Background

Spiraea L., a member of the tribe Spiraeeae of Amygdaloideae (Rosaceae), comprises approximately 90 species (http://www.plantsoftheworldonline.org), and is distributed across the northern temperate regions of the world. China is recognized as the center of diversity for this genus, with over 70 species, including 47 endemic species [1]. Spiraea is renowned for its ornamental value, particularly its attractive white, pink or purple flowers that bloom profusely in spring or summer, forming heads composed of numerous tiny blooms. Additionally, certain varieties of Spiraea with golden foliage, such as Spiraea × vanhouttei ‘Pink Ice’ and ‘Gold Fountain’, have been developed. To fully exploit their horticultural value, a comprehensive taxonomic and phylogenetic study of Spiraea taxa is necessary. However, the phylogeny of Spiraea has not been extensively investigated thus far.

The traditional infrageneric classification of Spiraea is primarily based on inflorescence types, including panicles, compound corymbs, corymb, umbel, and fascicle, as established by Poyarkova [2], Rehder [3], and Yü and Kuan [4]. For Chinese Spiraea taxa, four sections (i.e., sects. Spiraea, Calospira, Chamaedryon, and Glomerati) and ten series within the genus have been widely accepted [4]. Previous phylogenetic analyses have revealed issues with traditional classifications based on inflorescence morphology, as the majority of sections are considered paraphyletic or polyphyletic [5,6,7,8,9,10]. However, most nodes in these phylogenetic trees could not receive strong support values based on limited molecular fragments, highlighting the need for the development of robust molecular markers within Spiraea.

The plastid, an essential organelle in plant cells, plays a crucial role in plant growth and development [11]. Compared to nuclear and mitochondrial genomes, the plastid genome possesses unique characteristics, such as uniparental inheritance, absence of recombination, conserved genome structure and gene content, and moderate evolutionary rate. These features make the plastid genome highly valuable for resolving phylogenetic relationships at various taxonomic levels [12]. Furthermore, the assembly of plastid genomes is relatively straightforward due to their high cellular copy numbers and the development of assembly algorithms [13]. As a result, the number of complete plastid genomes has significantly increased in recent years, thanks to the advancement of high-throughput DNA sequencing technologies. Plastid genomes have proven to be invaluable in elucidating the phylogenetic backbone of Rosaceae and uncovering its evolutionary history [14]. The number of plastid genomes within the Rosaceae family has also rapidly increased in recent years (e.g., Fragaria [15, 16], Rubus [17], and Rosa [18, 19]), providing a stronger understanding of phylogenetic relationships at the species level and the phylogeographical patterns within the family. However, a limited number of plastid genomes of Spiraea species have been reported [20,21,22,23], and no solid divergence times of the genus have been inferred. Therefore, it is crucial to conduct comparative analyses with more Spiraea plastid genomes to better identify the evolutionary pattern of the plastid genome within Spiraea and assess its phylogenetic relationships.

In this study, we assembled the complete plastomes for 34 Spiraea accessions representing 27 Asiatic species using several available Spiraea plastomes from GenBank. Our objectives were to (1) examine the variation in plastome structure and features of Spiraea, (2) identify the hypervariable loci across these Spiraea plastomes, (3) infer the phylogenomic relationships of Spiraea, and (4) determine the divergence time of Spiraea. This study will contribute to a better understanding of the evolution of Spiraea based on plastome sequences.

