By utilizing long PacBio HiFi reads, the mitogenomes of diploid (2x) and tetraploid (4x) C. indicum were assembled with published C. indicum mitogenome (NCBI accession number: MH716014.1) used to extract sequences from whole genome data. In addition the mitogenome of C. indicum var. aromaticum was also assembled with Illumina reads. The mitogenome of C. indicum (2x) was assembled into a single circular genome, and was used to resolve the mitogenome of C. indicum (4x) and C. indicum var. aromaticum with multiple linkage structures ( Figure S1 ). The sizes of the three circular mitogenome were 192,408 bp with 45.51% GC content for C. indicum (2x), 193,563 bp with 45.43% GC content for C. indicum (4x), and 198,095 bp with 45.33% GC content for C. indicum var. aromaticum, similar with that in found in most Asteraceae species sequenced at present ( Figure 1 ). All three of these newly assembled mitogenomes contain 51 unique genes, including 33 protein coding genes (PCGs) genes, three rRNA genes, and 16 tRNA, with trnS-GCU having two copies and trnM-CAU having three copies in C. indicum (2x) and C. indicum (4x), but five copies in C. indicum var. aromaticum ( Table 1 ). Among the PCGs, eight genes contained introns, three of which (ccmFC, nad4 and rps3) contained one intron, and five others (cox2, nad1, nad2, nad5, and nad7) contained two or more introns. Among the three mitogenomes, all PCGs are single copy, except atp9 that has two copies in C. indicum var. aromaticum.

All three newly assembled chloroplast genomes maintain the classical dumbbell-shaped genome structure. The plastome sizes of the three samples were very similar at 151,033 bp for C. indicum (2x), 151,036 bp for C. indicum (4x) and 151,053 bp for C. indicum var. aromaticum with the same GC content at 37.47% for all these plastomes ( Figure S2 ). All three of these newly assembled plastomes of Chrysanthemum contain 109 unique genes, including 80 PCGs, 25 tRNAs, and four rRNAs with the IR region containing six protein coding genes, eight tRNAs and four rRNAs, with 13 PCGs containing introns, and two of these (rps12 and ycf3) have more than one intron. Overall, compared to mitogenomes, these three plastomes differ only slightly in a few regions and are generally conserved in sequence similarity.

In plants, the exchange of genetic material between cellular compartments (nucleus, mitochondria, and chloroplasts) is often referred to as Intracellular Gene Transfer (IGT) (Bergthorsson et al., 2003; Aubin et al., 2021). Characterizing these IGT events is key to understanding the genetics and evolution of organellar genomes. We did this in Chrysanthemum, by quantifying the sequence transfers from plastomes to mitogenomes ( Figure 4 ). Each accession had a total of 8-10 (from 5220-9176 bp in total length) fragments from a plastomic origin integrated into 8-12 (5212-9164 bp) locations in the mitogenome. The least number (8) of transferred sequences was in C. indicum (2x) and the most (10) in C. makinoi. The length of the transferred fragments ranged from the shortest at 79 bp to the longest of 2556 bp ( Table 2 ; Table S1 ). Correlation analyses indicated that the number of plastome transferred sequences and the size of the corresponding mitogenomes were positively correlated, as the number and length of plastome transfer sequences was greater for accessions with large mitogenomes ( Figure S5 ). The transferred sequences across the seven accessions were conserved in respect to both size and sequence of origin. Specifically all accessions contained seven completely transferred gene sequences (petL, petG, trnM-CAU, trnH-GUG, trnN-GUU, trnW-CCA, and trnP-UGG) and five partially transferred gene sequences (16S rRNA, rps11, ycf2, psaB, and psaA) ( Table 2 ). Interestingly, C. indicum var. aromaticum, C. boreale, and C. makinoi have partially transferred sequences of the 23S rRNA gene, but the lengths of these fragments vary from 135 bp in C. indicum var. aromaticum to 357 bp and 443 bp in C. makinoi, and 238 bp in C. boreale ( Table S1 ). These results suggest that the plastomic fragments were transferred to the mitogenome predating speciation, and that shared transfers specifically between C. indicum var. aromaticum, and other Chrysanthemum species may be the result of past hybridization, incomplete lineage sorting, or retention of ancestral transfers lost in other C. indicum species.

