The total lengths of the complete mitogenomes of N. pilipes, T. antipodiana and T. vitiana (previously N. vitiana) are 14,117 bp, 14,029 bp and 14,108 bp, respectively (Table 1; Table S2; Fig. 1). These three mitogenomes are shorter than those reported for T. clavata13. The lengths of Nephila and Trichonephila mitogenomes are similar to those reported for araneoid taxa ranging from 14,032 bp in Argiope perforata14 to 14,687 bp in Cyclosa japonica15 (NC_044696). The complete mitogenome of T. antipodiana has the smallest size compared to those of other araneoid taxa; the shortest so far reported is 14,032 bp in A. perforata. The gene arrangement in Nephila and Trichonephia mitogenomes is identical to those of other araneid spiders included in this study (Table S2; Fig. S1). All the present three mitogenomes (N. pilipes, T. antipodiana and T. vitiana) have 13 PCGs, two rRNA genes, 22 tRNAs, a non-coding A + T rich control region, and a large number of intergenic sequences (spacers and overlaps) (Table 1; Table S2; Fig. 1).
Besides, all three mitogenomes of N. pilipes, T. antipodiana, and T. vitiana are AT-rich (Table 2). These mitogenomes have negative values for AT skewness and positive values for GC skewness indicating the bias toward the use of Gs over Cs. Although an overall negative AT skewness value and positive GC skewness value are observed for the whole mitogenomes, they are variable for individual genes in different mitogenomes (Table 2). The A + T content for the N strand in the Nephila and Trichonephila mitogenomes is slightly higher than that for the J strand: with negative skewness value for the J strand and positive skewness value for the N strand (Table 2). The GC skewness value is positive for both the J and N strands, with the respective values for the J strand higher than those of the N strand.
The mitogenomes of both Nephila and Trichonephila are characterized by many more intergenic overlaps than spacers (Table 1; Table S2). The longest spacer in N. pilipes (19 bp) is between trnL1 and rrnL as well as between rrnL and trnV; that in T. antipodiana (24 bp) is between rrnL and trnV; that in T. vitiana (32 bp) between rrnL and trnV; and that in T. clavata (48 bp) between cox1 and cox2. The respective largest overlaps were: − 29 bp between trnW and trnY in T. vitiana; − 28 bp between trnE and trnF in N. pilipes; − 26 bp between trnR and trnE in T. antipodiana; and − 19 bp between nad3 and trnL2 in T. clavata.
A larger number of intergenic overlaps than spacers is also evident in the mitogenomes of other spiders: Tetragnatha maxillosa, and Tet. nitens (Tetragnathidae)16; Epeus alboguttatus (Salticidae)17; Wadicosa fidelis (Lycosidae)18; Ebrechtella tricuspidata (Thomisidae)19; Lyrognathus crotalus (Theraphosidae)20; and Cheiracanthium trivale (Cheiracanthidae), and Dystera silvatica (Dysteridae)21.
Protein-coding genes and codon usage
The A + T content for PCGs ranges from 69.7% for cox3 to 82.0% for atp8 in N. pilipes, 71.3% for cox1 to 83.4% for atp8 in T. antipodiana, 71.7% for cox3 to 81.4% for atp8 in T. vitiana, and 71.3% for cox3 to 83.4% for atp8 in T. clavata (Table S3). Interestingly, the AT skewness values are negative for the 13 PCGs in N. pilipes, T. antipodiana, and T. clavata; the AT skewness has both positive (nad4, nad4L, and nad5 PCGs) and negative values (the other PCGs) in T. vitiana. All the 13 PCGs in T. vitiana mitogenome have positive GC skewness value (Table S3). The mitogenomes of N. pilipes, T. antipodiana and T. clavata have negative GC values for nad1, nad4, nad4L and nad5 PCGs.
