Reproductive biology and variation of nuclear ribosomal ITS and ETS sequences in the Calligonum mongolicum complex (Polygonaceae)

Abstract To explore the biosystematics of the Calligonum mongolicum complex (Polygonaceae), the flowering phenological period, breeding and pollination characters and seed set of the complex (Calligonum Mongolicum Turze, Calligonum chinense A. Los., Calligonum gobicum A. Los., Calligonum pumilum A. Los. and Calligonum zaidamense A. Los.) were documented in the Turpan Eremophyte Botanical Garden, China. The sequences of the nuclear ribosomal ITS and ETS region were employed to differentiate the Calligonum mongolicum complex and other species in sect. Medusae. The results showed species of the Calligonum mongolicum complex occupied overlapping flowering periods and had consistent pollination agents. Their breeding systems are all self-compatible, tend to be out-crossing and they interbreed amongst each other (out-crossing index, OCI = 4).The crosses within and amongst species had high seed sets (44 - 65%). Phylogenetic analyses of Calligonum sect. Medusae and the network analysis of nrDNA (ITS and ETS) in the complex suggest interbreeding amongst “species” within the complex and provide evidence for taxonomically merging the five species in the complex. The detected hybridisation, occurring within the complex, suggests the need to improve traditional methods of ex situ plant conservation in botanical gardens for maintaining genetic diversity of Calligonum within and amongst species from different geographic areas.


Introduction
Calligonum L. is widely distributed in Northern Africa, Southern Europe and Western and Central Asia (Bao and Alisa 2003). It is the only genus in Polygonaceae that contains C 4 species (Pyankovet al. 2000) with rapid rates of evolution and diversifi cation (Mabberley 2008). Th e taxonomy of this genus is complex (Xu 1998) and that of the Calligonum mongolicum Turcz. complex is especially diffi cult. Calligonum mongolicum Turcz. is widely distributed from Xilinhot-Inner Mongolia in the east, Kyzyl Kum Desert in Uzbekistan in the west, Milan in Xinjiang in the south, Baitashan, Qitai and Karamay in Xinjiang in the north, with a longitudinal range of about 30° (Pavlov 1936;Drobov 1953;Baitenov and Pavlov 1960;Sergievskaya 1961;Kovalevskaja 1971;Shi et al. 2011). Calligonum pumilum A. Los., C. gobicum A. Los., C. chinense A. Los., C. alashanicum A. Los., C. zaidamense A. Los. and C. roborowskii A. Los. (1927) of the complex occur within the geographic range of C. mongolicum (Losinskaja 1927;Bao and Grabovskaya-Borodina 2003). All of these more narrowly ranged species were merged into C. mongolicum based on the variation of their fruit characters and the chromosome numbers (Soskov 1975a(Soskov , 1975b. However, these species are currently recognised in the Flora of China treatment according to their fruit morphology (Bao and Grabovskaya-Borodina 2003;Mao et al. 1983). Nevertheless the fruits are overall similar, making it diffi cult to distinguish the species of the complex (Soskov 2011, Mao and Pan 1986; Table 1). Analyses of the reproductive biology of the complex are important for resolving the taxonomy and exploring the evolutionary processes (Stebbins 1950;Grant 1992Grant , 1994Oldfi eld 2009).
Studies on the reproductive biology of Calligonum are rare. Kang et al. (2011) assessed information from four taxa (Calligonum calliphysa Bunge, C. rubicundum Bge., C. densum Borszcz and C. ebinuricum Ivanova) which were selected from each section (four sections in Calligonum) and revealed that all the investigated species were selfcompatible but there was no hybridisation amongst them. A few examples of hybridisation were mentioned such as between Calligonum dubjanskyi Litv. and C. bubuyri B. Fedtsch. ex Pavl., between C. acanthopterum and C. leucocladum and between C. acanthopterum Borszcz. and C. leucocladum (Schrenk) Bunge (Soskov 1975b). Th ese reported hybrids occurred between species within a section, including sect. Peterococus and sect. Medusae. Th e taxonomic relationships of the genus have been tested by the applications of several molecular techniques, such as the RAPD markers (Ren et al. 2002) and other chloroplast DNA markers (trnL-F, matK, atpB-rbcL, psbA-trnH, psbK-psbl and rbcL) (Tavakkoli et al. 2010;Sanchez et al. 2011;Abdurahman and Sabirhazi 2012;Sun and Zhang 2012;Li et al. 2014), but the markers employed so far have been ineffi cient for resolving the taxonomic problems in Calligonum. It was expected that reproductive biology and faster-evolving nuclear DNA sequences (Sang 2002;Zimmer and Wen 2012) might shed some light on the taxonomy of the genus.
Th e Calligonum mongolicum complex is almost exclusively diploids with 2n (2x) = 18, except C. roborowskii with 2n (4x) = 36 , although a polyploid count was reported as 2n (3x) = 27 (Shi et al. 2013) in an individual of C. mongolicum. Table 1. Diff erences in fruit characters among species of the Calligonum mongolicum complex according to the treatment in Flora of China, the monograph of Soskov (2011) and the observations by Shi et al. (2011). * NRR = Number of rows of bristles in each rib. Th e situation is markedly diff erent in other species of the Calligonum sect. Medusae which are polyploids with the most frequent chromosome number 2n (4x or 6x) = 36 or 54 (Wang and Yang 1985;Wang and Guan 1986;Shi and Pan 2015). Th e above chromosomal data indicate the signifi cant role of polyploidy in the evolution of the sect. Medusae of Calligonum. Th e fl owering phenology, characters of breeding systems and pollination and fruit set of the C. mongolicum complex (C. mongolicum, C. pumilum, C. chinense, C. alashanicum and C. zaidamense) have been documented by the authors, leaving out the tetraploid C. roborowskii (see also Wen et al. 2016). Th e phylogeny of Calligonum sect. Medusae has been reconstructed using nuclear ribosomal markers (ITS and ETS). Th e new data will be used to discuss the taxonomic implications of the species complex and the conservation strategy of Calligonum in botanical gardens.

