Diversification of the old world Salsoleae sl (Chenopodiaceae): molecular phylogenetic analysis of nuclear and chloroplast data sets and a revised …
ABSTRACT A first comprehensive phylogenetic analysis of tribe Salsoleae s.l. (Salsoloideae: Chenopodiaceae) is presented based on maximum parsimony and maximum likelihood analysis of nuclear ribosomal internal transcribed spacer and chloroplast psbB-psbH DNA sequences. Our data strongly support (1) the sister relationship of Camphorosmeae to the Salsoleae s.l.; (2) splitting of Salsoleae s.l. into two monophyletic tribes, Salsoleae s.s. and Caroxyloneae tribus nova; (3) the current status of most monotypic or oligotypic genera in Salsoleae; and (4) polyphyly of the Botschantzev and Freitag (among others) circumscriptions of Salsola, which falls into 10 (on average) monophyletic genera/lineages. Three well-supported genera are described as new (Pyankovia, Kaviria, and Turania), and four previously described genera are resurrected (Caroxylon, Clima-coptera, Kali, and Xylosalsola). Salsola s.s. include a group of central and southwest Asian and north African species that consists of Salsola sect. Salsola s.s., Salsola sect. Caroxylon subsect. Coccosalsola, Salsola sect. Obpyrifolia, Fadenia, Hypocylix, Seidlitzia, and Darniella. All species of tribe Caroxyloneae investigated so far have C 4 photosynthesis of the NAD-malic enzyme subtype, while the majority of the species of Salsoleae s.s. are known to be of the NADP-malic enzyme subtype.
- [show abstract] [hide abstract]
ABSTRACT: Salsola arbusculiformis is identified as a C3–C4intermediate species based on anatomical, biochemical and physiological characteristics. This is the first report of a naturally occurring intermediate species in the Chenopodiaceae, the family with the largest number of C4species amongst the dicots. In the genus Salsola, most species have Salsoloid anatomy with Kranz type bundle sheath cells and C4photosynthesis, while a few species have Sympegmoid anatomy and were found to have non-Kranz type bundle sheath cells and C3photosynthesis. In the cylindrical leaves of C4Salsola with Salsoloid type anatomy, there is a continuous layer of distinct, chlorenchymatous Kranz type bundle sheath cells surrounded by a single layer of mesophyll cells; whereas species with Sympegmoid type anatomy have an indistinct bundle sheath with few chloroplasts and multiple layers of chlorenchymatous mesophyll cells. However, S. arbusculiformis has intermediate anatomical features. While it has two-to-three layers of mesophyll cells, characteristic of Sympegmoid anatomy, it has distinctive, Kranz-like bundle sheath cells with numerous chloroplasts and mitochondria. Measurements of its CO2compensation point and CO2response of photosynthesis show S. arbusculiformis functions as an intermediate species with reduced levels of photorespiration. The primary means of reducing photorespiration is suggested to be by refixing photorespired CO2in bundle sheath cells, since analysis of photosynthetic enzymes (activity and immunolocalization) and14CO2labelling of initial fixation products suggestsminimal operation of a C4cycle.Annals of Botany - ANN BOT. 01/2001; 88(3):337-348.
- American Journal of Botany - AMER J BOT. 01/1993; 80(9).
- [show abstract] [hide abstract]
ABSTRACT: Kranz anatomy, with its separation of elements of the C4 pathway between two cells, has been an accepted criterion for function of C4 photosynthesis in terrestrial plants. However, Bienertia cycloptera (Chenopodiaceae), which grows in salty depressions of Central Asian semi-deserts, has unusual chlorenchyma, lacks Kranz anatomy, but has photosynthetic features of C4 plants. Its photosynthetic response to varying CO2 and O2 is typical of C4 plants having Kranz anatomy. Lack of night-time CO2 fixation indicates it is not acquiring carbon by Crassulacean acid metabolism. This species exhibits an independent, novel solution to function of the C4 mechanism through spatial compartmentation of dimorphic chloroplasts, other organelles and photosynthetic enzymes in distinct positions within a single chlorenchyma cell. The chlorenchyma cells have a large, spherical central cytoplasmic compartment interconnected by cytoplasmic channels through the vacuole to the peripheral cytoplasm. This compartment is filled with mitochondria and granal chloroplasts, while the peripheral cytoplasm apparently lacks mitochondria and has grana-deficient chloroplasts. Immunolocalization studies show enzymes compartmentalized selectively in the CC compartment, including Rubisco in chloroplasts, and NAD-malic enzyme and glycine decarboxylase in mitochondria, whereas pyruvate, Pi dikinase of the C4 cycle is localized selectively in peripheral chloroplasts. Phosphoenolpyruvate carboxylase, a cytosolic C4 cycle enzyme, is enriched in the peripheral cytoplasm. Our results show Bienertia utilizes strict compartmentation of organelles and enzymes within a single cell to effectively mimic the spatial separation of Kranz anatomy, allowing it to function as a C4 plant having suppressed photorespiration; this raises interesting questions about evolution of C4 mechanisms.The Plant Journal 10/2002; 31(5):649-62. · 6.58 Impact Factor
DIVERSIFICATION OF THE OLD WORLD SALSOLEAE s.l. (CHENOPODIACEAE):
MOLECULAR PHYLOGENETIC ANALYSIS OF NUCLEAR AND CHLOROPLAST
DATA SETS AND A REVISED CLASSIFICATION
Hossein Akhani,1,* Gerald Edwards,yand Eric H. Roalson2,y
*School of Biology, University College of Science, University of Tehran, P.O. Box 14155-6455, Tehran, Iran; andySchool of Biological
Sciences and Center for Integrated Biotechnology, Washington State University, Pullman, Washington 99164-4236, U.S.A.
A first comprehensive phylogenetic analysis of tribe Salsoleae s.l. (Salsoloideae: Chenopodiaceae) is
presented based on maximum parsimony and maximum likelihood analysis of nuclear ribosomal internal
transcribed spacer and chloroplast psbB-psbH DNA sequences. Our data strongly support (1) the sister
relationship of Camphorosmeae to the Salsoleae s.l.; (2) splitting of Salsoleae s.l. into two monophyletic tribes,
Salsoleae s.s. and Caroxyloneae tribus nova; (3) the current status of most monotypic or oligotypic genera in
Salsoleae; and (4) polyphyly of the Botschantzev and Freitag (among others) circumscriptions of Salsola, which
falls into 10 (on average) monophyletic genera/lineages. Three well-supported genera are described as new
(Pyankovia, Kaviria, and Turania), and four previously described genera are resurrected (Caroxylon, Clima-
coptera, Kali, and Xylosalsola). Salsola s.s. include a group of central and southwest Asian and north African
species that consists of Salsola sect. Salsola s.s., Salsola sect. Caroxylon subsect. Coccosalsola, Salsola sect.
Obpyrifolia, Fadenia, Hypocylix, Seidlitzia, and Darniella. All species of tribe Caroxyloneae investigated so
far have C4photosynthesis of the NAD-malic enzyme subtype, while the majority of the species of Salsoleae
s.s. are known to be of the NADP-malic enzyme subtype.
Keywords: Caroxyloneae, Chenopodiaceae, classification, molecular phylogeny, Salsoleae, Salsoloideae.
Chenopodiaceae is a cosmopolitan, eudicot lineage especially
diverse in arid, semiarid, saline, and hypersaline ecosystems
(Ku ¨hn et al. 1993; Hedge et al. 1997). The family is extremely
variable in its ecomorphological and anatomical types and
modes of photosynthesis (Carolin et al. 1975, 1978; Gamaley
and Voznesenskaya 1986; Pyankov et al. 1992, 1997, 2001b,
2002; Akhani et al. 1997, 2005; Jacobs 2001; Voznesenskaya
et al. 2001b, 2002; Kadereit et al. 2003; Schu ¨tze et al. 2003;
Edwards et al. 2004; Akhani and Ghasemkhani 2007). Many
members of this family are succulent and late flowering and
fruiting, which has historically made collections difficult to
identify, with many specimens lacking the necessary charac-
ters for species identification. Additionally, the high levels of
diversity in the deserts of central Asia and the Middle East
have created a limitation on investigation and collection ac-
tivities because of poor representation in Western herbaria.