The PCGs of Nephila and Trichonephila mitogenomes are characterized by four start codons: ATA, ATT, TTG and TTA in N. pilipes, T. antipodiana and T. vitiana; ATA, ATT, ATG and TTG in T. clavata (Table 1; Table S2). Two complete stop codons (TAA and TAG) are present in the Nephila and Trichonephila mitogenomes. In addition, T. clavata has a truncated incomplete T stop codon. ATT is the commonest start codon in N. pilipes (8 PCGS), while ATA is the commonest in T. antipodiana, T. vitiana and T. clavata (each with 6 PCGs).
Nephila pilipes has identical start/stop codons with the other three Trichonephila mitogenomes for atp8 (ATT/TAA), atp6 (ATA/TAA) and nad3 (ATT/TAA); T. antipodiana and T. vitiana for cox1 (TTA/TAA), cox2 (TTG/TAA), cox3 (TTG/TAA) and nad6 (ATT/TAA); and T. vitiana and T. clavata for nad4L (ATT/TAA). The mitogenomes of T. antipodiana, T. vitiana and T. clavata have identical start/stop codons for nad4 (ATA/TAA). T. vitiana and T. clavata have identical ATA/TAA codons for nad5 (ATA/TAG in N. pilipes and ATT/TAA in T. antipodiana). The nad2 PCG in N. pilipes and the three other Trichonephila mitogenomes have different start and/or stop codons (Table 1).
The most common start codon with ATA in other spiders includes Tet. maxillosa (5 PCGs) and Tet. nitens (5 PCGs)16; D. silvatica (6 PCGs)21; E. alboguttatus (5 PCGs)17; W. fidelis (5 PCGs)18; and E. tricuspidata (7 PCGs)19. Spiders with ATT as the most common start codon include: C. trivale (5 PCGs)21; L. crotalus (6 PCGs)20; Araneus ventricosus (Araneidae) (7 PCGs)22; Argiope ocula (Araneidae) (4 PCGs)23; Habronattus oregonensis (Salticidae) (6 PCGs)24; and Argyroneta aquatica (Cybaeidae) (6 PCGs)25. In six species of Dysteridae spiders, ATA is the commonest start codon in only one species (Parachtes teruelis); the other species have ATT as the commonest start codon26.
TAA is the commonest stop codon in N. pilipes (9 PCGs), T. antipodiana (10 PCGs), T. vitiana (11 PCGs), and T. clavata (9 PCGs), excepting: TAG for cob, nad1 and nad5 in N. pilipes; nad1, nad2 and cob in T. antipodiana; cob and nad1 in T. vitiana; and nad2, nad6, cob and nad1 in T. clavata (Table 1; Table S2).
TAA has been reported to be the most common stop codon in A. ventricosus (9 PCGS)24, Neoscona scylla (Araneidae) (12 PCGs)27, Tet. maxillosa (8 PCGs) and Tet. nitens (10 PCGs)16, E. alboguttatus (8 PCGs)17, Evarcha coreana (Salticidae) (9 PCGs)28, W. fidelis (7 PCGs)18, E. tricuspidata (5 PCGs)19, Uroctea compactilis (Oecobiidae) (6 PCGs)29, C. triviales (7 PCGs) and D. silvatica (7 PCGs)21, L. crotalus (8 PCGs)20, H. oregonensis (5 PCGs)24, A. aquatica (4 PCGs and 6 truncated T)25, Mesabolivar sp. 1 (Phocidae) (8 PCGs) and Mesabolivar sp. 2 (11 PCGs)30, and E. alboguttatus (8 PCGs)17.
In the present study, truncated incomplete stop codon (T) is detected only for cox3 in T. clavata (Table 1; Table S2). No incomplete stop codon has been reported for L. crotalus20. Truncated stop codons are however not uncommon in the animal world. Examples of spider mitogenomes with incomplete T stop codons are: E. tricuspidata19; Tet. maxillosa and Tet. nitens16; A. perforata14; A. ocula23; A. ventricosus22; E. alboguttatus17; E. coreana28; Neoscona nautica31; N. scylla27; H. oregonensis24; Mesabolivar sp. 130; C. triviale21; D. silvatica21; U. compactilis29; A. aquatica25; W. fidelis18.