Materials and methods
Five species of the Calligonum mongolicum complex (C. mongolicum, C. pumilum, C. chinense, C. alashanicum and C. zaidamense) were selected by the authors, leaving out the tetraploid C. roborowskii. Th ese selected species were brought to Turpan Eremophytes Botanical Garden (TEBG) from their natural habitats during 2011 to 2013 and were planted in the germplasm garden of Calligonum (Table 2, Qi and Pan 2010;Shi et al. 2013).

Collection of phenological information
Phenological information of the Calligonum species was collected from fi eld investigations. Th e phenological observations were made once every two days during the growing period, according to the method of the Chinese Phenological Observation Standard (Zhu and Wan 1973). Th e investigated fl owering phenological periods included fl ower bud appearance, beginning of fl owering, fl ower blooming, end of fl owering and fruit maturity. Th e starting date of a species' growing period was expressed in the day of year (calculated from 1 January of the current year and thereafter). Five plants from each species in the fi eld were randomly selected to document the fl owering phenology and they were observed every day in the blooming and fruiting periods from 2011 to 2013.

Pollen morphology
Scanning electron microscopy (SEM) was used to document the micromorphology of pollen. Samples were dehydrated and were then placed on aluminium stubs using double-sided adhesive tape and sputter coated with gold in a Hitachi E-1010 Ion Table 2. Voucher information for the samples used in the study.

Species
Pop.

individuals (fl owers in an individual) Location
Num. in DNA analysis

Species
Pop.

individuals (fl owers in an individual) Location
Num. in DNA analysis Coordinates

ITS ETS
Sputter Coater, following Wen and Nowicke (1999). Th e materials were subsequently observed and photographed under a Hitachi S-4800 scanning electron microscope. Pollen sizes from both polar view (P) and equatorial view (E) were measured using 10 grains of each sample.