Materials and methods

Plant materials, DNA extraction, and sequencing

In this study, we collected a total of 27 Asiatic Spiraea species and 34 accessions, representing four sections (i.e., sects. Calospira, Chamaedryon, Glomerati and Spiraea) of two subgenera (i.e., subgs. Protospiraea and Metaspiraea) in Spiraea. Among these accessions, 25 were newly collected from Hangzhou Botanical Garden. Voucher specimens were formally identified by Qiao-Ling Zhang and deposited at the Herbarium of South China Botanical Garden (IBSC), Chinese Academy of Sciences (CAS). Silica-gel dried leaves were used for DNA extraction by the cetyltrimethylammonium bromide (CTAB) method [24]. Paired-end (PE) sequencing libraries were constructed and sequenced on an MGISEQ-2000 (MGI, Shenzhen, China) at the Beijing Genomics Institution (BGI, Wuhan, China). Additionally, twelve plastid genomes of Spiraea were directly downloaded from GenBank (Accessed on February 5, 2023), which belong to nine Spiraea species and three outgroups (i.e., Malus baccata (L.) Borkh., Rosa cymose Tratt. and Sibiraea angustata (Rehd.) Hand.-Mazz.) for phylogenetic inferences. Among all accessions, four Spiraea species (Spiraea japonica L.F., S. × vanhouttei (Birot) Zabel, S. cantoniensis Lour., and S. prunifolia Siebold & Zucc.) have multiple individuals. The detailed sampling information is summarized in Table S1.

Plastid genome assembly, annotation, and comparison

Plastomes of 25 Spiraea accessions were newly assembled using GetOrangelle [25] with the reference plastome of Spiraea mongolica (NC_051992). These complete plastid sequences were annotated using the Plastid Genome Annotator (PGA) [26] and manually checked and adjusted with the aid of Geneious 2019 [27]. The circular physical maps were drawn using OrganellarGenome DRAW (ORDRAW) [28].

The total length of the plastomes, as well as the lengths of the large single copy (LSC), small single copy (SSC), and inverted repeat regions (IRA and IRB), gene content, and GC content of all Spiraea species were analyzed using Geneious. The relationships among the genome size and LSC length, IR length, intron length, and intergenic spacer length of Spiraea species were tested using least squares linear regression analysis in R 4.2.2 [29]. The boundary information between the IRs and SC regions was analyzed in Geneious.

Divergent hotspot analyses

The plastid genome sequences were aligned using MAFFT v7.308 [30]. The nucleotide polymorphism (Pi) of plastid genomes was calculated using the sliding window method with a length of 600 bp and a step size of 200 bp in DnaSP v6 [31]. The sequence heterogeneity within 34 Spiraea plastid genomes was visualized using mVISTA with the Shuffle-LAGAN model [32].

Phylogenetic analyses and molecular dating

Phylogenetic analysis was performed on all 34 Spiraea plastid genome sequences, with M. baccata (MK571561), R. cymosa (NC_051550), and Si. Angustata (NC_054238) used as outgroups. The 79 protein-coding genes (PCGs) and whole plastid genome sequences (including all coding and non-coding regions) were aligned separately using MAFFT and concatenated into a sequence matrix using AMAS [33]. Maximum likelihood (ML) and Bayesian inference (BI) trees were each inferred using RAxML v8.2.10 [34] and MrBayes v3.2.6 [35]. The ML tree was constructed using RAxML under the GTRGAMMA model with 1,000 bootstrap replicates. BI was performed with MrBayes in the GTR + G + I model on XSEDE (the CIPRES Science Gateway, [36]). The Markov chain Monte Carlo (MCMC) chains were run for 10 × 106 generations, and stationarity was assessed using TRACER v1.7.1 (effective sample size (ESS) > 200) [37]. The first 25% of runs were discarded as burn-in, and the remaining trees were used to construct the consensus tree. Bayesian posterior probabilities (BPPs) were used to estimate support for each branch. The phylogenetic tree file was visualized in FigTree v1.4.3 [38].

To estimate the divergence time of Spiraea, we retrieved the other 86 plastid genomes of Rosales taxa from GenBank (Table S1). A total of 78 PCGs extracted from 120 plastid genomes were extracted and aligned separately by MAFFT. Maximum likelihood trees were constructed for each of these 78 PCGs using RAxML, and molecular clock-like genes (clock-like genes) were screened from the 78 gene trees by using the SortaDate package [39]. The top-ten clock-like genes were selected and aligned using MAFFT, and then concatenated using AMAS for further divergence time estimation.