In general, the incomplete T stop codon in spiders involve the nad genes. Other incomplete stop codons may also be present in spider mitogenomes. Both T and TA stop codons are present in Mesabolivar sp. 130 and two species of Neoscona31. H. appenicola and five species of Parachtes have TA stop codon for two to four PCGs, while only H. appenicola and three species of Parachtes have T stop codon in one or two PCGs26. Incomplete TT stop codon has been reported for nad4L in C. triviale21. Incomplete stop codons are presumed to be completed by post-translational polyadenylation32.
The frequency of individual amino acid varies among the congeners of Trichonephila as well as the genera Nephila and Trichonephila (Fig. 2). However, the most frequently utilized codons are highly similar in these mitogenomes. The predominant amino acids (with frequency above 200) in all the four mitogenomes are isoleucine (Ile), leucine2 (Leu2), methionine (Met), phenylalanine (Phe), serine2 (Ser2), and valine (Val) (Table S4).
Analysis of the relative synonymous codon usage (RSCU) reveals the biased usage of A/T than G/C at the third codon position (Fig. 2). The frequency of each codon is very similar across all the four spider mitogenomes. The Ka/Ks ratio (an indicator of selective pressure on a PCG) is less than 1 for all the 13 PCGs in Nephila and Trichonephila mitogenomes, indicating purifying selection (Fig. 3; Table S5). Similar finding has been reported for 17 spider mitogenomes20. The sequence of the Ka/Ks ratio (cox1 < cox2 < cob < cox3 < nad1 < nad4 < atp6 < nad5 < nad4L < nad3 < nad2 < nad6 < atp8) in Nephila and Trichonephila species differs from that of (cox1 < nad1 < cox2 < nad5 < cob < cox3 < nad4 < atp6 < nad4L < nad3 < nad2 < nad6 < atp8) reported for 17 spider mitogenomes20. The cox1 gene with the lowest Ka/Ks ratio in spider mitogenomes, representing fewer changes in amino acids, supports its use as a molecular marker for species differentiation and DNA barcoding33,34.
Ribosomal RNA genes
Of the two rRNA genes in Nephila and Trichonephila mitogenomes, rrnS is much shorter, ranging from 693 bp in N. pilipes to 702 bp in T. antipodiana, while rrnL ranges from 1042 bp in T. antipodiana to 1050 bp in T. vitiana (Table 1, Table S2). As in other araneid spiders, rrnL is located between trnL1 and trnV and rrnS between trnV and trnQ (Fig. 1; Fig. S1).
Both the rRNA genes of the complete mitogenome are AT-rich (Table 2). The AT skewness value is variable among the mitogenomes: positive for both rrnL and rrnS in T. antipodiana and T. clavata; negative for both genes in T. vitiana; and negative for rrnL but positive for rrnS in N. pilipes. The GC skewness value is negative for rrnL and positive for rrnS in N. pilipes, T. antipodiana and T. clavata mitogenomes; it is positive for rrnL and negative for rrnS in T. vitiana.
Most spiders have longer rrnL than rrnS gene: Tet. maxillosa and Tet. nitens16; C. triviale and D. silvatica21; E. coreana28; W. fidelis18; A. perforata14; L. crotalus20; E. tricuspidata19, and A. aquatica25. Some spiders have similar length for rrnL and rrnS: for example, the length of rrnL and rrnS is the same (1722 bp) in N. nautica and N. doenitzi31.
Transfer RNA genes
The tRNAs of the whole Nephila and Trichonephila mitogenomes are AT-rich (Table 2), with positive AT skewness value in T. antipodiana and negative value in N. pilipes, T. vitiana and T. clavata; the GC skewness value is positive for all the four mitogenomes.
Most of the tRNAs in Nephila and Trichonephila mitogenomes have aberrant clover-leaf secondary structure, including truncated aminoacyl acceptor stem and mismatched (lacking well-paired) aminoacyl acceptor stem (Fig. 4).
