Specimens of five species were collected using an entomological net. Among them, C. diminutus, C. cloueti, M. sinae, and A. sichuanus were collected in the Dayaoshan Mountains in Guangxi Zhuang Autonomous Region, and S. femoralis was collected at Yingjiang County in Yunnan Province, China. Specimens were immediately preserved in 95% ethanol in the field after collected and then stored at –24 °C in the laboratory. The specimens were identified based on morphological characteristics under a Leica M205 FA stereomicroscope. Total DNA was extracted from muscle tissues using the Ezup Column Animal Genomic DNA Purification Kit (Shanghai, China) following the manufacturer’s instructions.

DNA sequencing and de novo assembly of each mitogenome were performed by Beijing Aoweisen Gene Technology Co. Ltd (Beijing, China). 22 tRNA genes were identified using the MITOS webserver, with the parameters of the Invertebrate Mito genetic code (Bernt et al. 2013). Their secondary structures were plotted manually from the MITOS predictions using Adobe Illustrator. Every sequence of tRNA genes was manually checked separately. The PCGs were identified as open reading frames corresponding to the 13 PCGs. The rRNAs and control regions were identified by the boundaries of the tRNA genes. The tRNA secondary structures were identified using tRNAscan-SE (Lowe and Chan 2016). Mitogenome maps (Suppl. material 1: Fig. S1) were produced using Organellar Genome DRAW (OGDRAW) (Greiner et al. 2019). The Base composition and relative synonymous codon usage values were determined using MEGA 6.0 (Kumar et al.2016). Strand asymmetry was calculated using the formulae AT-skew = (A – T) / (A + T), and GC-skew = (G – C) / (G + C) (Perna and Kocher 1995). In the control region (CR), tandem repeat elements were detected by Tandem Repeats Finder (Benson 1999). The heterogeneous analysis of the 13 PCGs and two rRNAs datasets were performed using AliGROOVE 1.06 (Kück et al. 2014), and the nucleotide diversity (Pi) and the ratio of Ka/Ks of PCGS were calculated with DnaSP v. 5 (Librado and Rozas 2009).

Phylogenetic trees for A. sichuanus, C. diminutus, C. cloueti, M. sinae, S. femoralis, and 10 other buprestid species belonging to four subfamilies were reconstructed by three separate datasets (13 PCGs, 2 rRNAs, and 13 PCGs + 2 rRNAs) using different best-fit models (Table 4). The mitogenomes of Heterocerus parallelus (Heteroceridae) and Dryops ernesti (Dryopidae) were used as outgroups, as they are phylogenetically distant from Buprestidae in the suborder Polyphaga (Xiao et al. 2019). The phylogenetic analyses were performed using PhyloSuite v. 1.2.2 (Zhang et al. 2020). Nucleotide sequences of the 13 PCGs and 2 rRNAs of all 17 mitogenomes were aligned using ClustalW (Thompson et al. 1994) and trimmed using trimAl v. 1.2 (Capella-Gutiérrez et al. 2009). The best-fit model for three datasets was determined by ModelFinder based on Bayesian information criterion. The maximum-likelihood (ML) and Bayesian inference (BI) methods were used to reconstruct the phylogenetic trees by IQ-tree v. 1.6.8 (Guindon et al. 2010) and MrBayes v. 3.2.6 program respectively (Ronquist et al. 2012). Bayesian analyses were run with two independent chains spanning 2,000,000 generations, four Markov chains, sampling at every 100 generations, and a burn-in period of 0.25 for each chain. The phylogenetic trees were edited and visualized by Figtree v. 1.4.3.

The complete mitogenomes of the buprestids A. sichuanus, C. diminutus, C. cloueti, M. sinae, and S. femoralis have the following GenBank accession numbers attributed to them: OK189519, OK189521, OK189520, OK189522, OK349489. The mitogenomes of these five species contained the 37 typical mitochondrial genes (13 PCGs, 22 tRNAs, and 2 rRNAs) and a control region (CR) (Table 2). The composition and arrangement of the mitochondrial genes in these species (Table 2) were highly similar as those in most other buprestid species (Duan et al. 2017; Cao and Wang 2019a, 2019b; Xiao et al. 2019; Chen et al. 2021; Peng et al. 2021).