The concatenated matrix was used to estimate the divergence time using BEAST v2.7 [40]. The yule tree prior was used with an optimized relaxed clock (ORC) and GTR + GAMMA was selected for the molecular clock model. We ran four independent analyses, each consisting of 400 million generations and sampled every 10,000 generations. The results from these analyses were combined using the LogCombiner package. Convergence was assessed using TRACER, based on the effective sample size (ESS) of each parameter with a likelihood > 200, following the method described in [41]). After discarding the first 25% of trees as burn-in, the maximum clade credibility (MCC) tree was generated using the TreeAnnotator package. The mean nodal heights and age estimation with 95% highest posterior densities (HPDs) were visualized using FigTree.

Twelve reliable fossils from Rosaceae were used for calibration, as described by Xiang et al. [42]. The fossil information and calibration settings are shown in Table S2. Notably, Spiraea-like fossils (fruits and leaves) have been found in the Early Eocene upland floras of the Okanogan Highlands of northeastern Washington State and British Columbia, Canada [42, 43]. Therefore, the stem age of Spiraea was estimated using a standard lognormal prior distribution with a mean of 0.25, a deviation of 0.3, and an offset of 48 Ma, which roughly matched the hypothesized age (49–50 Ma).

To examine the general patterns of diversification through time of Spiraea, lineage-through-time (LTT) plots were calculated using APE v3.5 [44]. To account for uncertainty in the dating estimates, the 1,000 random trees from the 4,000 converged trees from BEAST were used to calculate the 95% confidence interval (CI), and all outgroups were removed.

Selection pressure analyses

We excluded PCGs with lengths less than 300 bp. A total of 50 PCGs were chosen and aligned using the MUSCLE algorithm [45]. Stop codons were removed from the aligned sequences using MEGA 11 [46]. The resulting aligned sequences were then converted to PML format using PhyloSuite v1.2.2 [47].

To calculate the non-synonymous substitution rate (dN), synonymous substitution rate (dS), and ratio values (ω, dN/dS) of each protein-coding gene, we employed the Site Model (M0) using EasycodeML v1.4 [48]. The PCGs were categorized into seven functional groups based on the classification by Wicke et al. [49]: ATP synthase, NADPH dehydrogenase, cytochrome b/f complex, photosystem, ribosomal proteins, RNA polymerase, and other genes.

Likelihood ratio tests (LRT) were conducted to determine the presence/absence of selection. The log-likelihood statistic was calculated as 2∆L = 2(L1- L0). This value was then compared with a χ2 distribution with degrees of freedom calculated from the difference in the number of parameters between the models. L0 represents the log-likelihood under the null model, while L1 represents the likelihood under the alternative model. Model M0 assumes a single value of ω for all sites and lineages, while model M3 includes a discrete distribution of ω values. Model M1 assumes two site classes, and was compared with M2, which allows for three site classes including positive selection. Models M7 and M8 were also compared, with M7 assuming a beta distribution for variation in ω and M8 allowing for positive selection. Positive selection can only be inferred when models M2, M3, or M8 indicate codons with a ω ratio > 1 and the likelihood ratio test for positive selection is significant at P < 0.05. Bayesian empirical Bayes (BEB) was used to identify sites potentially under positive selection.

Results

Structural features and genome composition of plastid genomes in Spiraea

In this study, we analyzed the plastid genomes of 34 individuals from the Spiraea genus, representing 27 different Asiatic species. We obtained nine Spiraea plastid genome sequences from GenBank. The complete plastid genomes exhibited the typical quadripartite circular structure, consisting of a large single-copy (LSC) region, a small single-copy (SSC) region, and two inverted repeat (IR) regions (Fig. 1). The length of the genomes ranged from 153,822 bp (S. japonica, MZ981784) to 158,637 bp (S. insularis), while the LSC region length ranged from 82,226 bp (S. japonica) to 86,997 bp (S. insularis). The shortest SSC region was 26,333 bp, found in four species: S. ovalis, S. henryi, S. nipponica, and S. trichocarpa, while the longest SSC region was 26,492 bp in S. martini. The IR region length ranged from 18,700 bp (S. laeta) to 18,957 bp (S. ovalis). The GC content of the Spiraea plastid genome ranged from 36.6% to 36.9%. This information is summarized in Table S3. Linear regression analysis showed a significant correlation between genome size and the length of the SC region (R2 = 0.99, P < 0.01) and intergenic spacer (IGS) length (R2 = 0.21, P < 0.01), but no correlation with the IR region (R2 = 0.03, P = 0.31) or intron length (R2 = 0.02, P = 0.44) (Fig. 2).

