Introduction

Chloroplast (cp) is a specialized eukaryotic organelle and its genetic materials are mainly maternally inheritance, in which a core set of genes have originated from the cyanobacterial ancestor and are mostly involved in photosynthesis and metabolic processes1,2,3,4. Chloroplast genome has a small size and is roughly 120–180 kilobases in length5. The advancement of modern sequencing technologies has boosted the study of chloroplast genetics and genomics. Insights into chloroplast genome sequences have revealed considerable sequence and structural variations occurred within and between plant species. For example, three types of mutations, including gene/intron loss, inverted repeat changes and inversions in the land plant chloroplast genomes can lead to the gene order changes and are often referred to as structural changes or rearrangements5. To date, chloroplast genomes have been widely utilized as markers for studying the species identification, phylogenetic and population analyses6,7,8.

Filipendula Mill. (Rosoideae, Rosaceae) is a perennial herbaceous plant and contains approximately 15 species, which generally grows in the high mountain of the temperate regions9. The geographic distribution of Filipendula mainly covers East Asia, Europe and North America10. Filipendula species have long been utilized for medicinal purposes and most published papers have focused on the medicinal properties of these plants11,12. Their aerial parts (leave and flower) and underground organs (roots) are good resources of bioactive substances, including tannins, polyphenolic acid and essential oils, which have antioxidant, anticancer, anti-inflammatory, gastroprotective, anti-hyperalgesic, anti-genotoxic, and hepatoprotective effects13,14. Besides, the leaves of Filipendula can be processed into the herbal tea in Russia and other Siberia countries, which is used to relieve influenza and gout, to clean wounds and eyes15.

The classification of genus Filipendula is confused all the time16. Juzepczuk16 has divided this genus into three subgenera and two sections mainly based on the indigenous species. Afterward, Shimizu17 revised the former taxonomic system and classified 15 species of the whole genus into two monotypic subgenera (Hypogyna T. Shimizu and Filipendula) and one large subgenus (Ulmaria Moench) with four sections (Ulmaria Hill, Albicoma Juz., Sessilia T. Shimizu and Schalameya Juz.). In 1967, Sergievskaya amended the two former systems and divided the genus into four subgenera, including three subgenera of Shimizu’ system and subgenus Aceraria of Juzepczuk’s system16. Of these four subgenera, only Shimizu’s sect. Ulmaria was retained within subgenus Ulmaria and the remaining sections were transferred into subgenus Aceraria. In the last taxonomic revision of the genus, Schanzer9 divided the genus into four sections: Hypogyna, Schalameya, Albicoma and Filipendula mainly based on the morphological and geographic data. Therefore, the four systems are incongruent with each other to a certain extent and the names of some species in the different systems are still used.

To date, limited studies have been documented on Filipendula diversity and phylogenetic analysis. Only few studies have reported that isozymes18 and microsatellites19 can be used as markers to assess genetic variations in F. vulgaris. Investigations of the phylogeny of Rosoideae or Rosaceae have revealed that Filipendula as monophyly is sister to the rest of the subfamily Rosoideae20,21,22. Several evidence have revealed that the species in the basal lineage exhibited the unique chloroplast structure. For example, a single inversion as the powerful phylogenetic marker identified the basal members of the Asteraceae23. In a second case, two inversions and an expansion of the IR clarified the basal nodes in leptosporangiate ferns24. Whether did this phenomenon occur between Filipendula and other genera of the subfamily Rosoideae? However, the basic knowledge of the chloroplast genome in Filipendula is absent and the chloroplast phylogeny of Filipendula species has not been reported until now. Moreover, the infrageneric phylogenetic relationships of Filipendula was only analysed using one nucleotide segment (ITS)10. Therefore, the present study aimed to provide the unprecedented chloroplast genome data for comparative analysis, to reconstruct the infrageneric phylogeny of Filipendula based on eight cp genomes (F. vestita, F. ulmaria, F. palmata (including two varieties, F. palmata var. palma and F. palmata var. glabra), F. angustiloba, F. vulgaris, F. camtschatica and F. multijuga) and to explore evolutionary history of this genus.

