Characterizing the Complete Mitochondrial Genome of Arma custos and Picromerus lewisi (Hemiptera: Pentatomidae: Asopinae) and Conducting Phylogenetic Analysis.

We characterized the mitochondrial genome (mitogenome) and conducted phylogenetic analyses of 48 Hemiptera species by sequencing and analyzing the mitogenome of Arma custos (Fabricius) and Picromerus lewisi (Scott). The complete mitogenomes of the two predators were 16,024 bp and 19,587 bp in length, respectively, and it contained 37 classical genes, including 13 protein-coding genes (PCGs), 2 ribosomal RNA genes (rRNAs), and 22 transfer RNA genes (tRNAs), and a control region. Most PCGs in these predators use ATN as the start codon. This research revealed that the genes of the two natural enemy species have an A + T content of 75.40% and all tRNAs have a typical cloverleaf structure, with the exception of trnS1, which lacks a dihydrouridine arm. This is the first study to compare the mitochondrial genetic structure of two predatory insects; the mitochondrial genetic structure of individual predatory insects has been sequenced in previous studies. Here, phylogenetic analysis on the basis of amino acid and nucleotide sequences of 13 mitochondrial PCGs using Bayesian inference and maximum likelihood methods were conducted to generate similar tree topologies, which suggested that the two predators with close genetic relationships belong to Asopinae subfamily. Furthermore, the monophyly of the Pentatomoidea superfamily is well accepted despite limited taxon and species sampling. Finally, their complete mitogenome provided data to establish a predator-prey food web, which is the foundation of effective pest management. Our results also enhanced the database of natural enemy insects.

Mitogenomic sequences are widely used in biogeographic, molecular, and systematic studies (Cameron 2014). Mitogenomic studies involve exploring the origin and evolution of insects, explaining the species and evolution of the system, and revealing the geographical distribution of intraspecific polymorphism. This relationship provides several genome-level characteristics, including genomic diversity change, modes of control of transcription and replication, and RNA secondary structures (such as cloverleaf structure). The rate of base substitution is higher than that for most nuclear genes; without rearrangement during cell meiosis, these characteristics have made mitochondrial DNA a focus genetic marker for evolutionary studies (Curole and Kocher 1999, Saccone et al. 1999, Abascal et al. 2006).

The most important true bugs in agriculture and forestry are phytophagous species that attack cultivated crops (Song et al. 2013, Ogburn et al. 2016, Du et al. 2017). In contrast, cultivated crops and forests are infested with many species-preying pests (Sengonca et al. 2008, Zou et al. 2012, Zhang et al. 2019). The suborder Heteroptera contains >40,000 species (Zhao 2013). Previous studies have indicated that there are >300 species in 63 genera worldwide in the subfamily Asopinae (Hemiptera: Pentatomidae) (Ricardo et al. 2019). A. custos and P. lewisi are large, euryphagous, and widely distributed natural enemy insects that usually forage on the larvae and adults of various Chrysomelidae (Chen et al. 2007, Zhang et al. 2016, Yang et al. 2019, Wang et al. 2021), Lepidoptera (Gao et al. 2012; Jiang 2018; Tang et al. 2018, 2019; Xin et al. 2018; Wang et al. 2019; Fu 2021), and Hemiptera (Zhang et al. 2013a, Wang et al. 2017). These observations indicate that A. custos and P. lewisi could be adopted as useful biological controls to prevent pest infestation of cultivated crops, forests, and grass. Due to the importance of biological control, it is essential to obtain accurate and reliable phylogenic identification and construction to understand the biological properties, such as predation ability and life cycle, of these predator species. To date, only three mitogenomes have been sequenced for Asopinae species in the subfamily Asopinae (Zhao et al. 2017a, 2018, 2020), which impacts our knowledge on the phylogeny and diversity of Pentatomidae. Particularly, there are no complete mitogenomic sequences for large predatory stink bugs of Asopinae, such as A. custos and P. lewisi in NCBI.