A maximum likelihood (ML) phylogenetic tree of 26 Asteraceae species (Platycodon grandifiorus was set as outgroup) with complete mitogenome and plastome data was constructed based on 80 plastome genes and 32 mitogenome genes, respectively ( Figure 5 ). In our study, the phylogenetic topology at tribe level was generally consistent with previous research (Zhang et al., 2021). However, the internal phylogenetic relationships within the Anthemideae were somewhat different. On the mitogenome tree, C. boreale grouped with Artemisia giraldii, while on plastome tree, it grouped with Chrysanthemum. This may be due to the fact that the four mitogenome genes (atp9, mttb, nad2 and nad6) in C. boreale show greater similarity with that in A. giraldii, especially the atp9 gene which was identical between the two species. In a previous study, it was suggested that C. boreale may be an early diverging species within Chrysanthemum, and thus may have retained more ancestral characters (El-Twab and Kondo, 2007). From the phylogenetic tree all four genomes of C. indicum clustered together in both plastome and mitogenome analyses. The phylogenetic analysis also suggested that the maternal donor of the tetraploid C. indicum may be the diploid C. indicum, while C. indicum var. aromaticum appears to have undergone some degree of divergence from the other C. indicum.

With the set model parameters, we obtained two time tree result based on plastome estimation and mitogenome estimation ( Figures S6, S7 ). From the results, the 95% intervals obtained for plastomes are more convergent, while those for mitogenomes are broader, which may be related to the mutation rate and the number of mutation sites in mitochondria ( Figure S6 ). Here we take the plastome results as the basis. From the results, we learnt that the C. indicum was born at 1.43 Mya, and C. indicum var. aromaticum and C. indicum were separated at 0.81 Mya; while tetraploid C. indicum and diploid C. indicum were separated at 0.38 Mya ( Figure S7 ).

Plant mitogenomes have undergone tremendous structural changes during evolution, which have resulted in a near 200-fold variation in mitogenome size among plants (from ~66 Kb of Viscum scurruloideum to ~12 Mb of Larix sibirica) and frequent rearrangements with low gene density, in contrast to plant plastomes which are conserved in genome size from 115-165 Kb and containing a dumbbell-shaped genome structure (Putintseva et al., 2020; Wu et al., 2020). Owing to the release of a large amount of plant whole genome sequencing data, an increasing number of mitogenomes have been assembled and subjected to comparative studies, in which many mitogenomes have been found to exhibit a reticulated complex genomic structure. The diversity of many mitochondrial genomic structures is thought to be the result of recombination mediated by repetitive sequences (Wu and Sloan, 2019). However, the three mitogenomes of C. indicum newly assembled in this study show a simple monocyclic structure, along with a low number and percentage (0.9%-2.2%) of repetitive sequences, which may account for the structure and relatively small size at just over 40 Kb longer than their respective plastomes. In a comparison among four accessions of C. indicum, it was found that the mitogenome structure and sequence within species is relatively conserved, in contrast to large differences observed among mitogenomes within cultivated sorghum (Zhang et al., 2023). Meanwhile, only a 2.8 Kb segment of the mitogenome differed between accessions of C. indicum of different ploidy levels, two ORFs encoding proteins 496aa and 72aa in length were predicted on this segment, and the functions of these two ORFs have yet to be investigated by mitochondrial gene editing techniques (Kazama et al., 2019). However, when comparing among four different species in Chrysanthemum, it was found that the mitogenomes differed not only in size, but also in structure, mainly in genome sequence order. In conclusion, the relationship between mitogenome sequence and structural variation within Chrysanthemum and what effect this has on molecular function needs to be further investigated from a greater number of samples. What factors are driving structural conservation within C. indicum and divergence between species should also be investigated in greater depth.

