Background

Ostrya is a small genus belongs to the birch family Betulaceae and merely consists of eight species native in southern Europe, southwest and eastern Asia, and North and Central America [1,2,3]. These small deciduous trees, commonly called as hophornbeam or ironwood, are well known for their hard and heavy woods. Ostrya woods have been used for various purposes such as furniture, axles, fuel wood and charcoal [2, 4]. It had been long thought that there are five species distributed in China (http://www.efloras.org) until Lu et al. (2016) [5] found that Ostrya yunnanensis was nested within O. multinervis in both phylogenetic trees and morphological clustering. The two species were therefore recognized as the same species and renamed as O. chinensis [6]. Among the four species, O. japonica is mainly distributed in China, Japan and Korea, while the remaining three species, O. rehderiana, O. chinensis and O. trichocarpa, are endemic to China. The amount of these wild plants has been decreased rapidly due to overexploitation, habitat destruction and ecosystem deterioration. The famous endangered species, O. rehderiana, has been reported that only five mature trees survive and are conserved in the Tianmu Mountain, Zhejiang [7]. Although Ostrya species own excellent wood and are regarded as ecologically and economically valuable foundation for future forests, the interspecific delimitation and relationships remain unresolved. Previous study failed to discern the four Ostrya species based on four chloroplast (cp) fragments (rps16, trnG(UCC) intron, trnH–psbA, and trnL–trnF) while the ITS sequence variations did [5]. It has been suggested that the traditionally accepted cpDNA barcodes own limited resolution in delimitating closely related species due to the insufficient informative variations [8,9,10]. By contrast, the complete chloroplast genomes (plastomes) were proven to provide more valuable information in revealing phylogeny of plants [11,12,13,14,15,16]. And with the development of Next-generation sequencing technology, it has recently become cost-effective and it is easier to sequence the complete plastomes in plants than before [12, 17, 18]. The plastomes with more informative sites may increase resolution of the previously unresolved phylogenetic relationships based on a few cpDNA markers [15, 19]. Therefore, in this study, we re-investigate phylogenetic relationships of the four Chinese Ostrya species based on plastomes. We selected multiple individuals per species to cover the most of geographic distributions of the species to reduce the sampling bias. We also extracted three datasets, i.e., the whole plastomes, non-coding regions and protein-coding genes, for phylogenetic analyses and compared these plastome trees with the ITS tree. Specifically, the aims of our study are to: 1) investigate whether the three plastome datasets result in consistent phylogenies and whether they can discern the four species and 2) test if phylogeny based on plastome datasets congruent with the ITS tree.

Methods

Plant materials, DNA extraction and sequencing

We chose in total 33 individuals from 17 populations representing all the four Ostrya species that occur in China. For each species, we selected samples that cover most of its geographic distribution (Additional file 1: Table S1). The fresh leaf materials were collected in the field and preserved with silica gel immediately. Total genomic DNA was extracted from 20 mg silica gel-dried leaves on the basis of the modified CTAB procedure [20]. According to the basic protocol, we prepared the end-repaired, phosphorylated and A-tailed DNA fragments and ligated with index adapters. The library construction and whole-genome sequencing were accomplished at Beijing Genomics Institute (Shenzhen, Guangdong, China). The libraries were sequenced on the Illumina Hiseq 2000 platform.

Genome assembly and annotation

Raw reads of all samples were trimmed with a filter standard (Q < = 5 or N base content > 10%). Afterwards, we used Trimmomatic [21] to further filter the data. We downloaded about 3000 published chloroplast genome sequences from NCBI and built an index of them. All the high quality reads were then mapped to the index using Bowtie2 [22] and sorted by Samtools v.1.2 [23]. We used the software bam2fastq (https://gsl.hudsonalpha.org/information/software/bam2fastq) to generate the fastq format files from the bam file. Then, we assembled the plastomes using Velvet v1.2.10 [24] and filled up the gaps by GapCloser v1.12 (https://sourceforge.net/projects/soapdenovo2/files/GapCloser/). Moreover, we used Plann v.1.1.2 [25] to annotate the plastomes using O. rehderiana as a reference. Then we corrected the start codons, stop codons and intron/exon boundaries manually using Sequin v.15.10 (http://www.ncbi.nlm.nih.gov/Sequin/) as well as a visual software Geneious v.R.8.1.4 [26]. Ultimately, the circular images were performed through OGDRAWv.1.1 (http://ogdraw.mpimp-golm.mpg.de/) [27].