Controlled crossing experiments and observations on fruit and seed sets
Th e breeding systems of the C. mongolicum complex were examined by a hand-pollination test. More than 1600 buds were marked and bagged before opening during the period 2011 to 2013. Each fl ower of an individual plant was randomly assigned to one of the following treatments with each treatment, except hybridisation, including about 30 fl owers in each taxon: i) autonomous pollination: no treatment but just bagging to test self-pollination naturally; ii) selfi ng: test for self-compatibility by bagging and undertaking pollination from the same fl ower; iii) geitonogamous selfi ng: emasculation, bagging and pollination in the same individual but using diff erent fl owers, to test for self-compatibility; iv) crossing: emasculation, bagging and pollination from another individual that was located more than 2m from the recipient v) apomixis: emasculation, bagging but no pollen; vi) natural pollination: emasculation, no bagging; vii) autonomous pollination via geitonogamy: bagging the whole branch; viii) hybridisation: emasculation and cross-pollinations with four other species, each species included 100 fl owers. Th e stigma receptivity time was about 12 hours; and the pollen viability was about 12-24 hours (XS Kang, W Shi and BR Pan, unpublished data).

DNA extraction, amplification and sequencing
Nineteen (19) individuals of six species, C. mongolicum, C. pumilum, C. chinense, C. alashanicum, C. zaidamense and C. calliphysa were sequenced and 24 ITS sequences of Calligonum from GenBank were downloaded (Table 2). Young green branches of each species were collected from natural populations in China (Table 2). Th e samples were collected from adult individuals with green healthy branches (with no signs of parasitism or of drought stress). Th ey were dried in silica gel and kept in a freezer at -25°C. Voucher specimens of the studied material were deposited in the Herbarium of Institute of Ecology and Geography in Xinjiang (XJBI). Total genomic DNAs were extracted from fresh or silica gel dried assimilating branches following the protocol of Doyle and Doyle (1990). In this study, the protocols were followed for obtaining ITS sequences in plants by Wen and Zimmer (1996), Stanford et al. (2000) and Feliner and Rossello (2007). Th e ETS primers were newly designed for the study with the forward primer ETScalli1: 5'-GTTACTTACACTCC-CCACAACCCC-3' and the reverse primer as18SIGS: 5'-GAGACAAGCATATGAC-TACTGGCAGGATCAACCAG-3'. Th e DNA amplifi cations via a polymerase chain reaction (PCR) were performed using 10 ng of genomic DNA, 4 pmol of each primer, 0.5 U Taq polymerase (Bioline, Randolph, MA, USA) and 2.5 mM MgCl 2 in a volume of 25 μL using a PTC-225 Peltier thermal cycler. Th e PCR cycling parameters were as follows: 95 °C initial heating for 5 min, 40 cycles of 94 °C for 30s, 55°C for 45s for ITS (60°C for 40s for ETS) and 72°C for 60s and 72 °C for 10 min for fi nal extension. Th e PCR products were purifi ed using EXO-SapIT (US Biological, Swampscott, MA, USA) and sequenced in both directions using PCR primers. Th e ABI Prism Big Dye Terminator Cycle Sequencing Ready Reaction kit (Applied Biosystems, Foster City, CA, USA) was carried out for cycle sequencing with mixing in a 10 μL reaction volume including 5 ng of primer, 1.5 μL of sequencing dilution buff er and 1 μL of cycle sequencing. Th e conditions were as follows: 35 cycles of 96 °C for 30s denaturation, 50 °C for 30s annealing and 60 °C for 4 min elongation. An ABI 3730xl DNA analyser (Applied Biosystems, Foster City, CA, USA) was used for separating the sequencing products. Both strands of DNA with overlapping regions ensured that each base was double-checked. We assembled the electropherograms and generated the consensus with Sequencher 4.5 (GeneCodes, Ann Arbor, MI, USA).
Sequences were initially aligned using MUSCLE 3.8.31 (Edgar 2004), followed by manual adjustments using GENEIOUS 8.1.2 (Kearse et al. 2012). Th e newly generated sequences from the 20 samples of Calligonum were deposited in GenBank (Table 2). Th e jModeltest 2.1.7 (Posada 2008, Darriba et al. 2012) was used to show the best-fi t model of sequence evolution for each data. Th e Bayesian inferences were run according to the model chosen by the Akaike information criterion (AIC) method. Phylogenetic relationships were inferred using both maximum-likelihood estimation (ML) in RAxML (Stamatakis 2006) and Bayesian inference (BI) in MrBayes 3.1.2 (Ronquist and Huelsenbeck 2003). Bayesian analyses were conducted using the combined ITS and ETS data sets, using partitions of the respective models from the jModeltest. Th e ML analyses also used the partition of the two markers. Th e bootstrap analysis (Felsenstein 1985) was executed with 1000 replicates, with a maximum of 100 trees saved per replicate. Th e Bayesian inference was run with 2,000,000 generations and the Markov chain Monte Carlo (MCMC) run had one cold and three incrementally heated chains. For each dataset, all Bayesian analyses produced split frequencies of less than 0.01 and convergence between the paired MCMC runs were repeated twice to avoid spurious results. Th e remaining trees were used to construct majority-rule consensus trees after discarding the fi rst 2000-5000 trees as burn-in before stationary conditions were established. A neighbour-net analysis was conducted using the uncorrected p-distance between individuals and the programme SplitsTree 4.13.1 (Huson and Bryant 2006). Branch support was tested using bootstrapping with 1000 replicates.