Results

Characterization and structural analyses of eight Filipendula cp genomes

In this study, eight assembled cp genomes from seven Filipendula species in which F. palmata had two varieties (Fig. 1), had an average size of 154,522 bp (ranging from 154,205 bp-154,633 bp) and 36.63% GC content (Table 1). These eight cp genomes were divided into four regions and two copies of an inverted repeat (IR) separated large and small single copy regions (LSC and SSC), respectively (Fig. 2). The four regions formed the typical circular structure and varied a little in size, in which the LSC region had a largest size, ranging from 82,851 bp to 83,295 bp, followed by the IR region (from 27,093 bp to 27,286 bp) (Fig. 2 and Table 1). In addition, a total of 126 genes were annotated in each Filipendula cp genome except for F. camtschatica, including 81 protein-coding genes (PCGs), 37 tRNA and 8 rRNA genes. It was worth noting that the gene number had reduced by one because rpl14 was not found in F. camtschatica (Table 1). The majority of PCGs were involved in the photosynthesis and metabolism (Table S1). Of all genes, 16 duplicated genes were identified in the IR region, and 16 genes (petB, petD, atpF, ndhA, ndhB, rpoC1, rps16, rps12, rpl16, trnA-UGC, trnG-UCC, trnI-GAU, trnK-UUU, trnL-UAA, trnV-UAC, ycf3) had the introns, in which two genes (ycf3 and rps12) had two introns and the rest of them had one intron (Table S1).

Figure 1
figure 1

Photograph images of Filipendula palmata var. palmata (A-C), F. vestita (D and E) and F. ulmaria (F and G). Photos by Jie Cai, Ting Zhang and Ji-Dong Ya from Kunming Institute of Botany, Chinese Academy of Sciences.

Full size image
Table 1 Summary of complete plastomes of Filipendula species.
Full size table
Figure 2
figure 2

Circular map of Filipendula chloroplast genome. The inner grey circle indicates the GC content of each genome position. Genes in the inner circle of the genomic map are transcribed clockwise and vice versa.

Full size image

To further analyse the structure of eight Filipendula cp genomes, multiple alignments were conducted, and the results indicated that there were an identical gene order and orientation across these tested Filipendula species (Figure S1), which was consistent with the result of circular map of Filipendula cp genome (Fig. 2). Early findings have indicated that the variations of IR play an important effect on the stability of plastome structure5,25. In this study, a comprehensive comparison of the IR/SC boundaries were analysed among eight Filipendula cp genomes. The result indicated that the boundaries of IR/LSC were very conserved, LSC/IRb/a (JLB/A) boundaries were flanked rps19 and trnH with a length of 8 bp away from the 5’ end and 3’ end of these two genes, respectively (Fig. 3). In contrast, the IR/SSC junctions showed the few differences. ψycf1 and ycf1 separately span the boundaries of IRb/SSC (JSB) and IRa/SSC (JSA). Two flanked distances of the junction point between ψycf1 and JSB or ycf1 and JSA exhibited the different lengths in these two genes because the lengths of ψycf1 and ycf1 occurred a few changes in Filipendula species (Fig. 3). Therefore, the nearly unchanged IR might facilitate the stability of plastome structure of this genus. Altogether, these results demonstrated that the cp genome structure was evolutionarily conservative in Filipendula.

Figure 3
figure 3

Comparison of the border regions of four chloroplast genome parts among Filipendula species.

Full size image

However, we found that Filipendula cp genomes exhibited the structural differences when compared with those of other genera of Rosoideae. At first, Filipendula cp genomes had a smaller gene number and three genes (rps4, rpl2 and rpl32) were absent when compared to other genera of Rosoideae (Fig. 4). In addition, the gene order in three sequence blocks (ndhC and trnT-UGU, rps12 and accD, trnS-GGA and trnfM-CAU) of other genera of Rosoideae plants were highly conserved, whereas those of Filipendula cp genomes significantly differed (Fig. 4). Further analysis indicated that a minimum of three inversions occurred within cp genomes of Filipendula species (Fig. 4). Besides, Filipendula species had a plesiomorphic gene order similar to other genera of Rosoideae plants in two blocks of psbM and trnG-GCC, trnV-UAC and rbcL. However, these two blocks had the obvious changes in location within the cp genomes of Filipendula when compared to those of other genera of Rosoideae plants (Fig. 4). Such transpositions of these blocks caused to the divergent chloroplast gene order between Filipendula plants and other genera of Rosoideae plants (Fig. 4). Therefore, the cp genomes of Filipendula species exhibited the considerable differences in structure from those of other genera of Rosoideae plants: a minimum of 3 inversions, transpositions of two blocks within the LSC and gene losses.

Figure 4
figure 4

Structural variations between 15 representative genera of Rosoideae and Filipendula plastomes.