The diversity of photosynthetic types and leaf anatomies in
this family, particularly the discovery of two anatomical
types that perform C4 photosynthesis without Kranz anat-
omy in one species of Suaeda (Borszczowia) and two species
of Bienertia (Freitag and Stichler 2000, 2002; Voznesenskaya
et al. 2001b, 2002; Akhani et al. 2003, 2005), has attracted
considerable interest in this intriguing group.
The classification of Chenopodiaceae and its phylogenetic
relationships with other families have been explored by a
number of researchers using morphological and molecular
markers (Scott 1977a, 1977b, 1978; Cue ´noud et al. 2002;
Kadereit et al. 2003, 2006; Pratt 2003; Schu ¨tze et al. 2003;
Mu ¨ller and Borsch 2005; Shepherd et al. 2005; Kapralov
et al. 2006). The Salsoloideae subfamily has been circum-
scribed variously, but in recent years it has either included
tribes Sarcobateae, Suaedeae, and Salsoleae (Ku ¨hn et al.
1993) or been restricted to the tribes Camphorosmeae, Sclero-
laeneae, and Salsoleae (Kadereit et al. 2003). The Salsoloideae,
here defined as including the Salsoleae s.l. and Camphoros-
meae (including Sclerolaeneae) clades (Kadereit et al. 2003;
Pratt 2003; Kapralov et al. 2006), has been demonstrated to
be monophyletic (Kapralov et al. 2006). The monophyly of
the two Salsoleae clades in relation to the Camphorosmeae,
however, has been both questioned and poorly supported
in past studies (Pyankov et al. 2001a; Kadereit et al. 2003;
Kapralov et al. 2006).
Tribe Salsoleae includes one-third of all known genera cur-
rently recognized in the family Chenopodiaceae (32 of 98 gen-
era; sensu Ku ¨hn et al. 1993), but it is a poorly understood
lineage. Species concepts in the tribe have varied widely
among researchers, with some recognizing a large number of
species separated by relatively minor morphological differ-
ences (the Russian ‘‘splitters’’; Botschantzev 1970, 1972, 1974a,
1975b, 1976, 1977, 1981a, 1982, among others; Pratov 1986)
and others circumscribing fewer ‘‘metaspecies’’ (the European
‘‘lumpers’’; Freitag 1997), resulting in between 300 and 400
species accepted in the tribe (Botschantzev 1969a, 1969b,
1969c, 1970, 1971, 1972, 1974a, 1974b, 1975a, 1975b,
1975c, 1975d, 1976, 1977, 1980, 1981a, 1981b, 1982,
1986, 1989; Ku ¨hn et al. 1993; Freitag 1997). It is unclear
1Author for correspondence; e-mail email@example.com.
2Author for correspondence; e-mail firstname.lastname@example.org.
Manuscript received August 2006; revised manuscript received February 2007.
Int. J. Plant Sci. 168(6):931–956. 2007.
? 2007 by The University of Chicago. All rights reserved.
which school might better reflect phylogenetic relationships
and monophyletic lineages, but we here test several of these
concepts by sampling multiple individuals/populations within
some of these species groups. However, several of these groups
will require more detailed population genetic and morpho-
logical studies for an understanding of species boundaries.
This tribe is Old World in distribution, with its main center
of diversity in central Asian and Middle Eastern deserts and
subdeserts, with radiations into the Mediterranean, north
and south Africa, and Australia. Some species have also been
introduced into the New World. They are mostly leaf- and
stem-succulent halophytic, xerohalophytic, xerophytic, and ru-
deral plants with diverse traits, particularly in photosynthetic
pathways and concurrent anatomical structures (Butnik et al.
1991, 2001; Akhani et al. 1997; Pyankov et al. 1997, 2001b,
2002; Voznesenskaya et al. 1999, 2001a, 2001b; Akhani and
Ghasemkhani 2007). The potential synapomorphies for the
tribe are the presence of scarious winged perianth segments
in fruit, with possible loss in some species, and utricule with
a spiral embryo. Apparently, the winged fruiting perianth is a
most successful device for wind dispersal in desert areas.
However, the presence of a wing does not occur in all genera
and may be replaced by small protuberances or may be com-
pletely absent. In this latter group, zoochory and hydrochory
dispersal mechanisms seem likely.
Generic boundaries in Salsoleae have been the subject of a
long-standing controversy (Meyer 1829; Moquin-Tandon 1840,
1849; Bunge 1862; Bentham and Hooker 1880; Volkens
1893; Iljin 1936; Ku ¨hn et al. 1993; Hedge et al. 1997). Salsola
has had a controversial subgeneric classification, and its mono-
phyly has been questioned, as has the recognition of such genera
as Climacoptera (Botschantzev 1956, 1969b; Pratov 1986),
Halothamnus (¼Aellenia) (Iljin 1936; Botschantzev 1981b),
Darniella(Brullo1984),Fadenia(Aellenand Townsend 1972),
and Xylosalsola, Nitrosalsola, and Newcaspia (Tzvelev 1993).
Tables 1 and 2 summarize the complicated historical nomen-
clature of Salsoleae and the genus Salsola, at least for those
species included here. In the first phylogenetic analysis of
Salsoleae using internal transcribed spacer (ITS) sequences,
Pyankov et al. (2001a) revealed that Salsola is likely to be
polyphyletic, and similar results were found using rbcL se-
quences (Kadereit et al. 2003). The limited sampling of Salso-
leae in both of these studies, however, leaves many questions
regarding phylogenetic relationships and generic circumscrip-
tion in the tribe unanswered.
We use maximum parsimony and maximum likelihood
analyses of nrDNA ITS and cpDNA psbB-psbH spacer se-
quences to elucidate phylogenetic relationships in Salsoleae
to test generic monophyly. Further, we suggest a new generic
classification of the Salsoleae to more closely reflect phyloge-
Material and Methods
Most of the studied plants were collected by H. Akhani
during intensive collections since 1988 from Iran, Turkmeni-
stan, Turkey, and the United Arab Emirates. Some collections
were dried in silica gel during field studies, and additional
specimens were obtained from the herbaria GAZ (Gazy
Herbarium, Ankara, Turkey) K, LE, M, and MSB (Ludwig-
Maximilians-Universita ¨t, Mu ¨nchen, Germany) (table B1). Other
sources of samples included cultivated species in the greenhouse
of Washington State University (table B1) and nine ITS se-
quences previously published (Pyankov et al. 2001a; Kadereit
et al. 2003). Outgroups were chosen from representatives of
major lineages of the Suaedoideae and Salicornioideae (six
species in total); these lineages together have been demon-
strated to be the sister group to the Salsoloideae s.l. (Kadereit
et al. 2003; Kapralov et al. 2006). Because of amplification
failure or lack of material, we could not include the following
monotypic Salsoleae s.l. genera: Sevada Moq. (Moquin-
Tandon 1849), Iljinia Korovin ex V. Komarov (Iljin 1936),
Halarchon Bunge (1862), Physandra Botsch. (Botschantzev
1956), Traganopsis Maire et Wilczek, Nucularia Battand, and
Lagenantha Chiov. We also could not include the ditypic ge-
nus Choriptera Botsch. (¼Gyroptera). Further studies will be
necessary to resolve the phylogenetic position of these genera.
New sequences have been deposited in GenBank (acces-
sions EF453380–EF453632). The data matrix and resultant
trees have been deposited in TreeBase (accession S1737).
DNA was isolated using a modified 23 CTAB buffer
method (Doyle and Doyle 1987). Templates of the nrDNA
ITS region were prepared using the primers ITS5HP (59-AGG
TGA CCT GCG GAA GGA TCA TT-39; Suh et al. 1993) and
ITS4 (59-TCC TCC GCT TAT TGA TAT GC-39; White et al.
1990). Polymerase chain reaction (PCR) amplifications fol-
lowed the procedures described by Roalson et al. (2001). The
chloroplast psbB-psbH spacer region was amplified using
the primers psbB-psbH-f (59-AGA TGT TTT TGC TGG TAT
TGA-39) and psbB-psbH-r (59-TTC AAC AGT TTG TGT
AGC CA-39; Xu et al. 2000). PCR amplifications followed
the procedures described by Schu ¨tze et al. (2003).
