The Phylogeny of Calligonum and Pteropyrum (Polygonaceae) Based on Nuclear Ribosomal DNA ITS and Chloroplast trnL-F Sequences

INTRODUCTION

Calligonum possesses 80 species with xeromorphic shrubby characteristics distributed throughout Southern Europe, North Africa and Western and Central Asia as its main biodiversity center (Brandbyge, 1993; Mabberely, 1990). This genus is well distinguished from other genera of the family by the higher number of stamens (12-15) and four carpels/stigmas as well as C4 photosynthesis (Sage, 2004; Pyankov et al., 2000; Brandbyge, 1993). Eighteen species including six endemic ones have been identified among the flora of Iran (Mozaffarian, 2004; Rechinger and Schiman-Czaika, 1968). According to fruit morphology the genus has been divided into three sections: Calligonum (with bristled fruit), Pterococcus (with winged fruit) and Calliphysa (with membranous saccate fruit) (Rechinger and Schiman-Czaika, 1968). The genus Pteropyrum has 4 or 5 species in south west Asia and the Middle East, of which 3 species are distributed in Iran. The members of this genus like that of Calligonum are shrubs but have achenes with only 3 membranous wings. Pteropyrum in particular Calligonum are typical arid and hot desert plants in active sand dunes and playing the key role in the stability of desert natural vegetation ecosystem (Ren and Tao, 2004; and personal observations).
The climatic distribution pattern of Calligonum is similar to that of C4 large shrubby chenopod species (Haloxylon persicum, H. ammodendron and Salsola richteri) with NADP-ME metabolism and salsoloid assimilation organ anatomy (Pyankov et al., 2000).
Tavakkoli et al. (2008) conducted a morphology based-phylogenetic analysis of these taxa to test their relationships and monophyly. Sanchez and Kron (2008) using cpDNA sequences (rbcL and matK), and then Sanchez et al. (2009) using those genes as well as ndhF plus nrDNA ITS regions showed that monophyly of Calligonum is controversial. Their analyses revealed that Calligonum, Pteropyrum and Pteroxygonum form a weakly to moderately supported clade.
In this study, therefore, phylogenetic analyses were performed using both nrDNA ITS (ITS1, 5.8S and ITS2) and cpDNA trnL-F sequence data to address the following questions:
1) Are Calligonum and Pteropyrum both monophyletic? 2) Are Calligonum and Pteropyrum closely related taxa? 3) Is the current sectional classification of Calligonum supported? and 4) are these genera related to Atraphaxis?

MATERIALS AND METHODS

Taxon sampling: A total of 27 accessions representing 26 species were included in phylogenetic analyses using nrDNA ITS region and 28 accessions representing 27 species were analyzed for cpDNA trnL-F regions. Eleven species including Calligonum (6 species), Pteropyrum (3 species) and Atraphaxis (2 species) were sequenced for both regions newly. The remaining sequences were obtained from GenBank. The leaf/branchlet material was taken mostly from herbarium specimens deposited at the herbarium of the Research Institute of Forests and Rangelands (TARI). In some cases, the materials were collected from the field. Information on the accessions used in this study is presented in Table 1. According to Lamb Frye and Kron. (2003), Triplaris americana was chosen as an outgroup.

DNA extraction, amplification, and sequencing: Total genomic DNA was isolated from dried leaf or branchlet (only for Calligonum) of samples using the modified cetyl trimethylammonium bromide (CTAB) method of Doyle and Doyle (1987). The amplification of nrDNA ITS and trnL-F regions by polymerase chain reaction (PCR) were performed in a 25 ml reaction mixture, containing 16 ml of sterile water, 2 ml of 2.5 mM MgCl2 , 2.5 ml of 10X Gene Taq universal buffer (Cinnagen, Iran), 2.5 µl of 2.5 mM dNTPs mixture (Wako Nippon Gene, Japan), 0.5 ml of each primer (5 pmol/ml), 0.2 ml (4 Units) of Taq DNA polymerase (Cinnagen, Iran), and 1-1.3 µl of genomic DNA template (approximately 20 ng) using primer pair “ITS1F” (Navajas-P΄erez et al., 2005) and ITS4 (White et al., 1990). The trnL-F region was amplified using the universal “c” and “d” primers of Taberlet et al. (1991). The PCR condition, performed in a DNA thermal cycler (Primus 96, MWG, Germany), was 2.5 min at 95°C for initial denaturation followed by 38 cycles of 1 min at 95°C, 45 sec at 53°C for annealing, 2 min at 72°C for extension, followed by a final 7 min incubation at 72°C. The quality of PCR products were checked by electrophoresis on an 0.8% (w/v) agarose gel (using 1X TAE as the gel buffer) stained with ethidium bromide and then visualized under UV light. Each region was sequenced using the ‘Big dye terminator cycle sequencing ready reaction kit’ (Applied Biosystems, USA). with the appropriate primers in an ABI Prism 377 DNA sequencer (Applied Biosystems, USA).

