Phylogeny and maternal donors of Elytrigia Desv. sensu lato (Triticeae; Poaceae) inferred from nuclear internal-transcribed spacer and trnL-F sequences

Background Elytrigia Desv. is a genus with a varied array of morphology, cytology, ecology, and distribution in Triticeae. Classification and systematic position of Elytrigia remain controversial. We used nuclear internal-transcribed spacer (nrITS) sequences and chloroplast trnL-F region to study the relationships of phylogenetic and maternal genome donor of Elytrigia Desv. sensu lato. Results (1) E, F, P, St, and W genomes bear close relationship with one another and are distant from H and Ns genomes. Ee and Eb are homoeologous. (2) In ESt genome species, E genome is the origin of diploid Elytrigia species with E genome, St genome is the origin of Pseudoroegneria. (3) Diploid species Et. elongata were differentiated. (4) Et. stipifolia and Et. varnensis sequences are diverse based on nrITS data. (5) Et. lolioides contains St and H genomes and belongs to Elymus s. l. (6) E genome diploid species in Elytrigia serve as maternal donors of E genome for Et. nodosa (PI547344), Et. farcta, Et. pontica, Et. pycnantha, Et. scirpea, and Et. scythica. At least two species act as maternal donor of allopolyploids (ESt and EStP genomes). Conclusions Our results suggested that Elytrigia s. l. species contain different genomes, which should be divided into different genera. However, the genomes of Elytrigia species had close relationships with one another. Diploid species were differentiated, because of introgression and different geographical sources. The results also suggested that the same species and the same genomes of different species have different maternal donor. Further study of molecular biology and cytology could facilitate the evaluation of our results of phylogenetic in a more specific and accurate way.


Background
Triticeae in Poaceae includes not only the most economically important cereal crops (wheat, barley, and rye) but also forage grasses and ecological species in grasslands. Approximately 450 Triticeae species exist worldwide [1][2][3]. Given the wide variety of biological mechanisms and genetic systems, this tribe represents an excellent group for research on plant systematics, genetic diversity, and speciation [4,5].
Polyploidization and hybridization are the two main mechanisms in plant speciation and evolution [22,23]. The changes in the cell size, genome size, gene expression, genomic repatterning, epigenetic effects and retrotransposon activation are caused by the polyploidization and chromosome doubling [5,[22][23][24][25][26]. As a result of these changes, stabilization of hybrid condition and full fertility may occur. And the establishment of phenotypes in nature could be enhanced. Therefore, polyploids could adjust to match with the new ecological niches or become more competitive than parental donors [5,23,26,27]. The evolution of polyploidization alone and/or the combined effects of hybridization and polyploidization may lead to complex lineages, requiring an explanation of the phylogenetic relationship [27]. Molecular genetic analysis bears significance in elucidating phylogenetic relationships and genome evolution patterns in taxa for these kinds of plant groups [27,28]. The analysis of Molecular phylogenetic exploits DNA sequences elucidated the history of revolution and origins of species in Triticeae. This illustrates their hybridization events and parental lineages contains their formation, and identifies the polyploidization mode [29][30][31][32][33][34][35][36][37][38]. Reproducibility and simplicity represent the main qualities that make DNA sequencing a suitable choice for identification of phylogenetic relationships among taxa and genomes [28,39]. nrITS sequences were widely applied to explain genomic and phylogenetic relationship at a low taxonomic levels [40][41][42] and Triticeae species containing E, H, Ns, P, St and Xm genomes in Elymus, Hordeum, Psathyrostachys, Agropyron Pseudoroegneria and Leymus [5,27,33,[40][41][42][43]. Chloroplast DNA (cpDNA) sequences, including intron of trn-L and intergenic spacer of trnH-psbA, trnL-trnF, and trnS-trnG, are also widely used to identify maternal donors of polyploids with extra ability to analyze phylogenetic relationships among relevant species [41,[44][45][46][47].
The present study analyzed sequence data of one ITS region of nuclear DNA and one chloroplast gene (the intergenic region of trnL-trnF) from 18 species (subspecies) in Elytrigia s. l. with 21 species of related genera in Triticeae. The objectives are as follows: (1) to investigate phylogenetic relationships among species in Elytrigia s. l. and related genera; (2) to elucidate interspecific relationships among Elytrigia s. l. species; (3) to study phylogenetic relationships among species with different genomes and genome combinations; and (4) to discuss putative maternal donors for ESt genome species.

