Plant Syst. Evol. 247: 233–239 (2004)
DOI 10.1007/s00606-004-0155-x
Studies in the tribe Spermacoceae (Rubiaceae-Rubioideae):
the circumscriptions of Amphiasma and Pentanopsis
and the affinities of Phylohydrax
M. Thulin1 and B. Bremer1,2
1
2
Department of Systematic Botany, EBC, Uppsala University, Uppsala, Sweden
The Bergius Foundation at the Royal Swedish Academy of Sciences, Stockholm, Sweden
Received October 18, 2003; accepted February 24, 2004
Published online: June 11, 2004
Springer-Verlag 2004
Abstract. Phylogenetic analysis of rbcL cpDNA
sequences for 34 members of the tribe Spermacoceae s. lat. indicates that the African genera
Amphiasma, Conostomium, Manostachya, Pentanopsis and Phylohydrax form a strongly supported
clade, characterised by basal placentation and
heterostyly. This is a new position for the sea
shore genus Phylohydrax that has previously been
associated with Diodia and Spermacoce in Spermacoceae s. str. Amphiasma is not monophyletic
as the two samples of A. gracilicaule included
form a strongly supported clade with Pentanopsis
fragrans. A new taxonomy is therefore proposed,
where the previously monotypic Pentanopsis is
circumscribed as a genus of two species in northeastern tropical Africa, whereas Amphiasma is
treated in its original sense as a genus of about
eight species confined to south-central tropical
Africa. The new combination Pentanopsis gracilicaulis is made.
Key words: Rubiaceae, Spermacoceae, Amphiasma,
Conostomium, Hydrophylax, Manostachya, Pentanopsis, Phylohydrax, chloroplast DNA, rbcL, morphology, phylogeny, taxonomy.
Fifty-two genera were included in the tribe
Spermacoceae by Bremer and Manen (2000),
32 of these on the basis of molecular data,
and the remaining 20 on the basis of
morphological similarity. With this wide
circumscription the tribe includes, for example, all investigated taxa of the formerly
generally recognised tribes Hedyotideae and
Knoxieae. However, in the published cladograms (Bremer and Manen 2000: Figs. 3, 4)
only a small part of these genera are shown.
In the present study we include a relatively large sample of taxa from the Spermacoceae s. lat. and focus on a clade
including the African genera Amphiasma,
Conostomium, Pentanopsis (all formerly Hedyotideae), and Phylohydrax (formerly Spermacoceae s. str.). These four genera were
included also in the analysis of Bremer and
Manen (2000), but they were absent from the
published cladograms. Here we add five
further sequences from taxa belonging in
this clade, with the particular aim of investigating the circumscriptions of Amphiasma
and Pentanopsis, as well as the phylogenetic
position of the enigmatic Phylohydrax that
was segregated from Hydrophylax by Puff
(1986).
234
M. Thulin and B. Bremer: Studies in the tribe Spermacoceae (Rubiaceae)
Material and methods
Results
The sampling strategy was to include representative taxa from different parts of the tribe Spermacoceae, but with a concentration on the
Amphiasma/Pentanopsis clade, according to the
analyses in Bremer and Manen (2000). Based on
general morphological similarities and type of
placentation we also included genera that we
suspected could be close to Amphiasma and
Pentanopsis (Dibrachionostylus, Manostachya, and
Mitrasacmopsis). Two samples of Amphiasma
gracilicaule were included, one from a plant
closely similar to the type, and one from a more
large-flowered form. We also added a sequence of
Hydrophylax. Altogether 33 taxa representing 26
genera of Spermacoceae were sampled. As outgroup we used one taxon from outside the
subfamily Rubioideae (Luculia) and five genera
representing other tribes within Rubioideae (Bremer and Manen 2000). Silica gel-dried or herbarium material was used in the DNA investigations.
DNA was extracted and amplified according to
Bremer et al. (1995) and sequencing was
performed with a MegaBACE 1000 (Amersham
Biosciences) following the protocol of the
manufacturer. The sequences of the coding gene
rbcL were manually aligned by using the reading
frames of the corresponding amino acid sequences. The rbcL matrix comprises 40 sequences.