The PCR products were electrophoresed using a 0.8% agarose
gel in a 0.53 TBE (pH 8.3) buffer, stained with ethidium bro-
mide to confirm a single product, and purified using the PEG
precipitation procedure (Johnson and Soltis 1995). Sequencing
was performed using an ABI Prism 3730 genetic analyzer. Direct-
cycle sequencing of purified template DNAs followed manu-
facturer’s specifications, using the ABI Prism BigDye Terminator
Cycle Sequencing Ready Reaction Kit (PE Biosystems).
The two ITS sequencing primers provide sequences for
overlapping fragments that collectively cover the entire spacer
and 5.8S rDNA regions along both strands. The two psbB-
psbH sequencing primers provide near-complete overlap
along both strands. Sequencing of ITS and psbB-psbH used
the same primers that were used for amplification.
Automated DNA sequencing chromatograms were proofed
and edited and contigs were assembled using Sequencher 4.0
(Gene Codes). The ITS sequences were truncated to include
only ITS1, 5.8S, and ITS2. The psbB-psbH sequences were trun-
cated to include the 39 end of the psbB coding region, the psbB-
psbT intergenic spacer, the psbT coding region, the psbT-psbN
intergenic spacer, the psbN coding region, and the psbN-psbH
intergenic spacer. Identification of the terminal ends and
spacer boundaries of ITS1, 5.8S, ITS2, and the psbB-psbH
INTERNATIONAL JOURNAL OF PLANT SCIENCES
Historical Classifications of Salsoleae s.l. from 1829 to Present
Ku ¨hn et al.
(B) Vertical seeds
Classification of Salsola and its segregates is given separately in table 2. Other tribes are mentioned only when a member of Salsoleae was classified under that tribe. Genera
marked with a question mark are genera of questionable status, as dealt with in this article. Several other infratribal units are not listed here. Other authors: Botschantzev (1967, 1975a,
1975b, 1975c, 1975d: Salsoleae, subtrib. Sevadinae [Sevada, Lagenantha, Fadenia, Choriptera, Gyroptera], 1977: Agathophora [=Halogeton subgen. Agathophora]).
aThe monograph dealt only with Anabaseae.
Historical Classification of Salsola and Segregate Genera
Botschantzev (various articles)a
Hedge 1997; Rilke 1999
Sect. Caroxylon subsect.
(No species included
in our analysis)
Sect. Kali (Mill.)
Sect. Caroxylon subsect.
(as S. boissieri)9
Iljin ex Pratov9
Iljin ex Pratov3
7 Kaviria gen. nov.
sect. Nitraria Ulbr.)
1975b, 1975c, 1975d)
9 Pyankovia gen. nov.
10 Salsola s.s.
(1969a, 1969b, 1969c)
11 Turania gen. nov.
(1969a, 1969b, 1969c)
Sect. Caroxylon subsect.
Coccosalsola (1976, 1989)
(sub S. schweinfurthii)10
(sub S. obpyrifolia)
Sect. Caroxylon subsect.
Sect. Irania (1986)
S. androssowi s.l.11
Subtribe Sevadinae (1975c)
Sect. Caroxylon subsect.
(No species included
in our analysis)
Only those taxa that were used in our phylogeny are listed. For a full checklist, see appendix A and references. Superscript numbers correspond to the numbers of the accepted
genera in this article (right column). Other classifications not listed included Brullo (1984): Darniella;10Galushko (1976): Caspia; and Woloszczak (1885): Hypocylix.10
aReferences are given by section. Publication years of the relevant Botschantzev articles are given in parentheses.
gene regions was based on comparisons with other species
of Chenopodiaceae (Kapralov et al. 2006). Sequences were
aligned using Clustal X (Thompson et al. 1997) with gap open-
ing penalty of 10.00 and gap extension penalty of 1.00 for
both pairwise and multiple comparisons. The resultant align-
ment was then checked by eye for necessary minor corrections.
Alternate alignment parameters did not result in significantly
different topologies (data not shown). Gaps were not coded as
binary characters because of the complex nature of the gaps
in these data sets and the additional problem that they can-
not be integrated into the maximum likelihood analyses.
ITS and psb-psbH regions were analyzed separately and
in combination with both maximum parsimony (MP) and
maximum likelihood (ML) analyses. All analyses were per-
formed using PAUP* 4.0b10 (Swofford 2001). MP analyses of
the individual and combined data sets used heuristic searches
(ACCTRAN; 1000 random addition cycles, tree-bisection-
reconnection [TBR] branch swapping, limit of 10,000 re-
arrangements per addition sequence replicate). Swapping was
run to completion for all random addition replicates. Clade
support was estimated using 1000 heuristic bootstrap repli-
cates (100 random addition cycles per replicate, TBR branch
swapping, limit of 10,000 rearrangements per addition se-
quence replicate; Felsenstein 1985; Hillis and Bull 1993).
ML analyses employed heuristic searches (TBR branch
swapping). Clade support was estimated using 100 heuristic
bootstrap replicates (10 random addition cycles and 100 to-
tal rearrangements per replicate, TBR branch swapping;
Felsenstein 1985; Hillis and Bull 1993). ML analysis of the
ITS data set employed the general time-reversible model with
proportion of invariant sites (I) and gamma shape (G) param-
eters and empirical base frequencies (six substitution types:
A=C¼1:4064, A=G¼2:5332, A=T¼1:7413, C=G¼0:7280,
C=T ¼ 3:5703, G=T ¼ 1:0000; I ¼ 0:2193; G ¼ 0:9803; A ¼
0:2084, C ¼ 0:2519, G ¼ 0:2849, T ¼ 0:2548). ML analysis
of the psbB-psbH genetic region employed a five–rate class
transversion model with I and G parameters and empirical base
frequencies (five substitution types: A=C ¼ 0:9657, A=G ¼
1:3422,A=T ¼ 0:2773,C=G ¼ 0:7521,C=T ¼ 1:3422, G=T ¼
1:0000; I ¼ 0:2417; G ¼ 0:9036; A ¼ 0:2982, C ¼ 0:1616,
G ¼ 0:1794, T ¼ 0:3608). ML analysis of the combined data
set employed a four–rate class transition model with I and G
parameters and empirical base frequencies (four substitution
types: A=C ¼ 1:0000, A=G ¼ 1:9407, A=T ¼ 0:8031, C=G ¼
0:8031, C=T ¼ 2:5593, G=T ¼ 1:0000; I ¼ 0:2700; G ¼
0:7111; A ¼ 0:2613, C ¼ 0:2239, G ¼ 0:2239, T ¼ 0:2909).
These models were chosen based on the results of analysis us-
ing DT_ModSel (Minin et al. 2003). The DT_ModSel analy-
sis uses a Bayesian information criterion to select a model
using branch-length error as a performance measure in a de-
cision theory framework that also includes a penalty for
New ITS and psbB-psbH sequences were obtained for 132
species/accessions belonging to tribe Salsoleae s.l. and six
species of Salicornieae and Suaedeae as outgroups. The
aligned ITS data matrix was 743 base pairs (bp) long with
511 variable sites (68.8%), of which 400 (53.8%) were parsi-
mony informative. Because of poor sequencing reads of some
regions, three sequences are missing a portion (104–182 bp)
of the 59 end of the ITS 1 spacer, eight sequences are missing
a portion (92 bp) of the 59 end of the ITS 2 spacer, and 14 se-
quences are missing a portion (4–82 bp) of the 39 end of the
ITS 2 spacer. The aligned psbB-psbH data matrix was 741 bp
long with 270 variable sites (36.4%), of which 144 (19.4%)
were parsimony informative. Because of poor sequencing reads
of some regions, nine sequences are missing a portion (1–113 bp)
of the 59 end of the psbB-psbH spacer region, and 21 sequences
are missing a portion (1–121 bp) of the 39 end of the psbB-psbH
MP analysis of the ITS Salsoleae data set resulted in 1451
most parsimonious trees (length ¼ 3445 steps, consistency in-
dex ½CI? ¼ 0:303, retention index ½RI? ¼ 0:741, rescaled con-
sistency index ½RC? ¼ 0:224). The ITS ML analysis resulted in
a single tree (?lnL ¼ 17800:03435, where L ¼ likelihood).