Full size image

Repeats in plastome may be associated with the endpoints of inversion5. In present study, four types of repeats (palindromic repeats, forward repeats, reverse repeat and complement repeats) were detected in Filipendula cp genomes. The total number of repeats varied from 273 to 321 (Fig. 5), which outnumbered other species of Rosaceae (i.e. Sorbus)26. Filipendula camtschatica had the most abundant repeats, including 152, 6, 5 and 158 forward, reverse, complement and palindromic repeats, respectively (Fig. 4). Similarly, forward and palindromic repeats became two major repeat types in other six Filipendula species (Fig. 4). Although complementary repeats (2–6) and reverse repeats (5–8) had the small number, they were observed in each Filipendula species (Fig. 5). The majority of the repeats were found in intergenic regions (Table S2). Some repeats were found in coding or intron sequences of several genes, such as trnG-UCC, trnG-GCC, trnL-UAA, accD, psaA, psaB, clpP, ycf1, ycf2, ycf3, ycf4, petB, ndhF and trnL-UAA (Table S2). Interestingly, all the genes except trnG-UCC, ycf1, ycf2, petB and ndhF were located in three inversion and two transposition blocks (Fig. 5). Additionally, among six genes of reversion endpoints, only accD contained the repeats, none of repeats were observed in the remainder (ndhC, rps12, trnfM-CAU, trnS-GGA and trnT-UGU) (Table S2). It was worth mentioning that rps12 was duplicated in the endpoint of rps12-accD inversion in Filipendula (Fig. 5).

Figure 5
figure 5

Number of four type repeats examined in eight Filipendula chloroplast genomes.

Full size image

Genomic sequence divergence analysis in Filipendula

To better understand the sequence divergence of Filipendula species, eight whole plastomes were compared and used to analyse sequence identity with mVISTA program using the cp genome of F. angustiloba as a reference. The results indicated that the whole cp genomes of Filipendula species were relatively conserved, in which the LSC region exhibited the highest divergence, whereas the IR regions were the most conserved (Figure S2). In addition, the high sequence divergence mainly occurred in noncoding regions, whereas only several genes (i.e., accD, clpP, ycf1 and ycf2) were found to be divergent in their coding regions (Figure S2).

SSRs are a class of short tandem repeats (1–6 bp) and highly polymorphic markers, which are widely distributed in the plastomes in plants and commonly used for species identification and phylogenetic analyses27,28,29. In this study, the mono-, di-, tri-, tetra-, penta- and hexa-nucleotide repeat units were analysed. Filipendula cp genomes were found to contain 105 (F. vulgaris) to 123 (F. camtschatica) SSRs (Fig. 6A, Table S3). Most of the SSRs were mononucleotide repeats (66.07%, 56.91%, 60.91%, 65.77%, 66.07%, 63.72%, 58.93% and 63.81% in F. angustiloba, F. camtschatica, F. multijuga, F. palmata var. glabra, F. palmata var. palmata, F. ulmaria, F. vestita and F. vulgaris, respectively), which mainly made up of A and T nucleotides (Fig. 6A,B). Dinucleotide repeats were the second abundant SSRs with the major constitution of AT/AT nucleotides (Fig. 6A,B). Trinucleotids and tetranucleotide repeats were small in number, but both repeats were observed in each Filipendula plastomes (Fig. 6A,B, Table S3). By contrast, pentanucleotids and hexanucleotids were found in only few Filipendula plastomes. For example, few pentanucleotide repeats were only found in F. multijuga, F. camtschatica and F. vulgaris and one hexanucleotide repeat was only found in F. vulgaris (Fig. 6A).

Figure 6
figure 6

Frequency of six SSR types (A) and distribution of SSR sequences (B) examined in eight Filipendula chloroplast genomes.

Full size image

Besides, sliding window analysis was conducted to reveal the highly variable regions in eight Filipendula cp genomes. The average value of nucleotide diversity (Pi) over the entire cp genome was 0.005, indicating the whole cp genome was relatively conserved (Fig. 7). This result was consistent with the mVISTA result (Figure S2). In addition, we found that the high variability mainly occurred in noncoding regions. Four mutational hotspots with pi values greater than 0.02 were identified, namely ψycf1-ndhF, rps12-trnV-GAC, ndhF-trnL-UAG, and trnV-GAC-rps12 (Fig. 7). Of four variable regions, rps12-trnV-GAC and trnV-GAC-rps12 from two IR regions had the highest pi value. Based on these results, the noncoding regions exhibited the higher variability and divergence than the protein-coding regions. And then, selective signatures were determined by the ratios of non-synonymous (Ka) to synonymous (Ks) substitution rates on the 76 unique protein sequences. Our results demonstrated that the ratios of Ka/Ks of the majority of genes in these Filipendula species were less than 1, suggesting these PCGs were under strong purifying selection (Table S4). Two genes (matK and rps8) with Ka/Ks ratios more than 1 were under positive selection (Table S4).