IGT occurs frequently and continuously during nucleoplasmic coevolution between different genomes in plant cells (Zhang et al., 2020). In this study, we assembled the mitogenomes and plastomes of three C. indicum accessions and examined the sequence transfers from plastome to mitogenome in combination with four published organellar genomes of species in Chrysanthemum. It was found that the number and the size of these transferred sequences within Chrysanthemum showed a positive correlation with the size of the mitogenome. The transferred genes were conserved among the accessions, and the size of the transferred fragments containing these genes was also generally conserved. All samples contained seven completely transferred gene sequences from the plastome (petL, petG, trnM-CAU, trnH-GUG, trnN-GUU, trnW-CCA, and trnP-UGG) and five partially transferred genes (16S rRNA, rps11, ycf2, psaB, and psaA). These results suggest that these sequence transfers occurred prior to the divergence of Chrysanthemum species and have been retained through time albeit in different locations of the genome given rapid turnover in structure from recombination. Interestingly, all four accessions of C. indicum have the same transferred gene, except for C. indicum var. aromaticum that has an additional 135 bp transferred sequence containing a partially transferred sequence of the 23S rRNA gene. This transferred fragment shared between C. indicum var. aromaticum and the other species in Chrysanthemum may suggest that this lineage is early diverging or that introgression or incomplete lineage sorting may have been involved in the evolution of this variety.

We observed different topologies between phylogenetic analysis based on plastome and mitogenome sequences. Overall, the results obtained from chloroplasts are consistent with previous studies based on nuclear genes, while mitochondria tend to exhibit conflicting resolutions compared with nuclear gene results in adjacent clades (Zhang et al., 2021). This may be due to the slow mutation rate in sequences that have been observed in plant mitogenomes (Duff and Nickrent, 1999; Xi et al., 2013; Grewe et al., 2014). Additionally, the high frequency of RNA editing sites in mitochondrial genes imposes difficulties for mitochondrial gene tree construction (Drouin et al., 2008; Knoop, 2011). For example, in angiosperms, the number of editing sites in the mitogenome often ranges from 200 to 500, while in the plastome, there are only 30 to 50 editing sites (Malek et al., 1996; Knoop, 2011; Zhang et al., 2021). Furthermore, the editing sites in the mitochondrial genome are not only evident in the coding region, but also in tRNA, introns, and 5’ and 3’ untranslated regions (Malek et al., 1996; Oldenkott et al., 2014). Our study supports the view that mitochondrial genes may not be appropriate for plant phylogenetic inferences at some levels of taxonomic organization (Malek et al., 1996; Drouin et al., 2008; Zhang et al., 2021).

The samples collected for this study were all from the Chrysanthemum indicum germplasm base of China Resources Sanjiu Medical & Pharmaceutical Co., Ltd in Yangxin County, Huangshi City, Hubei Province. The total genomic DNA was extracted using a Cetyltrimethylammonium Bromide (CTAB) method (Arseneau et al., 2017). The same DNA sample was used for Illumina sequencing and HiFi sequencing by using a Hiseq Xten PE150 sequencing platform (Emerman et al., 2017) and PacBio Sequel II sequencing platform respectively (Eid et al., 2009).

Mitogenomes from three Chrysanthemum accessions were assembled with short read (Illumina) or long-read sequencing (PacBio HiFi) data. Specifically, for the diploid- and tetraploid-Chrysanthemum, first, a total of three Gb of reads were randomly extracted from the PacBio HiFi data by using SeqKit v2.1.0 (Shen et al., 2016). Then the published mitogenome of C. indicum (NCBI accession number: MH716014.1) was used as a reference to extract mitochondrial reads by using minimap2 with the parameter settings’-cx map-hifi -H’ (Li, 2018). Flye-meta v2.8.3-b1695 (Kolmogorov et al., 2020) was used to assembly the mitogenomes with PacBio HiFi reads, with the master circle conformation mitogenome of C. indicum (2x) and the complex mitogenome conformation of C. indicum (4x) with multiple connections obtained. Next the reference and C. indicum (2x) mitogenomes were mapped to C. indicum (4x) to simplify it to a single circle in Bandage (Wick et al., 2015). For C. indicum var. aromaticum with short read data, a total of five Gb reads were randomly extracted. GSAT was used to assembly the mitogenome with the pipeline ‘graphShort’, then simplified the genome in the same method as above (He et al., 2022). The three single circles were annotated by GeSeq (Tillich et al., 2017)with the reference mitogenome of C. indicum (MH716014), and the trans-spliced genes checked manually.