MP analysis of the psbB-psbH data set resulted in 11,122
most parsimonious trees (length ¼ 600 steps, CI ¼ 0:595,
RI ¼ 0:766, RC ¼ 0:456). The psbB-psbH ML analysis re-
sulted in two tied trees (?lnL ¼ 4783:22431). Strict consen-
sus trees of the MP individual data set analyses and the ML
trees of individual data set analyses are available from the
corresponding authors. MP analysis of the combined data set
resulted in 231 most parsimonious trees (length ¼ 4067 steps,
CI ¼ 0:343, RI ¼ 0:740, RC ¼ 0:254; fig. 1). The combined
ML analysis resulted in a single tree (?lnL ¼ 23216:90496;
Analyses of individual data sets resulted in congruent esti-
mates of relationships, with slight differences associated with
unresolved branches and short branches with low bootstrap
support, particularly in the psbB-psbH analysis. Combined
analyses reflect the well-resolved portions of individual data
set analyses, and all branches are better supported in the
combined analysis than in either of the individual data set
analyses (trees from individual analyses in TreeBase). Given
our results that multiple alignments of individual data sets
produced congruent topologies and that there were no well-
supported conflicting branches, as well as the fact that the
clades we found are generally supported by morphological
characters, we do not consider the high levels of ITS variabil-
ity or alignment issues to reduce the ability of our analyses to
reconstruct robust phylogenetic hypotheses. MP and ML
analyses of the combined data result in congruent inferences
of relationships, with differences in resolution resulting in
slightly different placement of some species (figs. 1, 2). These
differences, however, are associated with branches with low
bootstrap support in one or both analysis types. In all analy-
ses, Salsola s.l. is clearly polyphyletic, with Salsola species
present in seven to 13 lineages or different clades, depending
on the resolution of the phylogenetic hypotheses (figs. 1, 2).
Several other genera are not monophyletic as currently cir-
cumscribed, namely, Anabasis, Halanthium, Halimocnemis,
Hammada, Gamanthus, and Climacoptera (figs. 1, 2). In
some cases, this is due to the misclassification of one or a
small number of species (e.g., Climacoptera brachiata; figs.
1B, 2B), whereas other cases, such as the polyphyly and
AKHANI ET AL.—PHYLOGENY AND CLASSIFICATION OF SALSOLEAE s.l.
retention index ¼ 0:740; rescaled consistency index ¼ 0:254). A, Outgroups and clades of Camphorosmeae and Salsoleae s.s. tribes. B,
Caroxyloneae tribe clade. Numbers above branches reflect maximum parsimony bootstrap numbers. Shaded boxes refer to species traditionally
placed in the genus Salsola. Generic abbreviations are as follows: A: ¼ Anabasis, B: ¼ Bassia, Bi: ¼ Bienertia, C: ¼ Climacoptera, Ca: ¼
Maximum parsimony combined data analysis strict consensus tree of 231 shortest trees (length ¼ 4067; consistency index ¼ 0:343;
Camphorosma, Ch: ¼ Chenoleoides, Co: ¼ Cornulaca, Cy: ¼ Cyatobasis, F: ¼ Fadenia, G: ¼ Gamanthus, Gi: ¼ Girgensohnia, H: ¼ Halotis,
Ha: ¼ Halimocnemis, Hal: ¼ Halocharis, Hala: ¼ Halanthium, Halo: ¼ Halogeton, Halot: ¼ Halothamnus, Halox: ¼ Haloxylon, Ham: ¼
Hammada, Ho: ¼ Horaninowia, K: ¼ Kochia, Ka: ¼ Kalidium, Ki: ¼ Kirilowia, L: ¼ Londesia, M: ¼ Maireana, Mi: ¼ Microcnemum, N: ¼
Nanophyton, No: ¼ Noaea, O: ¼ Ofaiston, P: ¼ Panderia, Pe: ¼ Petrosimonia, R: ¼ Rhaphydophyton, S: ¼ Salsola, Sa: ¼ Salicornia, Se: ¼ Seidlitzia,
Su: ¼ Suaeda, Sy: ¼ Sympegma, T: ¼ Traganum.
s.s. tribes. B, Caroxyloneae tribe clade. Numbers above branches reflect maximum likelihood bootstrap numbers. Shaded boxes refer to species
traditionally placed in the genus Salsola. Generic abbreviations follow those in fig. 1.
Maximum likelihood combined data analysis tree (?lnL ¼ 23216:90496). A, Outgroups and clades of Camphorosmeae and Salsoleae
interdigitation of Halanthium and Halimocnemis, are more
difficult (figs. 1B, 2B).
Because individual analyses were congruent with, although
less resolved than, the combined analyses, we will generally
refer to the combined MP and ML results in our discussion.
It seems clear from the analyses presented here that the Salso-
loideae is a monophyletic group with the Camphorosmeae
tribe the sister clade to Salsoleae s.l. (figs. 1, 2) and is well
supported by maximum parsimony bootstrap (mpbs ¼ 92%)
and maximum likelihood bootstrap (mlbs ¼ 75%). This rela-
tionship has been previously found (Kadereit et al. 2003;
Kapralov et al. 2006), although this branch was weakly sup-
ported by rbcL, and the multigene analysis of Kapralov et al.
(2006), while providing strong support (mlbs ¼ 86%), did
not have sufficiently extensive sampling to allow confidence
in this relationship. One previous study placed the Camphor-
osmeae as sister to a portion of the Salsoleae s.l. (our Carox-
yloneae), creating a paraphyletic Salsoleae s.l., but this result
was weakly supported (mpbs < 50%; Pyankov et al. 2001a).
Furthermore, the traditional Salsoleae is clearly composed of
two strongly supported clades, here referred to as the Salso-
leae s.s. (mpbs ¼ 100%; mlbs ¼ 100%) and the Caroxylo-
neae (mpbs ¼ 98%; mlbs ¼ 98%; figs. 1, 2; see taxonomic
revision in app. A). The occurrence of two well-supported
clades in Salsoleae s.l. has been found with analyses of ITS
sequences from 34 species (Pyankov et al. 2001a) and 12 spe-
cies in rbcL analysis (Kadereit et al. 2003). The latter authors
referred to these two clades as Salsoleae I and Salsoleae II.
Both clades are well distinguished by a number of characters
(table 3). A particularly distinguishing characteristic of the
Caroxyloneae clade is the vesicular and disjunct anther ap-
pendage, which seems to occur in most groups and may be
involved in attracting insect pollinators, which have been ob-
served frequently in nature (H. Akhani, personal observa-
tion). These connectives are absent or very small in members
of Salsoleae and have been noted as a minute appendage in
Halothamnus, Noaea, and Halogeton (Kothe-Heinrich 1993;
Hedge 1997); they are rarely large as those found in Raphy-
dophytum (Iljin 1936). The two clades are also distinguish-
able based on C4photosynthesis subtypes: all known species
of Caroxyloneae are of the NAD-malic enzyme subtype (ta-
ble 3), and except for one doubtful case (H. Akhani, unpub-
lished data), all Salsoleae are known to be of the NADP-malic
enzyme subtype (see also Pyankov et al. 2001a, 2001b).
Classification and Relationships in Clade Salsoleae
The Salsoleae s.s. tribe is more diverse than Caroxyloneae,
both morphologically and physiologically. In this tribe, C4,
C3, and C3-C4intermediate species occur, with strong mor-
phological features separating the tribe from Caroxyloneae
(table 3). Four primary lineages or clades in Salsoleae s.s. can
be distinguished: Sympegma, the Halothamnus clade, the
Kali clade, and the Salsola clade (figs. 1A, 2A). The mono-
typic Sympegma is sister to a clade composed of the rest of
the lineages of the Salsoleae s.s., although its separation is
only weakly supported (mpbs ¼ 62%; mlbs < 50%). The or-
dering of the other three clades is not strongly supported
(mpbs < 50%; mlbs < 50%), but both analyses suggest that
the Kali clade is sister to a clade composed of the Halotham-
nus and Salsola clades.
The monotypic Sympegma is restricted to central Asia
and is unique in the family in having terminal glomerate
Comparison of the Characters of the Two Major Clades of Salsoleae s.l.
Character CaroxyloneaeSalsoleae s.s.
