3 Results and discussion
3.1 Basic structure
The full length of the mitochondrial genomes of Prophantis octoguttalis and P. adusta were 15,197 bp and 15,714 bp, respectively, including 37 genes and non-coding regions (Figure 1). Four protein-coding genes (nad1 , nad4 , nad5 , andnad4l ), two rRNA genes (rrnL and rrnS ), and eight tRNA genes (trnQ , trnC , trnY , trnF ,trnH , trnP , trnL1 , and trnV ) were encoded from the minority strands. The remaining 23 genes were encoded from the majority of the strands (Table 2). The mitogenomes of both species were arranged in the same order as that of Bombyx mori (Linnaeus, 1758), which is the model organism in Lepidoptera (Dai et al., 2013). There were eight gene overlaps and 15 gene gaps in the mitogenome ofP. octoguttalis , while five genes overlapped and 18 gene gaps were found in the mitogenome of P. adusta .
Figure 1. Visualization of the mitochondrial genomes ofProphantis octoguttalis and P. adusta
The mitogenome sequences of both species showed obvious AT biases. The nucleotide content of the P. octoguttalis mitogenome was A: 41.0%, T: 40.5%, C: 11.0%, and G: 7.5%, and for the P. adustamitogenome was A: 40.8%, T: 40.7%, C: 11.0%, and G: 7.4%. The AT contents were 81.5% and 81.6%, respectively, which were much higher than the GC content. The AT skew was 0.006 and 0.001, and the GC skew was -0.189 and -0.196, respectively, showing a slight A skew and a significant C skew (Table 3).
3.2 Protein-coding genes and codon usage
Thirteen protein-coding genes were identified in the mitogenomes of P. octoguttalis andP. adusta . Among them, atp8 , atp6 , cox1 ,cox2 , cox3 , nad2 , nad3 , nad6 , andcytb were encoded by the majority strand, and the remaining four genes were encoded by the minority strand. In P. octoguttalis , there was a 7 bp overlap between atp8 and atp6 and 1 bp overlap between atp6 and cox3 . In P. adusta , there was only a 7 bp overlap between atp8 and atp6 . The start codons of all genes were typical ATN (ATT, ATA, ATG), except forcox1 , whose start codon was CGA. The stop codons of cox1and cox2 in P. octoguttalis were terminated by an incomplete stop codon T, and the remaining genes were terminated by TAA, which was the most frequent stop codon. Among the protein-coding genes, the AT content was 80.3% and 79.6%, respectively. The AT bias of these two species was more significant in the third codon, and the AT content of the third codon (83.2%, 85.8%) was higher than that of the first (73.1%, 82.7%) and second codons (74.9%, 79.8%). The AT skew of these two species was 0.01 and 0.003, and their GC skew was -0.173 and -0.181, respectively, showing a slight A skew and an obvious C skew.
Figure 2. Relative synonymous codon usage (RSCU) ofProphantis octoguttalis and P. adusta
The concatenated lengths of the 13 PCGs of P. octoguttalis and P. adusta were 11,196 bp and 11,219 bp, encoding 3721 and 3728 amino acids, respectively. Statistics on the relative synonymous codon usage (RSCU) of P. octoguttalisand P. adusta showed that the codons UUA(L), AUU(I), UUU(F), AUA(M) and AAU(N) were used most frequently. In P. octoguttalis , CUG, GUC, CCG, CGG, AGC and AGG do not participate in amino acid synthesis, while in P. adusta , CUG and AGG do not participate. The codons of amino acids with RSCU > 1 all contained A or U (Figure 2), and the preference of these codons indirectly reflected the AT preference of the base.
3.3 rRNA genes and tRNA genes
In the mitogenomes of P. octoguttalis and P. adusta , two rRNA genes were encoded by the minority strand, with concatenated lengths of 2092 bp and 2077 bp, respectively. The rrnL gene was located between the trnL1 and trnV genes, which were 1355 bp and 1341 bp long, respectively; the rrnS gene was located between thetrnV gene and the A+T-enriched regions, which were 737 bp and 736 bp long, respectively.
In the mitogenomes of these two species, there were 22 tRNA genes with concatenated lengths of 1468 bp and 1481 bp, respectively. A total of 14 genes (trnM , trnI , trnW , trnL2 , trnK ,trnD , trnG , trnA , trnR , trnN ,trnS1 , trnE , trnT , and trnS2 ) were encoded by the majority chain, and the remaining eight genes were encoded by the minority chain, with the length of each gene ranging from 64 bp (P. octoguttalis ) – 71 bp. Except for trnS1 (AGN), which lacked the DHU arm, the secondary structures of the remaining 21 tRNAs folded into a typical clover-leaf structure (Figure 3). There were G-U and U-U base mismatches in the tRNA genes, which mostly occurred in the DHU, AA acceptor, and anticodon arms.
The AT content of the RNA gene of these two species was more than 80%, showing an obvious AT bias. As for base skew, both species showed a slight A skew and an obvious C skew.
Figure 3. Secondary structure of tRNA of Prophantis octoguttalis and P. adusta
3.4 Non-coding regions
The mitogenome of P. octoguttalis had eight gene overlaps totaling 24 bp, with a maximum overlap length of 8 bp between the trnW andtrnC genes, and 15 gene spacings totaling 172 bp, with a maximum spacing length of 45 bp between the trnQ and nad2 genes. The mitogenome of P. adusta had five gene overlaps totaling 21 bp, with a maximum overlap length of 8 bp between the trnW andtrnC genes, and 18 gene spacings totaling 240 bp, with a maximum spacing length of 54 bp between the trnS1 and trnE genes.