Sixteen tRNAs in the Nephila and Trichonephila mitogenomes do not possess a TΨC arm: seven in N. pilipes and 10 each in T. antipodiana, T. vitiana and T. clavata (Fig. 4). There are also tRNAs with complete loss of TΨC stem (trnD in N. pilipes; trnV in T. antipodiana; and trnK in T. clavata) and complete loss of TΨC loop (trnR and trnQ in N. pilipes and trnK in T. vitiana).
Two tRNAs (trnA, trnS2) do not have DHU arm in all the Nephila and Trichonephila mitogenomes. Other tRNAs without DHU arm are: trnR in N. pilipes; and trnS1 and trnT in T. clavata. The complete loss of DHU loop involves trnQ in N. pilipes, trnN and trnV in T. antipodiana and T. clavata, and trnV in T. vitiana (Fig. 4).
Many tRNAs in spider mitogenomes have been reported to lack a well-paired aminoacyl acceptor stem, a TΨC arm, and a DHU arm35. None of the 22 tRNA sequences in H. oregonensis mitogenome have the potential to form a fully paired, seven-member aminocyl acceptor stem24. Mismatched aminoacyl acceptor stem has been reported to be a shared characteristic among spider mitogenomes35. It has been postulated that the missing 3ʹ acceptor stem sequence is post-translationally modified by the RNA-editing mechanism24. In A. aquatica mitogenome, the tRNAs are characterized by mismatched aminoacyl acceptor stem, and excepting trnS1 and trnS2 (both with only TΨC loop), the remaining tRNAs lack a TΨC arm25. The armless tRNA secondary structures are conserved across the family Dysderidae36.
The length of the non-coding control region in N. pilipes (498 bp), T. antipodiana (428 bp) and T. vitiana (511 bp) is much shorter than that of T. clavata (848 bp) (Table 1; Table S2). Spider mitogenomes with less than 800 bp for the control region include: N. nautica (455 bp) and N. doenitzi (566 bp)31; E. coreana (697 bp)20; T. nitens (690 bp)17; H. oregonensis (716 bp)24; U. compactilis (688 bp)29; and L. crotalus (356 bp)20. Examples of spider mitogenomes with greater than 800 bp are: Tet. maxillosa (864 bp)17; E. tricuspidata (859 bp)19; C. triviale (985 bp), D. sylvatica (954 bp)21; E. alboguttatus (968 bp)16; and A. aquatica (2047 bp)25.
The A + T content of the control region of Nephila and Trichonephila mitogenomes is AT-rich (Table 2), with negative AT skewness value in T. antipodiana and positive values in N. pilipes, T. vitiana and T. clavata (Table S3). The GC skewness value is positive for all four mitogenomes.
The control region of Nephila and Trichonephila mitogenomes is characterized by: (i) many simple tandem repeats and palindrome; (ii) long poly-nucleotide; and (iii) several stem-loop structures in these spider mitogenomes. The presence of 15 tandem repeats of ATAGA motif with TATATACATAT stretch (except one each with TAT, TATGTACATAT, and TATATACATAA) in T. clavata (Fig. 5) is a unique feature for this orb-weaving spider. Five 135-bp tandem repeats and two 363-bp tandem repeats have been identified in the putative control region of A. aquatica25. A long tandem repeat region comprising three full 215 bp and a partial 87 bp is present in the control region of W. fidelis mitogenome18.
An early study based on one nuclear (18S) and two mitochondrial (COXI and 16S) markers revealed that N. pilipes and N. constricta Karsch, 1879 formed a clade that was sister to all other Nephila species37. This finding was supported by molecular phylogenetic study based on three nuclear and five mitochondrial genes which indicated that the genus Nephila was diphyletic, with true Nephila (containing N. pilipes and N. constricta) and the other species (now genus Trichonephila according to Kuntner et al.1) being sister to the genus Clitaetra Simon, 188938. Large genetic difference (Fixed Differences, FD = 80%) between N. pilipes and other Nephila (now Trichonephila) species [N. edulis (Labillardière), N. plumipes (Latreille, 1804) and N. tetragnathoides (Walckenaer, 1841)] in Australasia had also been reported based on allozyme data4.
