Fresh leaves of six taxa of the genus Catalpa, namely, C. fargesii f. duclouxii (Guiding County, Guizhou Province, China), C. fargesii (Tianshui City, Gansu Province, China), C. bungei (Luoyang, Henan Province, China), C. ovata (Tianshui, Gansu Province, China), C. bungei (Jinsiqiu) (Luoyang, Henan Province, China) and C. fargesii f. duclouxii (Huangxinzimu) (Fuquan, Guizhou Province, China), were collected. Six complete chloroplast genome sequences were deposited in GenBank with accession numbers OL628864 to OL628869 (Table 1). The samples were stored in silica gel and transported to the laboratory for low-temperature preservation (−40°C). Specimens of six taxa of the genus Catalpa preserved at the Institute of Forestry, Chinese Academy of Forestry, Beijing, were also examined (Table 1). Total DNA was extracted following the method of Li et al. (2013) and purified by a Wizard DNA cleanup system (Promega, Madison, WI, United States). DNA quality was assessed by spectrophotometry, and integrity was evaluated using a 1% (w/v) agarose gel (Promega, Madison, WI, United States).

Total DNA was fragmented into 350 bp fragments by ultrasound. A paired-end library was constructed by a NEBNext Ultra™ DNA library prep kit, and PE150 sequencing was performed on the Illumina HiSeq XTen platform. The NGS QC toolkit was used for quality control and to filter the low-quality reads. We used the obtained data for de novo assembly of the whole chloroplast genome with the GetOrganelle v1.7.5 pipeline using the following settings: F embplant_pt, -R 15, -K85 and 105. Using the published chloroplast genome of Tecomaria capensis (GenBank sequence acceptance number MG831880) of Bignoniaceae as a reference sequence, the Plann program (Huang and Cronk 2015) was used to annotate the chloroplast genes of Catalpa species. Some genes with unsuccessful or incorrect annotations were manually added in Sequin software. The structure map of the genome was first drawn using OrganellarGenomeDRAW-a (http://ogdraw.mpimp-golm.mpg.de/index.shtml) (Lohse et al., 2013) and then edited using Adobe Illustrator CS5. All chloroplast genome sequences were uploaded to the NCBI GenBank database for future reference.

GMATA (Wang and Wang 2016) software was used to analyze SSRs in the chloroplast genomes of the six taxa of the genus Catalpa with the parameters set as 1-10, 2-4, 3-4, 4-3, 5-3 and 6-3, that is, mononucleotide SSRs with a repeat unit of 1 and a repetition number ≥10, dinucleotide SSRs with a repeat unit of 2 and a repetition number ≥6, trinucleotide SSRs with a repeat unit of 3 and a repetition number ≥5, and tetranucleotide, pentanucleotide, and hexanucleotide SSRs with a repeat unit of 4, 5, and 6, respectively, and a repetition number ≥3 (Thiel et al., 2003). Two SSR markers separated by less than 100 bp were considered a composite microsatellite. The REPuter program (Kurtz et al., 2001) was used to find forward (F), palindromic (P), reverse (R) and complementary (C) oligonucleotide repeats with a minimum repeat size of 30 bp and a similarity of 90%. The REPuter program overestimated repeats, and redundant repeats were found in large repeats as well as in duplicated tRNAs.

The six assembled chloroplast genomes were compared with MAFFT (multiple alignments using fast Fourier transform) v7 software (Katoh and Standley 2013), and then the results were manually adjusted with MEGA7 software (Kumar et al., 2016; Abdullah et al., 2021b). MEGA7 was used to quantify the mutation sites and parsimony-informative sites in the chloroplast genomes of Catalpa. Taking the C. bungei sequence as the reference, the Shuffle-LAGAN model in the mVISTA program (http://genome.lbl.gov/vista/mvista/submit.shtml) was used to analyze the whole genome of Catalpa. First, we manually checked for small inversions and removed them from the alignment to avoid false results. The intergenic spacer regions and protein-coding regions were extracted from the alignment in Geneious R8.1 (Kearse et al., 2012) and visualized in DnaSP v.6 to determine the nucleotide diversity of each region (Rozas et al., 2017; Abdullah et al., 2021a).