A similar pipeline as mitogenome assembly was used to assemble the plastome of diploid- and tetraploid-Chrysanthemum accessions. Given the high copy number of plastome sequences within a cell, one Gb of HiFi reads were randomly extracted for assembly. For C. indicum var. aromaticum the plastome was extracted from the de novo assembly results of GSAT from above. The plastomes were annotated by GeSeq (Tillich et al., 2017; Qu et al., 2019) and PGA software, using the reference plastome of C. indicum (NC_020320).

The program Blastn v2.11.0+ with parameter settings ‘-evalue 1e-6’ was used to identify syntenic sequences between the seven accessions from Chrysanthemum (Camacho et al., 2009). Syntenic sequences longer than 1 Kb were visualized with NGenomeSyn v1.41 (He et al., 2023).

The repeat types, F (forward), P (palindrome), R (reverse), and C (complement) of dispersed repeat sequences in the seven Chrysanthemum accessions organelles were detected using REPuter with parameter settings ‘-c -f -p -r -l 20 -h 3 -best 300’ (Kurtz et al., 2001). The MISA software was used to identify simple sequence repeats (SSRs) with 10, 6, 5, 5, 5, and 5 repeat units set as minimum thresholds for mono-, di-, tri-, tetra-, penta-, and hexa-motifs respectively (Beier et al., 2017).The results were visualized with ggplot2 (Villanueva and Chen, 2019).

To determine the plastid-derived sequences within Chrysanthemum, the seven mitogenomes were searched against their corresponding plastomes by Blastn v2.11.0+ with parameter settings ‘-evalue 1e-6’. Then the plastid-derived sequences were annotated. The location of plastid-derived sequences in both the mitogenome (destination) and plastome (origin) were visualized with Circos v0.69-9 (Krzywinski et al., 2009). The plastid-derived sequences in mitogenomes were annotated by GeSeq (Tillich et al., 2017)and PGA software, using the reference plastome of C. indicum (NC_020320).

To analyze the phylogenetic relationships among Asteraceae, the plastomes and mitogenomes of C. indicum (2x), C. indicum (4x), C. indicum var. aromaticum, and 23 other Asteraceae were used, with Platycodon. grandiflorus (Campanulaceae) as an outgroup. The CDSs of 80 chloroplast protein-coding genes and 32 mitochondrial protein-coding genes were extracted, concatenated, and aligned using MAFFT v7.490 (Katoh et al., 2002) with poorly aligned sections trimmed with TrimAL v1.4 (Capella-Gutiérrez et al., 2009). These datasets were then used to conduct two separate phylogenetic analysis using IQ-TREE v2.0 with 1000 ultrafast bootstrap replicates to assess branch support based on the auto-selected best-fit model ‘TVM + F + R2’, with FigTree v1.4.3 used for tree visualization (Minh et al., 2020).

All gene pairs alignments were converted into PAML format using EasyCodeML (Yang, 2007; Gao et al., 2019). All the sequences pairs was merge in one file. Molecular divergence times were estimated by placing soft boundaries on the split nodes using records from the timetree database (http://timetree.org/). For divergence time markers we used two nodes, one for the divergence of the outgroup from the Asteraceae, set at 67.7-93.0 Mya, and the other divergence time for the node represented by Arctium lappa from the node represented by Chrysanthemum indicum, set at 36.6-45.1 Mya. The MCMCtree module of PAML software was implemented to estimate the divergence time of each node (Dos Reis, 2022). Tracer was used to evaluate whether the results of MCMCtree have converged. The result of divergence time tree results are visualized using ggtree package (Xu et al., 2022).