Fig. 1
figure 1

The plastid genome map of Asiatic Spiraea species. The genes belonging to different functional groups are shown in different colors. Genes inside and outside of the external circle are transcribed in clockwise and counterclockwise directions, respectively. The inner circle represents the quadripartite structure, with two copies of the inverted repeat (IRA and IRB), an LSC, and an SSC region in black with GC content in dark gray and AT content in light gray

Full size image
Fig. 2
figure 2

The relationships between the length of different parts of plastome and plastome size within Spiraea. a Correlation of the IR length and plastome size. b Correlation of the intron length and plastome size. c Correlation of the SC length and plastome size. d Correlation of the IGS length and plastome size

Full size image

The Spiraea plastid genomes encoded a total of 113–114 genes, including 78–79 protein-coding genes (PCGs), 30 tRNA genes, and 4 rRNA genes (Table S3). Seventeen genes contained one intron, namely, atpF, clpP, petB, petD, ndhA, ndhB, rpoC1, rpl2, rpl16, rps12, rps16, trnA-UGC, trnI-GAU, trnG-UCC, trnK-UUU, trnL-UAA, and trnV-UAC. One gene, ycf3, contained two introns. Additionally, seventeen genes were located in the IR region, including ndhB, rpl2, rpl23, rps7, rps12, rrn4.5, rrn5, rrn16, rrn23, trnA-UGC, trnI-CAU, trnI-GAU, trnL-CAA, trnN-GUU, trnR-ACG, trnV-GAC, and ycf2 (Table S4). The psaA gene was absent in the accession of S. japonica var. japonica (MZ981784).

Plastid genome variations within Spiraea

Our study revealed a significant level of conservation in the genome structure, gene order, and gene content of all Spiraea plastid genomes. Figure S1 demonstrates that the inverted repeat (IR) region exhibited relatively stable characteristics compared to the large single copy (LSC) and small single copy (SSC) regions, with strikingly higher GC content and lower nucleotide and gap variations (Fig. 3a). The conservation of all rRNA genes was observed, along with certain protein-coding genes, such as psbA, atpA, rpoB, psbC, psbD, rps4, rbcL, accD, petA, psbB, ycf2, ndhD, and rps7.

Fig. 3
figure 3

Basic features of Spireae plastid genomes. a The nucleotide diversity (Pi), GC content, and gap proportion in the plastid genome sequences of Spiraea detected by sliding window analysis, using 600 bp windows and a 200 bp step size. b Nonsynonymous (dN) and synonymous (dS) substitution rates and dN/dS for each protein-coding gene within the Spiraea plastid genomes

Full size image

In the comprehensive analysis of 34 Spiraea species’ plastid genomes, we observed that the length of the rps19 gene ranged from 279 to 282 bp, all of which were located in the LSC-IRB region (Fig. S2). Among them, a segment of 108 bp to 252 bp extended into the IRB region, resulting in the duplication of the rps19 gene in the IRA region. The ycf1 gene was situated in the SSC-IRA region, with a length ranging from 1,046 bp to 1,064 bp in the IRA region. However, the total length of the ycf1 gene in S. martini was only 1,131 bp, with only 82 bp in the SSC region, while S. veitchii exhibited the longest length of 4,577 bp. Additionally, a duplication of the ycf1 pseudogene was observed in the IRB region. The position of the ndhF gene varied, with 22 Spiraea accessions harboring it in the SSC region, while it appeared in the IRB-SSC region in the remaining accessions. The conserved trnH-GUG gene, spanning 74 bp to 75 bp, was distributed in the SSC-IRA region (Fig. S2).