All coding sequences (CDSs) were manually extracted from the chloroplast genomes. MEGA5 was used to analyze the codon usage frequency in each of the six Catalpa species (Kumar et al., 2008). Relative synonymous codon usage (RSCU) reflects whether a plastid gene is in a selected state, and codons with an RSCU value >1 are defined as high-frequency codons.

Thirty-three chloroplast genome sequences, including six from Catalpa and 17 from other species of Bignoniaceae, Lentibulariaceae and Lamiaceae from GenBank, were used for phylogenetic analysis. All chloroplast genome sequences were aligned using MAFFT v7, and regions with ambiguous alignment were trimmed by Gblocks 0.91b (Castresana 2000).

Phylogenetic analysis was carried out using the maximum likelihood (ML) and Bayesian inference (BI) methods. The optimal model was identified as TVM + F + I + G4 by ModelFinder based on the Bayesian information criterion (BIC) standard (recommended by the software) (Dong Zhang et al., 2020). ML calculations were performed using IQ-tree, with sampling repeated 1,000 times. BI of the phylogenies was implemented in MrBayes (Nguyen et al., 2015). Markov chain Monte Carlo (MCMC) analysis was run for 10,000,000 generations. Trees were sampled every 1,000 generations, and the initial 25% were discarded as burn-in. Finally, the average standard deviation of the split frequencies <0.01 was verified.

For the six taxa within the genus Catalpa, 2,435,211,900-3,831,597,000 bases of raw data with coverage ranging from 3722X-6742X were obtained (Table 1). The chloroplast genome of Catalpa has a typical structure, with a highly conserved, circular, double-stranded gene sequence mainly consisting of two IR regions separating two single-copy regions, namely, the LSC region and SSC region (Figure 1). The chloroplast genome length of the six taxa of the genus Catalpa ranged from 157,765 bp (C. fargesii) to 158,355 bp (C. ovata), the length of the LSC region ranged from 84,599 bp (C. fargesii) to 85,004 bp (C. ovata), and the length of the SSC region ranged from 12,662 bp (C. fargesii) to 12,675 bp (C. ovata) (Table 1). The length of the IR regions ranged from 30,252 bp (C. fargesii) to 30,338 bp (C. ovata). Therefore, the length variation of the LSC region was greater than that of the SSC and IR regions, and gene length variation was mainly caused by variation in the LSC region. The GC content of the genome is an important index for assessing the genetic relationships between species. The GC content of the chloroplast genomes of the six taxa within the genus Catalpa was 38.1%. There were almost no differences in the chloroplast genomes of the six taxa within the genus Catalpa.

In the chloroplast genomes of the six taxa within the genus Catalpa, 113 genes were detected, 19 of which were located in the IR regions. The 113 genes included 79 protein-coding genes, 30 tRNA genes and four rRNA genes (rrn5, rrn4.5, rrn16, and rrn23), and all four rRNA genes were distributed in the IR regions, resulting in a much higher GC content in the IR regions than in the two single-copy regions (Figure 1; Table 2). The infA gene was a pseudogene in all species. According to their function, the detected genes can be divided into three categories. The first category included 47 genes related to photosynthesis, including Rubisco large subunit genes, genes for components in the photosynthetic electron transport chain and genes presumed to be NAD(P)H dehydrogenase subunits. The second category consisted of six genes involved in the biosynthesis of amino acids, fatty acids and other substances, as well as some genes with unknown functions. In the third group, most of the genes were tRNA genes, including RNA polymerase subunits, rRNA and ribosomal proteins and other products. These 60 genes were related to transcription and translation. Studies have shown that introns play an important role in gene expression regulation, and many introns can enhance the level of foreign gene expression at specific times and locations in plants, in turn controlling agronomic traits (Jiao et al., 2012). Fifteen of the 113 genes contained introns, 13 contained one intron, and ycf3 and clpP contained two introns. rps12 is a spliced gene with a 5′-terminal exon located in the LSC region and a 3′-terminal exon located in the IR regions.