Phenological data
Th e bisexual fl owers occur in groups of two to four in assimilating branches of the Calligonum species. Th e perianth has fi ve tepals, which are green or red with a broad white margin abaxially, ovate, unequal and persistent in fruits. Th e fl ower has 12-18 stamens and the fi laments are connate at the base. Th e pollen presentation pattern is gradual and, when pollen is viable, the stigmas also have receptivity (no dichogamy) (BR Pan, unpublished data).
Th e fi ve Calligonum species fl ower from mid-April to mid-May in the fi eld. Th e duration of C. mongolicum and C. gobicum for fl owering was generally from mid-April to early May, whereas that of C. pumilum, C. chinense and C. zaidamense was from late April to mid-May; individual species of C. mongolicum continued to fl ower sporadically until late May. Th us the blooming period was similar for Calligonum both in fi eld and in TEBG (Figure 1).
Th e blooming periods of the complex overlapped and the percentage overlap was about 80-100% (Figure1). Th e peak fl owering periods of C. mongolicum complex occurred at the same time in early May. Although fl owering was generally ending in early May, fl owering in some individuals of C. mongolicum was still at its peak until mid-May.

Floral visitors
Th e major pollinators for collecting pollen and nectar were Apismellifera L. and Halictus sp., both of which collected pollen in pollen baskets on their third legs and, occasionally, pollen also adhered to their chests and then contacted with the stigmas whilst feeding. Th ese species frequently visited nearby fl owers on the same plant individual and frequent visits on the same fl owers were also undertaken. Other recorded species were nectar thieves including some fl ies (Lasiopticus sp., Musca domestica and Calliphoravicina), butterfl ies (Plebejusargus) and others in Formicidae.

Breeding systems
Th e results of the pollination experiment suggested that species in the complex had analogous mating systems (Tables 3 & 4), as both geitonogamy and cross-pollination conducted by hand yielded better fruit sets compared with natural pollination. Th ey   Figure 2). Th ey interbred amongst each other (OCI = 4). Th e spontaneous self-pollination did not occur because when pollinators were excluded in the bagging treatment, no fruits were produced. It resulted in a very low (if any) fruit set in the self-pollination treatment.   Th e fruit set using geitonogamy treatment shows self-compatibility within each species. Th e apomixis did not occur in these species as exclusion of both pollinators and emasculation did not result in any fruit set.
Hybridisation experiments in the complex resulted in a fruit set and the results (in percentage terms) are shown in Table 4. Th e fl owering of the complex was synchronised. Th e pollen morphology of the fi ve species showed similarities in major pollen characteristics such as shape, size and exine characters (Figure 2, Table 5). Th e hybridisation experiments and interspecifi c hand pollination yielded some viable seeds (Table 5). Th e maximum of the fruit set is amongst the C. mongolicum (65±1.25) and the C. pumilum (65±2.14) themselves; the minimum is that between C. chinense and C. mongolicum (41±1.15). In general, the fruit set amongst the fi ve species is similar (p>0.05).