Genome comparison and genome repeat

In order to show interspecific variations, the alignments of all plastomes were plotted using the mVISTA program with LAGAN mode [28]. All validated plastome sequences were submitted to GenBank (Additional file 2: Table S2). We determined three kinds of repeats among plastomes of the four Ostrya species: dispersed, palindromic and tandem repeat. The former two types were visualized and located by an online program REPuter [29] with a minimal length of 30 bp and 90% sequence identities (Hamming distance equal to 3) between the two repeats [30]. Tandem Repeats Finder [31] was used to detect tandem repeats with the following parameters: Match 2, Mismatch 7, Delta 7, PM 80, PI 10, Minscore 50, MaxPeriod 500, with similarity 100%. We further verified all the repeats and removed the redundant segment manually.

Phylogenetic analysis

According to Grimm & Renner (2013) [32], the most closely related genus Carpinus could not be separated well from Ostrya and there were no published Ostryopsis plastome sequences available in GenBank. We therefore downloaded a plastome of Corylus chinensis (GenBank accession number: NC_032351.1) from GenBank as outgroup to reconstruct the phylogenetic tree of the four Ostrya species. In order to test whether different plastome regions resulted in consistent phylogenies during the phylogenetic reconstructions, we used three plastome datasets: (1) the whole plastomes, (2) the non-coding regions and (3) the protein-coding genes (PCGs). In addition, we downloaded from GenBank the same ITS dataset as in Lu et al. (2016) [5]. For whole plastomes, non-coding sequences and ITS datasets, we used their nucleotide sequences directly and aligned each dataset using MAFFT v.7 [33] and trimal [34]. The later software was implemented for alignment trimming. Afterwards, we used Perl scripts to delete the gaps of aligned sequences. For the PCGs dataset, we first extracted 76 PCGs shared by all 33 Ostrya plastomes and Corylus chinensis and translated their nucleotide sequences into amino acid (aa) sequences for phylogenetic analysis in order to reduce the tree artifacts due to high DNA divergence. Multiple alignment of aa sequence of each gene was achieved by T-coffee [35] followed by trimal [34] to do automated alignment trimming. After that, FASconCAT-G [36] was used to produce a concatenated alignment.

Phylogenetic analyses were performed by both maximum likelihood (ML) and Bayesian inference (BI) methods. For whole plastomes, non-coding regions and ITS datasets, GTR + I + G as the best-fit model selected by JModeltest v.2.1.1 [37] was used for both ML and BI analyses, which were conducted by RAxML v8.1.24 [38] and MrBayes 3.2.2 [39] respectively. For aa sequences of PCGs, Prottest v.3.4.2 [40] was applied to select best-fit aa substitution model for each gene. The selected best-fit models were listed in Additional file 3: Table S3. Then we performed ML analysis using RAxML v8.1.24 with specific aa substitution matrix for each partition by using the selected best-fit aa substitution model. As for BI analysis using MrBayes, preliminary searches (mcmc nchains = 1; ngen = 10,000,000) were made under mixed aamodel (preset aamodelpr = mixed) to identify the best-fit aa substitution model. The result showed that Cprev was the best-fit aa substitution model. Then a final BI analysis using Cprev as the fixed model for all partitions (prset aamodel = fixed(Cprev)) and unlinked model parameters for each partition were conducted in MrBayes.

The node supports were determined with 1000 bootstrap replicates in ML analyses. For BI analysis, two independent parallel runs and four chains (one cold and three hot) were running for 10,000,000 generations with trees sampled every 500 generations. Then we determined convergence by examining trace plots of the log likelihood values for each parameter in Tracer [41]. In addition, we calculated the distances among plastomes of the four Ostrya species using the web tool GGDC [42] (http://ggdc.dsmz.de/distcalc2.php) to further compare the inter- and intra-specific variation. As individuals from the same population were grouped together in the phylogenetic trees (see Results), only one individual from each population of the four species was selected for this analysis.

Results

Plastome features

Characteristics of Ostrya plastomes were conservative (Figs. 1 and 2) with the length ranged from 158,870–159,301 bp. The chloroplast genome structures of all the four Ostrya species were consistent with mostly known angiosperms with a typical quadripartite structure consisting of a pair of inverted repeats regions (IRa and IRb: 26,059–26,069 bp) divided by a large single-copy region (LSC: 88,007–88,229 bp) and a small single-copy region (SSC: 18,721–18,975 bp) (Fig. 1 and Table 1). The overall GC content was absolutely identical (36.5%; Table 1) across all plastomes. All plastomes possessed 122 unique genes, including 85 protein-coding genes (79 PCG species), 29 tRNA genes (24 tRNA species), and 8 ribosomal RNA genes (4 rRNA species, Table 1). Most genes appeared in a single copy, while 15 were replicated once on the IR regions, including 4 rRNA (4.5S, 5S, 16S, and 23S rRNA), 5 tRNA (trnI-CAT, trnL-CAA, trnV-GAC, trnR-ACG and trnN-GTT), and 6 PCG species (rpl2, rpl23, ycf2, ndhB, rps7 and ycf1; Additional file 4: Table S4). The rps12 gene was a unique trans-spliced gene with three exons. Expansion of IR regions into rps19 at the IRb/LSC boundary region occurred in all Ostrya taxa and the pseudogene ycf1 was located at the junction of SSC/IRa, which gave rise to the incomplete duplication of protein-coding gene within IRs. In addition, among annotated genes, nine genes (rps16, atpF, rpoC1, petB, petD, rpl16, rpl2, ndhB and ndhA) contained a single intron, and three genes (rps12, clpP and ycf3) had two introns. Intergenic spacers showed moderate genetic divergences. The intergenic spacers of trnQ-psbK, trnS-trnR and psbZ-trnG were identified as the most divergence hotspots (Fig. 2).