Life form Mostly annual, with some hemicryptophytes and subshrubs Mostly shrubby, subshrubby and even tree;
Alternate in basal genera, opposite in most
Leaves and stems, often articulated
Mostly with mucro or spine at apex, rarely obtuse
Mostly terete and filiform
(flat in Anabasis and Halothamnus)
Plants mostly glabrous, with a tuft of flexuose hairs
in nodes or axil of leaves and flowers
Hairs papillose, unicellular; axial hairs flexuous and
multicellular, always basifixed
Mostly absent; if present, small and nonvesiculose,
color usually concolor with anthers
Branches and leaves Alternate (except Pyankovia and one species of Petrosimonia)
Spines at leaf apex
Leaves and sometimes stems by age, never articulated
Mostly obtuse, rarely with spine or mucro at apex
Flat, linear, oblong, ovate
IndumentumStem, leaves, and perianths with long multicellular hairs,
at least when young
Hairs various; mostly articulate, spinulose, flattened,
bladderlike, basifixed, or medifixed
Present and mostly separated from theca and vesiculose,
discolor with anthers
Type of indumentum
Wing on fruiting
Wings mostly present; absent in some genera
Plants with concentration in temperate deserts on
nitrified soil and ruderalized habitats
Mostly central and southwest Asia,
northern and southern Africa
Wings always present
C4with a few C3and C3-C4intermediates
NADP-malic enzyme (with few exceptions)
Plants concentrated mostly in hot deserts rich in sand,
gravel, and gypsum
As for Caroxyloneae but absent in southern AfricaGeography
INTERNATIONAL JOURNAL OF PLANT SCIENCES
inflorescence consisting of several flowers surrounded by two
or more bracts. This species is most likely to be C3, as sug-
gested by its Kranz-less sympegmoid leaf anatomy (Carolin
et al. 1975) and a ?21.6& carbon isotope ratio. However,
the carbon isotope ratio is more positive than is typical for
C3species, which suggests that more studies on living plants
are necessary to exclude the possibility of it functioning as a
The monophyletic Halothamnus clade probably includes
21–23 species, although only four taxa are included here.
Most species are concentrated in southwest Asia (Iran and
Afghanistan) but also occur widely in central Asia, and one
species is found in east Africa (Somalia, Ethiopia, and Dji-
bouti) (Kothe-Heinrich 1993). The monophyly of this genus is
well supported (figs. 1A, 2A; mpbs ¼ 100%; mlbs ¼ 100%).
This genus can be defined by green annual branches, speci-
form inflorescence, indurated fruiting perianths that are pit-
ted in the abscission zone, presence of a hypogenous disk,
horizontal seeds, absence of Kranz anatomy in cotyledon
leaves, and presence of a short anther appendage clearly not
separated from the theca. All known species are C4with a
leaf anatomy lacking a hypodermis layer (Pyankov and Vakh-
rusheva 1989; Kothe-Heinrich 1993; Akhani et al. 1997).
The Kali clade is strongly supported in all analyses (mpbs ¼
97%; mlbs ¼ 90%). This clade assembles four previously
separate taxa in Salsoleae: the genus Traganum, Salsola sect.
Kali, Salsola sect. Sogdiana (sensu Rilke 1999), and Salsola
sect. Caroxylon subsect. Arbusculae p.p. (sensu Botschantzev
1976) or Salsola sect. Arbuscula p.p. (sensu Freitag 1997).
Besides sharing similar habitats in sandy deserts or coastal
sands, members of this clade can be defined by the combination
of morphological characteristics including succulent leaves,
nonjointed stems, apiculate to spiny leaf and bract apices,
winged fruiting perianths in most species (except Traganum),
and a cupulate or cylindrical corona above the wings of
fruiting perianths. The clade is divided into two well-
supported subclades (figs. 1A, 2A). The first of these is the
Kali subclade, previously classified as Salsola sect. Kali (sensu
Rilke 1999) or Salsola sect. Salsola s.l. p.p. (Iljin 1936), and
is strongly supported (mpbs ¼ 99%; mlbs ¼ 99%). Included
here are annual or perennial species with spiny leaf tips that
lack a leaf hypodermis. The most important feature charac-
terizing this clade is the green cortex of annual shoots, which
are associated with longitudinal striae. The striate lines are
chlorenchymotous tissue interrupted by cholenchyma tissue
(Rilke 1999). Because the genus Kali Miller is validly pub-
lished (Miller 1754), it is here used for this subclade. We
here designate Kali soda Moench (Methodus 331, 1794) as
the lectotype of Kali (see app. A) against two other possibili-
ties: Kali tragus (L.) Scop., Fl. Carniol. 1: 775, 1772 (¼Salsola
tragus L., Cent. Pl. 2: 13, 1756) and Kali rosacea (L.)
Moench, suppl. Meth. Plant. 115, 1804 (¼S. rosacea L., Sp.
Pl. 222, 1753). Kali soda was validly described under Salsola
kali L., Sp. Pl.: 222, 1753, 1 yr earlier than the description of
the genus Kali in 1754 (Rilke 1999), and it is among the oldest
names of species in this genus (1753).
The second subclade of the Kali clade is a heterogeneous
assemblage of small trees, shrubs, subshrubs, and annual spe-
cies distributed in extreme deserts of central Asia and north
Africa, primarily as components of sandy ecosystems. Its
monophyly is well supported (mpbs ¼ 89%; mlbs ¼ 80%); it
is composed of species previously placed in Salsola sect.
Sogdiana (Iljin) Rilke, Salsola sect. Androssowia Rilke (sensu
Rilke 1999), Salsola sect. Caroxylon subsect. Arbusculae p.p.
(sensu Botschantzev 1976), and the monotypic genus Traga-
num. Based on the topology of the tree and distribution of
morphological features in the group, three well-supported
genera are here distinguished: Xylosalsola, Traganum, and
Turania. Xylosalsola Tzvelev includes C4species of Salsola
sect. Caroxylon subsect. Arbuscula p.p. (Botschantzev 1976)
or Salsola sect. Arbuscula p.p. (sensu Freitag 1997). These
are shrubby species of central Asia characterized by long te-
rete linear leaves, solitary flowers, milky white and shining
young stems, overlapping fruiting perianths that form a
corona-like structure above the winged fruiting perianths,
and presence of a minute anther appendage (Botschantzev
1976; Freitag 1997).
The small genus Traganum includes two north African/
eastern Mediterranean species, one of which is sampled here.
These are small shrubs with woolly nodes and semiterete
leaves. The fruiting perianths are indurated throughout, have
two hornlike teeth, and lack a wing. The leaves reportedly
lack a hypodermis layer (Carolin et al. 1975), which sepa-
rates this genus from other lineages in this subclade of the
The third lineage of this subclade includes species previ-
ously classified in Salsola sect. Salsola (Iljin 1936) or Salsola
sect. Sogdiana (Iljin) Rilke and Salsola sect. Androssowia
Rilke (Rilke 1999). These central Asian annual species have
succulent flat or semiterete leaves with a short (0.5 mm) or
long (5 mm) spine at the apex, a leaf hypodermis layer, and
cupulate fruiting perianths that are somewhat connate at the
base and give the ovary a false-inferior appearance. Further-
more, they have filiform stigmas that are very long, up to
three to five times as long as the style. Given the clear mor-
phological circumscription of these species, we are here rec-
ognizing this clade as the genus Turania (see app. A).
The Salsola clade is a complex assemblage of genera in Sal-
soleae but is moderately supported in the phylogenetic hy-
potheses presented here (mpbs ¼ 72%; mlbs ¼ 65%). This
lineage includes taxa occurring from central and southwest
Asia to the north African and Mediterranean areas. The
monophyly of the clade is supported by several morpho-
logical features including presence of a spine or mucro on
leaf tips that sometimes appears as a caducous bristle and
most species being completely glabrous or having papillose
or tubercle-like hairs. Many groups are represented by oppo-
site leaves or branches, and stamens have no or very short
anther appendages. Many genera have been previously de-
scribed in this clade, and most of them are supported by the
phylogenetic hypotheses (figs. 1A, 2A).
The genus Salsola was typified by Salsola soda (Jarvis et al.
1993; see Rilke 1999 for details), which is nested within a
homogenous group of species that have been variously placed
in several genera in previous classifications. The monophyly
of this clade is well supported (mpbs ¼ 84%; mlbs ¼ 79%).