Eight of these have been produced for this study
(Fig. 1), and the other sequences have been
published before (Bremer et al. 1995, Manen and
Natali 1995, Bremer 1996, Bremer and Manen
2000).
Phylogenetic relationships were obtained by
parsimony analysis using PAUP (Swofford 1998),
version 4.0b10. Only informative characters were
analysed. The search method was heuristic with 100
replications of RANDOM stepwise additions of
sequences, the TBR branch swapping, and MULPARS options in effect. Support for the nodes was
calculated with jackknife analysis with 1000 replicates and TBR branch swapping and MULPARS
off.
The data matrix comprises 39 species (one
species, Amphiasma gracilicaule, is represented
by two accessions) and 1402 positions, 187 of
which are parsimony-informative. The heuristic analysis resulted in 18 most parsimonious
trees 649 steps long with a consistency index
CI ¼ 0.604 and a retention index RI ¼ 0.709. A
strict consensus tree is shown in Fig. 1. The
included taxa of Conostomium, Phylohydrax,
Manostachya, Amphiasma, and Pentanopsis,
form a strongly supported clade (99%), ‘‘the
Pentanopsis clade’’. Further, Conostomium is
strongly supported as monophyletic (100%),
as is Amphiasma plus Pentanopsis (99%), but
the two samples of A. gracilicaule form a
strongly supported clade (100%) with Pentanopsis, whereas A. benguellense is strongly
supported (94%) as sister to A. luzuloides.
Hydrophylax is nested with species of
Diodia and Spermacoce in a strongly supported
clade (99%) that is sister with strong support
(97%) to a clade with other taxa with solitary
ovules, these two clades together corresponding to the former Spermacoceae s. str. Spermacoce itself is, according to our analysis,
biphyletic with S. hispida in the former and S.
laevis in the latter of these clades.
Discussion
Puff (1986), when erecting the new genus
Phylohydrax for P. carnosa (in South Africa
and Mozambique) and P. madagascariensis (in
Madagascar and Tanzania), suggested that
this genus and Hydrophylax (with the remaining species H. maritima in southern Asia) are
derived from different portions of ‘‘an ancestral Diodia–Spermacoce-like stock’’. Our
results confirm that Hydrophylax maritima is
indeed closely related to Diodia and Spermac
Fig. 1. Strict consensus tree with jackknife values above branches. The asterisks ( ) indicate new sequences
(Amphiasma benguellense, Bamps et al. 4489 (UPS); Conostomium longitubum, Thulin 10855 (UPS);
Dibrachionostylus kaessneri, Strid 2464 (UPS); Hydrophylax maritima, Lundqvist 8945 (UPS); Manostachya
ternifolia, Malaisse & Robbrecht 2063 (K); Mitra-sacmopsis quadrivalvis, Richards 11069 (K); Pentanopsis
gracilicaulis (Amphiasma gracilicaule), Thulin et al. 10608 (UPS) & Thulin et al. 10512 (UPS)
M. Thulin and B. Bremer: Studies in the tribe Spermacoceae (Rubiaceae)
235
236
M. Thulin and B. Bremer: Studies in the tribe Spermacoceae (Rubiaceae)
coce. As Spermacoce comes out as biphyletic,
Hydrophylax may indeed even be nested within
Spermacoce. The delimitations between Spermacoce, Diodia, Hemidiodia, and various other
segregates in this complex, are weak (see, for
example, Verdcourt 1976). Our sampling is too
small to allow any conclusions to be drawn,
but surely the clade corresponding to Spermacoceae s. str. is in need of further study. As
pointed out by Puff (1986) several apparently
unrelated species of Spermacoce and Diodia
grow on tropical sea-shores and Hydrophylax
maritima, with its creeping stems rooting at the
nodes and indehiscent fruits, may represent an
extreme case of adaptation in this respect.
The Pentanopsis clade has not previously
been detected in phylogenetic analyses. Andersson and Rova (1999) showed a close
relationship between Conostomium and Amphiasma, but also Oldenlandia affinis was
nested in this clade. We are not able to
contradict this position for O. affinis, but
regard it as highly unlikely. Despite the strong
support for the Pentanopsis clade in our
molecular analysis, it is not easy to characterise the group morphologically. All members
seem to be heterostylous, but heterostyly is
found also elsewhere in Spermacoceae, for
example in species of Pentas and Oldenlandia.
