one. Different gray colors indicate different haplotypes within each lineage and circles are sized according to
their relative frequency. Each Milicia cluster (K1 to K5) is mentioned nearby the divergence phase it reaches with respect to the others. From the starting
point to phase 1, the two diverging groups inherit the polymorphism of their ancestor and are distinguishable just by allele frequencies. FST¼0.1 is
obtained after BNe/5 in the absence of homoplasy, Ne being the effective population size. If effective sizes are enough large, mutation will exert more
influence than genetic drift on the time to reach phase 2. At this step, some sequences can reveal a monophyletic group, whereas others may show
polyphyly. That is the case of the cluster K2 with respect to K3,K4 and K5. Much later, complete extinctions of lineages may occur leading, for example,
to one clear monophyletic group and possibly mutual allelic exclusivity (phase 3): this is observed in K1 with respect to the other genetic clusters.
Reciprocal monophyly is reached at a later phase (phase 4), after B5Ne generations (Rosenberg, 2003): no pair of Milicia clusters displays such a
Evolutionary history of the genus Milicia
¨nou et al
populations, diversity measures also showed that populations from
the western African zones exhibit higher genetic diversity than those
of the Lower Guinea and the Congolia (Vigneron, 1984; Lowe et al.,
2010). A phylogeographical study of Khaya senegalensis,adry-forest
tree species, also revealed that West African groups display higher
levels of genetic diversity than those from Sudan and Uganda (Karan
et al.,2012).
The present study investigated the pattern of genetic differentiation in
Milicia, based on nuclear microsatellites, plastid and nuclear
sequences. Five genetic clusters were identified and their origin could
be linked to past environmental disturbances. Milicia is a Tertiary
genus with two cryptic species exhibiting incomplete lineage sorting.
Existence of paraphyly in Milicia can be explained by the time
required for isolated species to become reciprocally monophyletic,
and this depends on effective population size and generation time that
are likely very high in Milicia species. Mutual allelic exclusivity was
observed and this tends to confirm the recognition of two species in
that genus, despite very scarce morphological differences. Interspecific
hybridization looks scarce at present, but this pattern requires further
The trnC-ycf6 and psbA-trnH plastid DNA sequences have been
deposited in GenBank under the accession numbers KF719985–
KF720330. The At103 nuclear DNA sequences are available in
GenBank under the accession numbers KJ129036–KJ129052. Nuclear
microsatellites data have been deposited in DRYAD at http://
The authors declare no conflict of interest.
We thank the project PPR 10.000 (Gembloux Agro-Bio Tech; University of
Liege), the Fonds Le
´opold III (asbl, Belgium), and the Fund for Scientific
Research of Belgium (FNRS) through grants FRFC no. 2.4.577.10 and FRFC
no. 2.4.576.07 for their financial support. We are indebted to David Monticelli,
Lambert Y Kouadio, Guillaume Koffi, Bonaventure Sonke
´, Michel Baudoin,
Shango Mutambue, Jean-Pierre Mate, Emilien Dubiez, Pierre Proce
`s, Bertin
Kasongo, Pat Stoffelen and Georges Mumbere for their various contributions.
We also acknowledge the logging companies Pallisco, SFID, Wijma
(Cameroon) and Precious Woods Gabon for facilitating field work, and the
National Botanical Garden of Belgium (Meise) for permitting sample
collection from herbaria.
Anderson EC, Thompson EA (2002). A model-based method for identifying species hybrids
using multilocus genetic data. Genetics 160: 1 217–1229.
Anhuf D, Ledru MP, Behling H, Da Cruz FW, Cordeiro RC, Van der Hammen Tet al. (2006).
Paleo-environmental change in Amazonian and African rainforest during the LGM.
Palaeogeogr Palaeocl 239: 510–527.
Armbruster WS, Di Stilio VS, Tuxill JD, Flores TC, Runk JLV (1999). Covariance and
decoupling of floral and vegetative traits in nine neotropical plants: a re-evaluation of
Berg’s correlation-pleiades concepts. Am J Bot 86:3955.
ATIBT (2010). Statistiques 2009. La Lettre de l’ATIBT 32:425.
´ville A (1959). La flore forestie
`re de la Co
ˆte d’Ivoire. Deuxie
`me e
´dition re
Tom e 1. CTFT: Nogent-sur-Marne: Seine (France).
Bandelt H-J, Forster P, Ro
¨hl A (1999). Median-joining networks for inferring intraspecific
phylogenies. MolBiolEvol16:3748.
Berg CC (1977). Revision of African Moraceae (excl. Dorstenia, Ficus, Musanga,
Myrianthus). Bull Jard Bot Natl Belg 47: 267–407.
Bickford D, Lohman DJ, Sodhi NS, Ng PK, Meier R, Winker K et al. (2006). Cryptic
species as a window on diversity and conservation. Trends Ecol Evol 22: 148–155.
Bizoux JP, Dainou K, Bourland N, Hardy OJ, Heuertz M, Mahy G et al. (2009). Spatial
genetic structu re in Milicia excelsa (Moraceae) indicates extensive gene dispersal in a
low-density wind-pollinated tropical tree. Mol Ecol 18: 4398–440 8.
Chen C, Durand E, Forbes F, Franc¸ois O (2007). Bayesian clustering algorithms
ascertaining spatial population structure: a new computer program and a comparison
study. Mol Ecol Notes 7:747756.
Chevalier A (1917). Le s ve
´taux utiles d’Afrique tropicale franc¸aise—La fore
ˆt et les bois
du Gabon. Challamel: Paris, France.
