Molecular Phylogenetics and Evolution 47 (2008) 932–949
Contents lists available at ScienceDirect
Molecular Phylogenetics and Evolution
journal homepage: www.elsevier.com/locate/ympev
Phylogenetic relationships, character evolution and biogeography of southern
African members of Zygophyllum (Zygophyllaceae) based on three plastid regions
D.U. Bellstedt a,*, L. van Zyl b, E.M. Marais c, B. Bytebier a, C.A. de Villiers a, A.M. Makwarela b, L.L. Dreyer b
a
Department of Biochemistry, University of Stellenbosch, Stellenbosch, Private Bag X1, Matieland 7602, South Africa
Department of Botany and Zoology, University of Stellenbosch, Stellenbosch, South Africa
c
Department of Forest and Wood Science, University of Stellenbosch, Stellenbosch, South Africa
b
a r t i c l e
i n f o
Article history:
Received 8 June 2007
Revised 19 February 2008
Accepted 23 February 2008
Available online 29 February 2008
Keywords:
Zygophyllaceae
Zygophylloideae
Augea
Fagonia
Tetraena
Zygophyllum
trnLF
rbcL
Phylogeny
a b s t r a c t
The plastid coding rbcL and non-coding trnLF regions of 53 of 55 southern African Zygophyllum species
were sequenced and used to evaluate the phylogenetic relationships within the southern African representatives of the genus. Published sequences of the same gene regions of Australian, Asian and North African Zygophyllum species were included to assess the relationships of the species from these regions to the
southern African species. The addition of Z. stapffii from Namibia, found to be conspecific with Z. orbiculatum from Angola, lead to a greatly resolved tree. The molecular results were largely congruent with a
recent sectional classification of the southern African species and supported their subdivision into subgenera Agrophyllum and Zygophyllum. Reconstruction of the character evolution of capsule dehiscence,
seed attachment and seed mucilage showed that these characters allowed a division of southern African
species into the two subgenera but that this could not be applied to species occurring elsewhere. Other
morphological characters were found to vary and unique character combinations, rather than unique
characters, were found to be of systematic value in sectional delimitation. The study suggests that
repeated radiations from the horn of Africa to southern Africa and Asia and back lead to the present distribution of the taxa in the subfamily Zygophylloideae. Although this study supports some of the recent
taxonomic changes in the group, the unresolved relationships between the proposed genera Tetraena and
Roepera and those retained as Zygophyllum species suggest that changes to the taxonomy may have been
premature.
Ó 2008 Elsevier Inc. All rights reserved.
1. Introduction
The genus Zygophyllum L. is widely distributed in the arid and
semi-arid areas of southern, northern and north-eastern Africa
(White, 1983), the Mediterranean region, central Asia and Australia
(Retief, 2000). In southern Africa it ranges from the southern parts
of the Eastern Cape, through the Western Cape and Northern Cape
Provinces to Namibia, southwestern Botswana and as far north as
southern Angola. It occurs in seven of the eight biomes defined
by Low and Rebelo (1996), with the greatest number of species
present in the Nama and Succulent Karoo biomes (Van Zyl,
2000). Together with the genus Salsola L. (Chenopodiaceae), it
ranks as one of the most drought resistant plant genera in these
arid areas. As a result, it is a vital food source for small stock and
game in the very arid areas of the Northern Cape Province of South
Africa and southern Namibia.
Various authors have attempted an infrageneric classification
for the genus Zygophyllum. Endlicher (1841) subdivided the genus
* Corresponding author. Fax: +27 21 8085863.
E-mail address: dub@sun.ac.za (D.U. Bellstedt).
1055-7903/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.ympev.2008.02.019
into subgenera Zygophyllum and Agrophyllum Endl. based on fruit
dehiscence, while Sonder (1860; South African species) and Engler (1931, all species) ignored this character in their attempts to
define a sectional classification. Engler (1931) recognized 17 sections worldwide. Van Huyssteen (1937) again recognized the taxonomic significance of flower and fruit characters and
circumscribed the subgenera Zygophyllotypus Huysst. (referred
to as subgenus Zygophyllum by Van Zyl, 2000) and Agrophyllum
based on fruit dehiscence. The sectional classification of southern
African members of the genus according to Van Huyssteen (1937)
is summarized in Table 1. Outside of southern Africa, she recognized the sections Fabago Tourn. ex Adans. (22 species from Asia
including the type species, Z. fabago L.) and Sarcozygium (Bunge)
Engl. (one species from Mongolia, Z. xanthoxylum Engl.) in subgenus Zygophyllotypus and the sections Mediterranea Engl. (11 species from northern and eastern Africa) and Roepera Engl. (one
species, Z. fruticulosum DC. from Australia) in subgenus
Agrophyllum.
In her revision of southern African Zygophyllum, Van Zyl
(2000) recognized 54 species of which 17 were new, five of
these have been published (Van Zyl and Marais, 1997, 1999),
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
933
Table 1
A comparison of the classifications of southern African Zygophyllum of Van Huyssteen (1937) and Van Zyl (2000)
Van Huyssteen (1937)
Subgenus Agrophyllum Endl.
§ Alata Huysst. subsection
Angustialata Huysst.
§ Alata subsections
Morgsana Huysst.
§ Bipartita Huysst.
§ Cinerea Huysst.
Van Zyl (2000)
6 species
§ Alata
3 species (3 species reduced to synonomy)
§ Annua Engl.
§ Bipartita
Z. simplex L. (transferred from § Alata) and 2 new species
10 species (3 species reduced to synonomy, Z. simplex L.
(transferred to § Alata), Z. chrysopteron Retief and 3 new species
Z. giessii Merxm. & A.Schreib., Z. longicapsulare Schinz
Z. morgsana L.
10 species including
Z. simplex L.
Z. cinereum Schinz =
Z. longicapsulare Schinz
§ Cinerea
§ Grandifolia Engl.
§ Prismatica Van Zyl
Subgenus Zygophyllotypus (referred to as subgenus Zygophyllum by Van Zyl (2000)
§ Capensia Engl.
24 species
§ Capensia
§ Morgsana (Huysst.) Van Zyl
§ Paradoxa Huysst.
Z. paradoxum Schinz, Z. cordifolium L.f.,
§ Paradoxa
Z. orbiculatum Welw. ex Oliv.
§ Grandifolia
Z. stapffii Schinz
and 12 remain to be published (Table 2). Based on a comprehensive analysis of morphological characters, she divided these species into the two subgenera, Zygophyllum and Agrophyllum on the
basis of capsule dehiscence, seed attachment and the presence of
spiral threads in the seed mucilage. Her classification is compared to that of Van Huyssteen in Table 1. The most important
changes made to Van Huyssteen’s (1937) classification were
the transfer of section Grandifolia (containing only Z. stapffii),
from Zygophyllum to Agrophyllum and of section Morgsana (containing only Z. morgsana) from subgenus Agrophyllum to subgenus Zygophyllum.
In a study to establish the familial boundaries and intrafamilial relationships of the family Zygophyllaceae, Sheahan and
Chase (1996) on the basis of sequence data from the plastid
rbcL gene of all representative genera, recircumscribed the family to comprise the subfamilies Zygophylloideae, Tribuloideae,
Seetzenioideae, Larreoideae and Morkillioideae. They re-defined
the subfamily Zygophylloideae to include the genera Zygophyllum (ca. 150 species), Fagonia L. (30 species) and the monotypic
genera Augea Thunb. and Tetraena Maxim. Subsequently, in an
expanded study based on the trnLF and rbcL sequences of 15
species of Zygophyllum, three of Fagonia, and the monotypic Augea and Tetraena, Sheahan and Chase (2000) attempted to
establish the relationships of these genera within the subfamily
Zygophylloideae. Only six species of southern African Zygophyllum were included in this analysis, of which three belong to
subgenus Agrophyllum and three to subgenus Zygophyllum sensu
Van Zyl (2000). Species from subgenus Agrophyllum grouped
with Asian and North African Zygophyllum species and with Tetraena; those from subgenus Zygophyllum grouped with Australian species. The genera Fagonia and Augea and two
Zygophyllum species from the horn of Africa, Z. hildebrandtii
Engl. and Z. robecchii Engl., formed a separate clade. The relationships between some of these clades were, however, poorly
supported.
Beier et al. (2003) expanded on the previous study. They used
a wider sampling as well as morphological data and trnL intron
sequence data to elucidate the relationships within the subfamily Zygophylloideae. In addition to the six southern African Zygophyllum species used by Sheahan and Chase (2000), they
included Z. morgsana and Z. foetidum and the monotypic genera
Augea and Tetraena. Beier et al. (2003) retrieved a monophyletic
subfamily Zygophylloideae, but a paraphyletic genus Zygophyllum
Z. stapffii Schinz
Z. prismatocarpum Sond. (transferred from section Bipartita)
and 2 new species
29 species
Z. morgsana L.
Z. cordifolium L.f., Z. fusiforme Van Zyl ined.,
Z. orbiculatum Welw. ex Oliv.
including the genera Fagonia, Augea and Tetraena. Within the
paraphyletic Zygophyllum, a number of strongly supported clades
were retrieved, but relationships between these groupings were
poorly supported. Their phylogeny also did not retrieve the subgenera Zygophyllum and Agrophyllum, as circumscribed by Van
Huyssteen (1937) and Van Zyl (2000). They used these results
to establish a new generic classification for the Zygophylloideae.
The generic classification was supported by a limited number of
morphological synapomorphies.
In the current study, the plastid trnL intron and trnLF spacer
sequences of a near complete sampling (53 out of 55 taxa) of the
southern African Zygophyllum species were combined with published trnL intron and trnLF spacer sequences (Sheahan and
Chase, 2000) of Zygophyllum species occurring elsewhere and
the phylogenetic relationships were analyzed. To confirm the
position of some species in the tree, the rbcL gene of a representative subset of species was also sequenced and the combined
rbcL and trnLF sequence data were analyzed. These trees were
used to assess the subgeneric and sectional boundaries defined
by Van Zyl (2000). Patterns of evolution of three seed characters
were traced onto the combined results. Additionally, phytogeographical trends within the subfamily Zygophylloideae were
assessed based on the combined tree. Our results were then
compared to the newly proposed generic classification of Beier
et al. (2003).
2. Materials and methods
2.1. Sampling of taxa
Fifty-two out of the 54 Zygophyllum species recognized by Van
Zyl (2000) and an undescribed species, referred to as Zygophyllum
spec. aff. maritimum, were sampled (Table 2). All subgenera and
sections sensu Van Zyl (2000) were represented.
2.2. Outgroups
Outgroup taxa were chosen on the basis of Sheahan and
Chase (1996, 2000) and Beier et al. (2003). These included Tribulus macropterus as the most distant outgroup (Tribuloideae),
Seetzenia lanata (Seetzenioideae), and the New World species
Bulnesia arborea, Guaiacum guatemalense and Larrea tridentata
(Larreoideae).