SSRs, also known as microsatellites, are composed of repeating units with a length of 1-6 bp. In this study, a total of 248 SSRs were detected in the 6 chloroplast genomes of Catalpa. In terms of distribution, 197 SSRs were located in the LSC region (79.44%), 21 SSRs were located in the SSC region (8.47%), and 30 SSRs were located in the IR regions (12.1%). Therefore, the distribution of SSRs in the chloroplast genome of Catalpa is uneven (Figure 2A). The largest number of SSRs observed among the taxa within the genus Catalpa was 51, and the smallest was 43. The remaining four species (C. fargesii f. duclouxii, C. fargesii f. duclouxii (Huangxinzimu), C. bungei (Jinsiqiu), and C. bungei) had 46 SSR loci (Figure 2C). The chloroplast genomes of the six taxa within the genus Catalpa included mono-, di-, tetra-, and pentanucleotide SSRs (Figure 2D). Trinucleotide SSRs were observed in only one Catalpa species, and none of the six species contained hexanucleotide SSRs. Among the 248 SSR sites in the chloroplast genomes of Catalpa (Figure 2B), 219 sites (78.78%) were composed of A/T, and only seven sites (2.52%) contained G/C, indicating an SSR base composition preference for A/T. These findings are consistent with previous reports that SSRs are typically composed of polyadenine (PolyA) and polythymine (PolyT) repeats (Cheng et al., 2015; Shen et al., 2016). Tetranucleotides accounted for the largest percentage of SSRs (21.94%), followed by dinucleotides and pentanucleotides (both 5.04%). There were differences in the number and distribution of SSRs among the six species within the Genus Catalpa, which may be due to the deletion and mutation of gene sequences during the evolution of Catalpa. We also analyzed oligonucleotide repeats by REPuter and found four categories: palindromic (P), forward (F), reverse (R), and complementary (C). The abundance of the repeats varied among species based on the type of repeat. In the chloroplast genomes of C. ovata, C. bungei, and C. fargesii, REPuter revealed 49 repeats (F = 23, R = 26, P = 0, and C = 0), whereas in those of C. bungei (Jinsiqju), C. fargesii f. duclouxi (Huangxinzimu), and C. fargesii f. duclouxi, 49 repeats (F = 24, R = 25, P = 0, and C = 0) were detected (Figure 2E). Most of the repeats were between 35 bp to 39 bp and 40 bp to 44 bp long (Figure 2F).

The expansion and contraction of IR regions in the chloroplast genome are important evolutionary events in plants and relatively common phenomena, ultimately causing changes in the size and gene content of the chloroplast genome (Huang et al., 2014). To explore the potential expansion and contraction of IRs, the distributions of IR and SC border regions in the chloroplast genomes of six taxa within the genus Catalpa were compared. Genes with a boundary distribution of JLB, JSB, JSA and JLA included rps19, rpl2, rps15, ndHF, ndhH and trnH (Figure 3). The JLB, JSB, JSA and JLA boundaries showed very similar gene distributions. In C. bungei, C. fargesii, C. fargesii f. duclouxii, C. fargesii f. duclouxii (Huangxinzimu) and C. bungei (Jinsiqiu), rps19 was located in the LSC region, 3 bp from JLB, while rps19 was located 31 bp from JLB. The distributions of genes at the JLB, JSB, JSA, and JLA borders were very similar among the six taxa of the genus Catalpa. The length of the rps15 gene was 228 bp in the IR region of C. bungei, C. fargesii, C. fargesii f. duclouxii, C. fargesii Bur f. duclouxii (Huangxinzimu) and C. bungei (Jinsiqiu) but 231 bp in the IR region of C. ovata. Only the trnH gene of C. ovata was 8 bp away from JLA, and the trnH gene was located in the LSC region and 13 bp away from JLA.

Using the chloroplast genome of C. bungei as a reference, the mVISTA tool was applied to perform multiple sequence alignment, and the sequence similarity results were visualized to determine the degree of differentiation. The chloroplast genome sequences of the six taxa within the genus Catalpa were highly similar and conserved. The variation in the LSC and SSC regions was significantly greater than that in the IR regions, the rRNA gene was highly conserved with almost no variation, and the sequence variation in coding regions was lower than that in noncoding regions. The gene regions with large variations were accD, psaI-ycf4 and ycf1-trnN (Figure 4). The conservation degree of other genes was very high, with most of the genes being more than 90% conserved.