Dainou K, Bizoux JP, Doucet JL, Mahy G, Hardy OJ, Heuertz M (2010). Forest refugia
revisited: nSSRs and cpDNA sequences support historical isolation in a wide-spread
African tree with high colonization capacity, Milicia excelsa (Moraceae). Mol Ecol 19:
Dainou K, Doucet JL, Sinsin B, Mahy G (2012a). Identite
´et e
´cologie des espe
`res commerciales d’Afrique centrale: le cas de Milicia spp. Biotechnol Agron
Soc 16: 229–241.
Dainou K, Laurenty E, Mahy G, Hardy OJ, Brostaux Y, Tagg N et al. (2012b). Phenological
patterns in a natural population of a tropical timber tree species, Milicia excelsa
(Moraceae): evi dence of isolation by time and i ts interaction with fee ding strategies of
dispersers. Am J Bot 99: 1453–1463.
Das M, Bhattacharya S, Basak J, Pal A (2007). Phylogenetic relationships among the
bamboo species as revealed by morphological characters and polymorphism analyses.
Biol Plantarum 51: 667–672.
Datwyler SL, Weiblen GD (2004). On the origin of the fig: phylogenetic relationships of
Moraceae from ndhF sequences. Am J Bot 91: 767–777.
Davies TJ, Barraclough TG, Chase MW, Soltis PS, Soltis DE, Savolainen V (2004). Darwin’s
abominable mystery: insights from a supertree of the angiosperms. Proc Natl Acad Sci
USA 101: 1904–1909.
De Queiroz K (2007). Species concepts and species delimitation. Syst Biol 56:879886.
Debout GD, Doucet JL, Hardy OJ (2011). Population history and gene dispersal inferred
from spatial genetic structure of a Central African timber tree, Distemonanthus
benthamianus (Caesalpinioideae). Heredity 106: 88–99.
Doyle JJ (1995). The irrelevance of allele tree topologies for species delimitation, and a
non-topological alternative. Syst Bot 20:574588.
Drummond AJ, Ho SY, Phillips MJ, Rambaut A (2006). Relaxed phylogenetics and dating
with confidence. PLoS Biol 4: 699–7 10.
Drummond AJ, Rambaut A (2007). BEAST: Bayesian evolutionary analysis by sampling
trees. BMC Evol Biol 7:214.
Duminil J, Heuertz M, Doucet JL, Bourland N, Cruaud C, Gavory F et al. (2010). CpDNA-
based species identification and phylogeography: application to African tropical tree
species. Mol Ecol 19: 5469–5483.
Durrieu de Madron L (2003). Accroissement diame
´trique du be
´et de l’iroko. Bois For
Tro p 275:8387.
Ebert SJ (2004). Silvicultural potential of Milicia excelsa–Working paper No. 18.I-TOO:
Freiburg, Germany.
Evanno G, Regnaut S, Goudet J (2005). Detecting the number of clusters of individuals
using the software STRUCTURE: a simulation study. Mol Ecol 14: 2611–2620.
´R, Nkolong E, Hubert D (2004). Plan d’ame
´nagement des unite
´s forestie
´nagement 10.041, 10.042 et 10.044 regroupe
´es. Pallisco: Douala, Cameroon.
Flot JF, Couloux A, Tillier S (2010). Haplowebs as a graphical tool for delimiting species: a
revival of Doyle’s ‘field for recombination’ approach and its application to the coral
genus Pocillopora in Clipperton. BMC Evol Biol 10:372.
Freeland JR, Kirk H, Petersen SD (2011). Mole cular Ecology. John Wiley & Sons, L td:
Oxford, UK.
Franc¸ois O, Durand E (2010). Spatially explicit bayesian clustering models in population
genetics. Mol Ecol Resour 10: 775–7 84.
Funk DJ, Omland KE (2003). Species-level paraphyly and polyphyly: frequency, causes,
and consequences, with Insights from Animal Mitochondrial DNA. Annu Rev Ecol Evol
S34: 397–423.
Gonmadje CF (2012). Diversite
´et bioge
´ographie des fore
ˆts d’Afrique centrale atlantique:
le cas du massif de Ngovayang (Cameroun). PhD thesis, Universite
´Montpellier 2:
Montpellier, France.
Goudet J (1995). FSTAT (Version 1.2): a computer program to calculate F-statistics.
Grant V (1949). Pollination systems as isolating mechanisms in Angiosperms. Evolution 3:
Hardy OJ, Charbonnel N, Fre
´ville H, Heuertz M (2003). Microsatellite allele sizes: a simple
test to assess their significance on genetic differentiation. Genetics 163: 1467–1482.
Hardy OJ, Vekemans X (2002). Spagedi: a versatile computer program to analyse spatial
genetic structure at the individual or population levels. Mol Ecol Notes 2: 618–620.
Hawthorne WD, Jongkind C (2006). Woody Plants of Western African Forests: A Guide to
the Forest Trees, Shrubs and Lianes from Senegal to Ghana. Royal Bota nic Gardens:
Kew, UK.
Hey J (2006). On the failure of modern species concepts. Trends Ecol Evol 21:447450.
Ingrouille M, Chase MW (2004). Becoming fruitful and diversifying: DNA sequence
phylogenetics and reproductive physiology of land plants. In: Hemsley A, Poole I (eds)
Evolution of Plant Physiology: From Whole Plants to Ecosystems.LondonElsevier
Academic Press: London, UK, pp 327–342.