The present phylogenetic trees based on 13 PCGs and 15 mt-genes (13 PCGs and 2 rRNA genes) reveal identical topology with very good nodal support based on ML and BI methods (Fig. 6, Fig. S2). The genera Nephila and Trichonephila form a clade distinct from other genera of Araneidae. T. antipodiana and T. vitiana are closer related in the lineage containing also T. clavata, while N. pilipes is distinctly separated from these Trichonephila species. The araneid subfamilies Araneinae (genera Araneus, Cyclosa, Hypsosinga and Neoscona), Argiopinae (genus Argiope), Cyrtarachninae (genus Cyrtarachna) and Cyrtophorinae (genus Cyrtophora) form a clade distinct from the Nephila–Trichonephila clade.
Araneinae does not form a monophyletic group, with the genus Cyclosa being basal to the other Araneinae genera (Araneus, Hypsosinga and Neoscona), as well as the monophyletic subfamilies Argiopinae and Cyrtophorinae (Fig. 6; Fig. S2). Argiopinae and Cyrtophorinae form a lineage distinct from the Araneinae lineages comprising Neoscona and (Araneus–Hypsosinga), Cyrtarachninae is basal to the above araneid subfamilies. A large, representative taxonomic sampling is needed to reconstruct a robust phylogeny.
Both the BI and ML trees based on two rRNA (rrnL and rrnS) sequences reveal identical clades as 15 mt-genes and 13 PCGs (Fig. 6; Fig. S2). However, the genera Araneus and Argiope do not form monophyletic lineages, and the genus Cyclosa is the most basal genus to the other araneid genera. This result indicates that the rRNA genes alone are not suitable for reconstructing phylogeny at the higher taxonomic level.
In a recent study based on 13 protein-coding genes of the complete mitogenome, Nephilidae (represented by T. clavata) is basal to the family Araneidae19. Our present study, with the inclusion of N. pilipes, T. antipodiana and T. vitiana (previously N. vitiana) as well as T. clavata and additional recently published mitogenomes of Araneidae supports the Nephila–Trichonephila clade being basal to other araneid subfamilies (Fig. 6; Fig. S2). The close affinity of T. vitiana with T. antipodiana and T. clavata indicates that it is a member of the genus Trichonephila and not Nephila as currently recognized2.
The close affinity between T. antipodiana and T. vitiana is also reflected by their genetic distance: 8.65% based on 13 PCGs and 8.62% based on 15 mt-genes. On the other hand, the genetic distance between T. vitiana and N. pilipes is 21.68% based on 13 PCGs and 21.56% based on 15 mt-genes. Based on 15 mt-genes, the genetic distance between Trichonephila species ranges from 8.62 to 13.41% (Table S6).
Studies based on morphological data and mitochondrial and nuclear gene sequences have indicated closer relationship of T. antipodiana with T. clavata than with N. pilipes37,38,39. Based on anchored hybrid enrichment (AHE) targeted-sequencing approach with 585 single copy orthologous loci, the genus Nephila is basal to the genera Herennia Thorell, 1877, Nephilengys L. Koch, 1872, Nephilingis Kuntner, 2013, Trichonephila and Clitaetra1. The genus Clitaetra is basal to the genera Herennia, Nephilengys, Nephilingis, and Trichonephila.
Mitochondrial genomes have been applied particularly to studies regarding phylogeny and evolution of insects40. A recent study on spider mitogenomes covered only 12 species of Araneidae: 1 species of Trichonephila, 2 species of Araneus, 2 species of Argiope, 1 species of Cyclosa, 1 species of Cyrtarachne, 1 species of Hypsosinga, and 4 species of Neoscona21. Our present study has added 1 species of Nephila, 2 species of Trichonephila, 2 species of Argiope, 1 species of Cyrtophora, and 1 species of Neoscona. The taxon sampling is however still very limited compared to the large number of Araneid species. Studies on the mitogenomes of T. komaci and T. plumipes as well as other Nephila and Trichonephila species and related taxa will provide a potentially more robust phylogeny and systematics.