Plant materials
This study included 18 species (subspecies) of Elytrigia Desv. sensu lato and 21 species (subspecies) of related genera in Triticeae. Table 1 lists names, accession numbers, genomes, origins, and GenBank accession numbers. Bromus cartharticus Vahl. and Bromus tectorum L. were used as outgroup [5,41,48]. Seed materials with W 6 and PI accession numbers were carefully offered by the American National Plant Germplasm System (Pullman, Washington, USA). We gathered seed materials with Y and ZY numbers. Voucher specimens and plants were deposited at the perennial nursery and herbarium of the Triticeae Research Institute, Sichuan Agricultural University, China.

DNA extraction, amplification, and sequencing
Total genomic DNA was extracted from leaves of single plants by slight modification of Cetyltrimethyl Ammonium Bromide (CTAB) procedure [49]. nrITS sequence and chloroplast trnL-F sequence were amplified with primers described in Table 2. A final volume of 20 μL of mixed reagents was obtained for each polymerase chain reaction (PCR); reagents included 2× Taq PCR Master-Mix (10× ExTaq polymerase buffer, 3 mmol/L MgCl 2 , 500 μmol/L deoxynucleotide, 100 mmol/L KCl, and 20 mmol/L Tris-HCl), 1 μmol/L of each primer, 20-40 ng of template DNA, and distilled deionized water. PCR reactions were performed in GeneAmp T100 Thermal Cycler (Bio-Rad Inc., USA) employing protocols listed in Table 3. PCR products were  electrophoresed on 1.0% agarose gels, and purified using EZNA™ gel extraction kit (Omega, GA, USA), and were cloned into pMD-19 T vector (TaKaRa) following the instructions of manufacturer. All sequences were derived from at least 3 independent clones for diploid species, and 5-8 independent clones for allopolyploid species. Sequencing was performed from both directions by Sunbiotech Company (Beijing, China) [36].

Phylogenetic analysis
Alignment of nrITS and trnL-F sequences were conducted by using Clustal W algorithm [50] with additional manual adjustment. Two data matrices with included nrITS were performed using Maximum likelihood (ML) in PAUP*4.0a (Swofford, D.L., Sinauer Associates, http://www.sinauer.com) and Bayesian inference (BI) in MrBayes version 3.1.2 [51]. Phylogenetic analyses based on trnL-F sequences were performed with MrBayes version 3.1.2. Evolutionary model employed for phylogenetic study was performed using Modeltest v3.7 with Akaike information criterion (AIC) [52]. Best-fit model was GTR + G for nrITS data. ML heuristic studies were carried out with 1000 random addition sequence replications and reconnection branch swapping algorithm and tree bisection [Dong 2013]. Similar to ML analysis, BI analyses of nrITS were perfomed with the alike evolutionary model. TVM + G was the optimal model for trnL-F data based on AIC in Modeltest v3.7. Observation of consistency and examined log likelihoods among all independent runs showed that burn-in periods very long enough for chains to become stationary [37]. Figures included nonsignificant bootstrap support (BS) of more than 50% and posterior probabilities of more than 70%.
Median-joining (MJ) network method was effectively employed to study detailed progenitor-descendant relationship among polyploidy species within tribe Triticeae [27,35,37,53]. MJ network analysis was conducted by the Network 4.6.1.3 program (Fluxus Technology Ltd., Clare, Suffolk, UK). For the purpose of preventing single insertion/deletion events from being counted as multiple mutational stages in MJ network study, gaps in aligned nrITS and trnL-F sequences were not included in the calculation [37].
With assumed nucleotide frequencies A: 0.2286, C: 0.2966, G: 0.2794, and T: 0.1954, nrITS data yielded a single phylogenetic tree (−Lnlikelihood = 2553.6868). Proportion of invariable sites = 0, and gamma shape parameter = 0.4121. Likelihood settings from optimal model (GTR+ G) were selected by AIC in Modeltest v3.7. Similar to that of ML analysis, Bayesian study supposed the Bromus catharticus Vahl.    same topology. The tree demonstrated in Fig. 2 corresponds to the ML tree with posterior probabilities (PP) above and BS below branches [48]. nrITS region from species were divided into five clades (Clades I-V). Clade I was divided into three groups, namely, A, B, and C. Group A (BS < 50 and PP < 70%) consisted of the In MJ analysis, each circular network node indicated a single sequence haplotype, and the node size is proportional to the number of isolates with that of haplotype [37]. Median vectors (standing for missing intermediates) present nodes that haven't sampled deduced by MJ network study, and the numbers along branches illustrate the mutation positions. Distinguishing colors indicated various species species that share similar haplotype circular network node. Either alternative genealogies or true reticulation events are represented by network loops in closely related lineages [37]. The MJ network depicted genealogical relationships among 45 nrITS haplotypes from 49 taxa (Fig. 3) [48]. We found that MJ network represented a consistent phylogenetic reconstruction with ML tree. Then, we determined the names and group names of similar clusters to synchronize the MJ network. In nrITS MJ network analysis, five clusters (Clusters N-I to N-V) formed one star-like radiation. Three clusters (Clusters N-III to N-V) represented three different types of haplotypes (St, P, and Ns types) of Elytrigia s. l. and its related genera. Cluster N-I was divided into subclusters N-A, N-B,  Table 1 and N-C with E and St types, and Pse. spicata PI 506259 (PSP1) was placed at the central branching point. Cluster N-II included St type of Pse. spicata PI 563870 (PSP2) and Pse. tauri (PTAU), P type of Ag. cristatum (ACRI), Et. pungens (EPUN), and Et. pycnantha (EPYC), F type of Er. distans (EDIS) and Er. triticeum (ETRI), W type of Au. pectinatum (APEC) and Au. retrofractum (ARET), and unknown type of Et. varnensis (EVA1).