Jacobs BF (2004). Palaeobotanical studies from tropical Africa: relevance to the evolution
of forest, woodland and savannah biomes. Philos Trans R Soc B-Biol Sci 359:
Jakobsson M, Rosenberg NA (2007). Clumpp: a cluster matching and permutation
program for dealing with label switching and multimodality in analysis of population
structure. Bioinformatics 23: 1801–1806.
Evolutionary history of the genus Milicia
¨nou et al
Johnson SD, Steiner KE (2000). Generalization versus specialization in plant pollination
systems. Tree 15: 140–1 43.
Kalinowski ST (2005). Hp-Rare 1.0: a computer program for performing rarefaction on
measures of allelic richness. Mol Ecol Notes 5: 1 87–189.
Karan M, Evans DS, Reilly D, Schulte K, Wright C, Innes D et al. (2012). Rapid
microsatellite marker development for African mahogany (Khaya senegalensis,Melia-
ceae) using next-generation sequencing and assessment of its intra-specific genetic
diversity. Mol Ecol Resour 12:344353.
Koffi GK (2010). Etude de la variabilite
´tique et de la phyloge
´ographie de Santiria
trimera (Burseraceae)-implications pour une conservation durable des fore
ˆts humides
d’AfriquePhD thesis, Universite
´Libre de Bruxelles: Brussels, Belgium.
Li M, Wunder J, Bissoli G, Scarponia E, Gazzani S, Barbaro E et al. (2008). Development
of COS genes as universally amplifiable markers for phylogenetic reconstructions of
closely related plant species. Cladistics 24: 727–745.
Lowe AJ, Harris D, Dormo ntt E, Dawson IK (2010). Testing putative Afr ican tropical forest
refugia using chloroplast and nuclear DNA phylogeography. Trop P la nt B io l 3:5058.
Maley J (1996). T he African rain for est - main characteris tics of changes in ve getation and
climate from the Upper Cretaceous to the Quaternary. In: Alexander IJ, Swaine MD,
Watling R (eds) Essays on the Ecology of the Guinea-Congo Rain Forest. The Royal
Society of Edinburgh: Edinburgh. Vol. 104B, pp 31–73.
Mallet J (2010). Why was Darwin’s view of species rejected by twentieth century
biologists? Biol Philos 25: 497–527.
Mayr E (1942). Systematics and the Origin of Species from the Viewpoint of a Zoologist.
Columbia University Press: New York.
Meng J-L, Zhou X-H, Zhao Z-G, Du G-Z (2008). Covariance of floral and vegetative traits in
four species of Ranunculaceae: a comparison between specialized and generalized
pollination systems. J Integr Plant Biol 50: 1161–1170.
Nepal M (2012). Phylogenetics of Morus (Moraceae) inferred from ITS and trnL-trnF
sequence data. Syst Bot 37: 442–4 50.
Nichols JD, Agurgo FB, Agyeman VK, Wagner MR, Cobbinah JR (1998). Distribution and
abundance of Milicia species in Ghana. Ghana J Forest 6:17.
Nosil P, Funk DJ, Ortiz-Barrientos D (2009). Divergent selection and heterogeneous
genomic divergence. Mol Eco l 18: 375–402.
Nyong’o RN, Cobbinah JR, Appiah-Kwarteng J (1994). Flowering and fruiting patterns in
Milicia excelsa and Milicia r egia Welw. Ghana J Forest 1:1929.
Odee DW, Telford A, Wilson J, Gaye A, Cavers S (2012). Plio-Pleistocene history and
phylogeography of Acacia senegal in dry woodlands and savannahs of sub-Saharan
tropical Africa: evidence of early colonisation and recent range expansion. Heredity
109: 372–382.
Ofori DA, Cobbinah JR (2007). Integrated approach for conservation and management of
genetic resources of Milicia species in West Africa. For Ecol Manage 238:16.
Ofori DA, Swaine M, Leifert C, Cobbinah JR, Price AH (2001). Population genetic
structure of Milicia excelsa characterized by using RAPD and nucleotide sequencing L.
Genet Resour Crop Evol 48: 637–647.
Ofori DA, Swaine MD, Cobbinah JR, Price AH (2003). Genetic diversity and bio-
diversity conservation guidelines for Milicia species in Ghana. Ghana J Forest 11:
Ouinsavi C, Sokpon N, Bousquet J, Newton CH, Khasa DP (2006). Novel microsatellite
DNA markers for the threatened African endemic tree species, Milicia excelsa
(Moraceae), and cross-species amplification in Milicia regia.Mol Ecol Notes 6:
Plana V (2004). Mechanisms and tempo of evolution in the African Guineo-Congolian
rainforest. Philos Trans R Soc B Biol Sci 359: 1585–1594.
Pons O, Petit RJ (1996). Measuring and testing genetic differenciation with ordered versus
unordered alleles. Genetics 144: 1237–1245.
Posada D (2008). jModelTest: phylogenetic model averaging. Mol Biol Evol 25:
Pritchard JK, Stephens M, Donnelly PJ (2000). Inference of population structure using
multilocus genotype data. Genetics 155: 945–959.
Rosenberg NA (2003). The shapes of neutral gene genealogies in two species:
probabilities of monophyly, paraphyly and polyphyly in a coalescent model. Evolution
57: 1465–1477.
Schnell R (1946). Sur quelques plantes a
`usage religieux de la re
´gion forestie
`re d’Afrique
occidentale. JSocie
´Africanistes 16: 29–38.