The control regions of the mitogenomes of these two species were located between the rrnS and trnM genes, with full lengths of 327 bp and 735 bp, respectively. Both sequences showed a clear AT bias, with an AT content of 96.0% and 96.7%, respectively, which was significantly higher than that of GC. The AT skew and GC skew of both sequences were negative, showing a slight T skew and an obvious C skew.
3.5 Phylogenetic relationships
The mitogenomes of 55 Lepidoptera species were used in this study, including eight subfamilies of Crambidae as ingroups, with four Pyralidae species, Helicoverpa armigera (Noctuidae) andBombyx mori (Bombycidae) as outgroups. Four phylogenetic trees of Crambidae were reconstructed using ML and BI analyses based on two datasets: PCG123 and PCG123RT (Figure 4). All phylogenetic trees showed the monophyly of Crambidae and was strongly supported (PP=1/BS=100).
Figure 4. Phylogenetic tree constructed with BI and ML analyses based on two datasets show the similar topology except for the “non-PS clade”: (A) the BI tree of dataset PCG123RT and the ML tree of dataset PCG123. (B) the “non-PS clade” of ML tree of dataset PCG123RT. (C) the “non-PS clade” of BI tree of dataset PCG123. The values around the nodes are posterior probability (PP) and bootstrap support (BS)
The eight subfamilies of Crambidae in all phylogenetic tree was divided into two major sister lineages, the “PS clade” and the “non-PS clade”, which was first defined by Regier et al. (2012). Spilomelinae and Pyraustinae were sister groups to each other (PP=1/BS=100), forming the “PS clade”, which was consistent with previous studies based on molecular data (Regier et al., 2012; Léger et al., 2020) or mitogenomic data (Yang et al., 2018b; Zhang et al., 2020; Jeong et al., 2021; Liu et al.,2021; Qi et al., 2021).
In Spilomelinae, all phylogenetic results showed that the monophyly of Trichaeini was well supported (PP=1/BS=100), and the relationships within Spilomelinae were (Trichaeini + Nomophilini) + ((Spilomelini + (Hymeniini + Agroterini)) + Margaroniini). With the exception of the newly sequenced species of Trichaeini, the phylogenetic relationship among the tribes was roughly consistent with Liu et al. (2021) that Agroterini, Hymeniini, Margaroniini and Spilomelini are grouped into one branch, sister to Nomophilini. Our results showed that Trichaeini and Nomophilini were related to each other as a sister group (PP=1/BS=86) and were first separated from the base of the subfamily Spilomelinae. This confirmed the results of Matsui et al. (2022) based gene fragments. However, in the phylogenetic tree in Mally et al. (2019) based on gene fragments, Trichaeini and (Steniini + Nomophilini) formed a sister group relationship, which was inconsistent with (Trichaeini + Nomophilini) + Steniini in Matsui et al. (2022). Therefore, more samples, especially those of the closely related species of Steniini and Nomophilini, are expected to sequenced for the complete mitochondrial genomes in the future research, in order to clarify the phylogenetic relationships among these three tribes.
The differences among the four phylogenetic trees constructed in this study were mainly concentrated in the “non-PS clade”, which consisted of the remaining six subfamilies (Acentropinae, Crambinae, Glaphyriinae, Odontiinae, Schoenobiinae, Scopariinae). The “non-PS clade” was divided into the “OG clade” and the “CAMMSS clade” (PP=1/BS=98) in the BI tree of dataset PCG123RT and the ML tree of dataset PCG123. The “OG clade” consisted of Glaphyriinae and Odontiinae, which were related to each other as sister groups, with a high to low support (PP= 1/BS=55) and the monophyly of Odontiinae was highly supported (PP=1/BS=100). Acentropinae, Crambinae, Schoenobiinae and Scopariinae formed the “clade CAMMSS” which presented two close relationships, Acentropinae and Schoenobiinae as sister group (PP=1/BS=81), Scopariinae and Crambinae as sister group with a high to low support (PP=0.93/ BS=57). This was consistent with the results of Regier et al. (2012) and Léger et al. (2020) based on molecular data. Meanwhile, this result also confirmed the mitogenome-based results of Qi et al. (2020); Jeong et al. (2021) and Liu et al. (2021), which were based on the ML and BI trees of dataset PCG123R, PCG12 and PCG12RT, the ML phylogenetic trees of dataset AA and the BI trees of dataset PCG123 and PCG123RT. The affinities of the subfamilies in the “CAMMSS clade”, which based on the ML tree of dataset PCG123RT in this study, were exhibited different topologies: Scopariinae + (Crambinae + (Acentropinae + Schoenobiinae)), which was consistent with the ML tree of dataset PCG123 and PCG123RT in Liu et al. (2021). In the BI tree of dataset PCG123, the phylogenetic relationship of the “non-PS clade” was: Odontiinae + ((Scopariinae + Glaphyriinae) + (Acentropinae + ((Schoenobiinae + Crambinae))),with low support, which was completely different from the above situation. The phylogenetic topology varies among the subfamilies within the “non-PS clade” in different datasets, probably due to with only one sample each in Schoenobiinae, Scopariinae and Glaphyriinae, thus causing a long branch attraction.
On the basis of the above analyses, our analyses confirmed the sister relationship of Pyraustinae and Spilomelinae with strong support. Trichaeini in this study robustly constitute a monophyletic group in Spilomelinae, with the relationships (Trichaeini + Nomophilini) + ((Spilomelini + (Hymeniini + Agroterini)) + Margaroniini). Within the “non-PS clade”, the monophyly of Acentropinae, Crambinae, and Odontiinae were well supported. The close relationship between Odontiinae and Glaphyriinae, between Schoenobiinae and Acentropinae, and between Scopariinae and Crambinae seemed to be more realistic.