934
Table 2
Sectional classification of the southern African members of the genus Zygophyllum sensu Van Zyl (2000), geographic distribution, voucher details, collection locality and database accession number
Section
Species
Subgenus Agrophyllum
§ Annua
Z. simplex L.
Z. spongiosum Van Zyl ined.
Z. inflatum Van Zyl ined.
Geographic
distribution
Voucher details
Collection locality
Database accession
number, trnLF
Database accession
number, rbcL
Africa & Asia
Chase 806 (K)
Bellstedt 854 (STE)
HK 1573 (WIND)
Northern Hemisphere
Orange River bed, Aussenkehr, Namibia
27 km from Otjinungwa on Kunene River along track to Rooidrom,
Kaokoveld, Kunene Region, Namibia
7.5 km from Ruacana Falls on road to Swartbooisdrift, Kunene Region,
Namibia
*AJ387974
EF 656004
EF 656006
*Y15031
EF655984
EF655985
EF 656005
nd
NN & southern
Angola
NN & southern
Angola
HK 1490 (WIND)
Z. prismatocarpum Sond.
Z. patenticaule Van Zyl ined.
Z. pterocaule Van Zyl
SN & NC
SN & NC
SN & NC
Bellstedt 860 (STE)
Bellstedt 868 (STE)
Mucina 270806/25
(STE)
Orange River embankment, Rosh Pinah district, Namibia
Lorelei Farm, Rosh Pinah district, Namibia
Alexander Bay, Richtersveld, NC
EF 656009
EF 656008
EF 656007
EF655990
EF655989
nd
§ Bipartita
Z. applanatum Van Zyl
Z. clavatum Schltr. & Diels in
Schultze
Z. cylindrifolium Schinz
Z. segmentatum Van Zyl
Z. tenue R. Glover
Z. retrofractum Thunb.
SN
Coastal SN & NC
Bellstedt 870 (STE)
Bellstedt 878
20 km N of Rosh Pinah, Namibia
Lüderitz Peninsula, Namibia
EF 656012
EF 656010
EF655988
EF655986
NWN
SN
SN
SN, NC & inland
WC
NC & inland WC
Inland WC
N & NC
Craven 3800 (WIND)
Bellstedt 861 (STE)
van Zyl 4593 (STE)
Marais 430 (STE)
Twyfelfontein Rock Art Site, Namibia
Boom Riverbed, West of Rosh Pinah, Namibia
S Namibia
Doring River Crossing, Ceres Karoo, South Africa
*AJ387966
EF 656015
EF 656017
EF 656014
*AJ133864
EF655987
nd
nd
Marais 427 (STE)
Bellstedt 799 (STE)
van Zyl 4588 (STE)
Doring River Crossing, Ceres Karoo, South Africa
Prince Albert, South Africa
S Namibia
EF 656013
EF 656016
EF 656011
nd
nd
EF655991
*AJ387967
*AJ133865
Z. chrysopteron Retief
Z. turbinatum Van Zyl ined.
Z. decumbens Delile var.
decumbens
Z. decumbens Delile
Horn of Africa
Thulin et al. 7981
UPS
§ Alata
Z. microcarpum Licht. ex Cham.
Z. rigidum Schinz
Z. longistiputalum Schinz
SN, NC & WC
SN & NC
SN
van Zyl 4591 (STE)
van Zyl 4590 (STE)
not collected
S Namibia
S Namibia
Southern Namibia
EF 656002
EF 656003
EF655983
EF655982
nd
§ Cinerea
Z. longicapsulare Schinz
Z. giessii Merxm. & A.Schreib.
Z. stapffii Schinz
SN
SN
Coastal NN
Bellstedt 879 (STE)
Bellstedt 874 (STE)
Mannheimer 6577
(STE)
Lüderitz Peninsula, Namibia
Arimas Farm, Witputs district, Namibia
Swakopmund, Namibia
EF 656001
EF 656000
**
EF655981
EF655980
**
SN, NC & WC
SN & NC
NN & southern
Angola
Marais 446 (STE)
Bellstedt 857 (STE)
Craven 5096 (WIND)
Koekenaap, South Africa
Aussenkehr, Namibia
30 km west of Foz de Cunene, Southern Angola
EF 656022
EF 656023
EF 655999
EF655993
nd
EF655979
§ Grandifolia
Subgenus Zygophyllum
§ Paradoxa
Z. cordifolium L.f.
Z. fusiforme Van Zyl ined.
Z. orbiculatum Welw. ex Oliver
§ Capensia I
Z.
Z.
Z.
Z.
Z.
teretifolium Schltr.
botulifolium Van Zyl
spinosum L.
pygmaeum Eckl. & Zeyh.
rogersii Compton
Inland WC
WC
Coastal NC & WC
NC & inland WC
Inland WC
Marais 447 (STE)
Marais 451 (STE)
Bellstedt 801 (STE)
Marais 424 (STE)
Marais 432 (STE)
Douse the Glim road, Van Rhynsdorp district, WC
Farm Bulslaagte, Ceres to Matjiesfontein Road, Tanqua Karoo, WC
Rondeberg Nature Reserve, North of Cape Town
Robertson, South Africa
Hartnekskloof, Ceres Karoo, WC
EF
EF
EF
EF
EF
656025
656026
656038
656046
656037
nd
nd
nd
nd
nd
§ Capensia II
Z.
Z.
Z.
Z.
divaricatum Eckl. & Zeyh.
namaquanum Van Zyl ined.
sessilifolium L.
spitskopense Van Zyl ined.
Coastal SC & EC
NC
Coastal WC
Coastal WC
Dold 4655 (GRA)
Marais 440 (STE)
Marais 434 (STE)
van Zyl 4606 (STE)
Coega River Mouth, EC
Studor’s Pass, Kamieskroon, NC
Farm Neulfontein, Morreesburg district, WC
Morreesburg District, WC
EF
EF
EF
EF
656031
656036
656047
656048
nd
nd
EF655997
nd
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
§ Prismatica
§ Capensia
III
§ Morgsana
cuneifolium Eckl. & Zeyh.
hirticaule Van Zyl
calcicola Van Zyl ined.
fulvum L.
porphyrocaule Van Zyl ined.
swartbergense Van Zyl ined.
fuscatum Van Zyl
flexuosum Eckl. & Zeyh.
Marais 455 (STE)
van Zyl 3894 (STE)
Dreyer s.n. (STE)
van Zyl 4605 (STE)
Bellstedt 800 (STE)
Bellstedt 798 (STE)
Bellstedt 892 (STE)
Bellstedt 794 (STE)
Lutzville District, WC
Witputz district, Namibia
Stilbaai, South Africa
Franschhoek Pass, WC
Hex River Pass, South Africa
Swartberg Pass, Prince Alfred district, South Africa
Betty’s Bay, South Africa
Langebaan, South Africa
EF 656024
*AJ387972
EF 656030
EF 656044
EF 656018
EF 656043
EF 656045
EF 656032
nd
*AJ133869
nd
nd
EF655992
EF655996
nd
EF655995
Z. lichtensteinianum Cham.
NC, WC & EC
Van Zyl 4594 (STE)
Laingsburg district, WC
EF 656020
nd
Z. incrustatum E.Mey. ex Sond.
Z. maritimum Eckl. & Zeyh.
Z. aff. maritimum
Z. debile Cham.
Z. cretaceum Van Zyl ined.
Z. foetidum Schrad. & J.C.Wendtl.
Bellstedt 509 (STE)
Dold 4656 (GRA)
Dold 4654 (GRA)
Bellstedt 796 (STE)
Bellstedt 856 (STE)
Marais 423 (STE)
Beaufort West, South Africa
Coega River Mouth, South Africa
Somerset Heights, Grahamstown, South Africa
20 km N of Oudtshoorn, South Africa
Aussenkehr, Namibia
Robertson, South Africa
EF
EF
EF
EF
EF
EF
656019
656035
656034
656041
656028
656039
nd
nd
nd
nd
nd
nd
Z. macrocarpon Retief
Z. maculatum Aiton
Z. schreiberanum Merxm. & Giess
Z. leptopetalum E.Mey. ex Sond.
Z. pubescens Schinz
Z. leucocladum Diels in Schultze
NC, WC & EC
Coastal WC & EC
Coastal WC & EC
WC & EC
SN & NC
Inland NC, WC &
EC
SN & NC
Inland WC
SN
SN, NC & WC
SN & NC
SN
not collected
Marais 433 (STE)
Bellstedt 871 (STE)
Marais 422 (STE)
Bellstedt 881 (STE)
van Zyl 4479 (STE)
SN
Hartnekskloof, Ceres Karoo, South Africa
20 km N of Rosh Pinah, Namibia
Klawer, South Africa
Tiras Farm, Aus district, Namibia
Witputz district, Namibia
nd
EF 656033
EF 656027
EF 656040
EF 656042
EF 656029
nd
nd
nd
nd
nd
nd
Z. morgsana L.
NC, WC & EC
Bellstedt 890 (STE)
Steinkopf, South Africa
EF 656021
EF655994
Augea capensis Thunb.
Bulnesia arborea Engl.
Fagonia cretica L.
Fagonia indica Burm.f.
Fagonia luntii Baker
Guaiacum guatemalense Planch.
Rydb. & Vail
Larrea tridentata Coult.
Seetzenia lanata (Willd.) Bullock
Inland WC
New World
North Africa
Horn of Africa
Horn of Africa
New World
Bellstedt 934 (STE)
Chase 641 (K)
Chase 3432 (K)
Collenette 10/93 (K)
Wieland 4504 (K)
Chase 640 (K)
Ceres Karroo, WC
EF 655998
*AJ387947
*AJ387942
*AJ387943
*AJ387944
*AJ387948
EF655978
*Y15017
*AJ133855
*Y15018
*AJ133856
*Y15019
New World
Koue Bokkeveld,
SA
Mongolia
Africa
Asia
Asia
Australia
Australia
Chase 636 (K)
Herman 3964 (K)
*AJ387951
*AJ387956
*Y15022
*Y15025
Sheahan 1994 (K)
Collenette 3/93 (K)
Chase 516 (K)
Chase 1700 (K)
Chase 2204 (K)
SR 417 Adelaide Bot
Gardens
Chase 2203 (K)
Thulin et al. 9012
(UPS)
Thulin et al. 8428
(UPS)
Ryding 1347 (K)
>Thulin et al. 7977
(UPS)
*AJ387959
*AJ387961
*AJ387968
*AJ387975
*AJ387970
*AJ387964
*Y15027
*Y15028
*Y15030
*AJ133872
*AJ133867
*AJ133862
*AJ387969
*AJ387971
*AJ133866
*AJ133868
*AJ387973
*AJ133870
*AJ387965
*AJ387963
*AJ133863
*AJ133861
Additional species
Tetraena mongolica Maxim.