Phylogenetic analysis
Th e aligned matrix with 45 accessions of nrITS and ETS is 807bp long. Th e Phi test did not fi nd statistically signifi cant (p= 0.0323) evidence for the presence of chimeric sequences in the nrITS and ETS data matrix. Th e nrITS and ETS sequence alignment used for phylogenetic tree reconstruction included 44 sequences: 43 from the in-group and one of C. caput-medusae as the out-group. Th e data sets included 20 newly generated nrITS, 23 ITS sequences from GenBank and 20 new ETS sequences (Table 2).
Th e model test suggested F81 for ETS (nucleotide frequencies A = 0.2023, C = 0.3494, G = 0.2778, T = 0.1706) and TPM2uf for ITS (nucleotide frequencies A = 0.1873, C = 0.3265, G = 0.3277, T = 0.1586; substitution rates: RAC = 0.3484, RAG = 3.4478, RAT = 0.3484, RCG = 1.0000, RCT = 3.4478, RGT = 1.0000). Th e Bayesian inference used the partition of ITS and ETS based on the respective models. Th e ML analyses used GTR+G as the model. Topologies inferred by the two phylogenetic tree reconstruction methods were congruent (Figure 3). Th e most morphologically distinctive C. caput-medusae from Central Asia was used as the out-group, the fi rst diverged clade in the analyses being C. arich (six accessions included, PP 1.00, BS 92) from western Asia, the remaining species forming a large clade A. Of interest, all species from the C. mongolicum complex formed a clade. Th e fi ve species of the C. mongolicum complex, C. ebinuricum and two other species C. roborowskii and C. taklamakan were distributed within the broad geographic region of the C. mongolicum complex, but C. roborowskii and C. taklamakan were of a more restricted distribution in the Taklamakan Basin of Xinjiang province, China. Th e three individuals of C. ebinuricum which form an independent clade, have specifi c fruit characters diff erent from the complex. Th e individuals of Calligonum mongolicum and C. pumilum each did not form a clade, but they were intermixed with C. alashanicum, C. zaidamense and C. chinense, C. roborowskii and C. taklamakan, forming a large clade C (Figure 3). It is of interest to note that the p-distance amongst taxa of Calligonum for the ITS and ETS region is as high as 11.364% between the out-group species C. caput-medusae and C. mongolicum JX259384. Within the clade C, the p-distance was as high as 0.564% between C. ebinuricum and C. mongolicum JX259384. A neighbour-net was constructed for the C. mongolicum complex using ITS and ETS sequences which also supported the complex in one branch (Figure 4).