Sequence alignment: Sequences were edited using BioEdit ver. 7.0.9.0 (Hall, 1999) and aligned using ClustalX (Larkin et al., 2007) followed by manual adjustment. Alignment of each dataset required the introduction of numerous single and multiple-base indels (insertions/deletions). Positions of indels were treated as missing data for all datasets.

Phylogenetic analyses
Maximum parsimony method: Maximum Parsimony (MP) analyses were conducted using the PAUP* program version 4.0b10 (Swofford, 2002) for phylogenetic analyses. The heuristic search option was employed for each of the datasets, using tree bisection-reconnection (TBR) branch swapping, with 1000 replications of random addition sequence and an automatic increase in the maximum number of trees. Uninformative characters were excluded from the analyses. Branch support values were calculated using a full heuristic search with 1000 bootstrap replicates (Felsenstein, 1985) each with simple addition sequence. Combinability of these two datasets was assessed using the partition homogeneity test (the incongruence length difference (ILD) test of Farris et al.,1995) as implemented in PAUP*. The test was conducted with exclusion of invariant characters (Cunningham, 1997) using the heuristic search option involving simple addition sequence and TBR branch swapping with 1000 homogeneity replicates.

Bayesian method: Models of sequence evolution were selected using the program MrModeltest version 2.3 (Nylander, 2004) as implemented in MrMTgui (Nuin, 2005) based on the Akaike information criterion (AIC) (Posada and Buckley, 2004). On the basis of this analysis, datasets were analyzed using the GTR+I+G and GTR+G models for nrDNA ITS and trnL-F sequences, respectively. The combined sequences for 25 taxa were analyzed as a single partition with the GTR+I+G model. The program MrBayes version 3.1.2 (Ronquist and Huelsenbeck, 2003) was used for the Bayesian phylogenetic analyses. Posteriors on the model parameters were estimated from the data, using the default priors. The analysis was carried out with 2 million generations, using the Markov chain Monte Carlo search. MrBayes performed two simultaneous analyses starting from different random trees (Nruns=2) each with four Markov chains and trees sampled at every 100 generations. The first 25% of trees were discarded as the burn-in. The remaining trees were then used to build a 50% majority rule consensus tree accompanied with posterior probability (PP) values using. Tree visualization was carried out using Tree View version1.6.6 (Page, 2001).

RESULTS

Individual sequence data: The nr|DNA ITS dataset is 749 nucleotide sites long, of which 305 sites are potentially parsimony informative. The length of the nrDNA ITS ranges from 507 bp in Atraphaxis spinosa to 607bp in 6 studied species of Calligonum. A 50% majority-rule consensus tree obtained form Bayesian analysis with posterior probabilities and bootstrap values is presented in Figure 1. This tree is, topologically, almost the same as the single most parsimonious tree resulting from the MP method (data not shown), except that the Fagopyrum/Parapteropyrum clade is part of a weakly supported trichotomy comprising the Calligonum/Pteropyrum and Persicaria clades (Fig. 1).
The trnL-F dataset consisted of 28 accessions with 1201 aligned nucleotide sites, of which 224 sites are potentially parsimony-informative. The length of the trnL intron rang from 513 bp in Pteropyrum aucheri to 644 bp in Atraphaxis spinosa and the the length of partial trnL-trnF intergenic spacer ranges from 162 bp in A. spinosa and A. suaedifolia to 288 bp in Pteropyrum aucheri. A 50% majority-rule consensus tree from Bayesian analysis is presented in Figure 2. This tree is, topologically, the same as the strict consensus of three most parsimonious trees (data not shown). In this tree, Calligonum and Pteropyrum are closely related sister taxa and, in turn, are well allied with the Rumex/Polygonum/Atraphaxis/ Fallopia clade.
The combined sequence data: The partition homogeneity test suggested that the nrDNA ITS and trnL-F datasets were not congruent (P=0.012). The main topological differences between the two gene phylogenies are the positions of the Calligonum/Pteropyrum clade and Pteroxygonum. In spite of these conflicts and following the suggestions of several other researches (Yoder et al., 2001; Reeves et al., 2001; Wiens, 1998; Seelanan et al., 1997) that the ILD test may be unreliable, we combined these data sets directly. The combined dataset was 1952 nucleotide sites long, of which 507 were parsimony informative. A 50% majority-rule consensus tree from Bayesian analysis of the combined dataset is presented in Figure 3. This tree is topologically similar to the trnL-F tree. This Bayesian tree is, however, topologically and statistically well resolved and supported than the strict consensus tree of three most parsimonious trees does (not shown). In Bayesian tree, Pteroxygonum is weakly allied with a clade of Fagopyrum and Parapteropyrum. Calligonum and Pteropyrum are again sister taxa and formed a sister group relationship with a clade of Rumex/Polygonum/Atraphaxis/ Fallopia.
DISCUSSION