Tribulus macropterus Boiss.
Z. fabago L.
Z. xanthoxylum Engl.
Z. glaucum F.Muell.
Z. billardieri DC.
Z. fruticulosum DC.
Melocarpum hildebrandtii (Engl.)
Beier & Thulin
Melocarpum robecchii (Engl.)
Beier & Thulin
Z. coccineum L.
Z. album L.f.
Australia
Horn of Africa
Horn of Africa
Middle East
Middle East
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
Coastal WC
SN
Coastal WC
Coastal WC
Inland WC
Inland WC
Coastal WC
Coastal WC
Z.
Z.
Z.
Z.
Z.
Z.
Z.
Z.
Abbreviations used: Namibia (N), Northern Namibia (NN), North Western Namibia (NWN), Southern Namibia (SN), South Africa (SA) South African provinces: Eastern Cape (EC), Northern Cape (NC), Western Cape (WC). Sequences
downloaded from GenBank are indicated with an *, nd = not determined. The sequences of Z. stapffii were found to be identical to those of Z. orbiculatum and are indicated with an **.
935
936
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
2.3. Molecular data
DNA was extracted using the CTAB method of Doyle and Doyle
(1987). For a few taxa, the extraction was done from fresh plant
material and for the remaining taxa silica gel dried material was
used. Polymerase chain reaction (PCR) amplification of the trnL
intron and the trnLF spacer region was achieved using the c and f
primers of Taberlet et al. (1991). PCRs were performed in a Hybaid
Thermal Cycler (Thermo Electron Corporation, Waltham, MA, USA)
in a total volume of a 100 ll containing 2.5 mM MgCl2, 1 JMR-455
buffer (Southern Cross Biotechnology, Cape Town, RSA), 1 U of
Super-Therm Taq polymerase (Southern Cross Biotechnology, Cape
Town, RSA), 200 lM of each of the dNTP’s and 0.5 lM of each primer. Amplification profiles were 35 cycles with 1 min denaturation
at 94 °C, 1 min annealing at 55 °C, 90 s extension at 72 °C, followed
by a final extension step of 6 min at 72 °C. The amplified products
were separated by agarose gel electrophoresis and purified from
excised gel slices using the Wizard PCR Prep kit (Promega Corp.,
Madison, USA). Sequences were obtained by cycle sequencing using
the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems Inc., Foster City, USA) using primers c and f and in some
instances d and e (Taberlet et al. 1991), when polyA runs gave
ambiguous base calling. Cycle sequencing was done in 10 ll reactions consisting of an estimated 100 ng of DNA, 2 ll 5 buffer
(400 mM Tris–HCl, 10 mM MgCl2 at pH 9), 3.2 pmol primer, 2 ll
of Terminator Ready Reaction Mix and water. The cycle sequencing
profile was 35 cycles consisting of 10 s at 96 °C, 30 s at 52 °C and
4 min at 60 °C. Excess terminator dye was removed by gel filtration
through Centri-Sep 96 Multi-well Filter Plates (Princeton Separation, Adelphia, USA). Sequencing reactions were subsequently analyzed on an ABI 377 sequencer (Applied Biosystems Inc., Foster City,
USA). Electropherograms were edited using Chromas v1.45 (Technelysium Pty., Tewantin, Australia). Any samples giving sequencing
ambiguities were resequenced until unequivocal sequences were
obtained. Sequences were imported into DAPSA (Harley, 1998),
combined with 21 Zygophylloideae trnLF sequences generated by
Sheahan and Chase (2000) (indicated in Table 2) and aligned. The
aligned data matrix was imported into BioEdit v7.0.1 (Hall, 1999),
which was used to generate a Nexus file. All gaps, polyA rich areas
and areas that could not be aligned unambiguously were excluded
from the matrix. These were the 192 bp deletion in the trnL region
identified by Sheahan and Chase (1996) which is not present in the
zygophyllid clade, and four further regions, one which was in the
trnL intron and was 15 bp in length, and three regions which were
67, 21 and 19 bp, respectively, in length in the trnLF spacer.
The plastid rbcL region of a subset of 19 Zygophyllum species as
indicated in Table 1 was amplified by PCR using the primers 1F or
32F and 1460R (Lledó et al., 1998). Polymerase chain reactions
(PCR) were performed in a Hybaid Thermal Cycler (Thermo Electron
Corporation, Waltham, MA, USA) in a total volume of a 100 ll containing 2.5 mM MgCl2, 1 JMR-455 buffer (Southern Cross Biotechnology,
Cape Town, RSA), 1 U of Super-Therm Taq polymerase (Southern Cross
Biotechnology, Cape Town, RSA), 200 lM of each of the dNTP’s and
0.5 lM of each primer. Amplification profiles were 30 cycles with
30 s denaturation at 94 °C, 30 s annealing at 50 °C, 60 s extension at
72 °C, followed by a final extension step of 6 min at 72 °C. Sequencing
was performed using the primers 1F, 636F and 1460R as described
above. Sequence alignment was performed using BioEdit with the
rbcL sequences of 23 species which Sheahan and Chase (2000) had
generated (indicated in Table 2). A reduced trnLF alignment, in which
only these taxa were included, was also generated.
2.4. Phylogenetic analysis
All data matrices were analyzed using the parsimony option in
PAUP 4.0b10 (Swofford, 2002), and all substitutions were weighted
equally. The combined trnL intron and trnLF spacer data matrix, the
rbcL data matrix and the combined trnL intron, trnLF spacer and
rbcL data matrices were analyzed treating all missing and gap characters as missing data. In each analysis, 1000 replicates were run
using TBR branch swapping, holding 10 trees with MulTrees on
and Maxtrees was set to 10,000. Clade support was calculated with
1000 bootstrap replicates using TBR branch swapping and MulTrees on. Bootstrap percentages P75% were considered as wellsupported.
2.5. Bayesian analysis
MrBayes 3.1.2 (Huelsenbeck and Ronquist, 2001; Ronquist and
Huelsenbeck, 2003) was used for a Bayesian approach. The two
gene regions (trnLF, rbcL) were treated as separate partitions. The
General Time Reversible model of nucleotide distribution with
gamma shape parameter and a proportion of invariant sites
(GTR + I + C) was selected for each partition with the help of
MrModeltest v2.2 (Nylander, 2004). The chains were run for 2.5
million generations and sampled every 100 generations. To check
that the log-likelihood distribution had become stationary, the
fluctuating values of the log-likelihood were plotted. The burn in
was estimated empirically on the basis of these plots. 11,000 samples (44% of the run) were discarded. Swapping among chains and
acceptance of proposed changes to model parameters were monitored to ensure that efficient mixing had occurred. Under the
default 0.2 temperature parameter for heating the chains, swapping of chains proved to be below the recommended value of
10% and we therefore lowered this value to 0.15 to get an acceptable rate (>10%) of chain swapping. Only posterior probabilities
P95 were considered statistically significant support.
2.6. Morphological data
Van Zyl (2000), in her revision of the southern African members
of genus Zygophyllum, evaluated morphological attributes of stems,
flowers, sepals, nectar discs, seeds and seed mucilage of all species.
Since fresh material for Z. orbiculatum (native to Angola) was not
available, she compiled the morphological characters for this species from descriptions by previous authors (Van Huyssteen,
1937) and the scant herbarium material of which the flowers were
in a poor condition. She followed Van Huyssteen (1937) and placed
Z. orbiculatum in section Paradoxa because of its unifoliolate leaves,
yet concluded that the position of Z. orbiculatum was uncertain because of a lack of floral details. However, the availability of recently
collected fresh material allowed us to reassess some of the morphological characters of Z. orbiculatum. Fresh material of Augea capensis was also examined, and leaf morphological characters were
reassessed.
The morphological character assessment of non-southern African species was based on Beier et al. (2003).
Patterns of the evolution of capsule dehiscence, seed attachment and seed mucilage were investigated by ancestral character-state optimizations using the parsimony criterium in
Mesquite (Maddison and Maddison, 2006). Characters were traced
onto the Bayesian inference tree with the highest likelihood score
of the combined trnLF and rbcL matrix.
3. Results
3.1. Molecular data
The trnL intron and the trnLF could be sequenced without problems. Despite slippage in the sequencing signal induced by polyArich regions in the sequences of some taxa, the sequences could be
fully interpreted. Thus complete sequences of these non-coding
937
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
Table 3
Tree statistics of the full trnLF data set
Gene region
Length of aligned
matrix
Phylogenetically
informative characters
Phylogenetically
uninformative characters
Tree
length
CI
RI
Nodes with P 75
bootstrap support
trnL intron including the
trnL 3’ exon
trnLF spacer (excluding
Augea)
682
126
68
390
0.679
0.882
20/72
345
105
57
320
0.741
0.884
18/71
1027
231
125
718
0.699
0.878
30/72
Full trnLF region
regions could be generated for all of the taxa with the exception
of A. capensis. For many species, more than one specimen was
sequenced and little intraspecies sequence variation was observed
even within the highly variable polyA regions (data not shown).
Like Beier et al. (2003) we were also only able to sequence the trnL
intron of A. capensis and not the trnLF intron. By using additional
newly synthesized primers that matched downstream conserved
areas, we deduced that the trnLF spacer and the trnF gene must
have been deleted from the plastid genome of A. capensis. Thus in
the trnLF alignment matrix, only the trnL intron sequence of A. capensis could be used for phylogenetic reconstruction, and the trnLF
region was treated as missing data. The rbcL gene of the reduced
taxon set could be sequenced and the electropherograms interpreted without problems.
Sequence comparisons of the trnL, trnLF and rbcL gene regions
of two specimens of Z. orbiculatum, collected in southern Angola,
revealed that they were identical to the distinctive trnL, trnLF
and rbcL sequences of Z. stapffii collected in the vicinity of Swakopmund in Namibia. Z. stapffii was described as a species with bifoliolate leaves and Z. orbiculatum as having unifoliolate leaves. Recent
fieldwork has revealed that Z. orbiculatum in southern Angola discards one of each of the pairs of bifoliolate leaves during periods of
drought, which gives the impression that it is unifoliolate (Craven,
personal communication). Schreiber (1963) and Van Zyl (2000)
expressed their doubts about the systematic position of Z. orbiculatum, and although no flowers were available in the field during
January 2006, the vegetative material and gene sequences strongly
suggest that Z. orbiculatum and Z. stapffii are conspecific. Since the
name Z. orbiculatum (published in 1868) has priority over the name
Z. stapffii (published in 1888), we will use the name Z. orbiculatum
hereafter.
