The genus Cicer has been traditionally classified into two subgenera (Pseudononis and Viciastrum) and four sections (Cicer,Chamaecicer,Polycicer and Acanthocicer) based on morphological traits and geographical distribu- tion (Popov, 1929; van der Maesen, 1972, 1987). Morphological homoplasy (i.e. life cycle and flower size) and lack of diagnostic synapomorphies hinder sectional monophyly based on morphology in the infrageneric classification of van der Maesen (1972, 1987). Therefore, molecular sequence data were the obvious choice to try to find a robust phylogeny on which to base future taxonomic classifications.
Despite recent intensive molecular studies on the genus Cicer (i.e. Kazan and Muehlbauer, 1991; Ladbi et al., 1996; Ahmad, 1999; Iruela et al., 2002; Sudupak et al., 2002, 2003; Javadi and Yamaguchi, 2004b), little attention has been given to the aspect of infrageneric classification of the genus at molecular level.
The chloroplast and nuclear genomes have been extensively surveyed to reconstruct plant phylogeny (e.g. Olmstead and Palmer, 1994; Sang, 2002).
In legumes, the chloroplast trnK/matK and trnS–trnG and also the internal transcribed spacer (ITS) region of nuclear DNA markers provided very useful information at lower taxonomic levels (e.g. Steele and Wojciechowski, 2003;
Frediani and Caputo, 2005; Kenicer et al., 2005). A combination of both chlo- roplast and nuclear DNA sequences can provide complementary information on the evolution of the genus. Analysis of the chloroplast genome will provide information about the maternal evolutionary relationships between the spe- cies, and the biparentally inherited nuclear DNA will provide independent data from which to infer evolutionary relationships.
In the present study the nuclear ITS and the chloroplast trnK/matK and trnS–trnG regions were used to investigate phylogeny of Cicer, which provides a frame for the reclassification of the genus. Twenty-nine Cicer species repre- senting two subgenera and all four traditionally recognized sections were sam- pled. We selected Lens ervoides (Brign.) Grande, Pisum sativum L. and Vicia sativa L. (tribe Vicieae) as outgroups.
Total genomic DNA was extracted from freshly collected leaf from plants grown in pots in Osaka Prefecture University greenhouse or leaf material from herbarium specimens using a cetyltrimethyl ammonium bromide (CTAB) pro- tocol with minor modification (Doyle and Doyle, 1987).
Polymerase chain reaction (PCR) amplification and sequencing reac- tions of the plastid and nuclear regions were performed using the following universal primers: the trnK/matK region was amplified using a primer pair of trnK685F/trnK2R* and trnK1L/matK1932R, and the region was sequenced using trnK685F, trnK2R, matK4La, matK1100L and trnK1L primers (Hu et al., 2000). ITS4 and ITS5 primers (White et al., 1990) were used both for amplifi- cation and sequencing of the ITS1 and ITS2 spacer along with the 5.8S gene.
The trnS–trnG region amplified and sequenced using trnS-F and trnG-R prim- ers (Xu et al., 2000). DNA amplifications, purification and cycle sequencing were performed following Javadi and Yamaguchi (2004a).
Phylogenetic analyses were performed with phylogenetic analysis using parsimony (PAUP*) 4.0b10 (Swofford, 2002) using the parsimony analysis. A heuristic tree search was conducted using MULTREES, STEEPEST DESCENT off, and ACCTRAN optimizations and tree-bisection-reconnection (TBR) branch swapping. Searches were replicated 1000 times. Clade support was assessed using bootstrap value (Felsenstein, 1985) implemented in PAUP*4.0b10 (Swofford, 2002) using 1000 replicates. An incongruence length difference (ILD;
Farris et al., 1995) test was conducted using PAUP* version 4.0b10 (Swofford, 2002) to determine the congruence between plastid and nr ITS data-sets. The test was performed with 100 replicates, using a heuristic search option with random taxon addition and TBR branch swapping.
The ILD test result (P = 0.1) indicates an acceptable degree of congruence among the three components of the combined data-set (Farris et al., 1995).
The combined nuclear and plastid data matrix consisted of 3850 characters, of which 420 were potentially informative. The combined parsimony analyses, with gaps included, produced 100 equally parsimonious trees of 1085 steps with a consistency index of 0.877 (0.792 excluding uninformative characters) and retention index of 0.878. The strict consensus tree is shown in Fig. 2.1.
The maximum parsimony analyses of combined nrDNA and cpDNA data indicate that three major clades correspond to biogeographic relation- ships in the genus Cicer and contradict the monophyly of each section (Fig.
2.1). Clade A (African group, bootstrap support 100%) includes African spe- cies,C. cuneatum (section Cicer) and C. canariense (section Polycicer). Clade B (west-central Asian group, bootstrap support 99%) contains species belong- ing to four sections mainly distributed at high altitudes of western and Central Asia and sister group species of eastern Europe that are growing at low alti- tudes (Aegean-Mediterranean group) (Fig. 2.1). Clade C (Mediterranean group, bootstrap support 99%) consists of six species in section Cicer (C. arietinum, C. echinospermum, C. reticulatum,C. bijugum,C. judaicum and C. pinnatifi- dum) and one in section Chamaecicer (C. incisum) which are grouped into two subclades (C1 and C2) and distributed in the eastern part of the Mediterranean region (Fig. 2.1).