After the chloroplast genomes were compared by software, three base mutations and nondimorphic mutations were excluded from subsequent analysis. The sequence length was 159,629 bp, with a total of 301 polymorphic sites (polymorphic, S), 36 parsimony-informative sites, and six haplotypes. The nucleotide diversity (Pi) of the sequences was 0.00059 (Table 3). The IR segment had the fewest mutation sites, with 18 polymorphic sites and one parsimony-informative site. The sequence had four haplotypes, and the sequence polymorphism of this region was only 0.00018. The patterns of SNPs, 60 transitions (Ts) and 149 transversions (Tv) were determined, and the overall Ts:Tv ratio was 0.403, indicating a preference for transversions (Figure 5). The high-frequency SNPs were C to T and G to A, and mutations from A to T and from T to A exhibited the lowest frequency.

We recorded higher average polymorphism for intergenic spacer regions (0.0021) than for protein-coding sequences (0.0011). The polymorphisms of all regions are shown in Figure 6. We ignored loci <200 bp and selected 5 polymorphic regions with nucleotide diversity >0.003, of which three belonged to intergenic spacer regions and one to a protein-coding region (trnH-psbA, rps2-rpoC2, rpl22, ycf15-trnl-CAA, and rps15). rpl22 showed a nucleotide diversity of 0.00378 and contained four substitutions with 459 missing data points. A similar approach was used for ycf15-trnl-CAA, selecting a 364 bp region, which had a nucleotide diversity of 0.00458 and contained three substitutions. The selected regions may act as suitable and cost-effective markers (Table 4). These polymorphic loci might be helpful for phylogenetic inference and population genetic studies of Catalpa species.

The base compositions and AT/GC contents of the six genomes of Catalpa were identical. Using CDSs of the chloroplast genome, we estimated the codon usage frequency of the six taxa of the genus Catalpa. The total number of codons detected in C. bungei and C. fargesii was 27,012 and 26,746, respectively, while the number in the other four species was 26,750 (Figure 7). The chloroplast genome of Catalpa encodes 20 amino acids at all codons. Leucine (Leu) was the most frequently used in the six taxa of the genus Catalpa, with a frequency ranging from 10.4% (2808) - 10.46% (2825), while cysteine (Cys) was the lowest in the six taxa of the genus, with a frequency reaching only 1.17% 315) - 1.18% (318). The RSCU values of the CDSs of Catalpa were calculated. Synonymous codon usage (RSCU value) refers to the relative probability of synonymous codons encoding corresponding amino acids for a specific codon, which eliminates the influence of amino acid composition on codon use. If there is no preference for the use of a codon, the RSCU value of the codon is equal to 1. When the RSCU value of a codon is greater than 1, it means that the codon is used more often than another, and vice versa. The results showed that the RSCU values of the six taxa included in the present study were similar. There were 30 codons with an RSCU value > 1, only one of which ended with G (UUG); the remaining 29 codons ended with A and T. The codons with an RSCU value < 1, except for UGA (stop codon) and CUA ending in A, ended in C or G. Therefore, the codon pairs ending with C and G in the Catalpa chloroplast genome have low bias, and they are nonpreferred codons. Due to usage frequency variation, the RSCU values of the chloroplast genome are a valuable form of evolutionary information resulting from mutation and selection that are essential in studying organismal evolution (Morton 2003).

The phylogenetic trees constructed using the ML and BI methods for the whole genome sequences of Catalpa chloroplasts had highly similar topologies (Figure 8). Strong bootstrap support and high posterior probabilities were recorded at all branch nodes. The seven species of Catalpa formed a monophyletic clade, with C. ovata diverging before the other six taxa included in the present study, with a high support rate. All taxa of the Bignoniaceae family were grouped together. In terms of the interspecific relationships of Catalpa, in the phylogenetic tree constructed based on the whole chloroplast genome sequences, C. ovata was located at the base and formed sister branches with the other six species, and C. speciosa formed sister branches with the other five species (C. fargesii f. duclouxii, C. fargesii, C. fargesii f. duclouxii (Huangxinzimu), C. bungei (Jinsiqiu) and C. bungei). C. fargesii f. duclouxii (Huangxinzimu) and C. fargesii Bur f. duclouxii formed a subbranch and were sister to C. fargesii.

