1 Introduction

Kalidium Moq. (Chenopodiaceae), are identified as Euhalophytes and divided into five species, Kalidium caspicum (L.), Kalidium gracile Fenzl, Kalidium cuspidatum (Ung.-Sternb.) Grub. var. cuspidatum, Kalidium foliatum (PALL.), Kalidium schrenkianum Bunge. ex Ung. -Sternb and one variant species, Kalidium cuspidatum (Ung.-Sternb.) Grub. var. sinicum A. J. Li, mainly distributed in Southeast Europe and Northwest Asia as shrubs (Kong, 1979; Krever et al., 1998). Previous studies have demonstrated that Kalidium species play an important role in maintaining the balance of grassland ecosystems and preventing soil erosion (Zhao et al., 2002). In comparative studies from three other Euhalophytes species (Suaeda sala, Atriplex centralasiatiea and Nitraria sibirica), Kalidium have proven to possess strong tolerance to saline-alkali soil and drought, as a dominant species in desert areas (Zhao et al., 2002; Zhou et al., 2009). In China, Kalidium is a succulent salt plant mainly used as forage grass for camels, horses and sheep, excellent for grazing herds in the winter. Traditional taxonomic methods for identifying Kalidium relied on morphological and phenotypic characters, which have limits in differentiating species (Kong, 1979). Therefore, developing a common DNA barcoding for species identification of Kalidium is required.

As an increasingly prevalent molecular technique to remedy the limitation of taxonomic research relying solely on morphological features, DNA barcoding has been used to facilitate accurate species-level identification using specific DNA regions (Kress and Erickson, 2007). A potential DNA barcode, CO1 (cytochrome c oxidase subunit 1), was successfully identified as a standard mitochondrial region to discriminate species in animal groups (Hogg and Hebert, 2004; Barrett and Hebert, 2005; Fu et al., 2011). Although previous studies have proposed many potential DNA barcodes, including a nuclear DNA locus (ITS) and several chloroplast DNA regions (matK, rbcL, trnH-psbA, atpF-atpH, et al.) has been widely applied in plant ecology and evolutionary studies (Kress and Erickson, 2007; Newmaster and Ragupathy, 2009; Moniz and Kaczmarska, 2010), universal DNA barcodes for plant species identification have not been very informative (Newmaster et al., 2006; Ford et al., 2009; Dong et al., 2015).

In this study, four candidate DNA loci were selected, one nuclear gene (ITS) and two previously used chloroplast genes (matK and rbcL) and one recently developed candidate chloroplast gene (ycf1b) (Dong et al., 2015), to screen their suitability as DNA barcodes for Kalidium. The major purpose of this study is to address the following two questions: (1) determine ideal DNA markers for species-level identification of Kalidium, (2) verify identification of the variant species of K.cuspidatum var. sinicum is consistent with that of traditional taxonomy relying only on morphology.

2 Material and methods 2.1 Plant samples

Leaf tissue of five species and one variant species (Figure 1) were collected within the full range of Kalidium in China. For each species, 20 to 30 individuals were sampled from each population; and all the samples were silica gel-dried. Three to four individuals were sampled from three to five populations for each species in order to avoid individual bias for this study. A total of 82 samples, representing geographic location and collection information are available in Table 1. We selected Salsola laricifolia Turcz. ex Litv and Chenopodium album L. (Chenopodiaceae) as outgroups.

2.2 DNA extraction, amplification and sequencing

DNA was extracted from leaves by the slightly modified SDS method (Jia et al., 2010). PCR amplification of four candidate barcodes was carried out on the Cycler^TM Thermal Cycler (Bio-Rad, USA). Detailed information on the amplification conditions of the tested four regions is available in Table 2. Direct sequencing PCR products were carried out by two directions. Five individuals of the ITSlocus contained segregating indels that prevented direct sequencing, PCR fragments were purified and sub-cloned into the pMD19-T vector (Takara, China) and three to five clones were then directly sequenced.

2.3 Sequence analysis

Sequence chromatograms were edited and aligned using Aligner v.5.1.0 (Codon Code Corporation, Dedham, MA), with all posterior probabilities > 0.8 and all polymorphic and heterozygous sites manually confirmed. We calculated K2P (Kimura 2-parameter) distances using MEGA 5 software to evaluate the intra-specific and inter-specific differences (Kumar et al., 2008). The distribution graphs of intra-specific and inter-specific K2P genetic distances of each DNA locus were made and compared to value barcode gaps using Origin 8.5 (Meier et al., 2006). The differentiation power of the four DNA barcodes was estimated using tree-based methods. Bootstrap support values were performed through 1, 000 random replicates.

3 Results 3.1 PCR amplification, sequencing and alignment

All four candidate loci exhibited high PCR success and sequencing success (100%) (Table 3). The 336 new sequences of the four DNA regions were obtained representing five species and one variant of Kalidium, which also included two outgroup taxa, S.laricifolia and C. album, respectively (Table 3). The results showed that the tested primers possessed prominent universality (Table 3). GeneBank accession numbers of all the sequences are listed in Table 4.