What can we tell about infrageneric classification of the genus Cicer from the present study? In section Cicer (subgenus Pseudononis), the monophyly of the six species in clade C supports the cpDNA trnT-F sequence results (Javadi and Yamaguchi, 2004a, 2005). Section Cicer is characterized by small flow- ers, imparipinnate leaves or rachis ending in a tendril, and erect or prostrate stems (van der Maesen, 1972, 1987). All these characters appear elsewhere in the genus. According to our results, section Cicer is polyphyletic and its traits will need to be reanalysed or weighted differently to define monophyletic subgroups around the type species of section Cicer.
The small section Chamaecicer (subgenus Pseudononis) with two species, C. chorassanicum (annual) and C. incisum (perennial), shows conflicting pat- tern in molecular data (clades C and B, Fig. 2.1). Non-monophyly of section Chamaecicer is also congruent with geographical distribution pattern of these taxa. The trifoliolate species C. chorassanicum is a taxon of north and central Afghanistan, and north and north-east Persia, and grows under dry conditions at high altitude where members of sections Polycicer and Acanthocicer (van der Maesen, 1972) are quite common. The placement of C. incisum (section Chamaecicer) in clade C is supported by its morphological traits and geograph- ical pattern. The appearance of section Chamaecicer in two distinct clades based on molecular data calls into question the integrity of the section.
SectionPolycicer (subgenus Viciastrum) has a widely distributed pattern that includes members of section Acanthocicer (subgenus Viciastrum) at high altitude in Persia and Central Asia. Morphologically, species assigned in section Acanthocicer share the tragacanthoid and spiny plant shape (van der Maesen, 1972, 1987), which is related to their dry habit. However, our results suggest that section Acanthocicer is not monophyletic; four species belong to section Acanthocicer (C. tragacanthoides,C. subaphyllum,C. macracanthum and C. pungens) and are grouped with members of section Polycicer (clade B, Fig. 2.1).
C. montbretii, C. floribundum, C. isuaricum and C. graecum (section Polycicer) form a well-supported monophyletic group at the base of clade B. They also share morphological features such as erect habit, entirely dentate margins of leaflet (except near the base), inflorescence with 2–5 flowers and arista with clavate leaflet at the tip (van der Maesen, 1972). Biogeographically, these four
species are distinct within the section in having relatively humid forest distribu- tions in eastern Europe and southern Turkey at lower altitudes (0–1700 m). This group was treated as a series, Europaeo-Anatolica by van der Maesen (1972);
now we distinguish it informally as the Aegean-Mediterranean group.
The two African species C. canariense (section Polycicer) and C. cuneatum (sectionCicer) form a highly supported basal clade (bootstrap support 100%) in the phylogenetic tree (Fig. 2.1), which is in agreement with the previous studies (Iruelaet al., 2002; Javadi and Yamaguchi, 2004a,b; Frediani and Caputo, 2005).
These two species are also morphologically distinct and share the synapomor- phic trait of a climbing habit. Geographically, they are growing in isolated areas
C.arietinum C.echinospermum C.reticulatum C.bijugum C.judaicum C.pinnatifidum
C.chorassanicum C.incisum
C.yamashitae C.atlanticum C.oxyodon C.pungens C.kermanense C.tragacanthoides C.microphyllum C.nuristanicum C.songaricum C.spiroceras C.subaphyllum C.stapfianum C.macracanthum C.multijugum
C.floribundum C.graecum C.isauricum C.montbretii C.canariense C.cuneatum Lens ervoktes Vicia sativa Pisum sativum Outgroup C.flexuosum SectionAcanthocicer
SubgenusCicer = Pseudononis SubgenusViciastrum
100
100
92 73
C 99
100
100 99
93
99
96 90 58
92
87
100
95 100
A 100
68 B
Mediterranean Group
West-central Asian group
Aegean mediterranean group
African group SectionCicer
SectionPolycicer SectionChamaecicer
Fig. 2.1. Strict consensus tree of 100 maximum parsimonious trees of combined nrITS, cp trnK/matK and trnS–trnG data (Tree length = 1085, CI = 0.877 and RI = 0.878).
in Africa. C. canariense, endemic to the Canary Islands, was first described by Santos Guerra and Lewis (1985) and was placed in a new monospecific sub- genusstenophyllum (Santos Guerra and Lewis, 1985). C. cuneatum is distrib- uted in Ethiopia, north-east Sudan, south-east Egypt and Saudi Arabia (van der Maesen, 1972, 1987). Our analyses suggest the exclusion of C. cuneatum and C. canariense from sections Cicer and Polycicer, respectively. It appears that this small group of species is well differentiated, with a number of molecular and morphological (vetch-like) synapomorphies making them appear quite different.
Sampling another African species, C. atlanticum, may also provide better insight into the differentiation of the African species in the genus Cicer.
The current molecular study from both chloroplast and nuclear regions of Cicer has demonstrated that the intuitive classification systems devised for the genus in the past (van der Maesen, 1972, 1987) inadequately reflect the natural groupings within the genus. In general, morphological features for delimiting sections show high homoplasy, but ecology and geographical habitat are the most important features to express relationships and are highly congruent with molecular data.