Four of the 13 PCGs (nad1, nad4L, nad4, and nad5), eight tRNAs (trnQ, trnV, trnL1, trnP, trnH, trnF, trnY, and trnC), and two rRNAS (rrnL and rrnS) are encoded on the N-strand, whereas the other 23 genes (9 PCGs and 14 tRNAs) are encoded on the J-strand. The mitogenome sequence of these five buprestid species ranged in size from 15,367 to 16,521 bp.

The mean A + T nucleotide contents of five complete mitogenomes were similar: 68.42% in C. diminutus, 69.27% in C. cloueti, 72.42% in M. sinae, 71.73% in A. sichuanus, and 73.23% in S. femoralis. The entire mitogenomes had a higher A + T contents of 68.42–73.23% (66.05–72.50% for PCGs, 70.95–74.03% for tRNA genes, 75.20–77.33% for rRNA genes, and 74.17–78.38% for the CR) than G + C contents, which is consistent with the typical base of buprestid mitogenomes. The overall AT skews in these five complete mitogenomes were 0.12, 0.11, 0.11, 0.12, and 0.12, respectively. These five species showed a positive TA skew, suggesting that a slight AT bias which are similar to those observed in other buprestid species (Duan et al. 2017; Cao and Wang 2019a, 2019b; Xiao et al. 2019; Chen et al. 2021; Peng et al. 2021).

The total lengths of PCGs in these five buprestid species ranged from 11,090 to 11,158 bp, accounting for 67.54–72.17% of the entire mitogenomes. Similar to the other buprestid mitogenomes, nad5 and atp8 were found to be the largest (1708–1723 bp) and smallest (156–159 bp) genes, respectively. The majority of PCGs strictly started with an ATN (ATA/ATT/ATC/ATG) start codon, except for the nad1 starting with TTG. All PCGs strictly terminated with TAR (TAG/TAA) or an incomplete stop codon T–. Similar to most previously sequenced members of Buprestidae, the AT skew (0.11–0.12) of these five PCGs (Table 3) were similar among the 15 buprestid species. Summaries of the numbers of amino acids in the annotated PCGs and relative synonymous codon usage are presented in Figs 1 and 2. Overall codon usage among the sequenced buprestid mitogenomes was found to be similar, with Leu2, Ile, Phe, Ser2, Gly, Met, and Val being the seven most frequently coded amino acids.

Figure 1. 

Numbers of different amino acids in the mitogenomes of the five buprestid species; the stop codon is not included. AS: Agrilus sichuanus, CC: Coraebus cloueti, CD: Coraebus diminutus, MS: Meliboeus sinae, and SF: Sambus femoralis.

Figure 2. 

RSCU (relative synonymous codon usage) of the mitogenomes of the five buprestid species; the stop codons are not included.

The nucleotide diversity (Pi) of the 13 PCGs among five species of Agrilinae is provided (Fig. 3), which ranged from 0.202 to 0.375. In these genes, nad2 (Pi = 0.375) presented the highest variability, followed by nad6 (Pi = 0.346), nad4 (Pi = 0.300), and nad5 (Pi = 0.290); cox1 (Pi = 0.20) exhibited the lowest variability. The ratio of Ka/Ks (Fig. 4) for each gene of the 13 PCGs was calculated. The values of nad4 and nad4L are distinctly higher than others, which suggests that the genes nad4 and nad4L have a relatively higher evolutionary rate.

Figure 3. 

Nucleotide diversity (Pi) of 13 PCGs among five newly sequenced Agrilinae mitogenomes.

Figure 4. 

The ratio of Ka/Ks of 13 PCGs among the 15 reported Buprestidae mitogenomes.