Scotti-Saintagne C, Dick CW, Caron H, Vendramin GG, Guichoux E, Buonanici A et al.
(2012). Phylogeography of a complex of lowland Neotropical rain forest trees (Carapa,
Meliaceae). JBiogeogr40: 676–692.
StatSoft (2004). Statistica. StatSoft France: Maisons-Alfort, France.
Vande Weghe JP (2003). Les fore
ˆts d’Afrique Centrale: la Nature et l’Homme. Latoo: Tielt.
Vigneron P (1984). Polymorphisme enzymatique et variabilite
´tique des provenances
ivoiriennes et congolaises de Terminalia superba Engl. et Diels. Bois For Trop 204:
Waples RS, Gaggiotti O (2006). What is a population? An empirical evaluation of some
genetic methods for identifying the number of gene pools and their degree of
connectivity. Mol Ecol 15: 1419–1439.
White MG (1966). A comparison of Chlorophora excelsa (Welw.) Benth and Hook (F.) and
C. regia A. Chev. (Fam. Moraceae). Commonw Forest Rev 45: 150 –153.
Yemane K, Taieb M, Faure H (1987). Limnogeologic studies on an intertrappean
continental deposit from the northern Ethiopian Plateau (371030E, 121250N). JAfr
Earth Sci 6: 91–101.
Zerega NJ, Clement WL, Datwyler SL, Weiblen GD (2005). Biogeography and divergence
times in the mulberry family (Moraceae). Mol Phylogen Evol 37:402416.
Supplementary Information accompanies this paper on Heredity website (
Evolutionary history of the genus Milicia
¨nou et al
... The clearest evidence of cryptic species occurs when molecular markers reveal well-differentiated gene pools distributed in sympatry and without admixed individuals, since genetic differentiation cannot result from spatial isolation but from intrinsic reproductive barriers in this context (e.g., Ikabanga et al. 2017;Lissambou et al. 2018). However, assessing when gene pools correspond to distinct species or to differentiated conspecific populations can be difficult and somewhat arbitrary depending on the criterion used to define species (De Queiroz 2007;Daïnou et al. 2014). This is particularly the case when putative species are distributed in allopatry because we cannot observe in situ whether they would interbreed freely or not if they had the opportunity to mate. ...
... The presence of diagnostic morphological traits matching discontinuities of molecular variation can then be used as a criterion to delimit species (e.g., Ikabanga et al. 2017;Lissambou et al. 2018). It must also be kept in mind that speciation is a long-term process that can be made more complex by hybridization between species (e.g., Daïnou et al. 2014Daïnou et al. , 2016Daïnou et al. , 2017. Molecular data in combination with morphological characters have been successfully used to resolve taxonomic relations within different tropical plant taxa (Dauby et al. 2010;Donkpegan et al. 2017;Ikabanga et al. 2017;Tosso et al. 2018; Monthe et al. 2018). ...
... Molecular data in combination with morphological characters have been successfully used to resolve taxonomic relations within different tropical plant taxa (Dauby et al. 2010;Donkpegan et al. 2017;Ikabanga et al. 2017;Tosso et al. 2018; Monthe et al. 2018). Although cryptic species have been rarely highlighted in plant taxa, many recent studies had pointed at hidden genetic diversity within African tree species Daïnou et al. 2014;Ewédjè et al. 2020). ...
Full-text available
In the pantropical mimosoid legume genus Parkia, taxonomic classification has remained controversial in Africa due to clinal phenotypic variations. Three species (P. biglobosa, P. bicolor, and P. filicoidea) are currently recognized, ranging from West to East Africa, with partially overlapping ranges across different floristic regions. However, additional taxa (species or varieties) have been suggested by different authors. To assess species boundaries of African Parkia and phylogeographic patterns within species, we genotyped 889 individuals using 10 microsatellite markers and compared our results with existing morphological descriptions. Bayesian genetic clustering confirmed the species boundaries assessed from morphological traits and did not reveal introgression but identified genetic discontinuities within each species. Six moderately differentiated genetic clusters were recovered in P. biglobosa (pairwise FST: 0.05–0.19), while P. bicolor and P. filicoidea, each, displayed four well-differentiated clusters (FST: 0.18–0.41 and 0.11–0.34, respectively). Within each species, genetic clusters occurred in parapatry. Parkia biglobosa clusters were congruent with the longitudinal clinal variation in leaflets sizes but 26% of individuals presented admixed genotypes. Genetic clusters in P. bicolor and P. filicoidea followed environmental gradients as well as phytogeographic subdivisions. They were also largely congruent with morphological discontinuities described in previous taxonomic studies and < 10% individuals showed admixed genotypes. We conclude that only one species (P. biglobosa) should be considered in the Guineo-Sudanian savanna. However, P. bicolor and P. filicoidea might each represent a complex of (sub)species. Thus, the hypotheses of cryptic species within P. bicolor and P. filicoidea should be further investigated by testing reproductive barriers.
... Recent phylogeographical studies of African rain forest plant species support the hypothesis that the forest cover fragmented under past climate changes, although additional investigations are needed to better understand when and how this happened (e.g. Lowe et al. 2000;Daïnou et al. 2014;Heuertz et al. 2014;Blatrix et al. 2017;Piñeiro et al. 2017). In particular, debates are still ongoing about the location of persistent forest fragments during glacial periods, especially in the Guineo-Congolian region where the Last Glacial Maximum is thought to have been more influential than in the other tropical regions (Anhuf et al. 2006). ...