trnL-F data
Comparison of all species studies showed that the length of trnL-F sequences ranged from 809 bp to 882 bp. Likelihood settings from optimal model (TVM + G) were chosed by AIC in Modeltest v3.7. Fig. 4 illustrates the BI tree with PP above branches. All trnL-F sequences from Elytrigia and its related genera species were similar. In the trnL-F MJ network analysis, 25 haplotypes were derived from 37 taxa. MJ network represented consistent phylogenetic reconstruction with BI tree. We determined Fig. 4 Bayesian inference tree inferred from the trnL-F sequences of Elytrigia sensu lato and its affinitive species. Numbers above branches indicate posterior probabilities (PP) ≥ 70% by BI clusters' names following the name of groups shown in the ML tree. The trnL-F MJ network was divided into four clusters (Clusters N-One to N-Four). All species containing E or St genome were clustered together with E or St diploid species in Cluster N-One. Cluster N-Two included F, P, and W types of haplotypes. Ns type of Psathyrostachys haplotype species and H type of Hordeum haplotype species were grouped, respectively, in Clusters N-Three and N-Four (Fig. 5).

Discussion
Phylogenetic relationships among species in Elytrigia s. l.
Elytrigia s. l. is distributed in subtropical and warm temperate regions of both hemispheres [6]. Classification and systematic position of Elytrigia remain controversial [7,8,[54][55][56]. Traditionally, the classification based on morphology and Elytrigia species contains E e , E b , E e E b , E e E e St, E b E e St, E e St, StSt, StH, and EStP genomes. However, Dewey [1] and Löve [8] showed that taxonomic treatment of Triticeae species should be depended on genomic constitution. Therefore, Elytrigia species must be reclassified. Several current studies used molecular biology to study phylogenetic relationships of Elytrigia s. l. species and its related genera [43,57,58]. Hsiao et al. [59] estimated phylogenetic relationships of 30 diploid species of Triticeae (Poaceae) from the nrITS region of nuclear ribosomal DNA. Results illustrated that each genome group of species is monophyletic and consisten with cytogenetic evidence, and Australopyrum (Tzvelev) A. Löve (W) is closely concerned with Agropyron Gaertn.
In this study, based on nrITS data, all Elytrigia s. l. species were classified in four groups (E, H, P, and St types) in the ML tree and MJ network. These results indicated that Elytrigia s. l. species contain different genomes. These findings also strongly support previous results. Genome species are not highly supported in Clades I-B, E e , and E b (BS < 50% and BI = 81%) (Fig. 2). This phenomenon provides evidence of close affinity between E e and E b genome species. Thus, these species are not homologous but homoeologous [63]. Our phylogenetic results also support previous cytological investigations reported by Löve [8,64,65], Yen and Yang [3], and Zhou [11].
Phylogenetic relationships between Elytrigia s. l. and related genera In the present study, in the ML tree and MJ network based on nrITS data, seven types of nrITS region (E, F, H, P, W, Ns, and St types) were obtained from the Elytrigia s. l. species and its related genera. In polyploidy species Et. repens (StStH), H type was clustered with Hordeum species in Clade IV (83% BS, 100% PP), and St type was clustered with Pseudoroegneria (Fig. 2). In this study, we failed to obtain E and St type nrITS sequences from Et. pycnantha (EStP) and Et. pungens (EStP),  Table 1 whereas P type was clustered with Agropyron in Clade II (86% BS and 97% PP). In the phylogenetic tree, Clades II and IV formed a monophyletic group, and results support the distant relationship between H genome and other genomes (E, P, St, and W) (Fig. 2). This finding also indicated that St genome is the origin of Pseudoroegneria, whereas P genome is the origin of Agropyron.