The tree statistics of the parsimony analyses of the trnL intron
(including trnL exon 2) and trnLF spacer data matrix are shown
in Table 3. Although the results of these analyses are not shown,
we analyzed the trnL intron data (i.e. including A. capensis) alone
and the trnLF spacer region alone (i.e. excluding A. capensis because
it does not possess a trnLF spacer region) and assessed the number
of nodes with P75% bootstrap support in these separate analyses
in comparison with the combined trnL intron and trnLF spacer
region analysis. In a phylogenetic analysis of n taxa, the number
of nodes in a completely resolved phylogeny is n 1. In Table 3,
the number of nodes with P75% bootstrap support is therefore
expressed as a fraction of the potential number of nodes in the
phylogeny. This allows a comparison of the informativeness of
the phylogeny based on the respective gene region and shows that
the combined trnL intron and trnLF spacer region give significantly
greater resolution than the trnL intron only, which has been used in
previous phylogenetic analyses (Beier et al., 2003). The parsimony
analysis of the combined trnL intron and trnLF spacer region
sequences retrieved >10,000 trees, yet due to the tree limit that
was set, only 10,000 were used to compute a strict consensus tree.
An examination of the trees generated showed that the large number of trees retrieved in the heuristic search was the result of the
many alternative trees generated among the members of the
southern African subgenus Zygophyllum due to the short branch
lengths in this clade. When compared to the parsimony strict consensus tree, the topology of the Bayesian tree resolved three more
nodes, but these were not significantly supported. The strict consensus tree and one of the shortest trees of the parsimony analysis
of the combined trnLF data set is shown in Fig. 1a and b,
respectively.
The tree statistics of and the proportion of supported nodes in
the parsimony analyses of the rbcL data matrix, the corresponding
reduced trnLF data matrix and the combined data matrix are
shown in Table 4. The parsimony analysis of the rbcL gene
retrieved 10 trees which were used to compute a strict consensus
tree. When compared to the parsimony strict consensus tree, the
topology of the Bayesian tree resolved one more node, but this
was not significantly supported. One of the shortest trees of the
parsimony analysis of the rbcL data set is shown in Fig. 2. The parsimony analysis of the reduced trnLF data matrix retrieved 28 trees
which were used to compute a strict consensus tree. When compared to the parsimony strict consensus tree, the topology of the
Bayesian tree resolved two more nodes, but these were not significantly supported. These trees are not shown. The parsimony analysis of the combined rbcL and trnLF dataset retrieved 37 trees
which were used to compute a strict consensus tree. The Bayesian
50% majority rule consensus tree resulting from the combined rbcL
and trnLF analysis and one of the shortest trees of the parsimony
analysis are shown in Fig. 3a and b, respectively. The Bayesian
Inference analysis for a double run (Nruns=2) of the combined rbcL
and trnLF dataset gave an average standard deviation of split frequencies of 0.005967 after 2.5 million generations. When compared to the parsimony strict consensus tree, the topology of the
Bayesian tree resolved the basal polytomy in the Zygophylloideae
into three deeper nodes, but these were not significantly
supported.
A comparison of the trees generated with the complete trnLF
data matrix, the reduced rbcL data matrix and the combined trnLF
and rbcL data matrix, all revealed the following groupings: (a) A
strongly supported monophyletic subfamily Zygophylloideae with
a paraphyletic genus Zygophyllum, in which the genera, Augea and
Tetraena were nested, was retrieved sister to a monophyletic subfamily Larreoideae. (b) Four strongly supported monophyletic
groups comprised of (i) subgenus Agrophyllum, (ii) genus Melocarpum (Engl.) Beier & Thulin and the genus Fagonia, (iii) southern
African and Australian members of subgenus Zygophyllum and
(iv) A. capensis and Z. orbiculatum. The Asian members of subgenus
Zygophyllum, Z. fabago and Z. xanthoxylum, were retrieved as a
monophyletic group in the reduced rbcL and combined trnLF and
rbcL data matrix, but were not retrieved as a monophyletic group
in the trnLF analysis.
When the topologies of the trees generated with the full trnLF,
the reduced rbcL and the combined trnLF and rbcL data matrices
were compared, different topologies of the five above mentioned
groupings were apparent. These topologies were, however, not
supported in any of the analyses, and hence no hard incongruences
between the trnLF, the rbcL and the combined trnLF and rbcL topol-
938
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
100/100
85/65
94/100
76/68
100/100
92/100
64/60
100/100
92/100
100/100
99/100
80/100
90/100
86/100
74/88
99/100
96/100
99/100
99/100
64/80
99/100
95/100
100/100
79/73
63/94
80/100
97/100
94/100
76/100
96/100
95/100
95/100
90/100
77/100
Tribulus
Seetzenia
Larrea
Bulnesia
Guaiacum
Z. xanthoxylum
Z. fabago
Augea capensis
Z. orbiculatum
M. hildebrandtii
M. robecchii
F. cretica
F. luntii
F. indica
Z. giessii
Z. longicapsulare
Tetraena mongolica
Z. microcarpum
Z. rigidum
Z. coccineum
Z. album
Z. simplex NA
Z. simplex SA
Z. inflatum
Z. spongiosum
Z. clavatum
Z. decumbens NA
Z. decumbens SA
Z. pterocaule
Z. patenticaule
Z. prismatocarpum
Z. applanatum
Z. chrysopteron
Z. retrofractum
Z. segmentatum
Z. turbinatum
Z. cylindrifolium
Z. tenue
Z. porphyrocaule
Z. incrustatum
Z. lichtensteinianum
Z. morgsana
Z. hirticaule
Z. schreiberanum
Z. cordifolium
Z. fusiforme
Z. cretaceum
Z. leucocladum
Z. glaucum
Z. billardieri
Z. fruticulosum
Z. botulifolium
Z. cuneifolium
Z. teretifolium
Z. calcicola
Z. divaricatum
Z flexuosum
Z. maculatum
Z. aff. maritimum
Z. maritimum
Z. namaquanum
Z. rogersii
Z. spinosum
Z. foetidum
Z. leptopetalum
Z. debile
Z. pubescens
Z. swartbergense
Z. fulvum
Z. fuscatum
Z. pygmaeum
Z. sessilifolium
Z. spitskopense
Tribuloideae
Seetzenioideae
Larreoideae
Asian species
SA species
§Grandifolia
Horn of Africa
species
Fagonia
§Cinerea
Asian species
§Alata
Asian species
§Annua
§Bipartita
§Prismatica
§Bipartita
§Capensia
§Morgsana
§Capensia
§Paradoxa
§Capensia
Australian
species
§Capensia
Fig. 1. (a) The strict consensus tree of the parsimony analysis of the trnL intron and trnLF spacer sequence data. Bootstrap percentages followed by Bayesian posterior
probabilities are shown below branches. The sections identified by Van Zyl (2000) to which the southern African species belong, the area of origin of non-southern African
species or the outgroup subfamilies are indicated next to the relevant species. (b) One of the shortest trees of the parsimony analysis of the trnL intron and trnLF spacer data.
Branch lengths are shown above branches. The main groupings within the subfamily Zygophylloideae are indicated: subgenus Agrophyllum; southern African (SA) and
Australian subgenus Zygophyllum species; Asian subgenus Zygophyllum species; genus Melocarpum and genus Fagonia; and Z. orbiculatum and Augea capensis. NA = northern
Africa, SA = southern Africa.
ogies were found. However, if the proportion of the nodes supported in the reduced trnLF, the reduced rbcL and the combined
trnLF and rbcL data matrix were compared, it became apparent that
the combined trnLF and the rbcL sequence data did not result in a
marked increase in resolution and node support of these groupings
relative to one another.
939
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
b
29
39
Tribulus
Seetzenia
32
49
Larrea
Bulnesia
Guaiacum
14
Augea capensis
21
7 Z. orbiculatum
9
M. hildebrandtii
14
M. robecchii
19
12 F. cretica
22
14
F. luntii
10
10
F. indica
12
Z. xanthoxylum
5 Z. fabago
6
8
Z. giessii
17
3
Z. longicapsulare
Tetraena mongolica
5
Z. microcarpum
11
Z. rigidum
Z. coccineum
5
Z. album
9
Z. simplex NA
5 Z. spongiosum
4 Z. simplex SA
4 Z. inflatum
7
4 Z. clavatum
4 Z. decumbens NA
11
4 Z. decumbens SA
3 Z. pterocaule
Z. patenticaule
Z. prismatocarpum
Z. turbinatum
Z. segmentatum
3 3 Z. applanatum
Z. chrysopteron
Z. retrofractum
3 Z. cylindrifolium
7
4 Z. tenue
4
Z. glaucum
6
Z. billardieri
8
4 Z. fruticulosum
Z. incrustatum
3 Z. lichtensteinianum
Z. morgsana
3 Z. cordifolium
Z. fusiforme
Z. teretifolium
Z. cuneifolium
Z. botulifolium
Z. hirticaule
Z. schreiberanum
Z. cretaceum
Z. leucocladum
Z. porphyrocaule
6 Z. leptopetalum
4
Z. pubescens
Z. debile
3
Z. swartbergense
Z. maculatum
5 Z. foetidum
Z. rogersii
Z. divaricatum
Z. namaquanum
Z. maritimum
Z. aff. maritimum
4 Z. spinosum
Z. calcicola
Z. flexuosum
Z. fuscatum
4 Z. fulvum
Z. pygmaeum
Z. sessilifolium
Z. spitkopense
11
11
41
38
15
23
12
5 changes
Fig. 1 (continued)
Augea &
Z. orbiculatum
Melocarpum
& Fagonia
Asian
Zygophyllum
Agrophyllum
SA & Australian
Zygophyllum
940
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
Table 4
Tree statistics of the reduced trnLF and rbcL data set
Gene
region
Length of
aligned matrix
Phylogenetically
informative characters
Phylogenetically
uninformative characters
Tree
length
CI
RI
Nodes with P 75
bootstrap support
trnLF
region
rbcL
947
221
113
650
0.728
0.839
26/42
1364
196
81
548
0.595
0.807
18/42
Combined
2311
417
194
1211
0.658
0.816
22/42
37
34
Tribulus
Seetzenia
13
31
100/100
Larrea
15
8
Bulnesia
23
Guaiacum
3 Z. rigidum
* Z. microcarpum
5
7
Tetra mongolica
78/
2
Z. album
2
100 3
* 3 Z. coccineum
6
Z. giessii
11
3
6
*
Z. longicapsulare
5
Z.
simplex
NA
7
5 2
Z. simplex SA
*
*
23
89/100
4
Z. patenticaule
4
Z. prismatocarpum
6
*
4
5
*
33
100/100
2
2
16
100/100
15
97/100
5
- /62
6
71/
100
Z. applanatum
2
Z. decumbens NA
Z. decumbens SA
6
Z. cylindrifolium
4
Z. clavatum
2
Z. segmentatum
2
M. robecchii
M. hildebrandtii
13
F. cretica
4
F. indica
13
*
F. luntii
13
100/100
25
11
86/100
6
- /83
Agrophyllum
Z. spongiosum
13
11
14
3
- /93
Augea capensis
Z. orbiculatum
Z. xanthoxylum
Z. fabago
2
Z. porphyrocaule
Melocarpum
& Fagonia
Augea &
Z. orbiculatum
Asian
Zygophyllum
Z. hirticaule
5
Z. swartbergense
21
100/100
Z. cordifolium
Z. morgsana
2
2 Z. flexuosum
* Z. sessiliflora
SA & Australian
Zygophyllum
Z. glaucum
2 Z. billardieri
5
Z. fruticulosum
5 changes
Fig. 2. One of the shortest trees of the parsimony analysis of the rbcL sequence data. Branch lengths are shown above branches. Branches that collapse in the strict consensus
are indicated with an arrow. Bootstrap percentages followed by Bayesian posterior probabilities are shown below branches. Where there was insufficient space to indicate
high branch support (P75% BS and P95% PP) this was indicated with an *. The main groupings within the subfamily Zygophylloideae are indicated: subgenus Agrophyllum;
southern African (SA) and Australian subgenus Zygophyllum species; Asian subgenus Zygophyllum species; genus Melocarpum and genus Fagonia; and Z. orbiculatum and Augea
capensis. NA = northern Africa, SA = southern Africa.