... The first major divide separating northern and southern populations of Lower Guinea is congruent with recent findings from other Central African tree species (e.g. Daïnou et al. 2010;Dauby et al. 2010;Duminil et al. 2010;Debout et al. 2011;Daïnou et al. 2014;Blatrix et al. 2017;Piñeiro et al. 2017) although the limit between northern and southern genetic clusters can range from 0.5°and 2.5°N orth according to the species (Hardy et al. 2013). This outcome suggests that surviving populations of B. toxisperma have been isolated long enough in each zone during the Pleistocene, and the current distribution results from their expansion since the end of the LGM. ...
Full-text available
How the Central African rain forests have been affected by climatic fluctuations of the Quaternary remains debated. Phylogeographical studies have shown that tree species from western Central Africa often display spatially congruent genetic discontinuities, supporting the hypothesis that the forest was previously fragmented. Extensive seed dispersal is expected to accelerate the admixture between gene pools but most of the species studied so far have presumably limited seed dispersal abilities. Here, we genotyped 15 nuclear and three plastid microsatellite markers in a low-density Central African tree species with long-distance seed dispersal: Baillonella toxisperma (Sapotaceae). While plastid markers revealed a weak structure in Cameroon, nuclear markers highlighted three genetic clusters: two distributed in Cameroon and separating Atlantic coastal forests from the inland forests, and one cluster occurring in Gabon. Substantial genetic differentiation with a phylogeographical signal was detected only between Cameroonian and Gabonese populations, suggesting two major genetic clusters located approximately North and South of the equatorial climatic hinge. Genetic differentiation was very low between the clusters within Cameroon. This pattern could be partially explained by the climate niche distribution modelling applied on the Last Glacial Maximum (LGM) which predicts a unique remnant population per country. The deep North-South differentiation in a species with long-distance seed dispersal supports the hypothesis that Central African rain forests have been fragmented at the height of the equator during a substantial part of the Quaternary.
... However, previous preliminary genetic studies suggested that this genus might contain several undescribed cryptic species (Dauby & al., 2010;Piñeiro & al., 2016Piñeiro & al., , 2017. We used, for the first time in this genus, multivariate analyses of morphological traits, which have been applied successfully to guide species delimitation in other tropical taxa (e.g., Daïnou & al., 2014Daïnou & al., , 2016Ikabanga & al., 2017). ...
... C.C.Berg from M. regia (A.Chev.) C.C.Berg (Daïnou & al., 2014). Unfortunately, no floral traits could be examined for these groups and fruits are rather similar. ...
Full-text available
Combining genetic and morphological markers is a powerful approach for species delimitation, much needed in tropical species complexes. Greenwayodendron (Annonaceae) is a widespread genus of trees distributed from West to East African rainforests. Two species and four infra‐specific taxa are currently recognized. However, preliminary genetic studies and morphological observations suggested the occurrence of additional species, undescribed to date. We tested species delimitation within Greenwayodendron by combining morphological and population genetics data. First, a visual inspection of about a thousand specimens suggested the existence of seven morphogroups: four of them occur in Central Africa and overlap in Gabon while three others are allopatric, occurring respectively in West Africa, East Africa, and the islands of São Tomé and Príncipe. Their morphological differentiation was confirmed by analysis of 27 morphological characters coded from 233 herbarium specimens. Second, after genotyping 800 samples at eight nuclear microsatellites, Bayesian clustering analyses (STRUCTURE) identified four genetic clusters corresponding to the well‐sampled morphogroups but failed to separate the three remaining morphogroups represented by few samples. However, we show that this is an inherent limit of the STRUCTURE algorithm, whereas factorial correspondence analysis (FCA) and pairwise FST and RST measures confirmed the genetic differentiation of all morphogroups. We considered that a clear genetic differentiation occurring between sympatric populations advocates for recognizing distinct species following the biological species concept. Our analyses highlight that the current taxonomic treatment of Greenwayodendron underestimates the total number of species. We identified two new species and support separation at the rank of species of two varieties (G. suaveolens subsp. suaveolens var. gabonica, G. suaveolens subsp. suaveolens var. suaveolens) and one subspecies (G. suaveolens subsp. usambaricum). The taxonomic status of specimens collected in São Tomé and Príncipe remains inconclusive, partly due to the limited fertile material available. Our study highlights the strength of combining morphological and population genetics data for discovering new taxa. Guidelines for using genetic clustering approaches in species delimitation are provided.
... Closely related species which belong to the same genus can thus commonly share ancestral polymorphisms. As a consequence, phylogenetic resolution is often poor in tree clades (Willyard et al. 2009;Grivet et al. 2013;Daïnou et al. 2014). ...
... Species non-monophyly can only be investigated by including multiple accessions per species in phylogenetic trees. Examples of such densely sampled phylogenies revealed discordance between morphological and genetic species delimitation for animals (Wiens and Penkrot 2002) and plants (Daïnou et al. 2014, Prata et al. 2018, thus reviving the discussion about whether species should be monophyletic (Rieseberg and Brouillet 1994). Different modes of speciation such as allopatric, parapatric, or sympatric result in different phylogenetic patterns (Rieseberg andBrouillet 1994, Knowles andCarstens 2007). ...
... In the case of incipient species, insufficient isolation and divergence may have precluded the formation of morphologically discrete and reciprocally monophyletic entities for the molecular loci considered. In such cases, combining systematics with population genetic approach is crucial to consider the balance between recognizing ongoing diversification (Moritz 2002;Daïnou et al. 2014) and the risk of over-splitting taxa into populations that may compromise both management and rescue, and ultimately their long-term persistence (Frankham et al. 2012). ...