Sliding window analysis demonstrated that nucleotide polymorphisms in Spiraea plastid genomes were more conserved in the IR region than in the SC region, and non-coding regions exhibited more pronounced variation than coding regions (Fig. 3a). The nucleotide polymorphism (Pi) ranged from 0 to 0.0138, with an average value of 0.0036. Relatively high Pi values (> 0.011) were observed in ndhF-rpl32, trnH-GUG-psbA, trnG-UCC-atpA, petN-psbM, trnT-UGU-trnL-UAA, rpl16, rpl32-trnL-UAG, and ycf1.

Selection pressure analyses

This study investigated the adaptive evolution of Spiraea by analyzing the ratio of dN/dS. A ratio greater than one indicates positive selection, a ratio less than one indicates purifying selection, and a ratio equal to one indicates neutral evolution. Analysis of 50 PCGs revealed that the average value of dN was 0.0074, with the highest value observed for rpl22 (0.0394) and the lowest being 0. The average value of dS was 0.0449, with the highest value observed for rps14 (0.1179) and the lowest values observed for ycf2 (0.0022). The average value of dN/dS was 0.2122, with the highest value observed for ycf2 (1.7749) and the lowest being 0.0001 (Fig. 3b).

Most genes exhibited a significantly larger dS compared to dN, indicating purifying selection. However, the dN/dS ratio of the ycf2 gene (1.7749) was greater than one, suggesting that this gene has undergone significant positive selection (P < 0.05; Tables S5-6). When categorizing the protein-coding genes based on their function, it was found that the other type genes had the highest dN value (0.0127), while photosystem genes had the lowest dN value (0.0016). In terms of dS, ribosomal protein genes had the highest value (0.0599), while RNA polymerase genes had the lowest value (0.0265). Except for the ycf2 gene whose dN/dS value was greater than 1, the rest of the functional genes exhibited values less than 1. Among these groups, other type genes had the highest dN/dS value (0.4261), while photosystem genes had the lowest (0.0401) (Fig. 4).

Fig. 4
figure 4

Boxplot of the values of the nonsynonymous (dN) and synonymous (dS) substitution rates and dN/dS in each functional gene group

Full size image

Phylogenetic analyses and divergence time estimates

For this study, a total of 34 Spiraea accessions and three outgroups were selected to construct a maximum likelihood (ML) tree and Bayesian inference using the protein-coding genes (PCGs) and whole plastid genome sequences. The ML tree, Bayesian inference, and BEAST MCC tree based on PCGs showed similar phylogenetic topographies, as shown in Fig. 5. Five main clades were identified with high support in the ML tree (Fig. 5a). However, the whole plastid genome sequences only recovered four of these main clades with full support values (Fig. S3). Clade II, identified by PCGs, could not be recovered, and its three branches within the clade were not strongly supported (Fig. S3). Overall, the traditionally defined subgenera and sections of Spiraea could not be supported by any datasets or phylogenetic methods (Fig. 5a-b; Fig. S3; Table S1).

Fig. 5
figure 5

Phylogeny, divergence time estimate, and LTT plot for Spiraea. a Cladogram of the maximum likelihood (ML) phylogenetic tree based on 79 protein-coding genes (PCGs). All main clades in both trees (ML and BI) are identical. Node labels are marked with the bootstrap values and Bayesian posterior probabilities (bootstrap value/posterior probability). The upper left shows the ML tree with branch length. b Maximum clade-credibility tree (MCC tree) from BEAST. Node labels represent estimated divergence times in millions of years. The confidence intervals are provided in Fig. S4. c Lineage-through-time plot for Spiraea

Full size image

To estimate the divergence date, the top-ten clock-like genes were filtered out using Sortadate. These genes were accD, atpA, matK, ndhA, ndhD, ndhF, ccsA, rpoB, rpoC2, and ycf1. Twelve reliable fossils from Rosaceae were used. The stem age of Spiraea was estimated to be 49.24 Ma (95% HPD: 48.64–50.09 Ma), with diversification occurring since the Early Oligocene (~ 30 Ma, Fig. S4). Most extant Spiraea lineages (29 out of 34) originated from the Pliocene period, and 25 taxa diverged from their most recent common ancestor (MRCA) since the Pleistocene (Fig. 5b). The LTT plot showed a clear shift in lineages since approximately 5 Ma, followed by a rapid accumulation after approximately 2 Ma (Fig. 5c).