This study presents the complete mitogenomes of the two natural enemy insects, A. custos and P. lewisi, which contribute to the establishment of the gene bank for natural enemy insects and provide data to support the establishment of a predator–prey food web. So far, a few large predatory stink bugs of the Reduviidae (Li et al. 2011, Kocher et al. 2014), Nabidae (Li et al. 2012a,b), and Anthocoridae (Du et al. 2014) families have been sequenced. This is the first study to use mitogenomes of the two species to preliminarily analyze and discuss their features at the mitogenomic level. We also discussed the architecture of their mitogenome and analyze the cloverleaf structure of tRNAs across the heteropterans. Finally, that the results of this study will provide a reference for the phylogenetic analysis and identification of Hemiptera, especially Asopinae.

Materials and Methods

Samples and DNA Extraction

A. custos and P. lewisi were supplied by Puchang Science Park at Suiyang County of The Guizhou Tobacco Company Zunyi Branch, Zunyi, Guizhou Province, China, in November 2020. Identification was done on the basis of traditional morphological characteristics. After starvation for 48 h, fresh tissues were preserved in 95% ethanol at −40°C at the Guizhou Provincial Key Laboratory for Agricultural Pest Management of the Mountainous Regions, Institute of Entomology, Guizhou University. Total DNA was extracted from the entire body using the Genomic Extraction Kit (Sangon Biotech, China) according to the manufacture’s protocol (ORI-GENE in Beijing, China).

Sequencing and Sequence Assembly

Agarose (1%) electrophoresis was used to evaluate the DNA quality before sequencing, ONT was used to sequence the complete mitogenomes at ORI-GENE (Beijing, China), then FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/), CLC Genomics Workbench 10.0 (CLC bio, Arhus, Denmark), and Cutadapt (http://cutadapt.readthedocs.org/) were used to retrieve and qualify the raw data. Then, the reads were compared against reference mitochondrial genomes (specifically, Dinorhynchus dybowskyi, Picromerus griseus, Reduvius tenebrosus, and Triatoma infestans) from NCBI using Minimap2 (Li et al. 2018) to get the objective reads. The objective reads were assembled using Unicycler v0.4.6 (Wick et al. 2017).

Genome Annotation and Analysis

We identified and predicted the tRNAs using MITOS web server (http://mitos.bioinf.uni-leipzig.de/index.py) (Perna and Kocher 1995 and Bernt et al. 2013), whereas the tRNAs that could not be identified were subjected to tRNAscan-SE 2.0 (http://lowelab.ucsc.edu/tRNAscan-SE/) to verify the prediction status (Bernt et al. 2013). The protein-coding genes (PCGs) were identified using ORF Finder (https://www.ncbi.nlm.nih.gov/orffinder/) in GenBank. The following formula was utilized to calculate the AT and GC skews: AT skew = (A − T)/(A + T) and GC skew = (.G − C)/(G + C) (Bernt et al. 2013). We then analyzed and computed the nucleotide composition and relative synonymous codon usage (RSCU) using MEGA 7.0 (Kumar et al. 2016).

Phylogenetic Analysis

Phylogenetic analyses were conducted based on the 46 complete mitogenomes of heteropteran species from NCBI. Two species (Idiocerus laurifoliae and Nilaparvata lugens) from Auchenorrhyncha were selected as the outgroup (Table 1). The 13 PCGs were compared with previous studies (Li et al. 2012a,b; Du et al. 2014). DNA alignment was inferred from the amino acid alignment of the 13 PCGs using MEGA v.7.0 (Kumar et al. 2016). JModelTest v.2.1 (Darriba et al. 2012) and ProtTest v.3.4 (Abascal et al. 2005) were used for model testing and selection for Bayesian inference (BI) and maximum likelihood (ML) analysis. MrBayes v.3.2 (Ronquist et al. 2012) were employed for BI analysis to analyze this dataset. Two simultaneous runs of one million generations were conducted for the matrix, and each set was sampled every 1,000 generations with a burn-in of 25%. The PHYML online web server was used for ML analysis to analyze the dataset and calculate the ultrafast bootstrap approximation approach for 10,000 replicates. Finally, the resulting phylogenetic trees were visualized using FigTree v.1.4.3 (Mousavi et al. 2014).