The parsimony analysis of the combined trnL intron and trnLF
spacer region retrieved some of the sections recognized by Van
Zyl (2000) as monophyletic. In subgenus Agrophyllum, monophy-
letic sections Cinerea, Alata, Annua and Bipartita were retrieved.
Although section Prismatica was retrieved as monophyletic in
the parsimony analysis, it nests within section Bipartita, render-
941
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
a
Tribulus
Seetzenia
Larrea
100/100
86/99
Bulnesia
Guaiacum
100/100
99/100
M. robecchii
M. hildebrandtii
99/100
F. cretica
100/100
100/100
F. indica
Melocarpum
& Fagonia
F. luntii
Augea capensis
99/100
100/100
Z. orbiculatum
Z. xanthoxyllum
98/100
Z. fabago
Zygophyllum
Z. glaucum
97/100
*/87
Augea &
Z. orbiculatum
99/100
*/63
Z. billardieri
Z. fruticulosum
100/100
Z. porphyrocaule
Z. hirticaule
63/99
70/71
63/99
Z. cordifolium
SA & Australian
Zygophyllum
Z. morgsana
*/70
Z. swartbergense
100/100
95/100
Z. flexuosum
Z. sessilifolium
100/100
Z. rigidum
Z. microcarpum
92/100
Tetreana mongolica
*/98
99/100
Z. album
Z. coccineum
98/100
Z. giessii
100/100
Z. longcapsulare
Z. simplex N
100/100
61/99
51/76
Z. simplex SA
Z. spongiosum
100/100
100/100
Agrophyllum
Z. patenticaule
Z. prismatocarpum
100/100
Z. clavatum
*/96
77/97
*/55
Z. decumbens N
Z. decumbens SA
Z. applanatum
52/95
52/93
Z. segmentatum
Z. cylindrifolium
Fig. 3. (a) The Bayesian 50% majority rule consensus tree of the combined trnLF and rbcL spacer sequence data. Parsimony bootstrap percentages followed by Bayesian
posterior probabilities are shown below branches. Where nodes collapsed in the parsimony strict consensus tree, but were supported in the Bayesian analysis, they were
indicated with an asterisk. The main groupings within the subfamily Zygophylloideae are indicated: subgenus Agrophyllum; southern African (SA) and Australian subgenus
Zygophyllum species; Asian subgenus Zygophyllum species; genus Melocarpum and genus Fagonia; and Z. orbiculatum and Augea capensis. (b) One of the shortest trees of the
parsimony analysis of the combined trnLF and rbcL spacer sequence data. Branch lengths are shown above branches. The areas in which the species occur are indicated with
bars; NA = Northern Africa; SA = Southern Africa.
ing it paraphyletic. In the Bayesian analysis, section Prismatica
was retrieved in a polytomy with members of section Bipartita,
thereby leaving the question of the paraphyly of section Bipartita
unconfirmed. In subgenus Zygophyllum, a monophyletic section
Paradoxa was retrieved but with the exclusion of Z. orbiculatum.
Z. orbiculatum as the sole representative of section Grandifolia
was retrieved in an isolated position sister to A. capensis supporting its sectional status. The parsimony analysis of the combined trnL intron and trnLF spacer region did not retrieve Van
Zyl’s (2000) sections Capensia and Morgsana as monophyletic
nor were the series she recognized within Capensia supported.
The parsimony analyses of the reduced rbcL and combined trnLF
and rbcL matrices, also retrieved the sections Cinerea, Alata, An-
nua, Bipartita, Prismatica and Grandifolia as monophyletic, but
section Prismatica was retrieved as part of a polytomy of members of section Bipartita in the combined trnLF and rbcL analysis
leaving the monophyly of section Bipartita unconfirmed. Other
groupings of Zygophyllum species occurring outside southern
Africa which were also retrieved as monophyletic in all of these
analyses were the Australian species belonging to section Roepera and the two north African and Asian species, Z. album and
Z. coccineum. The grouping of Van Zyl’s section Alata with the
two north African and Asian species, Z. album and Z. coccineum,
and the Asian species, Tetraena mongolica, was retrieved as
monophyletic in the complete trnLF analysis and the combined
rbcL and trnLF analysis.
942
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
b
67
73
Tribulus
Seetzenia
47
74
Bulnesia
33
Guaiacum
3
M. robecchii
27
3
M. hildebrandtii
26
F. cretica
38
14
F. indica
22
15
F. luntii
40
Augea
capensis
31
19
Z. orbiculatum
23
Z. xanthoxyllum
8
21
Z. fabago
4
Z. glaucum
7
10
5
Z. billardieri
10
9
Z. fruticulosum
3
41
Z. porphyrocaule
25
64
Larrea
40
17
76
22
Horn of Africa
Africa, Asia,
New World
SA
Asian
Australian
Z. hirticaule
3
Z. cordifolium
6
5
Z. morgsana
8
Z. swartbergense
SA
3 Z. flexuosum
4
Z. sessilifolium
21
6 Z. rigidum
Z. microcarpum
6
9
Tetraena mongolica
5
Z. album
9
6
Z. coccineum
13
Z. giessii
34
9
Z. longicapsulare
7
Z. simplex NA
10 7
11
Z. simplex SA
10
17
Z. spongiosum
10
Z. clavatum
3 6
Z. decumbens NA
4
Z. decumbens SA
11 Z. patenticaule
7
Z. prismatocarpum
5
Z. applatum
3 4
3 Z. segmentatum
13
Z. cylindrifolium
SA
Asian
SA
NA
SA
SA
NA
SA
10 changes
Fig. 3 (continued)
3.2. Results of the morphological assessment
Van Zyl’s (2000) morphological comparison of the southern
African members of subgenera Zygophyllum and Agrophyllum to
that of Z. orbiculatum (=Z. stapffii) is shown in Table 5. It shows that
there are a number of synapomorphies which allow the delimitation of the two subgenera in this region. Z. orbiculatum was also
shown to share more characters with subgenus Agrophyllum than
with subgenus Zygophyllum, which was the basis of Van Zyl’s
placement of Z. orbiculatum in subgenus Agrophyllum. It does, however, also reveal that Z. orbiculatum displays a number of distinc-
tive autapomorphies (as compared to both subgenera), in
particular those of capsule dehiscence and seed attachment. Van
Zyl (2000) considered both of these characters as taxonomically
informative and used them to place Z. orbiculatum (=Z. stapffii) in
section Grandifolia.
Morphological characters used to define sectional boundaries of
southern African Zygophyllum taxa are shown in Table 6. Within
subgenus Agrophyllum, sectional boundaries were defined by synapomorphies but these were often homoplasious. Within subgenus
Zygophyllum, morphological characters were much more uniform,
with the exception of leaf morphology, which was used to define
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
section Paradoxa, and floral morphology, which was used to define
the tetramerous section Morgsana. In all other southern African
Zygophyllum species flowers were pentamerous. Apart from in section Morgsana, tetramerous flowers were also present in Australian
members of subgenus Zygophyllum (Beier et al., 2003).
The examination of fresh material of A. capensis revealed that
many of the leaves on a single plant were not unifoliolate as previously interpreted from herbarium material (Sheahan and Chase,
1996; Beier et al., 2003) but rather were attached to the stem in
asymmetric pairs to produce sessile bifoliolate rather than opposite
unifoliolate leaves. This interpretation would have been difficult
from dried herbarium material, but showed that A. capensis shares
this character with most other members of the Zygophylloideae.
3.3. Patterns of character evolution
Patterns of evolution of the characters of seed dehiscence, seed
mucilage and seed attachment compared by means of character
optimization on the tree retrieved with the highest likelihood score
in the Bayesian analysis are shown in Figs. 4–6, respectively. As
this tree retrieved a monophyletic subgenus Zygophyllum and a
monophyletic Agrophyllum, capsule dehiscence (loculicidal versus
septicidal), the character used in the definition of these subgenera,
was fully congruent with the tree topology. The character optimization illustrates how capsule dehiscence via dorsal and ventral
sutures which occurs in Z. orbiculatum does not optimize within
the two subgenera of Zygophyllum. The optimization of seed
attachment revealed that seed attachment via an aril only occurred
in the southern African and Australian members of subgenus Zygophyllum but not in the Asian members of subgenus Zygophyllum
which exhibited seed attachment via a long funiculum that also
occurred in the rest of subgenus Agrophyllum. Intermediate seed
attachment via a short thick funiculum was present in Z. orbiculatum, Melocarpum hildebrandtii and Melocarpum robecchii. The optimization of seed mucilage showed that members of subgenus
Zygophyllum occurring in southern African and Australia and A.
capensis possessed threads of uniform width in their mucilage.
Seed mucilage in Asian members of subgenus Zygophyllum, subgenus Agrophyllum and Z. orbiculatum, contained wineglass-shaped
threads.
943
4. Discussion
4.1. Molecular data
In this study, the trnL intron and the trnLF spacer sequences of
52 southern African Zygophyllum species were combined with
those published by Sheahan and Chase (2000) to produce a tree
of the southern African members of the genus. Both gene regions
contributed almost equally to the number of phylogenetically
informative characters and resulted in a tree in which 41% of the
nodes were well-supported. The inclusion of Z. orbiculatum changed the topology considerably in comparison to the one of Sheahan
and Chase (2000) and Beier et al. (2003), and also improved resolution. Poorly sampled groups i.e. those from outside of the southern African area, showed poor resolution. This analysis retrieved a
monophyletic subfamily Zygophylloideae, within which the genera
Fagonia, Melocarpum, Augea and Tetraena are nested rendering
Zygophyllum paraphyletic. Although Van Zyl (2000) transferred Z.
orbiculatum (=Z. stapffii) from subgenus Zygophyllum to subgenus
Agrophyllum, it was retrieved as sister to A. capensis. The analysis
revealed four main groupings: subgenus Agrophyllum; southern
African and Australian members of subgenus Zygophyllum; A. capensis and Z. orbiculatum; and genus Melocarpum and genus Fagonia.