Discussion

Plastid genome features and molecular variation patterns of Spiraea

The plastid is a vital organelle in plant cells, playing a crucial role in plant growth and development [50]. While significant changes in plastome organization have been observed in various distantly related plant groups, such as conifers, the IR-lacking clade (IRLC) of Fabaceae, and some parasitic flowering plants [51], the plastid genome tends to remain relatively stable across major plant lineages. It typically consists of a quadripartite structure, with two copies of a large inverted repeat (IR) separated by small (SSC) and large (LSC) single-copy regions [51]. The typical plastome organization is also found in Spiraea (Fig. 1; Table S1). In our study, we examined the plastid genome of Spiraea and identified several variation patterns within these genomes.

First, we observed a positive relationship between the size of SC region and plastid genome size. While expansion/contraction of the IR regions is one major driver of genome size variation [51], this was not evident in Spiraea (R2 = 0.03, P = 0.31; Fig. 2a). However, there was a significant correlation between the plastid genome size and the length of the SC region (R2 = 0.99, P < 0.01; Fig. 2c). The gene content across all Spiraea plastid genomes was highly conserved, except for Spiraea japonica var. japonica (accession no. MZ981784), which had the smallest plastid genome within Spiraea and lacked an important photosynthesis-related gene, psaA. Additionally, the intergenic spacer (IGS) in the SC region may contribute to the size variation of the SC region in Spiraea (Fig. 2d).

Second, we identified several hypervariable hotspots within the Spiraea plastome, all of which were localized in the SC regions (Fig. 3a). The pattern of lower IR substitution rates compared to the SC regions has been observed in Spiraea plastid genomes (Fig. 3a; Fig. S1), as well as in other plant lineages [51]. This decreased substitution rate in the IR is likely a result of a copy-dependent repair mechanism [52, 53].

Previous studies have used nrDNA ITS sequence data and two plastid DNA loci (trnL-trnF and matK) to investigate the phylogeny of Spiraea [6,7,8,9]. However, these methods have shown limited resolution in resolving the phylogeny of Spiraea [6,7,8,9], indicating low sequence divergence. Our results partially support this hypothesis, as trnL-trnF and matK exhibit relatively low sequence divergence (each with Pi values of 0.0043 and 0.0083) compared to highly variable hotspots (such as hotspots with Pi values > 0.011) (Fig. 3a). Therefore, hypervariable loci are needed to increase the phylogenetic resolution in low-level phylogenetic or phylogeographic studies in Spiraea.

We conducted an investigation into the hypervariable hotspots across the plastid genomes within the Spiraea genus. These findings provide valuable genetic resource information for future phylogenetic and phylogeographic studies. We identified eight top hypervariable loci with Pi values larger than 0.011, namely, ndhF-rpl32, trnH-GUG-psbA, trnG-UCC-atpA, petN-psbM, trnT-UGU-trnL-UAA, rpl16, rpl32-trnL-UAG, and ycf1.

Many of these loci, such as ndhF-rpl32, trnH-GUG-psbA, trnT-UGU-trnL-UAA rpl16, rpl32ni-trnL-UAG, and ycf1, have been previously identified in studies conducted by Taberlet et al. [54], Shaw et al. [55,56,57], and Dong et al. [58]. However, it is important to consider the characteristics of these variable regions themselves. Factors such as fragment size and mutation rate (substitution saturation) can have a detrimental effect on phylogenetic analysis. For example, the trnH-psbA locus is controversial due to its high mutation rate and significant length variation caused by insertions, deletions, simple sequence repeats, and localized inversions in some plant lineages [59,60,61]. These factors can impact the accuracy of sequence alignment and subsequently affect the precision of phylogenetic inference and species identification. Within Spiraea, the trnH-psbA alignment matrix has 89 indels, indicating its unstable size. Additionally, the ycf1 locus also exhibits a high degree of sequence variability and has been proposed as a candidate barcode for land plants. However, its high mutation rate and absence in certain plant lineages, such as Poaceae [58, 62], can lead to alignment ambiguities similar to those observed in trnH-psbA [58, 63].