In spite of the morphological synapomorphies that strongly
support this clade, species of this clade have been placed
by various authors in several sections of Salsola, including
sect. Salsola (Rilke 1999), sect. Obpyrifolia (Botschantzev
AKHANI ET AL.—PHYLOGENY AND CLASSIFICATION OF SALSOLEAE s.l.
and Akhani 1989), sect. Caroxylon subsect. Coccosalsola
(Botschantzev 1976, 1989), and sect. Coccosalsola (Freitag
1997), or have been classified into other genera, including
Seidlitzia Bunge ex Boiss. (Iljin 1954), Hypocylix Woloszczak
(Woloszczak 1885), Darniella Maire & Weiller (Brullo 1984),
Neocaspia Tzvelev (Tzvelev 1993), Caspia (Galushko 1976),
Fadenia (Aellen and Townsend 1972), and Anabasis p.p. The
most obvious synapomorphy of Salsola s.s. as treated here is
the presence of clusters of two to six flowers (or, rarely, one)
in the axil of each floral leaf. Further characters include ab-
sence or presence of a very minute anther appendage, pres-
ence of a hypogynous disk or staminode (much reduced in
S. soda), presence of a leaf hypodermis, cylindrical and ob-
tuse leaves (more often obpyriform, at least in the juvenile
state or in bracts) that are opposite in most species, and fruit-
ing perianths with well-developed wings (reduced in S. soda).
A surprising result is the inclusion of the tropical African
monotypic genus Fadenia in Salsola (Aellen and Townsend
1972). Fadenia zygophylloides is known from Kenya, Ethio-
pia, and Somalia and was previously separated from all other
species of Salsoloideae by the fruiting perianths having longi-
tudinal membranous crests. This species has been previously
classified in subtribe Sevadinae (Botschantzev 1967, 1975c;
Boulos et al. 1991). Although we have not sampled all spe-
cies of the complex, there is little doubt that all species
treated under the genus Darniella by Brullo (1984) and the
genus Seidlitzia by Iljin (1954) belong to Salsola s.s.
With the exclusion of Anabasis setifera, the rest of the ge-
nus Anabasis forms a well-supported monophyletic group
(mpbs ¼ 99%; mlbs ¼ 96%). The monophyly of this clade is
supported by the combination of several morphological fea-
tures, including a perennial and shrubby habit, a thick basal
caudex (mostly woolly), opposite leaves and branches, verti-
cal seeds, fleshy utricle that resembles a berrylike fruit in sev-
eral species, articulated branches, vestigial leaves in most
species (in Anabasis eugeniae, the leaves are developed), and
the presence of a multilayered epidermis and sunken stomata
(Bokhari and Wendelbo 1978). Ecologically, most species are
extreme xerohalophytic species and frequently grow on halo-
gypsum soils. The genus Anabasis is distributed from south-
west Europe and north Africa to the Red Sea coast (Ethiopia)
and southwest and central Asia.
The monophyly of the genus Halogeton is strongly sup-
ported by all analyses (mpbs ¼ 100%; mlbs ¼ 100%). This
is a small genus of approximately five species, including both
annual (in temperate salines and ruderal places) and peren-
nial species (in warm and hot deserts). The genus is well
defined by the combination of fleshy cylindrical leaves termi-
nating in a persistent or caducous bristle, presence of three to
several flowers in the axil of each floral leaf, presence of a
papillose staminodial disk, presence of five wings on fruiting
perianths, and membranous perianth segments. Some authors
have removed the perennial species to the genus Agathophora
(Botschantzev 1977; Hedge 1997). Our results support a mono-
phyletic clade including both annual and perennial species;
however, because only one of each is sampled here, future
studies will be necessary to explore whether the two growth
forms form monophyletic sister lineages.
Girgensohnia, Cyatobasis, and two species of Hammada
form a clade in the analyses presented here, although it is
weakly supported. The central Asian and Persian genus Gir-
gensohnia includes approximately four or five species, three
sampled in this study, and forms a monophyletic well-
supported group (mpbs ¼ 96%; mlbs ¼ 86%). Morphologically,
the genus is well defined by an annual life form, opposite
leaves and branches, presence of an indumentum of scabrid
papillae, semiamplexicaule leaves with a scarious base and
spine-tipped apex, and vertical seeds. The species of Girgen-
sohnia are ruderal and sometimes weedy species on low salty
soils in the deserts of central Asia and Iran. The monotypic
central Anatolian genus Cyatobasis was described by Aellen
(1949), who distinguished it from Girgensohnia by characters
such as elongate styles, a noncapitate stigma, and connate leaf
base. Our analyses suggest that this species, together with
Hammada articulata and Hammada griffithii, forms a grade
leading to Girgensohnia s.s. These results and shared mor-
phological characters suggest that a wider circumscription of
the genus Girgensohnia, including Cyatobasis and Hammada
p.p. and probably the other species of Arthrophytum, is ap-
Analysis of two of the approximately six species of Cornu-
laca reveals a strongly monophyletic group (mpbs ¼ 100%;
mlbs ¼ 100%). The genus is characterized by a sturdy habit,
alternate branches, decurrent strongly spiny leaves and
bracts, presence of a dense white tuft of hairs among and at
the base of flowers, membranous perianth segments that be-
come indurated and coalescent in fruit, one (sometimes two)
terminal perianth spine, filaments connating into a tube, and
vertical seeds. Cornulaca species occur in central and south-
west Asia and northern Africa on sandy or dry soils and can
tolerate long periods of drought (H. Akhani, personal obser-
vation). The genus is sister to Horaninowia, another spiny
genus, but is clearly separated by other morphological char-
acteristics (see next paragraph).
Horaninowia is a well-supported monophyletic genus in
our analyses (mpbs ¼ 88%; mlbs ¼ 73%). There are approx-
imately seven spiny annual species that are characterized by
a green cortex, the presence of unicellular papillae, spiny-
tipped leaves and bracts, solitary flowers in hairy leaf axils,
exappendiculate anthers, perianths in fruit becoming hard-
ened in the upper middle, capitate or clavate stigmas, and
horizontal seeds (Carolin et al. 1975). Species of Horanino-
wia are restricted in their range to central Asia and Iran,
growing on sandy dunes or gravelly deserts. The phylogenetic
analyses and morphological features clearly support a close
relationship with Cornulaca (figs. 1A, 2A).
The traditional circumscription of Haloxylon (Iljin 1936)
includes only tall shrub to small tree species and is well sup-
ported in these analyses (mpbs ¼ 100%; mlbs ¼ 100%). The
two species, Haloxylon ammodendron and Haloxylon persi-
cum, are found in central and southwest Asia and occur on
sandy dunes or dry salty habitats close to the margins of playas,
where their long roots have access to underground salty water
(Le ´onard 1991; Akhani et al. 2003; Akhani 2004). The com-
bination of unique tree life form with articulated branches,
horizontal seeds, occurrence of an arista at the scalelike leaf
apex, presence of a hypodermis layer in assimilating shoots,
and the isopalisade cotyledon leaves without Kranz anatomy
(Pyankov et al. 1999), characterizes the genus. Bunge (1879)
and Hedge (1997) proposed a broader circumscription of this
INTERNATIONAL JOURNAL OF PLANT SCIENCES
genus, including species from other genera such as Arthrophy-
tum and Hammada, but this is not supported by the phyloge-
netic hypotheses presented here (figs. 1A, 2A).
The genera Hammada and Arthrophytum have been inter-
preted differently by different authors. Hedge (1997) and
Boulos (1996) considered them congeneric with Haloxylon.
The three species we have analyzed (Hammada salicornica,
Hammada articulata, and Hammada griffithii) are not closely
associated with Haloxylon and may form early lineages of
the Girgensohnia/Cornulaca/Horaninowia clade (fig. 2A), al-
though these relationships are not well supported and are
placed differently by the MP strict consensus (fig. 1A) but,
again, with little support.
The phylogenetic hypotheses presented suggest a possible
clade including the C3Salsola montana complex, Salsola ar-
busculiformis, Raphydophytum regelii, and the genus Noaea
(fig. 2A). This clade, however, is not strongly supported
and is not present in the MP strict consensus (fig. 1A). The
placement of the C3-C4intermediate Salsola arbusculiformis
(Voznesenskaya et al. 2001a) between Salsola montana and
Noaea, a C4genus, might demonstrate an interesting case of
transition in photosynthetic pathway across a clade. Given
the weak phylogenetic placement of S. arbusculiformis in the
phylogenetic hypotheses presented here, we are only infor-
mally recognizing this species as ‘‘Collinosalsola’’ and will
await further evidence of its phylogenetic position before for-
mally placing the species. The small subshrub Raphydophy-
tum is characterized by stiff and spinescent leaves that are
acicular and three-angular in cross section, with scabrid mar-
gins and a dilated base. The perianths bear wings near the
base, and the filaments produce a staminal tube with well-
developed semiorbicular lobes on the hypogynous disk.