Evidence for interbreeding of species in the Calligonum mongolicum complex
Species isolation is frequently caused by the temporal heterogeneity of blooming amongst sympatric species (Levin 1971;Adams 1983;Grant 1992Grant , 1994. Th e fl owering periods of fi ve species in the complex showed a high degree of overlapping, with some diff erences in peak blooming periods (also see cases in Wilson 1983;Burd 1995).
Th ese fi ve diploid species of Calligonum have similar pollen characters in both with spheroidal shape and tricolporate apertures with each other (Figure 2). Th e other species in Medusae also have the similar pollen characters but without specifi c pollen indexes (P&E) analysis (Qiu 1988;Gulinuer 2008). Th e hand-pollination tests suggested the fi ve species are self-compatible (geitonogamous, not autophilous). Furthermore, pollinators were necessary for the sexual reproduction in the complex, although some fruit sets were resulted with exclusion of pollinators. Th e results of test crosses suggest the existence of a strong internal hybridisation potential in each of these species.
Crossing compatibility between the species of the C. mongolicum complex is largely the same as that between individuals within the same species (Table 4). Th e crossing behaviour amongst them is consistent with the view from Soskov (1975aSoskov ( , 1975b by treating these various segregate species as one variable biological species of C. mongolicum. Lack of phylogenetic structure and nrDNA sequence variation as indirect evidence for interbreeding in the C. mongolicum complex Although phylogenetic inference based on nrITS needs to be considered carefully (Alvarez andWendel 2003, Feliner andRossello 2007), some conclusions may be drawn based on the ITS and ETS analyses of the target species. As shown by the ML and Bayesian trees of nrITS and ETS sequences (Figure 3), a striking divergence exists between C. arich (clade A) and other species. Yet species of the C. mongolicum complex had very similar or identical sequences (Clade C in Figure 3). Th e nrITS and ETS tree together with the network of ribotypes (Figure 4) suggest the lack of phylogenetic structure within the complex. Excluding C. arich (5 individuals), C. ebinuricum (3 individuals) can be easily diff erentiated from the C. mongolicum complex (13 individuals) (Figures 3 and 4). Th e intermixed patterns of sequences from diff erent "species" of the C. mongolicum complex may indicate past or present introgressive potential of the C. mongolicum complex and argues for the existence of hybridisation or interbreeding (if these "species" represent the same taxon).

Implications on taxonomy and conservation of Calligonum
Calligonum is one of the medium-sized genera of Polygonaceae with approximately 60-80 species and represents a rapid diversifi cation in the hot and arid deserts of Central Asia to western China (Mabberley 1990). Molecular analyses of both nrDNA ITS and some cpDNA sequences (trnL-F, matK, atpB-rbcL, psbA-trnH, psbK-psbL and rbcL) have not resolved relationships amongst species of Calligonum (Sanchez et al. 2011, Sun and Zhang 2012, Li et al. 2014. Our study showed that C. ebinuricum possesses highly distinct nrITS sequences (Figures 3 & 4); yet the ITS and ETS sequences of the C. mongolicum complex generated a topology with the species of the complex highly intermixed with each other in the tree. Th e authors' results both in this paper and in their previous studies , 2013, Shi and Pan 2015 argue for the merging of C. chinense, C. gobicum, C. pumilum and C. zaidamense with C. mongolicum as proposed by Soskov (1975aSoskov ( , 1975b. Detailed evidence was also recently presented on merging C. pumilum with the more widespread C. mongolicum (Shi et al. 2016). Detailed morphological comparisons of the other species in the complex will be pursued by the authors as was done for C. pumilum and C. mongolicum (Shi et al. 2016) and the phylogeographic structure of the complex will be further explored with phylogenomic methods .
Distributional ranges of some species in clade C ( Figure 3) do not overlap but are geographically close or adjacent to each other. Calligonum roborowskii (2n=36) grows at the edge of Taklamakan basin; C. taklamakan occurs in the central part of the basin; and the other species in the complex except C. mongolicum are confi ned to the south-eastern edge of the basin and C. ebinuricum is in North Xinjiang and also in Mongolia but never in South Xinjiang. According to their morphological comparisons (Gulinuer 2008, Kang et al. 2008, the taxonomic relationship of C. ebinuricum and C. taklamakan with other species needs further analyses. Th e fact that most of the collected seeds can germinate without any pre-treatment suggests that the fi ve Calligonum species produce enough seeds to renew the populations. On the other hand, the ex-situ conservation of genetic diversity for the long-term survival of species of Calligonum needs a new management strategy due to their reproductive biology and the potential for hybridisation/interbreeding (Kramer andHavens 2009, Swarts andDixon 2009). Special eff orts are needed to ensure isolation of genetic sources in ex situ conditions.