Full-text available
Background and aims – Conserving threatened biodiversity needs a robust knowledge and delineation of the taxonomical entities targeted for in or ex situ conservation. In the genus Armeria (Plumbaginaceae), the frequent correlation between geographical isolation and morphological differentiation led to the delineation of some taxa consisting of few populations. Armeria belgenciensis is a narrow endemic represented by one population in southern France, which recently experienced a strong demographic decline caused by repeated disturbances of its habitat and was the object of a demographic rescue. Within two kilometres of A. belgenciensis, the presence of A. arenaria subsp. peirescii, also represented by one small population, poses the question whether the two populations are genetically isolated and should be considered as separated taxa for conservation.Material and methods – The genetic structure of Armeria belgenciensis and the geographically close populations of Armeria arenaria were analysed based on 328 AFLP markers using model-based and model-free clustering methods. In addition, flowering phenology was estimated to address the possibility of pre-zygotic isolation between A. belgenciensis and A. arenaria subsp. peirescii. Key results – The genetic structure supported a close relationship between A. belgenciensis and A. arenaria subsp. peirescii, compared to other populations of A. arenaria. Despite its recent decline, the genetic diversity of A. belgenciensis was not particularly low compared to the other studied populations or other rare species in this genus. The existence of gene flow between these two populations is likely given their moderate differentiation and the detection of admixture by one of the clustering methods. The phenological study showed that cross-pollination is possible during a short time period. Conclusion – We propose to place these two populations under the name A. belgenciensis, but to keep them as two different management units until there is no evidence of a harmful effect due to small effective sizes.
... This supports the view that past climatic events have created genetic structures within many African tropical forest plant species, as obtained with several types of markers and species (e.g. Budde et al., 2013;Daïnou et al., 2014;Demenou et al., 2017;Demenou et al., 2016;Duminil et al., 2015;Faye et al., 2016;Hardy et al., 2013;Iloh et al., 2017;Migliore et al., 2018 and others). Our results also highlight the key role of the area surrounding the Cameroonian Volcanic Line for hosting ancient lineages, consistently with previous studies Dauby et al., 2014;Migliore et al., 2019). ...
Paleo-environmental data show that the distribution of African rain forests was affected by Quaternary climate changes. In particular, the Dahomey Gap (DG) – a 200 km wide savanna corridor currently separating the West African and Central African rain forest blocks and containing relict rain forest fragments – was forested during the mid-Holocene and possibly during previous interglacial periods, whereas it was dominated by open vegetation (savanna) during glacial periods. Genetic signatures of past population fragmentation and demographic changes have been found in some African forest plant species using nuclear markers, but such events appear not to have been synchronous or shared across species. To better understand the colonization history of the DG by rain forest trees through seed dispersal, the plastid genomes of two widespread African forest legume trees, Anthonotha macrophylla and Distemonanthus benthamianus, were sequenced in 47 individuals for each species, providing unprecedented phylogenetic resolution of their maternal lineages (857 and 115 SNPs, respectively). Both species exhibit distinct lineages separating three regions: 1. Upper Guinea (UG, i.e. the West African forest block), 2. the area ranging from the DG to the Cameroon volcanic line (CVL), and 3. Lower Guinea (LG, the western part of the Central African forest block) where three lineages co-occur. In both species, the DG populations (including southern Nigeria west of Cross River) exhibit much lower genetic diversity than UG and LG populations, and their plastid lineages originate from the CVL, confirming its role as an ancient forest refuge. Despite the similar phylogeographic structures displayed by A. macrophylla and D. benthamianus, molecular dating indicates very contrasting ages of lineage divergence (UG diverged from LG since c. 7 Ma and 0.7 Ma, respectively) and DG colonization (before and after the Last Glacial Maximum, respectively). The stability of forest refuge areas and repeated similar forest shrinking/expanding events during successive glacial periods might explain why similar phylogeographic patterns can be generated over contrasting timescales. Keywords African rain forestCameroon Volcanic Linecolonization originDahomey Gapphylogeographyplastid genome sequencing
... In the last few years, population genetic data have accumulated for a number of African rainforest trees and revealed well-differentiated parapatric genetic clusters in Central and West African rainforests for most species (e.g., Budde et al., 2013;Hardy et al., 2013;Daïnou et al., 2014Daïnou et al., , 2016Heuertz et al., 2014;Piñeiro et al., 2017;Demenou et al., 2018;Migliore et al., 2018). In general, this genetic structuring cannot be explained by current geographic barriers such as the main mountain chains (Cameroonian Volcanic Line, Cristal Mountains, and Chaillu massif) or major rivers in the region (Sanaga, Dja, and Oougué Rivers). ...