The number of ovules per locule varies from
numerous to one, but a unifying feature is that
the placentation always is more or less basal,
whereas in Spermacoceae s. str. the solitary
ovules are attached near the middle of the
septum. Manostachya was originally stated to
have ovules attached to the middle of the
septum (Bremekamp 1952), and this has been
repeated by, e.g. Verdcourt (1989). However,
in material studied by us, Manostachya has
basal placentation as the other members of the
Pentanopsis clade. Basal placentation is otherwise rare in Spermacoceae, and the only other
instances may be Carphalea, Chamaepentas,
Mitrasacmopsis, Pseudonesohedyotis and Dibrachionostylus. Of these genera Carphalea,
Mitrasacmopsis and Dibrachionostylus are included in this study, and they do not belong in
the Pentanopsis clade (see Fig. 1). The remain-
ing taxa, Chamaepentas and Pseudonesohedyotis, are both monotypic African genera and,
judging from morphology, they probably do
not belong in the Pentanopsis clade either.
Phylohydrax is clearly nested in the Pentanopsis clade, even if its sister group relationship is
uncertain. It is also, with its heterostylous
flowers, filiform stigma lobes, and nearly basal
attachment of the placenta, morphologically
quite at home in this clade, although the solitary
ovules and indehiscent fruits are unique. The
superficially similar Hydrophylax, instead, has
isostylous flowers, capitate or shortly 2-lobed
stigma, and the placenta attached near the
middle of the septum, all characters that are
compatible with Spermacoceae s. str. Obviously
the various similarities in habit between
Phylohydrax and Hydrophylax can be attributed to convergent adaptations to a life on sea
shores. For example, the indehiscent fruits of
both these genera may be adapted to dispersal
with water as proposed by Puff (1986).
The two species of Conostomium included,
C. quadrangulare (type of genus) and C. longitubum (type of C. subgen. Beckia), form a
strongly supported clade (100%). Conostomium
is morphologically characterised by its pollen
grains. Bremekamp (1952) described them as
porate, but Lewis (1965) showed that short colpi
are also present. The other members of the
Pentanopsis clade, as far as known, all have
colporate pollen with long and distinct colpi. A
possible exception is Phylohydrax, the pollen
grains of which were described as ‘‘plurizonocolpate’’ by Robbrecht in Puff (1986).
Amphiasma was proposed by Bremekamp
(1952) as a genus in south-central tropical
Africa with eight species from southern Tanzania in the north to Namibia in the south.
Four of these were described as new, whereas
the remaining species were previously placed in
Oldenlandia. Since then, one further species
has been described, A. gracilicaule (Verdcourt
1981). This was based on a single collection
from central Somalia and, on account of its
long-tubed, glabrous corolla and 4-colporate
pollen it was placed in a subgenus of its own,
A. subgen. Stonocomium. The other species of
M. Thulin and B. Bremer: Studies in the tribe Spermacoceae (Rubiaceae)
Amphiasma have corollas with a very short
tube, hairy in throat, and 3-colporate pollen
(Bremekamp 1952). Comparisons were also
made between A. gracilicaule on one hand and
Pentanopsis and Conostomium on the other
(Verdcourt 1981). In our analysis two forms of
A. gracilicaule have been included, one corresponding to the type specimen in flower size
and one with slightly larger flowers. Both these
forms group with strong support (100%) with
Pentanopsis, a monotypic genus also from the
Horn of Africa region and actually partly
sympatric with A. gracilicaule.
The type of Amphiasma, A. luzuloides from
southern Tanzania and Malawi, forms a
strongly supported clade (94%) with A. benguellensis from Angola. Morphological support
for this clade is 3-colporate pollen, in contrast
to the 4-colporate pollen in the clade with
P. fragrans and A. gracilicaule. These two
clades together have a strong support (99%) in
the analysis and all members have dorsiventrally flattened seeds with thin walls of the
testa cells. Of the other members of the
Pentanopsis clade, Conostomium has angular
seeds, Phylohydrax has seeds rounded in section, and Manostachya has dorsiventrally flattened seeds with thick outer walls of the testa
cells (Bremekamp 1952).