Figure 7
figure 7

Sliding window analysis of Pi values among cp genomes of seven Filipendula species. X-axis, position of the midpoint of a window; Y-axis, nucleotide diversity of each window. (Window length: 600 bp, step size: 200 bp).

Full size image

Phylogenetic and molecular dating analysis of Filipendula

The structural rearrangement of chloroplast genome is usually used for reconstructing phylogenies of plants5. Based on our results, the overall structure of cp genome was highly conserved in seven Filipendula species (including two varieties). Under this case, the high homoplasy of cp genome structure was not used for phylogenetic analysis. Nevertheless, the cp genome of genus Filipendula generated gene loss, transposition and inversion, whereas the other genera of Rosoideae were lack of these structural changes. Two previous studies have given an identical support for Filipendula as the first clade to split off the rest of Rosoideae in the nuclear and plastome trees21,30. Therefore, Filipendula was the basal clade in the Rosoideae probably because the gene loss, transposition and inversion mark an ancient evolutionary split in this subfamily.

Besides, sequence divergence generated a large number of genetic variations in eight Filipendula cp genomes, which can be used for reconstructing the phylogeny of Filipendula. In this study, all PCG sequences were used to infer the phylogenetic relationships within this genus by ML, MP, BI and ASTRAL methods. The results indicated that these trees formed the major identical topology (Fig. 8). The phylogenetic analyses revealed that Filipendula was the basal genus in Rosoideae, which was consistent with previous results20,21. Infrageneric relationship of Filipendula had been resolved two major clades (Fig. 8). One clade contained F. vulgaris (the type species) with high support values (i.e. 100% BS and 1.0 PP). The remaining species clustered into two sister clades (Fig. 8). One sister group contained F. camtschatica and F. multijuga and other five Filipendula species or varieties formed another group (Fig. 8). In the previous study, phylogenetic analyses also resolved two major lineages within Filipendula based on one nucleotide segment (ITS)10. However, F. occidentalis from North America was the basal species and the others from Asia and Europe clustered into two sister clades10. Therefore, we will greatly expand the sampling to better understand the phylogenetic relationships within Filipendula.

Figure 8
figure 8

The phylogenetic tree of seven Filipendula species and 29 species of three subfamilies of Rosaceae based on PCG data. Numbers at nodes correspond to the support values of ML, MP, BI and ASTRAL, respectively.

Full size image

The divergence time between Filipendula and other genera of Rosoideae was estimated at 82.88 Ma (82.04–83.77 Ma, 95% HPD) in the Cretaceous (Fig. 9). After that, Filipendula located on a long branch, implying this genus had an evolutionary history different from other genera of Rosoideae. As shown in Fig. 9, the age of the most recent common ancestor of Filipendula was estimated at about 9.64 Ma (9.11–10.17 Ma, 95% HPD) in the late Miocene. In contrast, intergeneric diversity times of other genera of Rosoideae occurred from 73.3 Ma to 4.35 Ma. Previous studies have demonstrated that the diversification of Rosaceae increased at two different periods. The rapid initial diversity occurred in the late Cretaceous and the second shift occurred in the early Oligocene onwards21. Based on our results, genus Filipendula had an early origin but late diversification during evolutionary process, which was apparently different from other genera of Rosoideae.

Figure 9
figure 9

Divergence times estimation of 40 species of Rosaceae based on PCG data. The divergence times are shown near each node. Blue bars represent 95% high posterior density for the estimated mean dates. C1-C4 in the blue green circle represent calibration points.