Table 1.

Classification of the taxas used in this study

Superfamily	Family	Species	NCBI no.	Length
Pentatomoidea	Pentatomidae	Graphosoma rubrolineatum	NC033875	15,633
Pentatomoidea	Pentatomidae	Halyomorpha halys	NC013272	16,518
Pentatomoidea	Pentatomidae	Eurydema gebleri	NC027489	16,005
Pentatomoidea	Pentatomidae	Dolycoris baccarum	NC020373	16,549
Pentatomoidea	Pentatomidae	Nezara viridula	NC011755	16,889
Pentatomoidea	Pentatomidae	Gonopsis affinis	NC036745	16,011
Pentatomoidea	Pentatomidae	Dalsira scabrata	NC037374	15,614
Pentatomoidea	Pentatomidae	Pentatomidae sp.	NC037074	15,498
Pentatomoidea	Pentatomidae	Eurydema maracandica	NC037042	15,391
Pentatomoidea	Pentatomidae	Carbula sinica	NC037741	15233
Pentatomoidea	Pentatomidae	Picromerus lewisi	MW355499.1.	19,587
Pentatomoidea	Pentatomidae	Arma custos	MW355500.1.	16,024
Pentatomoidea	Pentatomidae	Dinorhynchus dybowskyi	NC037724	15,952
Pentatomoidea	Pentatomidae	Picromerus griseus	NC036418	16,338
Pentatomoidea	Pentatomidae	Coptosoma bifaria	NC012449	16,179
Pentatomoidea	Pentatomidae	Megacopta cribraria	NC015342	15,647
Pentatomoidea	Pentatomidae	Coridius chinensis	JQ739179	14,648
Pentatomoidea	Pentatomidae	Eusthenes cupreus	NC022449	16,229
Pentatomoidea	Pentatomidae	Tessaratoma papillosa	NC037742	155,65
Coreoidea	Alydidae	Riptortus pedestris	NC012462	17,191
Coreoidea	Rhopalidae	Stictopleurus subviridis	NC012888	15,319
Lygaeoidea	Malcidae	Macroscytus gibbulus	NC012458	15,575
Lygaeoidea	Colobathristidae	Phaenacantha marcida	NC012460	14,540
Pyrrhocoroidea	Largidae	Physopelta gutta	NC012432	14,935
Pyrrhocoroidea	Pyrrhocoridae	Dysdercus cingulatus	NC012421	16,249
Aradoidea	Aradidae	Libiocoris heissi	NC030363	15,168
Aradoidea	Aradidae	Aneurus similis	NC030360	16,477
Aradoidea	Aradidae	Aneurus sublobatus	NC030361	16,091
Aradoidea	Aradidae	Brachyrhynchus hsiaoi	NC022670	15,250
Aradoidea	Aradidae	Neuroctenus parus	NC012459	15,354
Reduvioidea	Reduviidae	Triatoma infestans	NC035547	17,301
Reduvioidea	Reduviidae	Reduvius tenebrosus	NC035756	17,090
Reduvioidea	Reduviidae	Brontostoma colossus	NC024745	16,625
Reduvioidea	Reduviidae	Triatoma dimidiata	NC002609	17,019
Cimicoidea	Anthocoridae	Orius sauteri	NC024583	16,246
Cimicoidea	Anthocoridae	Orius niger	NC012429	14,494
Miroidea	Miridae	Lygus hesperus	NC024641	17,747
Miroidea	Miridae	Nesidiocoris tenuis	NC022677	17,544
Miroidea	Miridae	Lygus lineolaris	NC021975	17,027
Miroidea	Miridae	Lygus pratensis	NC037926	16,591
Gerroidea	Hydrometridae	Hydrometra greeni	NC012842	15,461
Gerroidea	Hydrometridae	Aquarius paludum	NC012841	15,380
Saldoidea	Saldidae	Saldula arsenjevi	NC012463	15,324
Notonectoidea	Notonectidae	Enithares tibialis	NC012819	15,262
Naucoroidea	Naucoridae	Ilyocoris cimicoides	NC012845	15,209
Gelastocoroidea	Gelastocoridae	Nerthra indica	NC012838	16,079
Nepoidea	Belostomatidae	Lethocerus indicus	NC027194	17,632
Nepoidea	Belostomatidae	Lethocerus deyrollei	KU237288	19,295
Fulgoroidea	Delphacidae	Nilaparvata lugens	NC021748	17,619
Membracoidea	Cicadellidae	Idiocerus laurifoliae	NC039741	16,811