Chromosome numbers are known for all
genera in the Pentanopsis clade, except from
Pentanopsis itself. Conostomium and Amphiasma have x ¼ 9 (Lewis 1965, Kiehn 1985) and
this base number characterises also many
related genera, such as Agathisanthemum,
Dibrachionostylus, Lelya, Oldenlandia, and
Pentodon, and it is the most common number
in the previous Hedyotideae. However, Manostachya is exceptional in having x ¼ 11 (Lewis
1965, Kiehn 1985), a common number in other
parts of Rubiaceae, and Lewis (1965) suggested its occurrence in Manostachya to be ‘‘a
relic’’. In our view it is more likely that x ¼ 11
in Manostachya is secondarily derived from
x ¼ 9, although we cannot suggest any particular mechanism for this. Most interesting is the
report of x ¼ 14 (2n ¼ 56) in Phylohydrax
(Kiehn 1985, Puff 1986). This base number is
237
characteristic for Spermacoceae s. str., and has
also been reported for Hydrophylax (Puff
1986). However, Kiehn (1985) gave the number as 2n ¼ 56 ± 2, and we believe that further
studies are needed to exclude the possibility
that Phylohydrax is a hexaploid with x ¼ 9.
Taxonomy of Amphiasma and Pentanopsis
With its present circumscription Amphiasma is
obviously not monophyletic and some taxonomic change is necessary. We have considered the following three options, under the
assumption that the two samples of A. gracilicaule are conspecific:
(1) to unite Amphiasma and Pentanopsis to
create a genus of about 10 species, two in
north-east tropical Africa and eight in
south-central Africa (some of the latter are
very similar according to Verdcourt 1976).
(2) to transfer A. gracilicaule to Pentanopsis to
create a genus of two species in north-east
tropical Africa. This would differ from
Amphiasma not only in having 4-colporate
(versus 3-colporate) pollen grains, but also
in, for example, its much larger flowers. The
corolla tube in such an extended Pentanopsis would be 12–37 mm long, whereas in
Amphiasma s. str. it is only 2–4 mm long.
(3) to place A. gracilicaule in a genus of its own
differing from Pentanopsis in its glabrous
corolla. In P. fragrans the corolla-lobes are
covered on the inside with short blunt hairs.
On balance we prefer the second option. By
this we avoid monotypic entities and get two
geographically confined genera that can be
very easily recognised by the great difference in
flower size. Amphiasma can then be circumscribed in the same way as it was before 1981,
whereas for Pentanopsis we propose a new
taxonomy that is outlined below.
PENTANOPSIS Rendle in J. Bot. 36: 28
(1898). Type: P. fragrans Rendle.
Amphiasma subgen. Stonocomium Verdc. in
Kew Bull. 36: 498 (1981). Type: A. gracilicaule
Verdc.
238
M. Thulin and B. Bremer: Studies in the tribe Spermacoceae (Rubiaceae)
Slender shrubs or subshrubs. Leaves
opposite, linear-subulate to lanceolate; stipules sheathing the stem, united with petiole,
with fimbriate margin, becoming ± woody
and persistent. Flowers fragrant, heterostylous, solitary or few together terminating
short shoots; pedicels short. Calyx-tube
obovoid; lobes 4, narrowly triangular to
linear, not united at the base. Corolla
white, with slender tube; lobes 4, induplicatevalvate in bud. Stamens 4, inserted in
corolla-tube; anthers oblong; pollen grains
4-colporate. Ovary 2-locular with numerous
ovules on peltate placentas inserted at the base
of the septum; style with filiform stigma-lobes
shortly exserted in long-styled flowers, included in short-styled flowers. Capsule obovoid or ellipsoid, with a short beak,
loculicidally dehiscent. Seeds flat, elliptic; testa
thin-walled, reticulate, with lumina of testa
cells smooth.
Genus of two species in north-eastern
tropical Africa.
Pentanopsis differs from Amphiasma in its
much larger flowers (corolla tube 12–37 versus
2–4 mm long), 4-aperturate (versus 3-aperturate) pollen grains, and by its stipules becoming more or less woody and persistent. Also,
the calyx-lobes are not united at the base,
whereas this generally seems to be the case in
Amphiasma.