However, the position of the Asian members of subgenus Zygophyllum, Z. fabago and Z. xanthoxylum, remained unresolved. The analysis supports the transfer of Z. morgsana from subgenus
Agrophyllum to subgenus Zygophyllum as proposed by Van Zyl
(2000). Within subgenus Agrophyllum, the molecular analysis
strongly support Van Zyl’s (2000) sections Cinerea, Alata, Annua
and Bipartita. Although section Prismatica was monophyletic, it
nested within section Bipartita. Within subgenus Zygophyllum, section Paradoxa was also retrieved but with the exclusion of Z. orbiculatum. The monophyly of sections Morgsana and Capensia could
not be confirmed. Van Zyl (2000) suggested a subdivision of section
Capensia into three series based on leaf morphology. This was not
supported by the present analyses. However, the molecular phylogeny does reveal a strongly supported Cape Floral Region (CFR)
clade. The lack of resolution in section Bipartita and in subgenus
Zygophyllum is due to a paucity of phylogenetically informative
characters in these clades.
Table 5
Morphological characters of southern African members of subgenera Agrophyllum and Zygophyllum and Z. orbiculatum (=Z. stapffii)
Subgenus Zygophyllum
Z. orbiculatum (=Z. stapffii)
Druse crystals absent in mesophyll
Druse crystals present in mesophyll
Subgenus Agrophyllum
Druse crystals present in mesophyll
Flowers large (petals 8–22 mm long 3–12 mm
wide), petals yellow, usually marked at base with
red or brown, with short claws
Flower size medium (petals 10–11 mm long 3–
4 mm wide), petals white, unmarked, obovate, with a
long claw
Flowers small (petals 2–10 mm long 0.5–3 mm wide),
usually white, rarely light yellow or orange, never
marked at base, with long claws
Sepals not articulate, not succulent
Sepals adnate at base, not articulate, leathery in
texture
Sepals articulate and usually succulent
Sepals persistent
Sepals persistent
Sepals not persistent
Nectar disc regularly angled never lobed
Nectar disc regularly angled, 10-lobed, lobes small
and directed out and downward, with nectaries
visible as groups of darker cells
Nectar disc angled and lobed, lobes arranged into pairs,
variously orientated
Nectar disc always papillate
Nectar disc smooth
Nectar disc smooth
Nectar disc uniformly level, not sloping or with raised
or sunken areas
Nectar disc sloping slightly towards its periphery
Nectar disc sloping towards its periphery, with raised
and sunken areas
Fruit a loculicidal capsule
Fruit dehiscing along both ventral and dorsal sutures
Fruit a septicidal capsule
Seeds oblong, not compressed, white aril present
Seeds sub-pyriform, compressed, funicula short and
thick
Seeds pyriform, sub-pyriform, compressed, funicula long
and narrow
Seed mucilage structured, with long spiral inclusions
of uniform width
Seed mucilage structured, with short spiral inclusions
that unravel at apex
Seed mucilage structured, with short spiral inclusions
that unravel at apex
Young stems usually flat on ventral side, with lateral
ridges, or round in cross section
Young stems with a flat ventral area but without
ridges
Young stems usually grooved
944
Table 6
Morphological character states used in the delimitation of sections of southern African Zygophyllum
Morphological characters
Indumentum
Leaf attachment &
number
Leaf shape
Stipules
Seed mucilage
Staminal scale morphology
Grandifolia
Glabrous
Bifoliolate, petiolate
Subrotund
Succulent, leathery, triangular or subrotund,
patent, 1 on ventral, 1 on dorsal side, caducous
Structured, short spiral
inclusions
Simple, apex acute, upper
margins lacerate
Pale cream, spathulate
Alata
White, two armed
hairs present or
glabrous
Bifoliolate, subpetiolate
or petiolate
Widely or narrowly obovate
Subulate, white, stiff, 2 on ventral, 2 on dorsal
side, semi-permanent
Structured, short spiral
inclusions
Simple , alternately
enfolding filament
Pale cream or pink,
spathulate
Cinerea
Silver-white, two
armed trichomes
Bifoliolate, petiolate
Orbiculate or obovate
Triangular or subulate?, reflexed and base
thickened, 2 on ventral, 2 on dorsal side,
caducous
Structured, short spiral
inclusions
Simple, apex acute, upper
margins lacerate
Petals narrowly obovate
or spathulate, cream or
pink
Annua
Glabrous
Simple, sessile
Succulent, obovoid or
globose
Triangular, membranous, apex sometimes
incused, 2 on ventral, 2 on dorsal side, semipermanent
Unstructured, jelly like
Biparted almost at base
Petals spathulate, white,
yellow or orange
Bipartita
Glabrous
Bifoliolate, petiolate
Cylindric, oblong, obovate
or subrotund
Widely triangular, membranous, 2 on ventral, 2
on dorsal side, semipermanent or caducous
Structured, short spiral
inclusions
Biparted almost at base
Pale cream, axillary
Prismatica
Glabrous
Simple, sessile
Suborbicular/obovate
Filamentous, minute, caducous, 2 on ventral, 2
on dorsal side
Structured, short spiral
inclusions
Biparted almost at bas
Pale cream, axillary
Paradoxa
Glabrous
Simple, with short
petiole or sessile
Subrotund or obovate
Navicular, membranous, reflexed, 1 on ventral,
1 on dorsal side, caducous
Structured, long spiral
inclusions of uniform
length
Simple, margins lacerate
Yellow marked at base
Capensia I
Glabrous
Bifoliolate, sessile
Terete or linear, adaxial
groove, sometimes
succulent
Triangular, 1 on ventral, 1 on dorsal side,
sometimes caducous
Structured, long spiral
inclusions of uniform
length
Simple, margins lacerate
Yellow (in one case
pink) marked at base
Capensia II
Glabrous
Bifoliolate, sessile
Obovate or cuneate
Triangular or semi-circular, 1 or 2 on ventral, 1
on dorsal side, sometimes caducous
Structured, long spiral
inclusions of uniform
length
Simple, margins lacerate
Yellow marked at base
Capensia II
Glabrous
Bifoliolate (trifoliolate
in Z. schreiberanum),
petiolate
Obovate or elliptic
Triangular , 1 or 2 on ventral, 1 or 2 on dorsal
side, sometimes caducous (in Z. schreiberanum
leaflike)
Structured, long spiral
inclusions of uniform
length
Simple, margins lacerate,
sometimes bordered with
papillae
Yellow marked at base
Morgsana
Glabrous
Bifoliolate, petiolate
Obovate or subrotund, apex
rounded, base narrow or
obtuse
Triangular, membranous, reflexed, 1 on ventral,
1 on dorsal side, caducous
Structured, long spiral
inclusions of uniform
length
Simple oblong or obovate,
margins lacerate
Yellow, marked at base,
4-merous
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
Section
945
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
(a) Fruit of Z. simplex,
subgenus Agrophyllum;
dehiscence septicidal
(b) Fruit of Z. porphyrocaule,
subgenus Zygophyllum;
dehiscence loculicidal
Tribulus
Seetzenia
Bulnesia
Guaiacum
Larrea
Augea capensis
Z orbiculatum
Z xanthoxyllum
Z fabago
Z billardieri
Z fruticulosum
Z glaucum
Z porphyrocaule
Z cordifolium
Z morgsana
Z hirticaule
Z flexuosum
Z sessilifolium
Z swartbergense
Z giessii
Z longicapsulare
Z simplex NA
Z simplex SA
Z spongiosum
Z clavatum
Z decumbens NA
Z decumbens SA
Z segmentatum
Z cylindrifolium
Z applanatum
Z patenticaule
Z prismatocarpum
Tetraena mongolica
Z album
Z coccineum
Z rigidum
Z microcarpum
M robecchii
M hildebrandtii
F cretica
F indica
F luntii
Augea &
Z. orbiculatum
Asian
Zygophyllum
SA & Australian
Zygophyllum
Agrophyllum
Melocarpum
& Fagonia
(c) Fruit of Z. orbiculatum;
dehiscence via dorsal
and ventral sutures
Fig. 4. Character evolution of fruit dehiscence as reconstructed on the tree retrieved with the highest likelihood score in the Bayesian analysis of the combined analysis of rbcL
and trnLF using Mesquite (Maddison and Maddison, 2006). Insert (a) illustrates the fruit of Z. simplex, typical of subgenus Agrophyllum, and a detached mericarp after
septicidal dehiscence. Insert (b) illustrates the fruit of Z. porphyrocaule ined. which dehiscences loculicidally. Insert (c) illustrates the fruit of Z. orbiculatum, which dehiscences
via dorsal and ventral sutures during dry weather conditions. The alternating exocarp and endocarp sections are shown with seeds still attached to the endocarp.
The rbcL phylogeny also retrieved the four main groups found in
the analysis of the trnL intron and trnLF spacer sequences and
added a fifth group of Asian members of subgenus Zygophyllum,
Z. fabago and Z. xanthoxylum. Although the relationships between
these groups were resolved, these were not supported, as was also
found by Sheahan and Chase (2000). The sister relationship between A. capensis and Z. orbiculatum was, however, confirmed,
which was important, as the trnLF spacer is absent in this taxon,
and therefore only the trnL data contributed to the combined trnLF
data set. The parsimony statistics for the rbcL results in comparison
to those of the trnLF phylogeny, revealed that homoplasy was high-
er in the rbcL matrix. The number of phylogenetically informative
characters in the rbcL matrix (196) was also lower than in the
reduced trnLF data matrix (221), which increased to 231 if all taxa
were included.