All known species of Noaea are included in our phyloge-
netic analyses, and its monophyly is well supported (mpbs ¼
100%; mlbs ¼ 100%). This genus is characterized by alter-
nate branches, leaves spiny tipped or cuspidate at the base
with broad white membranous margins, and vertical seeds.
All three species grow in temperate and cold-temperate des-
erts or montane and submontane steppe vegetation, which is
not typical for C4species.
The S. montana complex was classified in Salsola sect. An-
chophyllum by Iljin (1936), sect. Caroxylon subsect. Arbus-
culae by Botschantzev (1976), and sect. Arbuscula by Freitag
(1997). The complex includes subshrubby species that differ
from species of previously mentioned Salsola s.l. groups in hav-
ing not only green young stems but also a sympegmoid leaf
anatomy (Akhani and Ghasemkhani 2007), filaments attached
to the disk without staminodes, and anthers divided only to
two-thirds of their length. The precise phylogenetic position of
this strongly supported clade (mpbs ¼ 100%; mlbs ¼ 100%) is
not clear, and we are therefore here treating this complex as the
informal taxonomic entity ‘‘Oreosalsola’’ (see app. A).
The S. montana species complex represents an assemblage
of microspecies (Salsola maracandica Iljin, Salsola oreophila
Botsch., Salsola masenderanica Botsch., Salsola botschant-
zevii Kurbanov, Salsola flexuosa Botsch., Salsola tianschanica
Botsch., Salsola lipschitzii Botsch., Salsola junatovii Botsch.,
and S. montana Litw.), which are collectively included in a
broadly defined S. montana by Freitag (1997). We have here
examined three populations in this complex, one from Golestan
National Park (S. montana), one from the Alborz mountains
(S. masenderanica), and ‘‘Salsola touranica,’’ an undescribed
but likely distinct entity from the Touran Protected Area of
Iran. Members of this species complex need to be studied fur-
ther in order for us to understand where species boundaries lie
and whether one or eight or more species should be recognized.
Salsola divaricata was included in Salsola sect. Caroxylon
subsect. Coccosalsola by Botschantzev (1976, 1989). This
shrubby species is endemic to the Canary Islands and is distinc-
tive in having opposite leaves, mature leaves that are triangular
in cross section, and leaves with one layer of hypodermis, two
layers of palisade parenchyma, scattered peripheral vascular
bundles, and a central aqueous tissue. Morphologically, it is
very similar to species of Salsola s.s., but this species does not
strongly group with Salsola s.s. Given its unclear phylogenetic
position and the need to sample the similar C3Mediterranean/
north African/central Asian species Salsola genistoides, Salsola
webbii, and Salsola pachyphylla, no nomenclatural changes are
Classification and Relationships in Clade Caroxyloneae
Three major clades can be distinguished in Caroxyloneae,
which are here labeled as the Caroxylon clade (mpbs ¼ 88%;
mlbs¼93%), the Kaviria clade (mpbs ¼ 86%; mlbs ¼ 87%),
and the Climacoptera clade (mpbs ¼ 77%; mlbs ¼ 83%). Two
of the three clades can be divided further into two or more
monophyletic lineages, which in most cases correspond with
traditional classifications of Salsoleae genera. However, the re-
lationship and generic circumscription of several closely related
annual genera in this clade, including Halanthium, Halimocne-
mis, Halotis, Gamanthus, Climacoptera, Piptoptera, Halocharis,
Halarchon, Petrosimonia, and Physandra, has been controver-
sial (Pratov 1986; Akhani 1996; Hedge 1997; Assadi 2001; Gho-
badnejhad et al. 2004). These euhalophytic and xerohalophytic
species are endemic to the Irano-Turanian area, primarily in tem-
perate deserts of central Asia, Afghanistan, and Iran. Except Pet-
rosimonia, with connate cagelike anther appendages, all species
are characterized by large, often showy and colorful (white, yel-
low, or purple) vesciculate anther appendages, which apparently
act as an attractor for insect pollinators and may also contribute
as a wind-dispersal device for anthers and pollen grains, depend-
ing on the species.
The Caroxylon clade includes a large group of species tra-
ditionally classified as Salsola sects. Caroxylon p.p. (subsect.
Caroxylon, subsect. Vermiculatae), Cardiandra, Irania, and
Malpigipila and two species of sect. Belanthera (Salsola can-
escens and Salsola carpatha). The monophyly of this clade is
well supported (figs. 1B, 2B). This is the most widespread
lineage of Salsoleae s.l., with ca. 140 described species, being
found in central Asia, Arabia, and northern and southern
Africa (Botschantzev 1968, 1969a, 1969c, 1970, 1972, 1974a,
1974b, 1975b, 1975d, 1980, 1986; Freitag 1997). Our phy-
logeny includes 19 species covering most known lineages and
geographic areas. The clade is morphologically heterogeneous,
although the presence of an acute anther appendage, winged
perianth segments, a gibbous leaf base, and a staminal disk pro-
vides a combination of characters that distinguishes the clade.
The oldest generic name for species in this clade is Caroxylon
Thunb. It was reduced to a section of Salsola by several
AKHANI ET AL.—PHYLOGENY AND CLASSIFICATION OF SALSOLEAE s.l.
subsequent authors (Fenzl 1851; Iljin 1936) but recognized at
the generic level by Tzvelev (1993), which is supported by our
data. Relationships within Caroxylon are generally weakly sup-
ported (figs. 1B, 2B) and will require further study to clarify.
The Kaviria clade includes species traditionally classified in
Salsola sect. Belanthera (excluding S. canescens, S. carpatha, and
other microspecies classified in subsect. Kochioides by Botsch-
antzev [1968, 1980]), the oligotypic genus Halocharis, and the
monotypic genus Nanophyton (figs. 1B, 2B). The combination
of the C3cotyledon leaves, the absence of a leaf hypodermis, and
the presence of an acute triangular anther appendage that is sep-
arated from the thecae circumscribes this clade well, and mor-
phological and ecological features clearly separate each of these
three lineages from each other (table 4). Salsola sect. Belanthera
is here named Kaviria (see app. A) after the Persian term ‘‘Kavir,’’
a name used to refer to the Iranian Great Desert.
The Climacoptera clade is composed of a lineage including
Petrosimonia and Ofaiston, which is sister to the remainder
of the clade (figs. 1B, 2B). This is an exclusively Irano-Turanian
species group that predominantly occurs in annual halophytic
and xerohalophytic communities of central and southwest
Asia. Petrosimonia and Ofaiston are quite distinct morpho-
logically, and Petrosimonia is supported as monophyletic in
all analyses. Ofaiston is characterized by having only one or
two stamens and three to five perianth parts (tepals), strongly
keeled bracteoles, and small wings on two tepals (Iljin 1936).
In Petrosimonia, the anther appendages are connate at the
apex, creating a cagelike structure, and wings are completely
absent from the membranous tepals.
The remainder of the Climacoptera clade is strongly sup-
ported (mpbs ¼ 100%; mlbs ¼ 100%). All of the genera
within this clade are para- or polyphyletic, including Clima-
coptera, Gamanthus, Halanthium, and Halimocnemis (figs.
1B, 2B). The majority of the members of several genera form
strongly supported clades, and clear lineages can be defined
in several cases. If Climacoptera brachiata is excluded from
the rest of the genus, Climacoptera forms a strongly sup-
ported monophyletic genus (mpbs ¼ 100%; mlbs ¼ 100%).
Climacoptera was segregated from Salsola by Botschantzev
(1956); however, it was considered as Salsola sect. Physurus
by Freitag (1997). The presence of five winged perianths in
fruit, strongly fleshy, glaucous, and mostly decurrent floral
leaves, a main central erect stem, and an interrupted Kranz
layer on the adaxial leaf surface define the genus. The num-
ber of species in this lineage is unclear, as different authors
have recognized from as many as 42 species (Pratov 1986) to
as few as six (Freitag 1997). Preliminary evidence from Iran
(H. Akhani, unpublished data) suggests that approximately
eight to 10 species are distinguishable, as opposed to only
two species recognized in Flora Iranica by Freitag (1997).
