Phenological information of the Calligonum species was collected from field investigations. The 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). The investigated flowering phenological periods included flower bud appearance, beginning of flowering, flower blooming, end of flowering and fruit maturity. The 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 field were randomly selected to document the flowering phenology and they were observed every day in the blooming and fruiting periods from 2011 to 2013.

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 Sputter Coater, following Wen and Nowicke (1999). The 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.

The 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 flower of an individual plant was randomly assigned to one of the following treatments with each treatment, except hybridisation, including about 30 flowers in each taxon: i) autonomous pollination: no treatment but just bagging to test self-pollination naturally; ii) selfing: test for self-compatibility by bagging and undertaking pollination from the same flower; iii) geitonogamous selfing: emasculation, bagging and pollination in the same individual but using different flowers, 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 flowers. The 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).

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). The samples were collected from adult individuals with green healthy branches (with no signs of parasitism or of drought stress). They 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). The ETS primers were newly designed for the study with the forward primer ETScalli1: 5’-GTTACTTACACTCCCCACAACCCC-3’ and the reverse primer as18SIGS: 5’-GAGACAAGCATATGACTACTGGCAGGATCAACCAG-3’. The DNA amplifications 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 MgCl₂ in a volume of 25 µL using a PTC-225 Peltier thermal cycler. The 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 final extension. The PCR products were purified using EXO-SapIT (US Biological, Swampscott, MA, USA) and sequenced in both directions using PCR primers. The 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 buffer and 1 µL of cycle sequencing. The 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). The newly generated sequences from the 20 samples of Calligonum were deposited in GenBank (Table 2). The jModeltest 2.1.7 (Posada 2008, Darriba et al. 2012) was used to show the best-fit model of sequence evolution for each data. The 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. The ML analyses also used the partition of the two markers. The bootstrap analysis (Felsenstein 1985) was executed with 1000 replicates, with a maximum of 100 trees saved per replicate. The 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. The remaining trees were used to construct majority-rule consensus trees after discarding the first 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.

The bisexual flowers occur in groups of two to four in assimilating branches of the Calligonum species. The perianth has five tepals, which are green or red with a broad white margin abaxially, ovate, unequal and persistent in fruits. The flower has 12-18 stamens and the filaments are connate at the base. The pollen presentation pattern is gradual and, when pollen is viable, the stigmas also have receptivity (no dichogamy) (BR Pan, unpublished data).

The five Calligonum species flower from mid-April to mid-May in the field. The duration of C. mongolicum and C. gobicum for flowering 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 flower sporadically until late May. Thus the blooming period was similar for Calligonum both in field and in TEBG (Figure 1).

Figure 1. 

The phenological phases of the Calligonum mongolicum complex. 1 C. mongolicum 2 C. chinense 3 C. gobicum 4 C. pumilum and 5 C. zaidamense.

The blooming periods of the complex overlapped and the percentage overlap was about 80–100% (Figure 1). The peak flowering periods of C. mongolicum complex occurred at the same time in early May. Although flowering was generally ending in early May, flowering in some individuals of C. mongolicum was still at its peak until mid-May.

The 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. These species frequently visited nearby flowers on the same plant individual and frequent visits on the same flowers were also undertaken. Other recorded species were nectar thieves including some flies (Lasiopticus sp., Musca domestica and Calliphoravicina), butterflies (Plebejusargus) and others in Formicidae.

The 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. They also had similar pollen characters and indices (P & E) (Table 5 & Figure 2). They interbred amongst each other (OCI = 4). The 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. The fruit set using geitonogamy treatment shows self-compatibility within each species. The apomixis did not occur in these species as exclusion of both pollinators and emasculation did not result in any fruit set.

Figure 2. 

Equatorial view of pollen grains of the Calligonum mongolicum complex under SEM micrographs.1 C. mongolicum 2 C. chinense 3 C. gobicum 4 C. pumilum and 5. C. zaidamense.

Hybridisation experiments in the complex resulted in a fruit set and the results (in percentage terms) are shown in Table 4. The flowering of the complex was synchronised. The pollen morphology of the five species showed similarities in major pollen characteristics such as shape, size and exine characters (Figure 2, Table 5). The hybridisation experiments and interspecific hand pollination yielded some viable seeds (Table 5). The 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 five species is similar (p>0.05).

The aligned matrix with 45 accessions of nrITS and ETS is 807bp long. The Phi test did not find statistically significant (p= 0.0323) evidence for the presence of chimeric sequences in the nrITS and ETS data matrix. The 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. The data sets included 20 newly generated nrITS, 23 ITS sequences from GenBank and 20 new ETS sequences (Table 2).

The 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). The Bayesian inference used the partition of ITS and ETS based on the respective models. The ML analyses used GTR+G as the model. Topologies inferred by the two phylogenetic tree reconstruction methods were congruent (Figure 3). The most morphologically distinctive C. caput-medusae from Central Asia was used as the out-group, the first 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. The five 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. The three individuals of C. ebinuricum which form an independent clade, have specific fruit characters different from the complex. The 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. mongolicumJX259384. Within the clade C, the p-distance was as high as 0.564% between C. ebinuricum and C. mongolicumJX259384. 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).

Figure 3. 

Maximum likelihood tree for 43 (in-group) Calligonum nrITS and ETS sequences produced with RAxML. Numbers adjacent to (relevant) nodes represent maximum likelihood value and Bayesian posterior probabilities. Branches marked with an asterisk collapse on the maximum likelihood strict consensus tree of the same dataset. The branch marked with a number sign collapses on the Bayesian majority rule consensus tree of the same dataset.

Figure 4. 

Neighbour-net analyses of the Calligonum mongolicum complex, C. ebinuricum, C. calliphysa and closely related taxa based on uncorrected p-distances. Numbers indicate bootstrap values over 1000 replicates.

Species isolation is frequently caused by the temporal heterogeneity of blooming amongst sympatric species (Levin 1971; Adams 1983; Grant 1992, 1994). The flowering periods of five species in the complex showed a high degree of overlapping, with some differences in peak blooming periods (also see cases in Wilson 1983; Burd 1995).

These five diploid species of Calligonum have similar pollen characters in both with spheroidal shape and tricolporate apertures with each other (Figure 2). The other species in Medusae also have the similar pollen characters but without specific pollen indexes (P&E) analysis (Qiu 1988; Gulinuer 2008). The hand-pollination tests suggested the five 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. The 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). The crossing behaviour amongst them is consistent with the view from Soskov (1975a, 1975b) by treating these various segregate species as one variable biological species of C. mongolicum.

Although phylogenetic inference based on nrITS needs to be considered carefully (Alvarez and Wendel 2003, Feliner and Rossello 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). The 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 differentiated from the C. mongolicum complex (13 individuals) (Figures 3 and 4). The intermixed patterns of sequences from different “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).

Calligonum is one of the medium-sized genera of Polygonaceae with approximately 60–80 species and represents a rapid diversification 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. The authors’ results both in this paper and in their previous studies (Shi et al. 2011, 2012, 2013, 2016, 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 (1975a, 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 (Wen et al. 2015, Zimmer and Wen 2015).

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 confined 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. The fact that most of the collected seeds can germinate without any pre-treatment suggests that the five 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 and Havens 2009, Swarts and Dixon 2009). Special efforts are needed to ensure isolation of genetic sources in ex situ conditions.