1. Leaves ± scabrid; corolla-tube 18–37 mm
long;
corolla-lobes ± densely
covered
with short hairs on the inside.1. P. fragrans
– Leaves glabrous; corolla-tube 12–14 mm
long; corolla-lobes glabrous on the inside.
2. P. gracilicaulis
1. Pentanopsis fragrans Rendle in J. Bot. 36: 29
(1898). Type: Somalia, ‘‘Wagga’’ Mt, Lort
Phillips s.n. (BM holotype).
Conostomium brevirostrum Bremek. in
Verh. Kon. Nederl. Akad. Wetensch., Afd.
Natuurk., sect. 2, 48: 129 (1952), syn. nov.
Type: Somalia/Ethiopia border, between Dolo
(‘‘Doloun’’) and ‘‘Batta’’, Ruspoli & Riva 1094
(FT holotype).
Conostomium squarrosum Bremek. in Kew
Bull. 11: 169 (1956). Type: Ethiopia, between
Gorrahi and Wardere, Ellis 138 (K holotype).
Shrub, 0.3–2.5 m tall. Leaf-blades linear to
lanceolate, 6–40 · 0.8–5 mm, acute at the apex,
± densely scabrid with short hairs above to
subglabrous, usually with revolute margins;
stipule sheath 2–7 mm long, with 1–4 up to c.
1 mm long fimbriae. Pedicels 1–4 mm long.
Calyx-tube 1.5–2.5 mm long; lobes narrowly
triangular to linear, 2–11 mm long, minutely
ciliate. Corolla white or tinged purplish outside;
tube 18–37 mm long, sparsely hairy inside;
lobes 5–15 mm long, acute, sparsely to densely
covered with short blunt hairs on the inside.
Anthers 1–2 mm long in long-styled flowers,
2.1–3 mm long in short-styled flowers. Style 2lobed, 16–22 mm long in short-styled flowers,
21–30 mm long in long-styled flowers. Capsule
4.5–7 mm long. Seeds 1.5–2 mm long.
Deciduous or evergreen bushland, usually
on rocky ground; 145–1500 m. Somalia, eastern Ethiopia, northern Kenya.
Representative specimens. Ethiopia. Bale
Region: 42 km SE of Ghinir on road to Imi,
655¢N, 4057¢E, 31 May 1983, Gilbert, Ensermu & Vollesen 7978 (K, UPS). Somalia.
Nugaal Region: 49 km SE of Sinujiif, 810¢N,
4910¢E, 6 May 2001, Thulin, Abdi Dahir,
Abdulkadir Khalid & Ahmed Osman 10463
(K, UPS). Kenya. Northern Frontier Prov.:
Dandu, 14 Apr. 1952, Gillett 12789 (K).
2. Pentanopsis gracilicaulis (Verdc.) Thulin &
B. Bremer, comb. nov.
Amphiasma gracilicaule Verdc. in Kew
Bull. 36: 498 (1981). Type: Somalia, slopes E
of Gawen village, 519¢N, 4818¢E, Gillett,
Hemming & Watson 22249 (K holotype, EA
isotype not seen).
Shrub or subshrub, 0.5–1.5 m tall. Leafblades linear-subulate, 20–70 · 0.5–1 mm,
acute at the apex, glabrous, with revolute
margins; stipule-sheath 1–3 mm long with 2–4
up to c. 0.5 mm long fimbriae. Pedicels 1.5–
4 mm long. Calyx-tube 1.5–2.5 mm long; lobes
narrowly triangular, 1–4 mm long. Corolla
white or tinged greyish-brownish outside, gla-
M. Thulin and B. Bremer: Studies in the tribe Spermacoceae (Rubiaceae)
brous; tube 12–14 mm long; lobes 7–11 mm
long, acute. Anthers c. 1.8–2 mm long in longstyled flowers, short-styled flowers not seen.
Style 2-lobed, 11–14 mm long in long-styled
flowers. Capsule 4–6 mm long. Seeds 1.2–
1.5 mm long.
Acacia-Commiphora bushland or open
plains, in shallow soil over limestone; 180–
300 m. North-eastern and central Somalia.