Monophyly and relationships of Calligonum and Pteropyrum: Phylgenetic analyses of nrDNA ITS, trnL-F and combined nrDNA ITS-trnL-F datasets revealed that Calligonum and Pteropyrum are each monophyletic and closely related taxa. In a recent cladistic analysis of morphological data, these closely related taxa also appeared to be monophyletic (Tavakkoli et al., 2008). The previous studies using non-molecular evidence suggested their closest relationships. such evidences include vegetative morphology: polyachanthic life-form; stem anatomy: lignified secondary walls in the collenchyma, (Haraldson, 1978); floral characters: formation of commissural veins, non-fused petaloid tepals, irregular tepal epidermal cells and presence of papillae at the base of the stamen filament (Hong et al., 1998; Ronse-Decraene and Akeroyd, 1988); fruit morphology: armed achenes with wings/bristles (Brandbyge, 1993; Rechinger and Schiman-Czeika, 1968); and pollen morphology: microreticulate/perforate exine sculpturing (Hong, 1995; Nowicke and Skvarla, 1979).
Sanchez and Kron (2008) and Sanchez et al. (2009), based on combined rbcL-matK-ndhF sequences, have reported that Calligonum is not monophyletic with both Pteropyrum (P. aucheri and P. olivierii) and the monotypic Pteroxygonum (Pt. giraldii) nested within it. But, on the nrDNA ITS and the combined cpDNA-nrDNA ITS trees of Sanchez et al. (2009), Calligonum formed a well supported clade and weakly allied with Pteropyrum solely and Pteropyrum-Pteroxygonum, respectively. On the chloroplast tree resulting from their analysis Pt. giraldii is of a long branch (due to ndhF sequences, see below).
It is worthy to note that our alignment of both matK and rbcL sequences retrieved from GenBank (determined by Sanchez et al., 2009), shows that Pt. giraldii has almost the same sequences as that of four out of five Calligonum species examined, while Calligonum microcarpum has the same ones as that of Petropyrum. But, ndhF sequences of Pt. giraldii is completely different from that of Calligonum species. This indicates that Sanchez et al. (2009) might, most probably, determine the sequences of these taxa mistakenly. Meanwhile, the nrDNA ITS sequences of both Calligonum and Pteropyrum determined by Sanchez et al., are full of base-calling errors. All available evidence including growth habit (climbing stem with petiolate leaves), lower stamen number (usually 8), achene morphology (with three sharp horns at the base), chromosome base number (x=20), tepal venation (trifid) and perhaps the C3 photosynthetic pathway indicate that Pteroxygonum has no affinity with Calligonum (Sun et al., 2008; Li and Grabovskaya-Borodina, 2003; Ronse-Decraene and Akeroyd, 1988; Haraldson, 1978). This is consistent with our molecular analysis, and the nrDNA ITS tree of Sun et al. (2008), Pteroxygonum is a distinct lineage, sister to Polygonum (=Persicaria).
On the basis of growth anatomy (Haraldson, 1978) and floral characters (Hong et al., 1998; Ronse-Decraene and Akeroyd, 1988), it has been hypothesized that Calligonum and in particular Pteropyrum is related to Atraphaxis L. However, neither pollen morphological data (Hong, 1995; Nowicke and Skvarla, 1979) nor our molecular phylogenetic analyses support these hypotheses. rbcL (Lamb Frye and Kron, 2003), matK (Kim and Park, 2005), the cpDNA (rbcL-matK-ndhF) and nrDNA ITS (Sanchez et al., 2009) and the present nrDNA ITS and trnL-F phylogenies show that Atraphaxis related to Polygonum (and Polygonella) as well as Fallopia. On the other hand, Hong (1995) and Hong et al. (1998) based on pollen morphology and tepal surface morphology suggested a close relationship between Pteropyrum and the monotypic genus Parapteropyrum (P. tibeticum), endemic to the Xizang Plateau (Southeast Tibet) of China. Whereas, in our phylogenetic analyses and Sanchez et al.’s (2009) nrDNA tree, such relationship was not appeared. Parapteropyrum tibeticum is instead well allied with Fagopyrum (PP=100, BS=100).
In short, Calligonum and Pteropyrum have no a single relative genus, but as appeared in our trnL-F and the combined phylogenies as well as in Sanchez et al.’s (2009) ones, they are related with a clade of Rumex/Polygonum/Atraphaxis/ Fallopia.
 