Premise: Few studies have addressed the evolutionary history of tree species from African savannahs. Afzelia contains economically important timber species, including two species widely distributed in African savannahs: A. africana in the Sudanian region and A. quanzensis in the Zambezian region. We aimed to infer whether these species underwent range fragmentation and/or demographic changes, possibly reflecting how savannahs responded to Quaternary climate changes. Methods: We characterized the genetic diversity and structure of these species across their distribution ranges using nuclear microsatellites (SSRs) and genotyping-by-sequencing (GBS) markers. Six SSR loci were genotyped in 241 A. africana and 113 A. quanzensis individuals, while 2800 high-quality single nucleotide polymorphisms (SNPs) were identified in 30 A. africana individuals. Results: Both species appeared to be mainly outcrossing. The kinship between individuals decayed with the logarithm of the distance at similar rates across species and markers, leading to relatively small Sp statistics (0.0056 for SSR and 0.0054 for SNP in A. africana, 0.0075 for SSR in A. quanzensis). The patterns were consistent with isolation by distance expectations in the absence of large-scale geographic gradients. Bayesian clustering of SSR genotypes did not detect genetic clusters within species. In contrast, SNP data resolved intraspecific genetic clusters in A. africana, illustrating the higher resolving power of GBS. However, these clusters revealed low levels of differentiation and no clear geographical entities, so that they were interpreted as resulting from the isolation by distance pattern rather than from past population fragmentation. Conclusions: These results suggest that populations have remained connected throughout the large, continuous savannah landscapes. The absence of clear phylogeographic discontinuities, also found in a few other African savannah trees, indicates that their distribution ranges have not been significantly fragmented during the climatic oscillations of the Pleistocene, in contrast to patterns commonly found in African rainforest trees.
... This supports the view that past climatic events have created genetic structures within many African tropical forest plant species, as obtained with several types of markers and species (e.g. Budde et al., 2013;Daïnou et al., 2014;Demenou et al., 2017;Demenou et al., 2016;Duminil et al., 2015;Faye et al., 2016;Hardy et al., 2013;Iloh et al., 2017;Migliore et al., 2018 and others). Our results also highlight the key role of the area surrounding the Cameroonian Volcanic Line for hosting ancient lineages, consistently with previous studies Dauby et al., 2014;Migliore et al., 2019). ...
Khaya (Meliaceae) is a widespread genus of trees distributed from the tropical forests of Africa to Madagascar and the Comoros. Khaya species are very close morphologically and some contain considerable phenotypic and ecological diversity, raising the question of their delimitation: this is the case of Khaya anthotheca s.l. (sensu lato, including K. nyasica), which is often lumped with K. grandifoliola and K. ivorensis. We tested species delimitation within K. anthotheca by combining population-genetic and morphological data. First, a visual inspection of over a thousand specimens suggested the existence of five morphogroups. Second, after genotyping 498 samples at 71 nuclear SNPs, Bayesian clustering analyses (STRUCTURE) first identified five genetic clusters (K1–K5), while further analyses subdivided K5 into two clusters. The six genetic clusters obtained are distributed in parapatry or allopatry from West Africa to East Africa, except for cluster K4, which is in sympatry with K3 in Central Africa. The partitioning of genetic clusters is also largely congruent with the morphogroups identified. The morphological differentiation of these clusters was confirmed by statistical analyses of 27 leaf characters. The clear-cut genetic differentiation between clusters, with few admixed genotypes even in contact zones, argues for the recognition of distinct species following the biological species concept. Our analyses highlight that the current taxonomic treatment of Khaya underestimates by one-half the total number of species. We identify two new species, circumscribe K. anthotheca and support the rehabilitation at the rank of species of three taxa considered to be synonyms of K. anthotheca (K. agboensis, K. euryphylla, K. nyasica). These results illustrate the strength of combining population-genetic and morphological data to delineate an African rainforest tree species complex.
Full-text available
This paper gives a historical overview of the African rain forest from its origins, towards, the end of the Cretaceous period. The areas around the Gulf of Guinea, in particular from Ivory Coast to Nigeria and especially Cameroon, Gabon and Congo, appear to have been already occupied at this time by wet tropical forest formations mainly composed of Angiosperms. In the course of the Tertiary period the combined effect of the equator being situated further north than now and the development of the Antarctic ice cap favoured the development of wet tropical conditions over a large part of North Africa. Towards the end of the Tertiary, the equator reached its present position and the northern hemisphere ice caps appeared, and these phenomena resulted in the disappearance of the forest formations spread across the north of Africa, and the concentration of these formations near the equatorial zone around the Gulf of Guinea and in the Congo-Zaire basin. From 800 000 years ago onwards the marked glacial variations at middle and high latitudes in both hemispheres, lowered temperatures in equatorial areas and brought arid climates at times of maximum glacial extension. -Author
We describe a model-based clustering method for using multilocus genotype data to infer population structure and assign individuals to populations. We assume a model in which there are K populations (where K may be unknown), each of which is characterized by a set of allele frequencies at each locus. Individuals in the sample are assigned (probabilistically) to populations, or jointly to two or more populations if their genotypes indicate that they are admixed. Our model does not assume a particular mutation process, and it can be applied to most of the commonly used genetic markers, provided that they are not closely linked. Applications of our method include demonstrating the presence of population structure, assigning individuals to populations, studying hybrid zones, and identifying migrants and admixed individuals. We show that the method can produce highly accurate assignments using modest numbers of loci—e.g., seven microsatellite loci in an example using genotype data from an endangered bird species. The software used for this article is available from
This chapter focuses on DNA sequence phylogenetics and reproductive physiology of land plants. In the light of the new DNA sequence derived phylogenetic trees, physiological traits have been shown to have patterns of distribution substantially the same as other types of characters. They potentially have as much value in classification as other characters. Among plants, self-incompatibility (SI) is almost uniquely an angiosperm phenomenon. The amazing physiological diversity of self-incompatibility between and within groups emphasizes the importance of self-incompatibility (SI) in the diversification of the angiosperms. It also reflects on the fundamental importance of the closure of the carpel and evolution of the style and stigma within angiosperms because it was these events that provided the locus from which many kinds of self-incompatibility could arise. The distribution of different kinds of SI across groups at different ranks in the taxonomic hierarchy appears complicated because of a hitherto unsuspected flexibility in SI response. The assessment of homology is complicated because a shared common origin of an SI mechanism (true homology) may be obscured by multiple loss of SI and the presence of other kinds of SI in the families. Gene duplication and transpacific evolution, so that lineages of different SI genes in an individual taxon pre-date the divergence of the taxon itself, further complicate evolutionary patterns.