In this study, the chloroplast genomes of some taxa of the genus Catalpa (C. fargesii f. duclouxii (Huangxinzimu) and C. bungei (Jinsiqiu)) were sequenced for the first time. The chloroplast genome size ranged from 157,765 bp (C. fargesii) to 158,355 bp (C. ovata), displaying six haplotypes. There were 113 genes in the chloroplast genome of all species, including 79 protein-coding genes, 30 tRNA genes and four rRNA genes. The two genes ycf15 and ycf68 were not annotated in this study, possibly because they are pseudogenes (Lu et al., 2016; Wang et al., 2020), consistent with results reported for other Catalpa species (Ma et al., 2020a; Ma et al., 2020b; Li et al., 2020; Wang et al., 2020). The accD, rpl32 and ycf2 genes are lost from the chloroplast genome in some cases (Jansen et al., 2008; Oliver et al., 2010; Dong et al., 2018), but these genes were present in the Catalpa chloroplast genome. The overall structure of the chloroplast genome of Catalpa is relatively conserved, and no major gene deletions or genome rearrangements were found. The total GC content was highly consistent among species, while genome size differed slightly but not significantly. The mVISTA results and nucleotide diversity tests revealed high degrees of similarity between the chloroplast genomes, implying that the chloroplast genomes of Catalpa are less diverged than those of other species (Li et al., 2018).

Variation in genome structure is another form of information that helps reveal the genetic diversity of species or aspects of their population biology or evolution. The most common SSRs in the chloroplast genome of C. ovata were mononucleotides mainly composed of A or T and rarely G or C. Microsatellites are very important for the study of population genetics. There were significantly fewer di-, tetra-, tri-, and pentanucleotide motif repeats and no hexanucleotide repeats in the six studied species, similar to the results of Rono et al. (Rono et al., 2020). The codon is crucial to the correct expression of genetic information. In general, the start codon sequences of chloroplast genomic DNA are ATG, ATT and ATA. There were two unique patterns in RSCU and usage frequency values based on six haplotypes of protein-coding genes. First, in addition to the stop codons, all codons showed a preference for ending in A or T, but the low-frequency codons were biased toward ending in C or G. Second, the two stop codons (UAA and UGG) showed no bias, consistent with the findings of previous studies (Rono et al., 2020; Wen et al., 2021). Overall, apart from codon usage, the SNPs and SSRs of the Catalpa chloroplast genomes were different and can be used as excellent resources for evaluating population genetic diversity. The chloroplast genomes of the six taxa of the genus Catalpa showed high genetic diversity.

There are 10 species of Catalpa worldwide, with several varieties. However, due to the low genetic differentiation of Catalpa, the systematic relationships among these taxa are not clear (Jia et al., 2012). The pollen morphology of C. bungei showed that some morphological characteristics of C. fargesii f. duclouxii are the same as those of C. fargesii but different from those of C. bungei, which is differentiated by its morphology. The use of several chloroplast markers, such as ndhF and nuclear ribosomal DNA, for phylogenetic reconstruction is sufficient to draw firm conclusions about the interspecies relationships within Catalpa (Li 2008). Therefore, sampling of additional genetic features is expected to improve phylogenetic resolution. The large-scale application of Illumina HiSeq technology has improved the ability to sequence entire chloroplast genomes so that these genome sequences can be used to analyze the close relationships between species (Sheng et al., 2021; Tian et al., 2021).