Third, we investigated the pattern of molecular evolution among plastid genes in Spiraea. The evolution of substitution rates among plastid genes has been extensively studied with the recent release of numerous plastid genomes. Generally, non-synonymous rates of photosynthesis-related genes differ significantly from those of housekeeping genes, indicating that photosynthesis-related genes are under stronger functional constraints [64,65,66]. Our results demonstrated that the photosynthesis-related genes in terms of dN are highly conserved, providing further support for this hypothesis (Fig. 4).

It is generally assumed that most plant species have low substitution rates of most genes in organelles, and these genes evolve under strong purifying selection. However, certain genes in the plastid have undergone positive selection, particularly those involved in photosynthesis and other metabolic pathways (e.g., [67]). The relationship between the high levels of non-synonymous rate evolution and potential selective factors, such as novel ecological conditions, is still a topic of debate [68]. In our study, we observed dN/dS ratios of 1.7749 for the ycf2 gene, which is likely experiencing strong positive selection (P < 0.05; Table S6). Despite its unclear function, ycf2 is known as the largest chloroplast gene in angiosperms and encodes a motor protein that generates ATP for inner membrane translocation [69] and plant cell survival [70]. Positive selection of the ycf2 gene has been observed in numerous species (e.g., [71,72,73,74]), indicating its adaptive importance in various environments and plant lineages.

Phylogenetic and evolutionary implications of plastid genomes within Spiraea

The inheritance of plastid genomes in plants is typically uniparental, with the mother being the primary contributor. This results in a smaller effective population size compared to nuclear genomes. Consequently, the coalescent time of plastid haplotypes is shorter, making them ideal for studying genetic variations and tracing plastid evolutionary history [75]. Plastid markers, such as rbcL, matK, and trnL-trnF, have been widely used in phylogenetic studies at different taxonomic levels, including ordinal, familial, tribal, and generic levels [12, 76]. They have also been used in lower-level studies, such as phylogeographic investigations [77, 78] and species identification [79, 80].

Previous studies on the phylogeny of Spiraea have utilized nrDNA ITS sequence data and limited plastid DNA loci (trnL-trnF and matK) [6,7,8,9]. However, the resolution of the phylogenetic relationships within Spiraea based on ITS and trnL-trnF is limited, and none of the sections defined by Rehder [3] and Yü and Lu [81] based on inflorescence morphology are monophyletic [8]. Yu et al. [9] largely supported this hypothesis, except for sect. Spiraea. The inclusion of Spiraea douglasii in sect. Spiraea is questionable and represents a clear difference between the two aforementioned studies [8, 9]. The Spiraea species sampled in our study represent all four sections found in China (Table S1; [1, 81]). Our phylogenomic analysis provides the most detailed phylogenetic backbone of Spiraea to date and strongly supports the notion that the traditionally defined sections of Spiraea are not monophyletic groups [8, 9], despite only including one species (S. salicifolia) from sect. Spiraea in this study. We also found a complex evolutionary history of Spiraea, as there are many shallow relationships within Clades I and II that could not be highly supported (Fig. 5; Fig. S3). Additionally, some taxa with multiple accessions (such as S. japonica, S. cantoniensis, and S. × vanhouttei) are recovered as paraphyletic or polyphyletic groups (Fig. 5; Fig. S3).

There may be two primary reasons why the traditional taxonomic classification and some species boundaries within Spiraea cannot be supported by plastid data. First, the phylogenetic relationships inferred from plastid data could not fully capture the evolutionary relationships among taxa, potentially due to the occurrence of pervasive hybridization/introgression events in Spiraea. Recent evidence suggests that ancient hybridization/introgression events may have resulted in cytonuclear conflicts [82]. Spiraea is a genus known to frequently undergo hybridization, with numerous hybrids arising through horticultural practices [83]. Although wild hybrids are less commonly reported, there have been instances of hybrids occurring across different sections of the genus [83, 84]. These findings indicate that reproductive isolation is not complete among Spiraea species, making hybridization a potentially common phenomenon. Consequently, a substantial amount of nuclear genetic data is necessary to further investigate the traditional classification framework and hybridization/introgression events.