The sequence number marked in red is the research subject of this paper.

Results

Mitogenomic Organization

The mitogenomic sequences of the two predator insects A. custos and P. lewisi were double-stranded closed circular molecules, with respective lengths of 16,024 bp and 19,587 bp (Fig. 1). The mitogenomes of the two predator insects contained 13 PCGs, 22 tRNA genes, and 2 rRNA genes. All genes (Table 2) are similar to those described in other hemipteran insects (Junqueira et al. 2004, Lee et al. 2009, Chen et al. 2019). The sequences were uploaded and deposited in GenBank under the accession numbers MW355499.1 and MW355500.1.

Fig. 1.

Mitochondrial genome structure of A. custos and P. lewisi. The orientation of gene transcription is indicated with arrows; pink arrows show tRNAs, yellow arrows show PCGs, red arrows show the rRNAs, and wheat-colored arrows show the control region. Single-letter amino acid abbreviations have been used to name the tRNAs. The sequence length was indicated with the ticks in the outer cycle.

Table 2.

Nucleotide composition of the mitogenomes of the two natural enemy species

	A. custos									P. lewisi
	Size (bp)	T	C	A	G	A + T%	G + C%	AT skew	GC skew	Size (bp)	T	C	A	G	A + T%	G + C%	AT skew	GC skew
Genome	16,024	33.70	13.60	42.20	10.50	75.40	24.70	0.112	−0.129	19,587	33.90	14.20	41.50	10.50	75.40	24.70	0.101	−0.150
PCGs	10,973	42.20	12.10	33.10	12.60	75.30	24.70	−0.121	0.020	10,991	41.00	13.70	31.70	13.60	72.70	27.30	−0.128	−0.004
rRNA	2,081	45.00	8.10	33.20	13.70	76.00	24.10	−0.151	0.257	2,280	42.90	8.60	33.10	15.50	76.00	24.10	−0.129	0.286
tRNAs	1,471	37.70	10.30	39.40	12.60	77.10	22.90	0.022	0.100	1,493	37.20	11.50	37.50	13.90	74.70	25.40	0.004	0.094
Control region	1,336	35.90	9.80	39.40	9.80	75.30	19.60	0.046	0.000	4,651	38.70	9.30	42.70	9.30	81.40	18.60	0.049	0.000

The nucleotide composition of the two insects revealed a dominant content of A + T in the complete mitogenome, with the same relative A + T content of 75.40% (Table 2). Moreover, the two mitogenomes shared similarly overlapping regions and intergenic spacers. The intergenic overlapping regions were 37 bp in length, containing 8 overlaps of length 1–8 bp; the largest overlapping region was situated between trnW and trnC, and the intergenic spacer regions were 200 bp in length, containing 20 size 1–27 bp spacer, with the longest spacer (27 bp) located between trnS2 and nad1 in A. custos. In P. lewisi, six intergenic overlapping regions were examined with varying lengths of 1 to 8 bp, and the largest overlapping region was at the same position (between trnW and trnC) as that for A. custos. There were four more intergenic spacers compared with A. custos, and the largest spacer (25 bp) region was also situated between trnS2 and nad1 (Table 3).