Infrageneric relationships in Calligonum and Pteropyrum: As mentioned in the introduction, based on fruit morphology, Calligonum has been divided into three sections: Calligonum (with bristled fruit), Pterococcus (with winged fruit) and Calliphysa (with membranous saccate fruit) (Rechinger and Schiman-Czaika, 1968). Our molecular analyses of both nrDNA ITS and trnL-F did not resolve relationships among proposed sections of Calligonum. This is almost consistent with the combined rbcL and matK phylogeny of Sanchez and Kron (2008, and http://users.wfu.edu/sanca5/Adriana_Sanchez/Adriana_Sanchez/Calligonum.html; but see Sanchez et al. 2009). Lacking the resolution among Calligonum taxa is due to very low nucleotide substitution in both nrDNA ITS and trnL-F sequences However, in the Bayesian tree of combined nrDNA ITS-trnL-F sequences, Calligonum junceum (sect. Calliphysa) and C. persicum (sect. Pterococcus) formed a weakly supported subclade (posterior probabilities (PP) = 0.55, see Fig. 3). Ren and Tao (2004), using randomly amplified polymorphic DNA (RAPD) analyses of 14 Chinese Calligonum species, showed that C. junceum is positioned far from the remaining species studied and species having bristled fruit (sect. Calligonum) were not grouped in a single cluster as were the winged fruit species (sect. Pterococcus). In contrast to Pteropyrum, Calligonum, as one of the big genera in Polygonaceae with approximately 80 species (Mabberely, 1990), represent a rapid diversification for a short time in hot and arid deserts of Western Central Asia. Its diversification may caused by several factors including C4 photosynthetic pathway, bristled/winged fruits a dispersal units and hybridization/introgression followed by tetraploidy (Sage, 2004; Pyankov et al., 2000; Aparicio, 1989; Mao et al., 1983; Pavlov 1970; Rechinger and Schiman-Czaika, 1968).
Pteropyrum aucheri and P. olivierii are morphologically very similar to each other except that in the former, the shape of the leaf is linear and in the latter spatulate (Rechinger and Schiman-Czeika, 1968). Both species are distributed in arid regions, but P. aucheri has penetrated more to desert areas (Mozaffarian, 2004, and personal observations). However, the present molecular phylogenies and morphological cladistic analyses (Tavakkoli et al., 2008) did not put them close to each other. Instead, P. olivierii is allied with P. naufelum, a newly described species distributed in Iraq (Al-Khayat, 1990) and Southwest Iran (Akhani, 2004).

CONCLUSION

The present molecular data provide strong support for the monophyly of Calligonum and Pteropyrum and their closest relationship. They have no affinity with Atraphaxis, instead allied with a clade in which, the genus is nested. However, infrageneric relationships in Calligonum, due to the paucity of informative nucleotide sites in both nrDNA ITS and trnL-F sequences and in other cpDNA genes such as rbcL and matK (Sanchez and Kron, 2008), is not resolved. Fast evolving genic regions including non-coding cpDNA fragments and single copy nuclear DNAs are clearly required to resolve phylogenetic relationship among Calligonum species.

Acknowledgements

This work was supported by a research grant from the Tarbiat Modares University.