Fig. 1
figure 1

Gene map of the Ostrya chloroplast genomes. Because all 33 platomes have identical gene content and gene number, here we only show one plastome map as an example. Genes inside and outside of the circle are transcribed counterclockwise and clockwise directions, respectively. Genes belonging to different functional groups are shown in different colors. The dark gray inner circle corresponds to the GC content and the light-gray circle corresponds to the AT content

Full size image
Fig. 2
figure 2

Visualization of alignment of the four Ostrya species chloroplast genome sequences using O. trichocarpa as the reference. Grey arrows above the alignment indicate the orientation of genes. The vertical scale indicates the percentage of identity, ranging from 50 to 100%. The horizontal axis represents the coordinates within the chloroplast genome. Genome regions are color coded as protein-coding, rRNA coding, tRNA coding, and conserved non-coding sequences, respectively

Full size image
Table 1 Characters of four Ostrya species chloroplast genomes
Full size table

Repeat sequences are considered to play a significant role in phylogenetic analysis and also make positive efficiency for genome rearrangement analysis [43, 44]. In this study, a total of 249 repeats were detected in Ostrya plastomes (Fig. 3 and Additional file 5: Table S5). All repeats were similar among the four species and their overall distributions were conserved (Fig. 3a). The length of repeated sequences was mainly concentrated from 30 bp to 44 bp (Fig. 3b). The number of the three kinds of repeats was also similar, with palindromic repeat exhibiting maximum amounts (36.4%) followed by dispersed (32.4%) and tandem (31.2%) types (Fig. 3c). Most repeats were detected in intergenic regions (57.1%), while only a minority of repeats were distributed in intron regions (5.2%). The remaining repeats (37.7%) were found in coding regions (e.g., ycf1, ycf2, atpA, rpl2, rpl23, rps19, rps7, rps12 and ndhB; Fig. 3d).

Fig. 3
figure 3

Analysis of repeated sequences in four Ostrya plastomes. a Number of three repeat types in the four chloroplast genomes; b Number of repeated sequences divided by length; c Frequency of three repeat types; d Frequency of repeats in different regions

Full size image

Phylogenetic analysis

Three plastome datasets (i.e. whole plastomes, PCGs and non-coding regions) and one nuclear ITS sequences were used in this study for phylogenetic analyses. The whole plastome dataset comprised 157,116 bp, 2209 of which were variable and 456 were parsimony-informative. The non-coding dataset comprised 64,479 bp, 1332 of which were variable and 244 were parsimony-informative. 76 PCGs shared by all Ostrya samples and outgroup were translated into an amino acid dataset, which contained 21,720 aa, 228 of which were variable and 58 were parsimony-informative. However, the ITS dataset contained 634 bp with only 31 parsimony informative sites (Table 2). It is obvious that the plastome datasets comprised more variable sites than the ITS dataset but the variations rates were far lower than ITS datasets (Table 2).

Table 2 Number of informative sites in different datasets
Full size table

Phylogenetic trees using ML and BI methods done on each dataset resulted in similar topologies and we obtained almost the same topological divergences for all samples based on the PCGs and non-coding region datasets (Fig. 4b and c). The phylogenetic tree derived from the whole plastomes dataset differed slightly from the other two plastome trees. In the former tree, phylogenetic relationships among five subclades (i.e. Ore1, Oja1, Oja2–5, Oja5–8 and Och4–5) were resolved, while relationships among these five subclades were not clear in the later two trees (Fig. 4a, b and c). In general, except for the samples of O. rehderiana that collected from the same population, the other three species were all non-monophyletic in the plastome trees (Fig. 4a, b and c). O. japonica was separated into four subclades, while the other two species (O. chinensis and O. trichocarpa) formed another two subclades, respectively. Moreover, some samples that were collected from adjacent regions regardless of species may be clustered together in the plastome trees. For example, population Otr1 of O. trichocarpa was geographically close to populations Och2 and Och3 of O. chinensis (Additional file 6: Figure S1), and three samples of Otr1 were grouped with samples from populations Och2 and Och3. The heatmap of plastome distances was consistent with such failure of delimitation of interspecific relationships (Additional file 7: Figure S2). However, the ITS tree contrasted greatly with that inferred from the plastomes and the four species were all well-delimitated with high support values in the ITS phylogenetic tree (Fig. 4d).