Table 3.

Organization of the mitogenomes of the two natural enemy species

A. custos							P. lewisi
Gene	Direction	Location	Size (bp)	Start codon	Stop codon	INC	Gene	Direction	Location	Size (bp)	Start codon	Stop codon	INC
trnI	J	1–68	68	_	_	0	trnI	J	1–67	67	_	_	0
trnQ	N	66–134	69	_	_	−3	trnQ	N	71–141	71	_	_	3
trnM	J	147–213	67	_	_	12	trnM	J	156–226	71	_	_	14
nad2	J	236–1,177	942	ATA	TAA	22	nad2	J	251–1207	957	ATT	TAA	24
trnW	J	1,185–1,251	67	_	_	7	trnW	J	1,225–1,291	67	_	_	17
trnC	N	1,244–1,307	64	_	_	−8	trnC	N	1,284–1,346	63	_	_	−8
trnY	N	1,325–1,389	65	_	_	17	trnY	N	1,352–1,418	67	_	_	5
cox1	J	1,401–2,937	1537	TTG	T-	11	cox1	J	1,426–2,962	1537	TTG	T-	7
trnL2	J	2,938–3,004	67	_	_	0	trnL2	J	2,963–3,026	64	_	_	0
cox2	J	3,023–3,683	661	ATT	T-	18	cox2	J	3,045–3,705	661	ATT	T-	18
trnK	J	3,684–3,755	72	_	_	0	trnK	J	3,706–3,779	74	_	_	0
trnD	J	3,763–3,827	65	_	_	7	trnD	J	3,783–3,848	66	_	_	3
atp8	J	3,837–3,986	150	ATA	TAA	9	atp8	J	3,858–4,007	150	ATA	TAA	9
atp6	J	3,980–4,654	675	ATG	TAA	−7	atp6	J	4,001–4,675	675	ATG	TAA	−7
cox3	J	4,658–5,446	789	ATG	TAA	3	cox3	J	4,683–5,471	789	ATG	TAA	7
trnG	J	5,450–5,514	65	_	_	3	trnG	J	5,473–5,536	64	_	_	1
nad3	J	5,515–5,868	354	ATA	TAA	0	nad3	J	5,538–5,891	354	ATA	TAA	1
trnA	J	5,875–5,937	63	_	_	6	trnA	J	5,911–5,979	69	_	_	19
trnR	J	5,951–6,014	64	_	_	13	trnR	J	5,986–6,052	67	_	_	6
trnN	J	6,020–6,088	69	_	_	5	trnN	J	6,055–6,126	72	_	_	2
trnS1	J	6,088–6,156	69	_	_	−1	trnS1	J	6,126–6,195	70	_	_	−1
trnE	J	6,158–6,225	68	_	_	1	trnE	J	6,195–6,260	66	_	_	−1
trnF	N	6,224–6,294	71	_	_	−2	trnF	N	6,259–6,328	70	_	_	−2
nad5	N	6,289–7,998	1,710	ATT	TAA	−6	nad5	N	6,339–8,048	1,710	ATT	TAA	10
trnJ	N	7,999–8,065	67	_	_	0	trnJ	N	8,049–8,118	70	_	_	0
nad4	N	8,073–9,407	1335	ATG	TAA	7	nad4	N	8,123–9,451	1329	ATG	TAA	4
nad4L	N	9,401–9,688	288	ATT	TAA	−7	nad4L	N	9,445–9,732	288	ATT	TAA	−7
trnT	J	9,703–9,767	65	_	_	14	trnT	J	9,735–9,801	67	_	_	2
trnP	N	9,768–9,831	64	_	_	0	trnP	N	9,802–9,865	64	_	_	0
nad6	J	9,843–10,313	471	ATA	TAA	11	nad6	J	9,870–10,349	480	ATG	TAA	4
cytb	J	10,320–11,456	1,137	ATG	TAA	6	cytb	J	10,357–11,496	1,140	ATG	TAG	7
trnS2	J	11,458–11,526	69	_	_	1	trnS2	J	11,499–11,567	69	_	_	2
nad1	N	11,554–12,477	924	TTG	TAA	27	nad1	N	11,593–12,513	921	TTG	TAA	25
trnL1	N	12,478–12,542	65	_	_	0	trnL1	N	12,514–12,580	67	_	_	0
rrnL	N	12,543–13,829	1,287	_	_	0	rrnL	N	12,581–13,864	1,284	_	_	0
trnV	N	13,827–13,894	68	_	_	−3	trnV	N	13,872–13,939	68	_	_	7
rrnS	N	13,895–14,688	794	_	_	0	rrnS	N	13,941–14,936	996	_	_	1
A + T-ricJ		14,689–16,024	1,336	_	_	0	A + T-ricJ		14,937–19,587	4,651	_	_	0