Remarks. The plants from the northernmost locality (Thulin et al. 10608) have larger
corollas and longer calyx-lobes than the other
material. However, the difference is not great
compared to the variation found within P.
fragrans. Also, the rbcL sequences from largeflowered and small-flowered plants are practically identical. However, the rbcL sequence
data is also inadequate to distinguish between
P. fragrans and P. gracilicaulis, and would be
compatible with a taxonomy recognising either
one, two or three species of Pentanopsis. The
taxonomy with two species proposed here is
therefore ultimately based on the clear discontinuities found in the morphological variation.
Additional specimens. Somalia. Bari
Region: 72 km from Qarxis on track to Marraya, 823¢N, 5008¢E, 15 May 2001, Thulin et
al. 10608 (K, UPS). Nugaal Region: 22 km on
track from Gaalogod to Garadeen, 743¢N,
4936¢E, 9 May 2001, Thulin et al. 10512
(UPS). Mudug Region: E of Gawen, c. 30 km
on road from Hobyo to Wisil, 519N, 4819¢E,
28 May 1989, Thulin & Abdi Dahir 6673 (E,
FT, K, UPS).
We thank Nahid Heidari for technical assistance, Royal Botanic Gardens, Kew, for DNA of
Manostachya ternifolia and Mitrasacmopsis quadrivalvis, and Johan Rova and an anonymous
reviewer for comments on the manuscript. The
study has been supported by grants from the
Swedish Research Council.
References
Andersson L., Rova J. H. E. (1999) The rps16
intron and the phylogeny of the Rubioideae
(Rubiaceae). Plant Syst. Evol. 214: 161–186.
239
Bremekamp C. E. B. (1952) The African species of
Oldenlandia L. sensu Hiern et K. Schumann.
Verh. Kon. Nederl. Akad. Wetensch., Afd.
Natuurk., sect. 2, 48: 1–297.
Bremer B. (1996) Phylogenetic studies within Rubiaceae and relationships to other families based
on molecular data. Opera Bot. Belg. 7: 33–50.
Bremer B., Andreasen K., Olsson D. (1995) Subfamilial and tribal relationships in the Rubiaceae
based on rbcL sequence data. Ann. Missouri
Bot. Gard. 82: 383–397.
Bremer B., Manen J. F. (2000) Phylogeny and
classification of the subfamily Rubioideae (Rubiaceae). Plant Syst. Evol. 225: 43–72.
Kiehn M. (1985) Karyosystematische Untersuchungen an Rubiaceae: Chromosomenzählungen
aus Afrika, Madagaskar und Mauritius. Plant
Syst. Evol. 149: 89–118.
Lewis W. H. (1965) Cytopalynological study of
African Hedyotideae (Rubiaceae). Ann. Missouri Bot. Gard. 52: 182–211.
Manen J. F., Natali A. (1995) Comparison of the
evolution of ribulose-l,5-biphosphate carboxylase (rbcL) and atpB-rbcL spacer sequences in a
recent plant group, the tribe Rubieae (Rubiaceae). J. Mol. Evol. 41: 920–927.
Puff C. (1986) Phylohydrax (Rubiaceae-Spermacoceae) – a new genus to accommodate the African
and Madagascan ‘‘Hydrophylax’’ species. Plant
Syst. Evol. 154: 343–366.
Swofford D. L. (1998) PAUP . Phylogenetic analysis using parsimony. Sinauer, Sunderland, Massachusetts, USA.
Verdcourt B. (1976) Rubiaceae (part 1). In: Polhill
R. M. (ed.) Flora of Tropical East Africa.
Crown Agents for Oversea Governments and
Administrations.
Verdcourt B. (1981) Notes on African Rubiaceae.
Kew. Bull. 36: 493–557.
Verdcourt B. (1989) Rubiaceae (Rubioideae). In:
Launert E. (ed.) Flora Zambesiaca 5(1). Flora
Zambesiaca managing committee: London.
Addresses of the authors: Mats Thulin
(e-mail: mats.thulin@ebc.uu.se), Department of
Systematic Botany, EBC, Uppsala University,
Norbyvägen 18D, SE-752 36 Uppsala, Sweden.
Birgitta Bremer, The Bergius Foundation at the
Royal Swedish Academy of Sciences, Box 50017,
SE-104 05 Stockholm, Sweden.