For four DNA loci, aligned sequence lengths possessed a relatively great range 681 bp for ITS to 902 bp for matK (Table 3). The DNA barcode containing the most variable sites was ITS (119), followed by ycf1b(35), rbcL (7) and matK(7). The distribution graphs of the intra-specific and inter-specific distance for the four DNA regions displayed the highest mean inter-specific divergences of ITS region (0.0437), and matK had the lowest mean inter-specific divergences (0.003) (Table 3).

3.2 DNA barcoding gap

Graphing the distribution of K2P distances is to evaluate the barcoding divergence between intraspecific and interspecific genetic distances for all DNA barcodes tested (Figure 2). We found a large barcoding gap only in the ITS region and the other three DNA sequences tested proved to have no such barcoding gap (Figure 2).

3.3 Phylogenetic analyses

Neighbor-Joining (NJ) trees were constructed with the single-locus or combined DNA loci to evaluate the discrimination power for the four DNA barcodes (Figures 3, 4). As the single locus analyses, the species discrimination power of the ITS locus was highest, with a success rate of 100%, followed by matK (33.3%), ycf1b (16.7%) and rbcL (16.7%) (Table 3). The tree-based method for the ycf1b locus identified the same species as that of the rbcL locus (Figure 4). For possible combination of the other three DNA loci excluding ITS region analyses, the results showed that the possible combinations of matK with the other two loci had the same performance as using matK alone (Figure 4). Due to the relatively low success of species identification for single locus or random combination of the tested three barcodes excluding ITS region, the confident tree obtained only from ITS (Figure 3).

NJ tree analysis of ITS sequences revealed that the 82 samples used in this study were significantly split into six clades (Figure 3). Single or any combination of all four DNA barcodes tested identified K.cuspidatum var. sinicum as a distinctive clade (Figures 3, 4). The analysis of all NJ trees identified K.cuspidatum var. sinicum as a single species with high bootstrap values.

4 Discussion and conclusions

The success rate of PCR amplification and sequencing has long been treated as a significant index to estimate DNA barcodes. In this study, all four DNA barcodes tested were universal with PCR amplification and sequencing success. Through sequences analysis, the ITS region exhibited high resolution, as identification power has been showed in Alnus (Ren et al., 2010) and Euphorbiaceae (Pang et al., 2010). For species level identification, ITS showed great potential for being an ideal DNA barcode for Kalidium due to its high inter-specific divergence and discrimination rate (100%) in the four DNA loci tested. rbcL, as one of the core DNA barcodes for plants, has performed well for mosses, ferns and angiosperms (Hollingsworth et al., 2009; Liu et al., 2011; Zheng et al., 2015). Also, previous studies have indicated that the rbcLregion showed relatively low inter-specific K2P distances on determining closely related species (Hasebe et al., 1995; Newmaster et al., 2008; Gong et al., 2015). Our results showed that rbcL had relatively low inter-specific distance and the lowest species differentiation rate, and as such is not suitable as a DNA locus for discrimination in Kalidium. Recently, Dong et al. (2015) demonstrated that ycf1b is a plastid genome region with relatively high variable sites and proposed ycf1b as a core DNA barcode for species identification of land plants. Our results showed that ycf1b had the same discrimination rate (16.7%) as rbcL and were unfit for potential DNA barcode in Kalidium. The matK region has proven to possess a relatively significant inter-specific K2P distances and good generality in some land plants (Kelly et al., 2010; Gong et al., 2015). Lahaye et al. (2008) identified matK gene showed prominent universality for identification of flower plants species. Asahina et al. (2010) demonstrated that matK rather than rbcL was better suited in identifying medicinal Dendrobium species due to a high-level resolution. The matK locus showed relatively high level of discrimination rate in Kalidium species compared with the other two candidates (rbcL and ycf1b, the matK locus was proposed as a candidate DNA barcode rather than an ideal barcode owning to its lower species level identification than the ITS region in Kalidium.

Numerous studies have demonstrated the limitation of species delimitation relying solely on morphological features. For example, Protoparmelia was more diverse than what was expected from traditional taxonomy, consisting of several previous unknown depicted species, and cryptic species-lineages (Singh et al., 2015). According to morphological and biogeographic information, Wood (2006) treated Dendrobium officinale and D. tosaenseas a common species, but by molecular technique identification results suggested that they should be identified as two different species (Asahina et al., 2010). Compared to K. cuspidatum var. cuspidatum, K. cuspidatum var. sinicum was identified as a variant species based solely on different morphological traits (Kong, 1979). According to our observations on morphological characteristics of these two species in the field, there were significant differences on growth of the shoots, the length and diameter of the infructescences. Also, the phylogenetic analyses of all four DNA barcodes tested in this study showed that K. cuspidatum var. sinicum clustered into a single clade with strong bootstrap values greatly separated from the other five species. Obviously, species identification of Kalidium relying solely on morphology has its limitations. Thus, we suggested that K. cuspidatum var. sinicum should be identified as a single species in Kalidium, rather than as a variant.

In conclusion, results showed that the ITS region was an ideal DNA barcode in Kadilium, and the variant, K. cuspidatum var. sinicum, should be treated as a single species.

Acknowledgments:

We are grateful to KuiBing Meng, DeCheng Liu and FengZhu Zhang for sample collections in the field. This work was supported by the Program for New Century Excellent Talents in the Ministry of Education in China (NCET-09-0446), and lzujbky-2012-k22 to YuXia Wu.