Sha et al. [48] studied phylogenetic relationships of Leymus based on trnH-psbA and indicated that Pseudoroegneria species are close to Lophopyrum bessarabicum. Evidence from meiotic chromosome pairing [71], morphological data [72], and DNA sequencing [44] suggested close relationship among Lo. bessarabicum, Lo. elongatum, and Pseudoroegneria. In the BI tree and MJ network based on trnL-F sequences, all Elytrigia species were categorized under Clade One with a number of zero-length branches; Elytrigia species are sisters with diploid species of Eremopyrum (F), Agropyron (P), and Australopyrum (W) (94% PP). H and Ns genome species formed the monophyletic group. These results indicated minimal differences in E and St genomes based on trnL-F sequence and close relationship of E, F, P, St, and W genomes, which are distant from H and Ns genomes. These findings support previous studies on morphology [66,67], molecular biology [5,44,[68][69][70], and cytogenetics [2,60,61].
Putative maternal donor and origin of Elytrigia species cpDNA is mostly inherited from the female parent in tall plants. Therefore, it can be used to determine maternal donor in polyploids [37,53,73]. In the trnL-F ML, the phylogenetic tree showed high numbers of zero-length branches, which are mainly related to multifurcating relationships. Tree-based study methods are unable to represent multifurcating relationships and the coexistence of ancestors with their derivatives [53,74]. Network approaches were designed to deal with such multifurcations [53,[73][74][75][76].
Previous studies based on cpDNA indicated that Pseudoroegneria (St genome donor) species are the maternal donor for species of Triticeae [41,44,77]. However, cytologically, Yen et al. [78] considered that rather than the St genome, the maternal donor of Kengyilia is the origin of P genome species. In this trnL-F-based BI tree, Et. bessarabica (2×) was clustered with polyploidy species Et. farcta, Et. nodosa (PI 547344), Et. pontica (PI 547313), Et. pycnantha, Et. scirpea, and Et. scythica (92% PP). MJ network analysis showed that diploid species Et. bessarabica (E), Et. pontica (PI 547313), Et. pycnantha (EStP), Et. scirpea (E), and Et. scythica (ESt) exhibit the same haplotype in Cluster N-One (Fig. 5). Combined with BI and MJ analyses, we can conclude that E genome-diploid species in Elytrigia served as maternal donor of E genome for Et. farcta, Et. nodosa (PI 547344), Et. pontica, Et. pycnantha, and Et. scythica. This conclusion agrees with results of Liao et al. [37]. Et. nodosa (P I547345 Ukraine) was not clustered with Et. bessarabica, and its haplotype differs from that of Et. nodosa (PI 547344 Turkey). Results showed that (1) at no less than two species served as maternal donor, indicating that formation of Et. nodosa occur multiple times. A similar conclusion was observed based on Et. caespitosa, Et. intermedia, Et. varnensis, and Kengyilia species [45,58,79]. (2) Different maternal donors in Et. nodosa are affected by altitude and climate conditions [80]. In the BI tree based on trnL-F sequence, we can conclude that Pseudoroegneria species (St genome donor) acted as maternal donor of Et. repens (StStH), whereas species of Agropyron Gaertn. (P genome donor) acted as maternal donor of Et. pungens (EStStP). However, E genome acted as maternal donor of Et. pycnantha (EStP). This result indicated that different species served as maternal donors that contributed to species containing the same genomes. Previous findings on Et. intermedia were similar to our results [58]. Other polyploidy species in Elytrigia and diploid species containing E or St genome formed zero-length branches in Clade One because of the close relation of E and St genomes (Fig. 4). Sources of maternal donor of these genomes remain to be identified.

Differentiation and relationship between E and St genomes
In Clade I, species containing E, St, and ESt genomes and those in Cluster N-I Pse. spicata appeared at the central part, indicating close relationship of St and E genome species (Figs. 2 and 3). These findings coincide with previous findings on morphology and molecular biology [44,72,81]. In the present study, E and St types were obtained from species containing ESt genome grouped with Elytrigia or Pseudoroegneria diploid species, respectively. This phenomenon showed that E genome was the origin of diploid Elytrigia species with the E genome. St genome was the origin of Pseudoroegneria. Results from morphology, genetics, and molecular biology indicate that species containing E, St, and ESt genomes are closely related with Elytrigia.