Full size image

Discussion

The chloroplast genome has been used as a powerful marker for investigating plant evolution and phylogenetic analysis5. Several cases have been published comparing genomes of taxa among which the structural changes occurred in cp genomes of the basal members at the different taxonomic levels2,5. In the present study, the cp genomes of several Filipendula species in the basal clade of subfamily Rosoideae were first analysed. Our results indicated that eight Filipendula cp genomes share several common features with those of other genera of Rosoideae. For example, they have a typical quadripartite structure of cp genomes and similar GC content with most land plants26. The cp genome organization was highly conserved in the tested Filipendula species, but these cp genomes exhibited the apparent changes in gene order and orientation when compared with those of other genera of Rosoideae. Our results revealed that gene loss, inversion and transposition contributed to the structural changes between Filipendula species and the remainder of Rosoideae. Therefore, the phenomenon of structural changes (rearrangements) occurred in Filipendula species of the basal linage of Rosoideae.

The expansion and contraction or loss of IR can disrupt gene order, whereas the stability of IR may facilitate the gene order conservation5. In this study, the nearly unchanged boundaries of IR/SC contributed to reduce gene order changes of Filipendula cp genomes. In addition, gene/intron loss is considered as one of three classes of gene order changes in the land plant cp genomes5. In the present study, the gene loss events might occur in the Filipendula cp genomes because several genes were not found in their cp genomes. However, one gene absence (rpl14) within Filipendula was restricted to individual species (F. camtschatica). Therefore, we inferred that gene loss events might have continued to occur during Filipendula plant diversification. Inversion is the most common mechanism leading to gene order changes5. In this study, three inversions and two transpositions mainly contributed to the changes of order and orientation of 39 genes, which were involved in almost all functional classifications. Most chloroplast genes are often under the control of an operon; therefore, transcriptional regulation of these genes might be affected by the changes of gene order and orientation. Usually, both endpoints of inversion occur in non-coding regions, in which no genes are disrupted. In the study, one inversion occurred with the endpoint in accD. This gene not only contained the sequence repeats, but also exhibited the high sequence polymorphism. Therefore, we inferred that accD might be active in contribute to genomic rearrangement and sequence divergence of Filipendula plants.

The structural rearrangements have led to the low levels of homoplasy of cp genomes between Filipendula and the remainder of Rosoideae. In this study, the sequences of all unique PCGs were used to construct the phylogenetic tree. Our results indicated that Filipendula was indeed located in the basal linage of Rosoideae, which was consistent with the previous results20,21,22. Besides, we found that the gene order and orientation were conserved between the other genera of Rosoideae and the representative species of Dryadoideae, Amygdaloideae. Therefore, we inferred that the structural rearrangement of cp genomes should be an independent evolutionary event within Filipendula after divergence from the other genera of Rosoideae. Meanwhile, Filipendula species showed the different diversity periods from other genera of Rosoideae, implying that Filipendula species might experience the different evolutionary process. In the present study, infrageneric relationship of Filipendula was highly supported using PCGs, which was a little different from the segment tree10. In previous study, more samples were used to construct the phylogenetic tree, including the species from North America, Asia and Europe10. Therefore, the comprehensive phylogenetic relationship might be better understood by sampling more species within Filipendula.

Materials and methods

DNA extraction, sequencing, assembly and annotation

A total of seven Filipendula species were used, including F. vestita, F. ulmaria, F. palmata (including two varieties, F. palmata var. palma and F. palmata var. glabra), F. angustiloba, F. vulgaris, F. camtschatica and F. multijuga in this study. Of them, the former five species were collected from the different provinces of China and the last two from UK and Japan, respectively. The raw sequencing reads of F. vulgaris was downloaded from the NCBI SRA database. The detailed information was shown in Table S5. All the voucher specimens we collected were deposited in Herbarium, Kunming Institute of Botany, CAS (KUN) or Royal Botanic Garden Edinburgh Herbarium (E). The silica gel-dried leave of each species was used to extract the genomic DNA by the modified CTAB method and the constructed libraries were sequenced by Illumina NovaSeq PE150 platform.

The high-quality reads of the cp genome data were de novo assembled into the contigs using SPAdes software31, which were further circulated using Bandage software32. The genome annotation was implemented by GeSeq software33 using Potentilla spp. as references with 65% and 80% similarity to proteins and tRNA (or rRNA) genes, respectively. The annotated and circular plastome was drawn using OGDRAW program34. Besides, the raw data of cp genome of F. vulgaris (type species) were downloaded (SRA: ERR5554718) and analyzed using the same method as described in six species. All methods of experimental research on plants were performed in accordance with the relevant institutional, national, and international guidelines and legislation.