Although it may have been expected that the combined analysis could have retrieved a greater number of supported nodes
than the single matrices upon which it was based, this was
not the case and therefore reflects the increased amount of
homoplasy in the combined analysis. However, the combined
trnLF and rbcL phylogeny also retrieved the five main groupings
in the subfamily Zygophylloideae but the relationships between
946
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
(a)
aril attached
to an oblong seed
(b)
Long funicle
attached to pyriform seed
(c)
Short, thick funicle
attached to subpyriform seed
Tribulus
Seetzenia
Bulnesia
Guaiacum
Larrea
Augea capensis
Z orbiculatum
Z xanthoxyllum
Z fabago
Z billardieri
Z fruticulosum
Z glaucum
Z porphyrocaule
Z cordifolium
Z morgsana
Z hirticaule
Z flexuosum
Z sessilifolium
Z swartbergense
Z giessii
Z longicapsulare
Z simplex NA
Z simplex SA
Z spongiosum
Z clavatum
Z decumbens NA
Z decumbens SA
Z segmentatum
Z cylindrifolium
Z applanatum
Z patenticaule
Z prismatocarpum
Tetraena mongolica
Z album
Z coccineum
Z rigidum
Z microcarpum
M robecchii
M hildebrandtii
F cretica
F indica
F luntii
Augea &
Z. orbiculatum
Asian
Zygophyllum
SA & Australian
Zygophyllum
Agrophyllum
Melocarpum
& Fagonia
Fig. 5. Character evolution of seed attachment as reconstructed on the tree retrieved with the highest likelihood score in the Bayesian analysis of the combined analysis of
rbcL and trnLF using Mesquite (Maddison and Maddison, 2006). Insert (a) illustrates seed attachment via a long funiculum to a pyriform seed and insert (b) illustrates seed
attachment via an aril to an oblong seed. The third character state (c) of seed with short thick funicles attached to a sub-pyriform seed (in Z. orbiculatum, Melocarpum
hildebrandtii and Melocarpum robecchii) is not illustrated.
these were poorly supported. Thus the trnLF and rbcL data will
have to be augmented by additional sequence data from other
genes to establish the true relationships within the subfamily
Zygophylloideae.
4.2. Morphological assessment and character evolution
The morphological analysis performed by Van Zyl (2000) supported the division of southern African members of the genus
Zygophyllum into the two subgenera Agrophyllum and Zygophyllum based on capsule dehiscence, a classification which was contended by previous authors (Endlicher, 1841; Van Huyssteen,
1937). The dehiscence of the fruit of Z. orbiculatum (=Z. stapffii)
is neither loculicidal as in southern African members of the subgenus Zygophyllum, nor septicidal as in southern African members of the subgenus Agrophyllum, but dehisces via ventral and
dorsal sutures. Based on capsule dehiscence, it therefore appears
that Z. orbiculatum occupies an intermediate position, although
on the basis of other characters it seems to have closer affinities
with subgenus Agrophyllum. This was then also the reason why
Van Zyl (2000) transferred it from subgenus Zygophyllum to subgenus Agrophyllum. In this study, a resolved, but not supported
tree in which a monophyletic subgenus Agrophyllum and a
monophyletic subgenus Zygophyllum were retrieved, was used
947
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
(a)
long, uniform
width spirals in seed
mucilage
(b)
Short, unravelling
Wineglass-shaped spirals
in seed mucilage
(c)
No threads in seed mucilage
(d)
Seed mucilage under outer layer
Tribulus
Seetzenia
Bulnesia
Guaiacum
Larrea
Augea capensis
Z orbiculatum
Z xanthoxyllum
Z fabago
Z billardieri
Z fruticulosum
Z glaucum
Z porphyrocaule
Z cordifolium
Z morgsana
Z hirticaule
Z flexuosum
Z sessilifolium
Z swartbergense
Z giessii
Z longicapsulare
Z simplex NA
Z simplex SA
Z spongiosum
Z clavatum
Z decumbens NA
Z decumbens SA
Z segmentatum
Z cylindrifolium
Z applanatum
Z patenticaule
Z prismatocarpum
Tetraena mongolica
Z album
Z coccineum
Z rigidum
Z microcarpum
M robecchii
M hildebrandtii
F cretica
F indica
F luntii
Augea &
Z. orbiculatum
Asian
Zygophyllum
SA & Australian
Zygophyllum
Agrophyllum
Melocarpum
& Fagonia
Fig. 6. Character evolution of seed mucilage as reconstructed on the tree retrieved with the highest likelihood score in the Bayesian analysis of the combined analysis of rbcL
and trnLF using Mesquite (Maddison and Maddison, 2006). Insert (a) shows a photograph of the long, spirals of uniform width found in the mucilage and insert (b) shows a
photograph of short, unraveling wineglass-shaped spirals found in the mucilage. The third character state (c) exhibits mucilage in which no threads occur. The fourth
character (d) state exhibits seed mucilage under the outer cell layer which occurs only in Seetzenia. Mucilage characters were not scored for Tribulus.
to illustrate the evolution of fruit and seed characters within the
subfamily Zygophylloideae. As fruit dehiscence was used to define the subgenera it was therefore logical that the character
optimization of loculicidal and septicidal fruit dehiscence should
fit perfectly onto the monophyletic subgenera in this tree. All of
the molecular phylogenies retrieve Z. orbiculatum as sister to A.
capensis and not in subgenus Agrophyllum. Thus the character
optimization of fruit dehiscence of Z. orbiculatum also illustrates
the unique fruit dehiscence of Z. orbiculatum, supporting the suggestion that it should be placed in neither subgenus Agrophyllum
nor subgenus Zygophyllum.
Character optimization revealed that seeds attached via an
aril covering the hilum to an oblong seed only occur in the clade
containing southern African and Australian members of subge-
948
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
nus Zygophyllum but not in the clade consisting of the Asian
members of subgenus Zygophyllum, which rather share seed
attachment via a long funiculum with the rest of the recognized
subgenus Agrophyllum. A. capensis, genus Fagonia and the rest of
subgenus Agrophyllum possess a seed attachment via a long
funiculum to a pyriform seed. Van Zyl (2000) described the seed
attachment of Z. orbiculatum as a short and thick funiculum and
the seed shape as sub-pyriform. Beier et al. (2003), who examined the morphology of the two species from the horn of Africa,
M. hildebrandtii and M. robecchii, found that they also possess
sub-pyriform seeds attached via short funicula. Seed attachment
is therefore variable in these clades. Van Zyl (2000) suggested
that the short, thick funiculum of Z. orbiculatum may be an adaption to drought conditions. It would ensure attachment to the
xerochastic fruit capsule, which is dispersed by wind when dry
and which, when it becomes wet, retains moisture for seed germination. As Z. orbiculatum, M. hildebrandtii and M. robecchii occupy similar habitats, this may be a plausible explanation. In
contrast, the aril present in southern African and Australian
members of subgenus Zygophyllum suggests myrmecochorous
seed dispersal, which is a typical mechanism of seed dispersal
in the CFR (Bond and Slingsby, 1983).
Character optimization of seed mucilage revealed that short,
funnel-shaped, spiral inclusions are found in Z. orbiculatum (Van
Zyl 2000) and members of subgenus Agrophyllum (with the exception of section Annua which has mucilage that lack these structures). In subgenus Zygophyllum and A. capensis, mucilage has
long rod-shaped spiral inclusions, while in M. hildebrandtii and M.
robecchii and Fagonia the mucilage has no structure (Beier et al.,
2003). The sister clades of the Zygophylloideae, i.e. the Larriodeae,
exhibited a seed mucilage state without threads which appears to
be the plesiomorphic state and to which section Annua in subgenus
Agrophyllum apparently reverted to.
The members of the Z. orbiculatum and A. capensis clade and
the Fagonia and Melocarpum clade therefore share different combinations of capsule dehiscence, seed attachment and seed mucilage characters in comparison to the combination of these
characters in the members of subgenus Agrophyllum or subgenus
Zygophyllum. This can now be explained as sorting of distinctive
character combinations in the major clades of the Zygophylloideae in which different combinations of these characters occur
and is established here for the first time. The character optimizations also showed that the characters of seed mucilage and seed
attachment were synapomorphies of the two subgenera in
southern Africa, but do not hold for the members of the subgenera occurring outside this area. The taxonomic importance of
fruit dehiscence, which has for long been used in the delimitation of the subgenera Zygophyllum and Agrophyllum is, however,
confirmed.
A number of other morphological characters (leaf morphology, stipule morphology, trichome types, etc.), were phylogenetically not informative. Leaves, for example, were bifoliolate in
most members of subgenus Zygophyllum, in many members of
Agrophyllum and in Z. orbiculatum, but were unifoliolate in M.
hildebrandtii and M. robecchii, section Paradoxa of subgenus Zygophyllum, section Annua and section Prismatica of subgenus Agrophyllum, and uni- to trifoliolate and rarely up to 7-foliolate in
genus Fagonia and trifoliolate in Z. schreiberanum of subgenus
Zygophyllum. It was previously thought that A. capensis was unifoliolate, but our results have revealed that it is partially bifoliolate. Variation in leaflet number and shape appears to have
been selected for as an adaptation to drought. This hypothesis
is supported by the fact that species with simple leaves all occur
in areas of high aridity along the west coast of southern Africa
from about 150 km north of Cape Town to southern Angola,
never more than 200 km from the coast. In sections Paradoxa
and Prismatica these leaves were often conduplicate to additionally reduce moisture loss. This is also supported by the leaf
structure in Z. qatarense Hadidi (from the Middle East), which
is bifoliolate after rain but becomes unifoliolate during periods
of drought (Ismael, 1983). Species of Zygophyllum with bifoliolate
leaves have developed mechanisms to shed leaflets in response
to drought (Sheahan and Chase, 1996; Van Zyl, 2000, see also
earlier comments about Z. orbiculatum). Some species have also
developed trichomes to reduce moisture loss such as in Fagonia
(Beier et al., 2003) and section Cinerea (Van Zyl, 2000). This may
explain why leaf shape, leaflet number and several other morphological characters have been found to be largely uninformative in phylogenetic reconstructions (Beier et al., 2003), as they
indicate environmental adaptations rather than historical
relationships.
4.3. Historical biogeography
Climatologically, the Northern Cape Province (South Africa),
Namibia and southern Angola have been exposed to a general
aridification since the Miocene as a result of the establishment
of the cold Benguela current (Marlow et al., 2000; Linder,
2003; de Menocal, 2004). This would have led to extinction of
many plant species in this area, leaving only the most droughtresistant species, such as members of genus Zygophyllum and
genus Salsola. Currently, the narrow coastal strip in Namibia,
the slightly inland (±100 km) transfrontier region between South
Africa and Namibia in the Richtersveld (SA), Hunsberge Mountains (Namibia) and the lower Orange River course, which
receives a slightly higher precipitation than the surrounding
areas, constitutes a refugium for the few plants that can survive
such a harsh environment (Midgley et al., 2005). Moisture in this
area is restricted to the extremely scant rainfall and coastal fog,
but this is sufficient for the survival of many Zygophyllum species, as this is the center of diversity for subgenus Agrophyllum
in southern Africa (Van Zyl, 2000).