Taxonomy of species with ESt and EStP genomes
Polyploidization and hybridization are long recognized as prominent forces in evolution of plant species, which feature consequences of genomic changes [22,23]. Genome relationship and differentiation are often vague in some species because of frequent introgression of alien genes, polyploidization and chromosome segments from wide hybridization [43]. Thus, classification is one of the most important issues that require understanding.
Previous studies indicated that Et. caespitosa, Et. intermedia, Et. nodosa, Et. scythica, and Et. varnensis contain ESt genomes, which belong to Trichopyrum [3,17,82,83]. Comparison of partial sequences of nrITS gene showed that a TTTT insert at positions 58-61 in nrITS sequence was detected for 11 species (Fig. 1). This finding indicated that introgression of E genome during polyploidization or different independent hybridization events may create the variants in polyploidy ESt species. In the ML tree and MJ network based on nrITS sequence, one group is formed by ESt genome species (Et. caespitosa, Et. intermedia, Et. nodosa, Et. scythica, and Et. geniculata ssp. pruinifera) and unknown genome species of Et. varnensis. This result indicated that these species should be classified under the same genus (Trichopyrum). Species containing ESt genomes were grouped with diploid species Et. elongata (E genome), suggesting that E genome may be derived from Et. elongata. In this study, diploid species of Et. elongata were differentiated. We selected two Et. elongata (Iran, France) with different origins, which are divided into Clades I-B and C (Fig. 2). A TTTT insert at positions 58-61 in the sequence was also detected for Et. elongata (W 6 21,859) (Fig. 1). This result indicates that ESt genome polyploid species and diploid Elytrigia species (E genome) displayed hybridization event, resulting in divided E genome.
Et. varnensis was reported by Löve to contain ESt genomes (2n = 12× = 84) [8]. Yang [84] showed that Et. varnensis is a tetraploid species. Diversity of species ploidy may be caused by chromosome variation under natural conditions. In this study, we discovered that Et. varnensis clustered with Et. pungens, Et. pycnantha, Ag. cristatum, Au. pectinatum, Au. retrofractum, Er. distans, and Er. triticeum (85% BS and 100% PP) (Fig. 2). We concluded that this species contains P or F genome. Another estimate indicated that St genome allopolyploid species possibly resulted from introgression of Eremopyrum or Agropyron during polyploidization. Results strongly support those of previous studies in cytogenetics [85].
Possible genome constitutions and taxonomic treatment of Et. lolioides Cytologically, Et. repens comprised StH genome. Et. lolioides is a polyploid species, and its genomic constitutions remains unknown [8,17,62,86]. In the present study, Et. lolioides was clustered with diploid Pse. libanotica (St genome); this result indicated that Et. lolioides possesses one St genome. Et. lolioides clustered with H. bogdanii, H. chilense (H), and Et. repens (H copy) with high statistical result (98% BS and 100% PP) (Fig. 2). Such finding also indicated that Et. lolioides contains H genome and is closely related to StH genome species Et. repens. Thus, we can conclude that genomic constitution of Et. lolioides includes St and H genomes and belongs to Elymus s. l.

Conclusion
This study analyzed the phylogenetic relationship among Elytrigia s. l. species on the basis of the nrITS sequence data. The results supported the conclusion that Elytrigia s. l. consists of various genomes (E, H, P, and St types), which should be classified as different genera. Analyses based on nrITS sequence data and chloroplast trnL-F region show that the E, F, St, P, and W genomes have a close intergroup relationship but are distant with the H and Ns genomes. This finding strongly supports previous studies on morphology, molecular biology, and cytogenetics. nrlTS sequence analysis demonstrated that the E genome of species Et. caespitosa, Et. caespitosa ssp. nodosa, Et. intermedia, Et. scythica and Et. geniculata ssp. pruinifera, which contains ESt genomes, originated from Et. elongata in Lophopyrum. However, differentiation was found in diploid species Et. elongata; this phenomenon was possibly due to diverse geographical origins or introgression. Et. lolioides, which is composed of unknown genomes, contains the H and St genomes and has a close genetic relationship with Et. repens and El. canadensis, which contain the St and H genomes. Accordingly, the genome of Et. lolioides is inferred to contain St and H. In this paper, polyploid species of Elytrigia s. l. was deduced based on trnL-F sequence, the female parent of Et. caespitosa ssp. nodosa (PI547344), Et. farcta, Et. pontica (PI547313), Et. pycnantha, Et. scirpea and Et. scythica is the diploid species of Elytrigia s. l. containing the E genome; the maternal donor of the polyploidy species Et. caespitosa ssp. nodosa (PI547345), Et. pontica (PI383583), Et. repens, Et. geniculata ssp. pruinifera is the St genome. Different maternal donors were also found in allopolyploid species. This result could be attributed to different growth environments, introgression, or incomplete separation of genome E lineage. Thus, different haplotypes were presented.