Comparative analysis of chloroplast genomes

Four different repeat types, including forward, palindrome, reverse and complement sequences were analyzed and six microsatellites (simple sequence repeats, SSRs), including mono-, di-, tri-, tetra-, penta-, and hexa-nucleotide repeat units were identified in seven Filipendula species as previously described26. After the plastid genome sequences of seven Filipendula species were aligned using MAFFT v.6.83335 using the default settings, multiple alignments were used to analyze infrageneric genome collinearity using the Mauve software36. In this study, the representative species from 15 genera of Rosoideae, including Agrimonia coreana, Alchemilla argyrophylla, Argentina phanerophlebia, Comarum salesovianum, Dasiphora fruticose, Drymocallis saviczii, Filipendula angustiloba, Fragraria chiloensis, Geum elatum, Hagenia abyssinica, Potentilla anserina, Rosa acicularis, Rubus amaabilis, Sanguisorba filiformis, Sibbaldia aphanopetala and Sibbaldianthe adpressa were used for intergeneric collinearity analyses. Moreover, the comparative review of the whole genome alignment of seven Filipendula species was visualized using the mVISTA program37 using F. angustiloba as reference sequence. The online IRscope tool38 was used to analyze the joint site information of cp genome.

The sliding window analysis of nucleotide diversity (Pi) of cp genome was performed by DNASP 5.0 program39. The window length and step size were set to 600 and 200 bp, respectively. The ratio of the nonsynonymous (Ka) and synonymous (Ks) substitution rates was used to calculate the selective pressure between orthologous genes of cp genome of Filipendula species. The coding gene sequences were selected from each cp genome and then aligned by using by MAFFT v.6.83335. The resulted alignment was used to calculate the ratio of Ka/Ks using KaKs_Calculator 3.040.

Phylogenetic analysis and divergence time estimates

A total of seven Filipendula species (including two varieties of F. palmata) were used to construct the phylogenetic tree. Four species from other families of Rosales (two species from Cannabaceae, and one species from Moraceae and Rhamnaceae, respectively) and 29 species of three subfamilies of Rosaceae (one representative species of 25 other genera of Rosoideae, 2 species from Dryadoideae and 2 species from Amygdaloideae) were used as outgroups (Table S6). In this study, 76 protein-coding gene sequences were extracted on the basis of their annotation. All these sequences were aligned using MAFFT v.6.83335. We constructed the phylogenies using the concatenated and coalescent methods. For the concatenated analysis, all aligned PCG sequences were concatenated to a single alignment dataset for phylogenetic inference using maximum likelihood analysis (ML), Bayesian inference (BI), maximum parsimony (MP) methods. Briefly, the maximum likelihood (ML) analysis was performed using RAxML v7.2.641 under GTRGAMMA model for 1000 bootstrap42. Bayesian inference (BI) was performed using MrBayes v3.1.243. The Markov Chain Monte Carlo (MCMC) analysis was run for 2 million generations. The trees were sampled at every 100 generations and the first 25% trees were discarded as burn-in. Finally, the majority-rule consensus tree was generated by the remaining trees with posterior probability (PP) values for each node. Maximum parsimony (MP) analysis was run in PAUP (v4.0b10)44, using heuristic search with 1000 bootstrap replicates and tree bisection-reconnection (TBR) branch swapping. For the coalescent analysis, each PCG was used to construct a ML tree as above described. All generated gene trees were used to estimate the species tree with ASTRAL45 in Phylosuit46.

From the best ML tree, we generated 1000 bootstrap replicates to produce a dated phylogeny with a 95% confidence interval (CI) on the age at the nodes using TREEPL47, following the guide by Maurin48. We considered 90 and 106.5 Ma as the minimum- and maximum-age calibrations for the stem of Rosaceae as suggested by Zhang et al.21. Three fossil calibrations were also used as minimum-age calibrations assigned to internal nodes (all outside our study clades) (Table S7).

Conclusions

The complete chloroplast genomes of seven Filipendula species were analysed in this study. The genome structure and gene content within Filipendula were rather conserved. However, gene loss, transpostion and inversion were observed in the cp genomes of Filipendula when compared with those of other genera of Rosoideae. Sequence divergence mainly occurred in noncoding regions, in which numbers of SSRs and four mutational hotspots were identified in each Filipendula species. The phylogenetic and molecular dating analyses showed that Filipendula was divergent from other genera of Rosoideae about 82.88 Ma (82.04–83.77 Ma, 95%HPD). And seven Filipendula species were split at 9.64 Ma (9.11–10.17 Ma, 95%HPD) into two major clades. The results provided the basis for the study of the evolutionary history and phylogenetic analysis of Filipendula.