In this study, we used plastome sequences to assess the phylogenetic relationships within Catalpa. The results revealed deep phylogenetic relationships in this genus. Dode (Dode 1907) described samples of C. fargesii f. duclouxii collected from Yunnan, China, and found that the collected samples differed from C. fargesii in that the undersurface of leaves and petioles were hairless. Rehder (Rehder 1913) considered C. fargesii f. duclouxii to be a variety of C. fargesii. Gilmour (Gilmour 1936) further elaborated on this view: C. fargesii f. duclouxii is a hairless variety of C. fargesii, with C. fargesii being more closely related than C. bungei. Chloroplast genome ndhF and ribosomal DNA internal transcribed spacer ITS (nrDNA ITS) sequences were used to study the interspecific relationships of Catalpa, and phylogenetic trees constructed with ITS and chloroplast sequences showed that C. fargesii f. duclouxii formed its own branch and formed a sister branch with C. bungei and C. fargesii (Li 2008). The results of this study show that C. fargesii f. duclouxii and C. bungei are more closely related, C. fargesii f. duclouxii and C. fargesii f. duclouxii (Huangxinzimu) form a branch, and the branches formed with C. bungei and C. fargesii are sister branches. The results of this study do not support the conclusion that C. fargesii Bur f. duclouxii is a variant of C. fargesii as proposed by Rehder (1913) and Gilmour (1936). It is suggested that C. fargesii f. duclouxii be treated as a species independent of C. bungei and C. fargesii. This study also showed that all the taxa of Bignoniaceae clustered into one group, and similar family groups formed sister branches. Catalpa has sufficient genetic information, and Tecomaria capensis (NC 037462) is closely related to Catalpa, consistent with the results of previous studies (Gilmour 1936; Li et al., 2020). The results of this study provide strong evidence for elucidating the evolutionary history of these species and deeply analyzing the evolutionary events of Catalpa and even Bignoniaceae. The further development of sequencing technology will help fully reveal the general characteristics and patterns of variation of the chloroplast genome and provide a foundation for resolving the differences between morphological and genetic classification and for obtaining an in-depth understanding of plant evolution (Xu et al., 2021). However, due to the limited number of published chloroplast genomes of taxa within the genus Catalpa, there are still many difficulties in phylogenetic studies of this group. In the future, more data will be needed to explore their phylogenetic relationships.

Not all genes are phylogenetically useful in resolving taxonomic discrepancies. Oligonucleotide repeats exist widely in the plastome (Abdullah et al., 2021a). Mono-nucleotide, Palindromic, and forward repeats were the most common repeated sequences (Meng et al., 2019). Oligonucleotide repeats are also reported among the mutational events in chloroplast genomes (Abdullah, et al., 2020a). They consist of small repeats that exist in duplicate form (Kurtz et al., 2001) and mostly reported in size from 14 bp to 50 bp in chloroplast genome, unlike simple sequence repeats, which are one to six nucleotide tandem repeat units (Henriquez et al., 2014; Menezes et al., 2018; Abdullah et al., 2020a; Shahzadi et al., 2020). The results of Oligonucleotide repeats in this study are completely consistent with those mentioned above. In the taxonomy of the Catalpa, ndhF and the nrDNA ITS region can be discussed lower level relationships of plant groups (Baldwin et al., 1995; Soltis et al., 1998; Li 2008). However, the discriminatory power of these markers in Catalpa molecular phylogenetic investigations or DNA barcoding is deficient (Li 2008). Therefore, Chloroplast genome sequences provide an opportunity to elucidate patterns of genome evolution and provide valuable genetic resources for further research. Mutation events are not generally randomly distributed in the chloroplast genome but are concentrated in certain areas, forming “hotspots” (Dong et al., 2012; Wang et al., 2021). Comparing chloroplast genome sequences is an effective strategy for identifying mutation hotspots, and these highly variable regions can be used as DNA barcodes to distinguish species within specific taxa (Kuang et al., 2011; Abdullah et al., 2020a) and germplasm resources (Zhou et al., 2018; Ge et al., 2019). On the basis of the current study results, specifically, the results on nucleotide diversity among six Catalpa species or varieties, we suggest using a set of five divergent regions (≥200 bp) to solve taxonomic discrepancies and provide barcodes for the genus Catalpa. Regions of the plastome showed different polymorphisms, and certain regions were more predisposed to mutations. In this study, we identified five hypervariable regions, namely, trnH-psbA, rps2-rpoC2, rpl22, ycf15-trnl-CAA and rps15. These five regions had nucleotide diversity values of 0.00574 to 0.00378 from highest to lowest. The chloroplast genome sequences of the six taxa within the genus Catalpa were highly similar and conserved, and the noncoding regions had more variation than the coding regions, consistent with the results of previous studies (Perry and Wolfe 2002; Xiao-Feng Zhang et al., 2020). These variable regions can also be used to evaluate the phylogenetic relationships and interspecific differences of Catalpa (Yildirim et al., 2013). In this study, chloroplast genome data provided effective markers for inferring the phylogenetic relationships within Catalpa.