Strand of the genes is presented as J for majority and N for minority strand. In the column for intergenic length, a positive sign indicates the interval in base pairs between genes, while the negative sign indicates overlapping base pairs between genes.

Protein-Coding Genes

The mitogenomes of these two natural enemy species included 13 PCGs, which is in the typical order for hemipteran insects (Hao et al. 2017, Li et al. 2017). The 13 PCGs of the A. custos mitogenome were 10,973 bp in length and were 10,991-bp long for P. lewisi (Table 2). Of the 13 PCGs, 9 (cox1, cox2, cox3, atp6, atp8, nad2, nad3, nad6, and cytb) are encoded by the major strand (J-strand), whereas the other four were encoded by the minor strand (N-strand) (Table 3). The total number of codons in PCGs was 3646 and 3652 (Supp Table S1 [online only]), respectively. Thus, the typical ATN were used as the start codon in most PCGs of A. custos and P. lewisi, except for the nad1 and cox1 gene, which likely uses TTG as the start codon (Supp Table S2 [online only]). The alternative start codons have been found in other insects, such as GTG in Vargula hilgendorfii (Ogoh and Ohmiya 2004, Li et al. 2014) and TTAA in Drosophila melanogaster (Ballard 2000). PCGs terminate in two natural enemy species with the stop codon TAN; however, cox1 and cox2 terminate through the single T (Supp Table S2 [online only]).

The RSCU of the two natural enemy species showed distinct bias and were compared to each other in Fig. 2. The most frequently used amino acids in mitochondrial PCGs of two natural enemy species were Leu (UUR), Ile, Phe, and Met (Fig. 3 and Supp Table S1 [online only]), among them, the most frequently used amino acid in mitochondrial proteins of A. custos and P. lewisi was Leu (UUR), with RSCU 4.38 and 3.75, respectively (Fig. 3 and Supp Fig. S2 [online only]).

Fig. 2.

Relative synonymous codon usage (RSCU) within A. custos and P. lewisi. Codon families are shown on the X-axis and the frequency of RSCU on the Y-axis.

Fig. 3.

Codon distribution in two natural enemy species, the color-filled violet blocks indicate A. custos, whereas the filled LT magenta blocks represent P. lewisi. The total number of the codons was presented as numbers at the Y-axis and codon families are shown at the X-axis.

tRNAs and rRNAs

The total length of the tRNAs of the mitogenomes of A. custos and P. lewisi was 1,471 bp and 1493 bp, respectively (Table 2), and the tRNA genes are between 63 bp and 74 bp in length (Table 3). 14 genes (trnA, trnE, trnD, trnG, trnK, trnI, trnL2, trnM, trnN, trnR, trnS1, trnS2, trnT, and trnW) were located on the J-strand, and other 8 genes were embedded in the N-strand (Table 3). We found that only trn S1 lacked the dihydrouridine (DHU) arm, and the remaining 21 tRNA genes can form a typical cloverleaf structure of two natural enemy species (Supp Figs. S1 and S2 [online only]).