Fig. 4
figure 4

Cladogram of Ostrya species using Bayesian inference (BI) analysis and Maximum likelihood (ML) analysis based on different datasets. Support values are shown for nodes as BI posterior probability (above branches) /ML bootstrap (below branches). The four species are shown by different colors. a – c plastome trees based on different regions; population information for each species is shown at the tips. d nuclear ITS (Internal transcribed spacer) tree; samples for each species are selected from different populations

Full size image

Discussion

In the present study, we determined whole plastid genomes of four Ostrya species, which provides significant genetic resources and facilitate to comprehend the plastid genome evolution. The exhibition of typical angiosperm quadripartite structure and highly conserved gene content, gene order and GC content in Ostrya plastomes were in accord with other members of Betulaceae [45,46,47,48,49]. The presence of repeats, especially in intergenic spacers, was often found associated with divergence regions [50, 51]. We found 57% of repeats occurred in intergenic spacers, which may play a role in maintaining interspecific divergence. In addition, the non-coding regions showed higher interspecific variations than the protein-coding regions as also found in many other groups, and phylogenetic analyses based on different regions may show inconsistence [19, 52], The conservatism of the Ostrya plastomes and well-aligned plastomes across different species can therefore facilitate the further phylogenetic analyses.

Previous studies based on a few cpDNA markers failed to resolve the interspecific relationships of Ostrya [53, 54], and surprisingly, our analyses based on three different plastome datasets (i.e. whole plastomes, PCGs and non-coding regions) with obviously more informative sites still failed to do so, which is contrast to most plastome studies that recovered high phylogenetic resolution and resolved relationships [55,56,57,58]. However, it should be noted that multiple individuals from different populations per species were included in our analyses, and these samples seem to be clustered by geography other than species (Fig. 4, Additional file 6: Figure S1). These results further indicate that phylogenetic studies based on plastome using only one individual per species are not appropriate, especially among closely related species. For example, using one individual from either population Oja9 or Oja1 of O. japonica will lead to very different interspecific relationships to other species (Fig. 4).

By contrast, ITS sequences with fewer informatic sites compared with the plastome datasets (Table 2) successfully discriminate the four species. The incongruence between nuclear ITS and cpDNA trees is common observed in many other plants [9, 59, 60] and usually can be explained as follows: firstly, the nuclear ITS sequence has higher mutation rate than chloroplast genome [61] and is biparental inherited with both pollen and seeds dispersion, while chloroplast of most angiosperm is maternal inherited and dispersed only by seeds [62, 63]. Chloroplast DNA with relatively low rates of intraspecific gene flow should be more introgressed, which lead to transfer of genetic material across species boundaries [64]. This may explain why some samples belonging to different species collected from adjacent regions are grouped together in the plastome trees (Fig. 4). In contrast, nuclear loci that experience high rates of intraspecific gene flow should enhance species delimitation [65]. We find the same pattern in our results that the ITS tree has much clearer species delimitation than the plastome trees (Fig. 4). However, alternative scenario may occur in other systems, for example, in Orychophragmus [66] and in Primula section Armerina [67], where cpDNA was highly effective in discriminating closely related species, but nrDNA failed. In addition, incomplete lineage sorting occurred during the fast-radiative speciation of this genus may also play a role in the cyto-nuclear discordance [60, 68]. However, our data is not appropriate to clearly discriminate among these possible scenarios, we therefore recognize that further studies involving whole genome sequences at the population level are needed to better understand their evolutionary histories.

Conclusions

Our phylogenetic analyses based on plastome datasets still failed to resolve the interspecific delimitation and relationships among four closely related Ostrya species, in contrast to the well-resolved phylogeny based on the ITS sequence variations. Such incongruence may result from incomplete lineage sorting and hybrid introgression during the divergences of the four species. Previous plastome studies using one individual per species usually recovered high phylogenetic resolution and resolved relationships, however, our results indicated that multiple samples from different populations should be considered when doing such phylogenetic studies, especially among closely related species. Finally, our findings shed light on some interesting evolutionary questions, e.g., what causes O. japonica split into four subclades in the plastome trees, and further genomic studies at the population level are necessary to gain a deep understanding of the evolution of these species.