Codon usage of PCGs and gene selection pressure
Total PCGs nucleotide length ranges from 10 848 bp to 11 142 bp, and the AT contents ranges from 60.7% to 81.0%. Most of the PCGs initiate with the typical start codon ATN and TTG, whereas the special start codons AAC, AAT, AAA and TCA are found for COX1 ; AAA for COX2 ; GTG for ND1 and ND4L ; AGG and AAA for ND2 and GTG for ND4 . The most frequently used stop codons are TAA and TAG, followed by the incomplete stop codons T and TA. The most frequently used codons are UUA (Leu2), UCU and UCA (Ser2), CGA (Arg), whereas AGC (Ser1), ACG (Thr), GCG (Ala) and CUG (Leu1) are the least used (Fig. 4). For each PCG, the Ka/Ks ratio is less than one, and the ATP8 has the highest Ka/Ks ratio (0.33-0.67), followed by seven genes (ND6, ND5, ND4, ND2, ND4L, ND1, ND3 ) with Ka/Ks ratios of 0.17-0.42. Complex IV (COX1, COX2 and COX3 ), Complex III (CYTB ) andATP6 have low Ka/Ks ratios with range from 0.01 to 0.17 (Fig. 5). These results imply all of these 13 PCGs experienced purifying selection, especially Complex IV and Complex III.
Phylogenetic relationships
Substitution saturation tests show no saturation for three datasets AA, PCG12 and PCG12 + rRNAs (Iss < Iss.cSym or Iss.cAsym, p< 0.05) (Table 2), which proposes that these three datasets be appropriate for phylogenetic construction based on ML and BI. Six trees generated using these three datasets and both ML and BI are slightly different in topology (Figs. 6-7; Figs. S3-S6). The Ciidae is located at base of all phylogenetic trees, followed by families Mordellidae + Ripiphoridae. The Mordellidae + Ripiphoridae, Mordellidae and Ripiphoridae all looks monophyletic (PP = 1; BP = 100) and the later two appear sister groups each other. All remaining families also appear monophyletic, and the family Aderidae seems sister with “Meloidae clade” + “Tenebrionidae clade”. Both “Tenebrionidae clade” and “Meloidae clade” looks monophyletic with both of them being sister groups (Fig. 6). In the “Meloidae clade”, the families Meloidae + Anthicidae look a monophyletic (PP = 1; BP = 100), and it appears sister to the “Oedemeridae clade”, and the Meloidae and Anthicidae look monophyletic (PP = 1; BP = 100), and a sister each other. In the “Oedemeridae clade”, the Oedemeridae, Pyrochroidae, Salpingidae and Scraptiidae seem monophyletic (PP = 1; BP = 100), and Salpingidae looks a sister with Scraptiidae. In the “Tenebrionidae clade”, the family Lagriidae and Tenebrionidae are monophyletic (PP = 1; BP = 90-100), and both of them are sister groups each other. In Lagriidae, the subfamily Adeliinae is based at the subfamilies Lagriinae and Statininae, both of which look monophyletic (PP = 1; BP = 100) and are sister groups each other. In Tenebrionidae, the subfamily Pimeliinae appears monophyletic group (PP = 1; BP = 100) and is located at the base of Tenebrionidae. The Alleculinae and Stenochiinae look monophyletic (PP = 1; BP = 100), and the subfamilies Tenebrioninae and Diaperinae appears polyphyletic groups.
Six trees generated using AA, PCG12 and PCG12 + 2 rRNAs datasets and both ML and BI are slightly different in topology. For AA dataset, the topologies using ML and BI are different in the positions of Prostomidae. Prostomidae and Tetratomidae are clustered as one clade in ML tree, but not in BI tree. For PCG12 dataset, two same topologies of trees from BI and ML differ from two topologies of AA dataset in phylogenetic relationship of the “Oedemeridae clade”. For PCG12 + 2 rRNAs dataset, the positions of major families are the same as the four topologies of AA and PCG12 datasets, with only a few differences in the “Oedemeridae clade”.
Divergence time
The AA dataset was used to estimate divergence time because AA had higher node support values than others in the initial phylogenetic assessment using Bayesian appraoch. Based on three fossil calibrations points of Protoripidius burmiticus (166.6 Mya),Camelomorpha longicervix (145 Mya) and Alphitopsis initialis (143.6 Mya) (filled red cycles in Fig. 8), the superfamily Tenebrionoidea was inferred to originate in the early Jurassic (192.6 Mya, 95% confidence interval (CI): 179.3-208.7 Mya), with most families subsequently diverging in the Jurassic and early Cretaceous (Fig. 8). The family Mordellidae and Ripiphoridae is among the earliest diverged families in the superfamily, and is estimated to originate at 115.7 and 126.6 Mya in the Cretaceous, respectively. In the “Meloidae clade”, the family Meloidae is estimated to be derived at 105 Mya in the Cretaceous, the Anthicidae at 123.8 Mya in the early Cretaceous, and the Oedemeridae at 100.9 Mya in the middle Cretaceous. In the “Tenebrionidae clade”, the Lagriidae is estimated to originate at 134.3 Mya in the early Cretaceous, and the Tenebrionidae at 128.9 Mya in the early Cretaceous. In the family Lagriidae, the subfamily Statiriinae diverged 97.6 Mya in the late Cretaceous, and the Lagriinae diverged 78.6 Mya in the late Cretaceous. In the family Tenebrionidae, the subfamilies Pimeliinae, Alleculinae and Stenochiinae originated at 71.1, 53.6 and 53.3 Mya in the Paleogene, respectively. All of these families/subfamilies are proposed to be monophyletic and confirmed in relationships in the phylogenetic analyses using different data or inferring methods, and others are not determined for their monophyly or relationships or have a few species included in the phylogenetic analyses, and therefore they are not given an inferring of divergence time.