The rrnL (16S) and rrnS (12S) genes, which have the same situation in the two natural enemy species adjacent to the trn L1 and trnV and control region, respectively, are embedded in the N-strand (Table 3). The lengths of the two genes in A. custos were 1,287 bp (rrnL) and 794 bp (rrnS), respectively, whereas in P. lewisi, the two genes were 1,284 bp (rrnL) and 996 bp (rrnS) in length, respectively (Table 1).

Control Region

The control region is considered that can regulate the replication and transcription of the mitogenomeis and a source of length variation in the mitogenomeis (Zhang and Hewitt 1997). In most Hemiptera mitogenomes, the control region has been determined in four motifs: a long sequence of Ts, the tandem repeat sequences, a stem-loop structure, and an (A + T)-rich region (Cook 2005, Zhao et al. 2018). For example, the control region of A. custos locates between trnM and rrnS genes with 1336 bp in length, and the A + T content is 75.30%. However, the P. lewisi has an extremely long control region (4,651 bp), which was more than three times long compared with A. custos, and the A + T content in control region was up to approximately 81.40% (Table 1), which was significantly higher than A. custos.

Phylogenetic trees were built on the nucleotide and amino acid sequences of the 13 PCGs of 48 taxas of Hemiptera, performing two inference methods, ML and BI (Fig. 4. and Supp Fig. S3 [online only]). The five Pentatomomorpha superfamilies (30 taxas) followed the following monophyletic relationships: Aradoidea + (Pentatomoidea + (Coreoidea + (Lygaeoidea + Pyrrhocoroidea))), which is inconsistent with previous studies (Li et al. 2011, Yuan et al. 2015a). At the top of phylogenetic trees, P. lewisi and Picromerus griseus form sister group relationships, the four species (P. lewisi, A. custos, Dinorhynchus dybowskyi, and Picromerus griseus) also form sister group relationships, which showed the close genetic relationship (Fig. 4 and Supp Fig. S3 [online only]). The sister groups’ relationship of four Nepomorpha superfamilies (5 taxa) is confirmed. One superfamily of Leptopodomorpha is monophyletic with a sister relationship to Nepomorpha. Two branches (10 taxa) (Reduvioidea and (Cimicoidea and Miroidea)) comprise a paraphyletic Cimicomorpha. One superfamily of Gerromorpha with 2 taxas is monophyletic, located in the bottom of these five infra-orders.

Fig. 4.

Phylogenetic trees based on the nucleotide sequences of the 13 PCGs in the mitogenomes. (A) ML tree built on the codon positions of 13 PCGs. Bootstrap support values are shown on each node. (B) BI tree built on the codon positions of 13 PCGs. Bayesian posterior probabilities are shown on each node. The two natural enemy species sequenced in this study are marked with blue.

Discussion

Organization and Characteristics of A. custos and P. lewisi Mitogenome

In this study, we sequenced and annotated the complete mitogenomes of A. custos and P. lewisi. The comparative analyses of 48 Hemiptera genomes showed that the gene content, gene arrangement, and base composition are highly conserved in Hemiptera (Song et al. 2013, Yuan et al. 2015a, Zhao et al. 2018). The mitochondrial genome length of A. custos and P. lewisi is 16,024 bp and 19,587 bp, respectively; the mitochondrial genome length of P. lewisi could be the longest, whereas the mitogenome of the Nilaparvata lugens was the longest with lengths of 17,619 bp in the NCBI (Zhang et al. 2013b). The difference in length is shown in the control region by comparing the organization of the mitogenomes of two natural enemy species (A. custos: 1,336 bp and P. lewisi: 4,651 bp). The longest control region length is 2,429 bp, also in Nilaparvata lugens (Zhang et al. 2013b), which is far less than in P. lewisi with 4651 bp now. Whether the rich A + T content of the control region impact on transcripting and replicating of the mitogenome, and indirectly affects the predation effect of natural enemy insects, more further studies are always demanded to verify the function of the conserved control region in mitogenomes of Asopinae. Despite differences in sequence length of two natural enemy species, the mitochondrial genome order of A. custos and P. lewisi was similar to that of known ancestral taxa regarding the organization and composition of genome (Hua et al. 2008, Wang et al. 2017).