Second, more detailed revisions of the infrageneric classification and species boundary delimitation of Spiraea are needed. It is evident that the traditional classification system within Spiraea is inconsistent with the results obtained from plastid data ([23]; this study), nuclear ribosomal DNA internal transcribed spacer (nrDNA ITS) sequences [6, 7], and their combined analyses [8, 9]. This inconsistency strongly suggests a problem with the traditional infrageneric classification of Spiraea. Traditionally, inflorescence morphology has been the main focus in these classification schemes [3, 81], but it is likely that this trait does not accurately reflect the true evolutionary relationships within Spiraea. Therefore, an integrative framework for infrageneric classification in the genus is needed, which should incorporate additional diagnostic traits such as seed and palynology evidence [9].

Ancient Spiraea taxa are believed to have originated in the Early Eocene based on macrofossil evidence [42, 43]. Our study is the first to identify the divergence sequence of the Asian Spiraea taxa using twelve reliable fossil calibrations of Rosaceae. We acknowledge that the crown age inferred in this study may have been underestimated due to our limited sampling. Nevertheless, our divergence time for the genus is much older than the result of Spiraea taxa in the Qinghai-Tibetan Plateau (QTP) [6], which can be attributed to the difference in molecular markers, divergence estimate methods, and the species included in the two studies ([6]; this study). We conclude that Spiraea diversified after the Early Oligocene (~ 30 Ma), followed by a rapid speciation process during the Pliocene and Pleistocene periods (Fig. 5b-c). The Pliocene–Pleistocene period played a vital role in the speciation process of Spiraea, as most extant taxa used in this study originated during this period (Fig. 5b-c), which is supported by fossil records from the Pliocene–Pleistocene period in Asia (data from PBDB: https://paleobiodb.org/#/). The pattern of rapid radiation during this period (especially the Pleistocene) has also been observed in many plant groups in East Asia [85, 86], likely due to the effects of glaciation and climatic oscillations during the Pleistocene. East Asia, being less influenced by Quaternary glaciations, had a lower extinction rate or higher speciation rate during those times, leading to a relatively high net diversification rate [87,88,89,90]. Biotic factors, such as hybridization/introgression, possibly related to climate oscillations, have also contributed to the emergence of young species during the Pleistocene period. In theory, climate oscillations during the period may have favored introgressive hybridization/introgression upon secondary contact following initial species divergence [91, 92]. Ancestral polymorphism of plastomes could have been purified through the extent of chloroplast-capture events when hybridization/introgression frequently occurred [93], obscuring species boundaries and creating the illusion of young species pairs. Further studies, including nuclear sequences and more samplings, are needed to investigate this hypothesis.

Conclusions

In this study, we sequenced, assembled, and analyzed the plastomes of 34 Asiatic Spiraea accessions. The Spiraea plastid genome exhibits typical quadripartite structures and encodes 113–114 genes, including 78–79 protein-coding genes (PCGs), 30 tRNA genes, and 4 rRNA genes. There is a significant correlation between genome size and the length of the SC region. Several hypervariable regions within the Spiraea plastome were identified, all of which were localized in the SC regions. Our phylogenomic analysis successfully provides a phylogenetic framework for Spiraea but does not support its currently defined section boundaries. Additionally, we discovered that the genus underwent diversification after the Early Oligocene (~ 30 Ma), followed by a rapid speciation process during the Pliocene and Pleistocene periods. Although not perfect, the plastomes of Spiraea provide invaluable insights into its phylogenetic relationships and evolutionary history. Further investigations utilizing other genomes, such as the nuclear genome, are urgently needed to enhance our understanding of the evolutionary history of the genus.