Climacoptera brachiata has been variously treated as a mem-
ber of Salsola sect. Heterotricha Iljin (Iljin 1936), Climacoptera
(Botschantzev 1956), Climacoptera sect. Heterotricha Iljin ex
Pratov (Pratov 1986), and Salsola sect. Belanthera (Freitag 1997).
Based on the characteristic opposite leaves (except uppermost
floral leaves), small obtuse anther appendage, and presence of a
spinulose indumentum with long smooth articulate hairs, it is
well separated from the other genera of the Climacoptera
Morphological and Anatomical Comparison of Genera in the Kaviria Clade
Character KaviriaHalocharis Nanophyton
HabitUndershrub, rarely annual,
erect to ascending
Annual, prostrate Pulvinate undershrub, with
stout woody base, erect
(flowers in each
floral leaf axis)
Spinulose, branched, scabrid,
Solitary to several
Articulated, scabrid hairs,
multicellular flexuous hairs
on axil of flowers
Terete, strongly succulent,
with one or a few bristlelike
hairs at apex
Multicelluar, smooth flexuous
hairs in leaf and flower axils
Leaf shape Terete to semiterete, succulent,
hairy or glabrous throughout
Semiterete to triangular in
section, spiny tipped
VB associated with
Winged or with small
Horizontal, rarely vertical
Hypogynous disk present, without
or with short interstaminal lobes
VB lacking sclerenchymatous
VB associated with
Enlarged and inflated in
fruit but without wing
Filament not narrowed to base,
located at hypogynous disk
with staminode lobes
Anthers divided to the apex,
appendage triangular, smooth
Filaments not narrowed to base,
hypgynous disk absent
AntherTriangular, appendage scabrous Appendage divided to the base,
appendage vesiculous or
Terete, not dentate
High salty clay soils
Flat, shortly dentate
Dry gravelly and slightly salty soils
Terete, not dentate
INTERNATIONAL JOURNAL OF PLANT SCIENCES
clade. Given the isolated phylogenetic position of this species
and its distinctive combination of characteristics, we feel it is
best treated as a monotypic genus, here named Pyankovia
(see app. A) in honor of the late professor Vladimir Pyankov.
The remainder of the species in the Climacoptera clade be-
long to four genera: Gamanthus, Halanthium, Halotis, and
Halimocnemis. The relationships and generic boundaries of
these genera have been debated (Akhani 1996; Hedge 1997;
Assadi 2001; Ghobadnejhad et al. 2004). The phylogenetic
results presented here suggest the possibility of four lineages,
although the relationships among these lineages and clade
membership are generally poorly supported (figs. 1B, 2B),
and whether there are consistent morphological characters
by which to define these clades is unclear. Further, there are
entanglements of the types of some genera (e.g., Gamanthus
pilosus nested within the Halanthium clade, separate from
the rest of Gamanthus). While a case could be made for ei-
ther rejecting the current lectotype of Gamanthus or renam-
ing the rest of the Gamanthus clade under a new name
(given its strong support), we consider the recognition of all
of these species within Halimocnemis to be the best option at
this time, at least until the generic boundaries and nomencla-
tural problems can be untangled. While this group is not
present in the MP strict consensus, it is present in the most
likely tree, albeit with low support (58%). Because Halimoc-
nemis and Halotis are very similar morphologically and were
previously merged by Hedge (1997), and the phylogenetic
hypotheses places Halimocnemis purpureum and Halotis pe-
dunculata among Halanthium species (see Akhani 1996;
Hedge 1997), we feel the combination of these genera to be
a reasonable compromise, despite the low branch support.
The monotypic genus Piptoptera was amplified only for
psbB-psbH, confirming that it is not a well-supported mem-
ber of any of the clades described above, which corresponds
with its morphological isolation. It is weakly placed among
Halanthium and Gamanthus species. This might suggest its
inclusion in the more broadly circumscribed Halimocnemis,
but formal inclusion in that genus will require further data,
particularly given the peculiar features of this genus of sturdy
habit with adpressed indumentum and development of two
large, circular perianthal wings.
This article is the result of a sabbatical leave of H. Akhani
supported by a grant of the University of Tehran. The field-
work was partly supported by the Geobotanical Studies in
Different Parts of Iran I–III research project, and portions of
this project were also supported by Civilian Research and
Development Foundation grant RB1-2502-ST-03. Five of the
sequenced species in this study were provided during a re-
search visit to Royal Botanical Gardens, Kew (Jodrell Labo-
ratory). H. Akhani acknowledges the financial support of
the Royal Society and the help of Mark Chase and other
staff members. We acknowledge Maraym Ghasemkhani for
her help; Ehsan Akhani, A. Beck, R. Khoshravesh, M. Dja-
mali, T. Eftekhari, C. Deigele, and H. Ziegler for providing
material from collections in Munich and Iran; C. M. Wilmot-
Dear (Kew) for correcting the Latin diagnoses; and Larry
Hufford and two anonymous reviewers for comments on a
previous version of the manuscript.
A Revised Classification of Salsoleae s.l.
Here we present a synopsis of generic circumscriptions and new combinations resulting from this study, where both strong
molecular and morphological support necessitate changes. To save space, we have included only the most necessary nomencla-
tural data. Therefore, most synonyms and citations are not included in this article. A detailed morphological, taxonomical, and
anatomical assessment of tribes Camphorosmeae, Caroxyloneae, and Salsoleae awaits future publication.
Tribe Salsoleae s.s.
Anabasis L., Sp. Pl. 223, 1753. Type: Anabasis aphylla L.
Includes Anabasis aphylla L., A. aretioides Moq. & Coss., A. articulata (Forssk.) Moq., A. brevifolia C. A. Mey., A. brachiata Fisch. & C. A.
Mey., A. calcarea (Charif & Aellen) Bokhari & Wendelbo, A. cretacea Pall., A. ebracteolata Korov. ex Botsch., A. ehrenbergii Schweinf. ex
Boiss., A. elatior (C. A. Mey.) Schrenk, A. eriopoda (Schrenk) Benth. ex Volkens, A. eugeniae Iljin, A. ferganica Drob., A. gypsicola Iljin, A.
haussknechtii Bunge ex Boiss., A. iranica Iljin, A. jaxartica (Bunge) Benth. ex Volkens, A. lachnantha Aellen & Rech. f., A. paucifolia M.
Pop. ex Iljin, A. pelliotii Danguy, A. macroptera Moq., A. prostrata Pomel., A. oropediorum Maire, A. salsa (C. A. Mey.) Benth. ex
Volkens, A. syriaca Iljin, A. tianschanica Botsch., A. truncata (Schrenk) Bunge, A. turkestanica Iljin & Korov., and A. turgaica Iljin &
Arthrophytum Schrenk, Bull. Phys. Math. Acad. Petrop. 3: 211, 1845. Type: A. subulifolium Schrenk.
Includes Arthrophytum gracile Aellen, A. iliense Iljin, A. balchaschense (Iljin) Botsch., A. lehmannianum Bunge, A. pulvinatum Litv., A.
subulifolium Schrenk, A. longibracteatum Korov., A. korovinii Botsch., and A. betpakdalense Korov. (Korovin and Mironov 1935).
Cornulaca Delile, Flore d’Egypte—explic. des planches 72, 1813. Type: Cornulaca monacantha Delile.
Includes Cornulaca alaschanica C. P. Tsien & G. L. Chu, C. aucheri Moq., C. ehrenbergii Asch., C. korshinskyi Litv., C. monacantha Delile,
and C. setifera (DC.) Moq. (Aellen 1950; Boulos 1992).
Girgensohnia Bunge ex Fenzl in Ledeb., Fl. Ross. 3: 835, 1851. Type: Girgensohnia oppositiflora (Pall.) Fenzl.
Includes Girgensohnia diptera Bunge, G. imbricata Bunge, G. minima E. Korov., and G. oppositiflora (Pall.) Fenzl.
Halogeton C. A. Mey. in Ledeb., Icon. Pl. Fl. Ross. 1: 10, 1829. Type: Halogeton glomeratus (M. Bieb.) C. A. Mey. Synonyms: Agathophora
(Fenzl) Bunge, Micropeplis Bunge.
AKHANI ET AL.—PHYLOGENY AND CLASSIFICATION OF SALSOLEAE s.l.