Notably, establishing traditional predator–prey food web methods includes direct field investigation (Luck et al. 1988), digestive tract anatomy (Walker et al. 1988), and indoor predation function response (Chen et al. 2007, Tang et al. 2019). However, with the development of molecular biology, research methods of the food web are becoming more diversified; a high-throughput sequence of insect intestinal contents can be used to quickly and efficiently establish a predator–prey food web (Schmidt et al. 2014), What we are doing now is releasing A. custos and P. lewisi into tobacco fields, harvesting them regularly, sequencing their gut contents, and then comparing them in a database in NCBI to get a network of their prey web. In this paper, the mitogenomes of A. custos and P. lewisi were sequenced, and they could provide data support in establishing a predator–prey food web, which is the foundation of effective pest management. Meanwhile, they could enrich the database of natural enemy insects.

12 PCGs of the 13 PCGs utilize the ATN as the starting codon, with the exception of cox1 starting with TTG in most bug mitogenomes (Yuan et al. 2015a,b; Wang et al. 2020). However, there is a difference in the two predator bugs that 11 PCGs have the same start codons ATN, 2 PCGs (nad1 and cox1) utilize TTG as the start codon compared with other bugs. Most PCGs use the TAA as the stop condon, nevertheless, in some insects, nad1, cox2, and some other genes use the single T or TAG as the stop condon (Dai et al. 2012, Song et al. 2013). The results in this study agree with previous reports; in the two predator bugs, most PCGs stop with TAA, two PCGs (cox1 and cox2) stop with a single T. However, one PCG (cytb) stop with TAG (Table 2 and Supp Table S2 [online only]).

In Hemiptera insects, most tRNAs have a classical cloverleaf secondary structure; however, the lack of DHU arm in the trnS1 was common in Hemiptera mitogenomes (Zhao et al., 2017b, Li 2018, Ren et al. 2019), or the DHU arm shows a loop structure in some tTNAs (Yuan et al. 2015b, Wang et al. 2017). In the mitogenomes of A. custos and P. lewisi, the DHU arm of trnS1 forms a simple loop. Like other insect mitogenomes (Ogoh and Ohmiya 2004; Li et al. 2014, 2017), rrnL and rrnS in two natural enemy species were layed between trnL1 (CUN) and trnV, and between trnV and the A + T-rich region, respectively (Fig. 1). The full length of the rRNA genes in A. custos was 2081 bp, with an A + T content of 76%, whereas the lengths of P. lewisi was 2280 bp, also having an A + T content of 76% (Table 1).

Phylogenetic Relationships Among Heteroptera

In Pentatomidae, the species, A. custos and P. lewisi, presents a sister position with Dinorhynchus dybowskyi and Picromerus griseus, all belonging to the subfamily Asopinae (Zhao et al. 2017a, 2018), suggesting that A. custos (Fallou) and P. lewisi Scott belong to Asopinae; this is consistent compared with the results of previous studies (Zhao et al. 2013, Ren et al. 2019).

In a previous study, the species Graphosoma rubrolineata was reported to belong to the subfamily Podopinae, mixed with species from the subfamily Pentatominae and the subfamily Asopinae in the phylogenetic trees (Zhao et al. 2017a, Li 2018). In this study, only the subfamily Podopinae mixed with species from the subfamily Pentatominae was considered; this followed that Asopinae was more closely related to Pentatominae (Fig. 4. and Supp Fig. S3 [online only]). The phylogenetic relationships Pentatomomorpha within 14 Superfamily including 48 species, based on mitogenome data, agree with those on basis of the classical morphological classification, this showed that using mitogenome is an effective way to compare the genetic relationships at classificatory levels.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.