The modern age of species concepts began in 1942, when Ernst Mayr gave concept names to several different approaches to species identification. A long list of species concepts then followed, as well as a complex literature on their merits, motivations and uses. Some of these complexities arose as a consequence of the semantic shift that Mayr introduced, in which procedures for identifying species were elevated to concepts. Much of the debate in recent decades over concepts, and over pluralism versus monism, can be seen as an unnecessary consequence of treating species identification criteria as if they were more fundamental concepts. Recently, biologists have begun to recognize both the shortcomings of a lexicon of multiple species concepts and a common evolutionary idea that underlies them.
We review commonly used population definitions under both the ecological paradigm (which emphasizes demographic cohesion) and the evolutionary paradigm (which emphasizes reproductive cohesion) and find that none are truly operational. We suggest several quantitative criteria that might be used to determine when groups of individuals are different enough to be considered ‘populations'. Units for these criteria are migration rate ( m ) for the ecological paradigm and migrants per generation ( Nm ) for the evolutionary paradigm. These criteria are then evaluated by applying analytical methods to simulated genetic data for a finite island model. Under the standard parameter set that includes L = 20 High mutation (microsatellitelike) loci and samples of S = 50 individuals from each of n = 4 subpopulations, power to detect departures from panmixia was very high (∼ 100%; P < 0.001) even with high gene flow ( Nm = 25). A new method, comparing the number of correct population assignments with the random expectation, performed as well as a multilocus contingency test and warrants further consideration. Use of Low mutation (allozyme-like) markers reduced power more than did halving S or L . Under the standard parameter set, power to detect restricted gene flow below a certain level X (H 0 : Nm < X ) can also be high, provided that true Nm ≤ 0.5 X . Developing the appropriate test criterion, however, requires assumptions about several key parameters that are difficult to estimate in most natural populations. Methods that cluster individuals without using a priori sampling information detected the true number of populations only under conditions of moderate or low gene flow ( Nm ≤ 5), and power dropped sharply with smaller samples of loci and individuals. A simple algorithm based on a multilocus contingency test of allele frequencies in pairs of samples has high power to detect the true number of populations even with Nm = 25 but requires more rigorous statistical evaluation. The ecological paradigm remains challenging for evaluations using genetic markers, because the transition from demographic dependence to independence occurs in a region of high migration where genetic methods have relatively little power. Some recent theoretical developments and continued advances in computational power provide hope that this situation may change in the future.
In continuation of de Ruiter's treatment of Myrianthus and Musanga (Bull. Jard. Bot. Nat. Belg. 46: 471-510.1976), the present paper gives a revision of the African representatives of 17 genera of the Moraceae. The area studied not only consists of the African Continent, but also includes Madagascar, the Comoro Islands, the Mascarenes, the Seychelles, and the Aldabra Islands. Several new combinations are made: Antiaris toxicaria ssp. africana (Engl.) C. C. Berg, A. toxicaria ssp. africana var. usambarensis (Engl.) C. C. Berg, A. toxicaria ssp. macrophylla (R.Br.) C. C. Berg, A. toxicaria ssp. madagascariensis (H. Perrier) C. C. Berg, A. toxicaria ssp. humbertii (Léandri) C. C. Berg, Broussonetia greveana (Baillon) C. C. Berg, Treculia africana ssp. madagascarica (N.E.Br.) C. C. Berg, and T. africana ssp. madagascarica var. sambiranensis (Léandri) C. C. Berg. Many names are brought into synonymy. Besides revising taxa, the present study aims to fill a gap in our knowledge between Asian Moraceae (studied by Corner, whose studies resulted in a new classification of the family) and the neotropical Moraceae, a subject of study by the present author. Therefore discussions about classification of the family and relationships of African Moraceae with moraceous taxa elsewhere are an essential part of the present paper.
Below the level of what is commonly accepted as the species in sexual organisms there may be no phylogeny, only reticulate relationships (tokogeny) among individuals. The conventional molecular systematics research program, which infers organismal relationships from the topologies of allele or haplotype trees, will be misled below this critical level, in part because allele trees are not expected to track tokogenetic relationships. Yet there appears to be little or no consensus about how to delimit taxa that are appropriate for phylogenetic analysis. Individual sexual organisms cannot be the terminal units of analyses that employ typical nuclear genes, for which heterozygosity is expected, because heterozygous individuals cannot be uniquely placed on allele trees. Methods using the topologies of multiple allele trees also are flawed, in part for the same reason but also because, for an unknown fraction of the nuclear genome, allele phylogenies are not expected to track organismal relationships. A non-topological approach, which does not assume "species" or "populations," is proposed here. This method identifies allele pools at individual loci by the distribution of alleles in overlapping sets of heterozygous individuals. Individuals possessing alleles at a given allele pool of a locus belong to the same group, called here a field for recombination. Allele pools at each locus are a representation of the overall gene pool, which in turn can be inferred from multilocus genotypes; individuals sharing a common gene pool belong to the same multilocus field for recombination (ml-FFR). Once such gene pools are defined, allele topologies at suitable loci can be used to infer phylogenies.