The length of rrnL genes ranged from 1258 bp (S. femoralis) to 1271 bp (A. sichuanus), whereas rrnS ranged from 696 bp (S. femoralis) to 718 bp (C. diminutus). The A + T content of the rRNA genes ranged from 75.20% (C. diminutus) to 77.33% (M. sinae) (Table 3). Compared with those in other sequenced buprestid mitogenomes, the rRNA genes in these five newly sequenced buprestid mitogenomes are highly conserved (Hong et al. 2009; Duan et al. 2017; Cao and Wang 2019a, 2019b; Xiao et al. 2019; Sun et al. 2020; Chen et al. 2021; Peng et al. 2021). These rRNAs were located between the CR and trnL1, and separated by trnV. The total lengths of the 22 tRNA genes ranged from 1414 bp (C. cloueti) to 1444 bp (C. diminutus), whereas individual tRNA genes typically ranged in size from 58 to 70 bp, among which, eight tRNAs were encoded on the N-strand and the remaining 14 encoded on the J-strand. The secondary structures of tRNAs showed a standard clover-leaf structure (Suppl. material 1: Figs S2–S6), except for tRNA-Ser (Fig. 5) which lacks or has an unusual dihydrouridine arm, which forms a loop commonly found in other insects (Xiao et al. 2011; Park et al. 2012; Yu et al. 2016; Yan et al. 2017; Yu and Liang 2018; Li et al. 2019). In A. sichuanus, the longest intergenic nucleotide (774 bp) was located between trnW and trnC, which is an interesting and specific phenomenon in Buprestidae. The degree of heterogeneity of the 13 PCGs dataset was higher than that of the two rRNAs dataset (Suppl. material 1: Fig. S7). Additionally, the heterogeneity in sequence divergences was slightly stronger for Coraebus than for other buprestid genera (Suppl. material 1: Fig. S7).

Figure 5. 

The predicted secondary structures of the tRNA-Ser in the mitogenomes of the five buprestid species.

The CR, also known as the A + T-rich region (Wolstenholme 1992), was the largest non-coding region and located between trnI and rrnS. The length of CR ranged from 910 bp (S. femoralis) to 1577 bp (M. sinae). The A + T content (74.17–78.38%) of the CR of these five species was found to be higher than that of the whole genome (68.42–73.23%), PCGs (66.05–72.50%), rRNAs (75.20–77.33%), and tRNAs (70.95–73.85%) (Table 3). Moreover, the compositional analysis revealed that the mitogenomes of the five buprestid species had a positive AT skew (0.02–0.18) in the CR. In these five species, only C. cloueti and C. diminutus had no tandem repeat element detected; however, those of A. sichuanus (20 and 40 bp), M. sinae (53 bp), and S. femoralis (265 bp) had different lengths.

Both ML and BI trees using three datasets produced identical topologies (Figs 6–8), (Buprestinae + ((Chrysochroniae + Polycestinae) + Agrilinae)), in terms of subfamily-level relationship. The monophyly of Buprestidae is corroborated again, as all the buprestid species converged together as an independent clade, and two outgroup taxa obviously separated from the buprestid clade. The target species C. diminutus, C. cloueti, Meliboeus sinae, Agrilus sichuanus, and Sambus femoralis, as well as other species of Agrilinae, converged together as an independent clade. And the target species, M. sinae, was most closely related to the genus Trachys with high value support (Figs 6–8) which is inconsistent with the previous studies (Kubáň et al. 2000; Evans et al. 2015). The relationship of Agrilinae clades obtained from 2 rRNAs and 13 PCGs + 2 rRNAs datasets are identical but with different topology from the 13 PCGs dataset. In the topology generated from the 13 PCGs dataset, S. femoralis and Agrilus were clustered into a single branch with high support value (Fig. 6, ML: 77, BI: 1) whereas, in the topology generated from the 2 rRNAs and 13 PCGs + 2 rRNAs datasets, S. femoralis split from base of the Agrilinae clades (Figs 7, 8). Based on these results the position of the genus Sambus in the tribe Agrilini was not suitable and suspect. The different tree topologies suggested that the rRNA genes were extremely valuable for the phylogenetic analysis of Agrilinae. Coraebini is the most diverse tribe in Agrilinae, and 10 subtribes are defined (Kubáň et al. 2000). The genus Meliboeus (Meliboeina) and Coraebus (Coraebina) in different clades suggested that the tribe Coraebini was polyphyletic, which is consistent with the previous study of Evans et al. (2015). The samples used in this study might be too limited for a comprehensive phylogeny of Buprestidae which still needs a deep study in the future.

Figure 6. 

Phylogenetic relationships of 15 selected buprestid species using both BI and ML analyses based on 13 PCGs of mitogenomes. The numbers on the branches show posterior probability (BI tree), whereas the values under branches are bootstrap (ML tree).

Figure 7. 

Phylogenetic relationships of 15 selected buprestid species using both BI and ML analyses based on 2 rRNAs of mitogenomes. The numbers on the branches show posterior probability (BI tree), whereas the values under branches are bootstrap (ML tree).