Drought resistance of species in subgenus Zygophyllum would
have allowed these species to survive in the Northern Cape, and
when the dry summers and wet winters established themselves
in the CFR, this climate change appears to have favoured a radiation of the subgenus into this area. Most of the members of subgenus Zygophyllum possess excellent drought resistance mechanisms
to survive the dry summers in the CFR. These include xerophytic,
often terete leaves and a flowering season in the winter and early
spring when moisture is still abundant. Seed attachment via a myrmecochoric aril which occurs commonly in the CFR, also appears to
have developed as an adaptation to survival in this area. As southern African members of subgenus Zygophyllum occur primarily in
the CFR, while the rest of the genus occurs outside of this area,
its evolution serves as an interesting example of the radiation of
a plant genus in the CFR. Although no attempts were made at dating our phylogeny, the long branch of the clade containing southern African and Australian members of subgenus Zygophyllum and
the short terminal branches within the subgenus, support a
hypothesis of relatively recent radiation of the subgenus in the
CFR. By contrast, the subgenus Agrophyllum, if gauged by the more
even and longer branch lengths between the nodes, is older, and
has shown a more constant rate of radiation outside of the CFR.
This concurs with the views of Engler (1896) and Van Huyssteen
(1937) who both considered subgenus Agrophyllum to be ancestral
in the genus. Interestingly, where species in section Bipartita have
also speciated into the CFR, the same pattern of short branch
lengths is found, possibly implying recent radiation.
In a phytogeographic study of the genus Fagonia, Beier et al.
(2004) concluded that this genus had evolved in the horn of
Africa because it is sister to M. hildebrandtii and M. robecchii,
D.U. Bellstedt et al. / Molecular Phylogenetics and Evolution 47 (2008) 932–949
and that the southern African members of genus Fagonia had migrated from the horn of Africa southward. This study reveals that
an ancestor of the rest of the Zygophylloideae may have evolved
in the horn of Africa and migrated from there to southern Africa
where many species of both subgenera of Zygophyllum have speciated in the Northern Cape Province, southern Namibia and
along the Namibian coast. Additionally the phylogeny shows evidence of repeated migrations of Zygophyllum species from southern Africa to the arid areas in the horn of Africa and Asia and
back. These migrations could have occurred during repeated
dry glacial periods (de Menocal, 2004) in which it is postulated
that a dry corridor existed from southern Namibia to the horn
of Africa (White, 1983).
4.4. Taxonomic implications
Sheahan and Chase (2000) found that Augea, Fagonia and Tetraena
were embedded in Zygophyllum, which therefore became a paraphyletic taxon. Accordingly, Beier et al. (2003) proposed a subdivision
for the Zygophylloideae based on morphological and trnL intron
data. The name Zygophyllum was retained for the clade that contains
the type species, Z. fabago, whereas other Zygophyllum species were
transferred to the genera Roepera and Tetraena; and Z. hildebrandtii
and Z. robecchii were transferred to Melocarpum. The genera Fagonia
and Augea, however, were retained. Beier et al. (2003) argued against
lumping all of the above genera into Zygophyllum because the Linnean names Zygophyllum and Fagonia had been in use since 1753, Augea since 1794 and Tetraena since 1889. Our findings support their
transfer of the horn of Africa species, Z. hildebrandtii and Z. robecchii
to Melocarpum. However, our evidence for a sister grouping of Z.
orbiculatum and A. capensis, does not support the renaming of Zygophyllum stapffii as Tetraena stapffii. Beier et al. (2003) did not examine
Z. orbiculatum (=Z. stapffii) but placed Z. stapffii in the newly circumscribed Tetraena on the basis of fruit dehiscence, which they indicated as a schizocarp. On the basis of the unique morphological
characters of Z. orbiculatum (=Z. stapffii), and its sister relationship
to Augea, we suggest that it ought to be put in a monotypic genus
(Marais, Bellstedt, Van Zyl and Craven, in prep.) being a true Kaokoveld endemic, and not, as previously classified (van Zyl, 2000), a
peculiar outlier of the subgenus Zygophyllum that occurs predominantly in the CFR (see Table 2).
Beier et al. (2003) advocated that the genus name Zygophyllum
be retained for a poorly supported clade (56% bootstrap value in
their trnL-based phylogeny) which includes Z. fabago, the type species of the genus, Z. xanthoxylum and a number of related species.
The evidence found in this study places doubt on assigning all
other Zygophyllum species to Tetraena and Roepera as our combined
rbcL and trnLF analysis retrieves a monophyletic grouping consisting of the Asian Zygophyllum species as well as the groupings renamed Tetraena and Roepera. In our opinion, the inclusion of
more taxa and more gene regions will be required to establish
the proper relationships between the members of the subfamily
Zygophylloideae, and until such time we advocate against changing the current taxonomy.
Acknowledgments
We express our sincere gratitude to Patricia Craven, Herta Kolberg, Colleen Mannheimer and Tony Dold for the collection of plant
material. Cape Nature, Northern Cape Department of Nature Conservation and Environmental Conservation, and the Namibian Ministry of Environment and Tourism are thanked for permission to
collect material. Funding for this research was provided by the
South African National Research Foundation and the University of
Stellenbosch.
949
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.ympev.2008.02.019.
References
Beier, B.A., Chase, M.W., Thulin, M., 2003. Phylogenetic relationships and taxomony
of subfamily Zygophylloideae (Zygophyllaceae) based on molecular and
morphological data. Plant Syst. Evol. 240, 11–39.
Beier, B.A., Nylander, J.A.A., Chase, M.W., Thulin, M., 2004. Phylogenetic
relationships and biogeography of the desert plant genus Fagonia
(Zygophyllaceae) inferred by parsimony and Bayesian model averaging. Mol.
Phylogenet. Evol. 33, 91–108.
Bond, W.J., Slingsby, P., 1983. Seed dispersal by ants in the shrublands of the Cape
Province and its evolutionary implications. S. A. J. Sci. 79, 231–233.
De Menocal, P.B., 2004. African climate change and faunal evolution during the
Pliocene–Pleistocene. Earth Planet. Sci. Lett. 220, 3–24.
Doyle, J.J., Doyle, J.L., 1987. A rapid DNA isolation procedure for small quantities of
leaf tissue. Phytochem. Bull. 19, 11–15.
Endlicher, S.L., 1841. Genera plantarum secundum ordines naturales disposita. Part
18: 1161. Fr. Beck, Vienna.
Engler, A. 1896. Über die geographische Verbreitung der Zygophyllaceen im
Verhältniß zu ihrer systematischen Gliederung. In: Abhandlungen Akademie
Berlin, pp. 1–36, plus 2 maps.
Engler, A., 1931. Zygophyllaceae, second ed.. In: Engler, A., Prantl, K. (Eds.), Die
Natürlichen Pflanzenfamilien, vol. 19 Engelmann, Leipzig, pp. 144–184.
Hall, T.A., 1999. BioEdit: a user-friendly biological sequence alignment editor and
analysis program for Windows 95/98/NT. Nucl. Acids Symp. Ser. 41, 95–98.
Harley, E.H., 1998. DNA and Protein Sequence Alignment (DAPSA). Computer
Program Distributed as Shareware by the Author, University of Cape Town, Cape
Town, South Africa.
Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogeny.
Bioinformatics 17, 754–755.
Ismael, A.M.A., 1983. Some factors controlling the water economy of Zygophyllum
qatarense growing in Qatar. J. Arid Environ. 6, 239–246.
Linder, H.P., 2003. The radiation of the Cape Flora, southern Africa. Biol. Rev. 78,
597–638.
Lledó, M.D., Crespo, M.B., Cameron, K.M., Fay, M.F., Chase, M.W., 1998. Systematics
of Plumbaginaceae based upon cladistic analysis of rbcL sequence data. Syst.
Bot. 23, 21–29.
Low, A.B., Rebelo, G., 1996. Vegetation of South Africa, Lesotho and Swaziland.
Department of Environmental Affairs and Tourism, Pretoria.
Maddison, W.P., Maddison, D.R., 2006. Mesquite: a modular system for evolutionary
analysis, Version 1.11. Available from: <http://mesquiteproject.org/>.
Marlow, J.R., Lange, C.B., Walter, G., Rosell-Melé, A., 2000. Upwelling intensification
as part of the Pliocene–Pleistocene climate transition. Science 290, 2288–2291.
Midgley, G.F., Reeves, G., Klak, C., Richardson, J., 2005. Late Tertiary and Quaternary
climate change and centers of endemism in the southern African flora. In:
Purvis, A., Gittleman, J., Brooks, T. (Eds.), Phylogeny and Conservation.
Cambridge University Press, Cambridge, pp. 230–242.
Nylander, J.A.A., 2004. MrModeltest v2. Program Distributed by the Author.
Evolutionary Biology Centre, Uppsala University, Uppsala.
Retief, E., 2000. Zygophyllaceae. In: Leistner, O.A. (Ed.), Seed plants of southern
Africa: families and genera. Strelitzia 10, 563–565.
Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phylogenetic inference
under mixed models. Bioinformatics 19, 1572–1574.
Schreiber, A., 1963. Die Gattung Zygophyllum L. in Südwestafrika. Mitteilungen aus
der Botanischen Staatssammlung, München 5, 49–114.
Sheahan, M.C., Chase, M.W., 1996. A phylogenetic analysis of Zygophyllaceae R. Br.
based on morphological, anatomical and rbcL DNA sequence data. Bot. J. Linn.
Soc. 122, 279–300.
Sheahan, M.C., Chase, M.W., 2000. Phylogenetic relationships within
Zygophyllaceae based on DNA sequences of three plastid regions, with special
emphasis on Zygophylloideae. Syst. Bot. 25, 371–384.
Sonder, O.W., 1860. Zygophyllaceae. In: Harvey, W.H., Sonder, O.W. (Eds.), Flora
Capensis, vol. 1. A.S. Robertson, Cape Town.
Swofford, D.L., 2002. PAUP*. Phylogenetic Analysis Using Parsimony (* and Other
Methods). Version 4.0b10. Sinauer Associates, Sunderland, Massachusetts.
Taberlet, P., Gielly, G., Pautou, G., Bouvet, J., 1991. Universal primers for
amplification of three non-coding regions of chloroplast DNA. Plant Mol. Biol.
17, 1105–1109.
Van Huyssteen, D.C., 1937. Morphologisch-systematische Studien über die Gattung
Zygophyllum. Dissertation, Berlin.
Van Zyl, L., 2000. A Systematic Revision of Zygophyllum in the Southern African
Region. Ph.D. thesis, University of Stellenbosch.
Van Zyl, L., Marais, E.M., 1997. Two new species of Zygophyllum (Zygophyllaceae)
from the Western Cape, South Africa. Bothalia 27, 129–133.
Van Zyl, L., Marais, E.M., 1999. Three new species of Zygophyllum (Zygophyllaceae)
from Namibia and Northern Cape, South Africa. Bothalia 29, 231–237.
White, F., 1983. The Vegetation of Africa. A Descriptive Memoir to Accompany the
UNESCO/AETFAT/UNSO Vegetation Map of Africa. Natural Resources Research,
vol. 20. UNESCO, Paris.