Introduction

Spiders of the genus Nephila Leach, 1815 and genus Trichonephila Dahl 1911 are members of the family Nephilidae1 or subfamily Nephilinae of Araneidae2. Before the taxonomic treatment by Kuntner et al.1, Trichonephila species were traditionally treated as members of the genus Nephila. Nephila and Trichonephila are large orb-weaving spiders, with Trichonephila komaci3 being the largest species ranging from some 33–40 mm in total length3. At different times, they have been treated as members of the family Nephilidae1,4,5, and members of the subfamily Nephilinae within the family Araneidae2,6,7.

Kuntner et al.1 listed two species of Nephila and 12 species of Trichonephila. In contrast, the World Spider Catalog2 recorded 10 species of Nephila and 12 species of Trichonephila. Recently, a new species Nephila nandiniae has been described from Bangladesh8. Kuntner et al.1,9 did not include the taxon Nephila vitiana (Walckenaer, 1847) in their studies. N. vitiana was treated as a valid species by Harvey et al.4 and listed as an accepted species in the World Spider Catalog, version 21.52. It is morphologically very similar to Trichonephila antipodiana (Walckenaer, 1841). Both taxa exhibit similar abdominal (opisthosomal) colour polymorphism in the adult females10,11. Furthermore, the juvenile spiders in both species possess very different colour patterns from the adults. However, adult female N. vitiana is easily distinguished from other members of the T. antipodiana species-group by the possession of a red-brown sternum4,12.

Nephila pilipes (Fabricius, 1793) is distributed from India to China, Vietnam, Philippines, and Australia2. T. antipodiana occurs in China, Philippines to New Guinea, Solomon Islands, and Australia (Queensland), whereas N. vitiana (T. vitiana in the present study) is confined to Indonesia, Fiji, and Tonga2.

Of the Nephila and Trichonephila taxa, only the complete mitochondrial genome of T. clavata (previously N. clavata) has been published13 and is available in the GenBank. There is no report on the phylogenomics of T. vitiana. In view of the lack of study on the mitogenome of the Nephilidae/Nephilinae, and the conflicting systematic status, we sequenced and annotated the complete mitogenomes of N. pilipes, T. antipodiana and T. vitiana to determine their features and phylogenetic relationship. Therefore, this study aims to elucidate the relationship of Nephila and Trichonephila species and support the taxon vitiana as a valid species of the genus Trichonephila.

Results and discussion

Mitogenome features

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).

Table 1 Gene order and features of mitochondrial genome of Nephila pilipes (NP), Trichonephila antipodiana (TA), Trichonephila vitiana (TV, previously N. vitiana) and Trichonephila clavata (TC, NC_008063). CR, control region; size in bp; minus sign indicates overlap.
Full size table
Figure 1
figure 1

Complete mitogenomes of Nephila pilipes, Trichonephila antipodiana, T. vitiana (previously N. vitiana) and T. clavata with BRIG visualization showing the protein-coding genes, rRNAs and tRNAs. GC skew is shown on the outer surface of the ring whereas GC content is shown on the inner surface. The anticodon of each tRNAs is shown in parentheses. Figure generated by BRIG Development version (0.95-dev.0004) (http://brig.sourceforge.net/).

Full size image

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.

Table 2 A + T content (%), AT and GC skewness of the Nephila and Trichonephila mitogenomes. NP, Nephila pilipes; TA, Trichonephila antipodiana; TV, Trichonephila vitiana (previously N. vitiana); TC, Trichonephila clavata NC_008063.
Full size table

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).

Figure 2
figure 2

Amino acid frequency (A) and relative synonymous codon usage (B) of PCGs in the Nephila and Trichonephila mitogenomes generated using MEGAX (https://www.megasoftware.net/). NP, Nephila pilipes; TA, Trichonephila antipodiana; TV, Trichonephila vitiana (previously N. vitiana); TC, Trichonephila clavata.

Full size image

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.

Figure 3
figure 3

Box plot for pairwise divergence of Ka/Ks ratio (mean ± SD, and range) for 13 PCGs of Nephila (n = 1) and Trichonephila (n = 3) mitogenomes generated using DnaSP6.0. (http://www.ub.edu/dnasp/).

Full size image

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).

Figure 4
figure 4

Cloverleaf structure of the 22 inferred tRNAs in the mitogenomes of Nephila and Trichonephila mitogenomes obtained from MITOS web-server (http://mitos.bioinf.uni-leipzig.de/index.py).

Full size image

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.

Control region

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.

Figure 5
figure 5

Fifteen tandem repeats of ATAGA motif with TATATACATAT stretch (except one each with TAT, TATGTACATAT, and TATATACATAA) in the control region of Trichonephila clavata checked using Tandem Repeats Finder (http://tandem.bu.edu/trf/trf.html).

Full size image

Phylogenetic analysis

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.

Figure 6
figure 6

Bayesian inference phylogenetic tree based on (a) 13 PCGs and 2 rRNA genes, (b) 13 protein-coding genes, and (c) 2 rRNA genes of the whole mitogenomes of Nephila, Trichonephila and other araneid taxa with Tetragnatha taxa as outgroup. Trichonephila vitiana (previously Nephila vitiana); Numeric values at the nodes are Bayesian posterior probabilities. Figures generated by Mr Bayes v.3.1.2 (https://nbisweden.github.io/MrBayes/download.html).

Full size image

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.

Conclusion

The whole mitogenomes of N. pilipes, T. antipodiana and T. vitiana (previously N. vitiana) possess 37 genes (13 protein-coding genes, two rRNA and 22 tRNA genes), a non-coding control region and intergenic spacer and overlap sequences. Most of the tRNAs have aberrant clover-leaf secondary structure, including loss of TΨC stem and DHU arm as well as truncated and mismatched (lacking well-paired) aminoacyl acceptor stem. The gene arrangement is identical to those of other araneid mitogenomes. Based on 13 protein-coding genes (PCGs) and 15 mitochondrial genes (13 PCGs and two rRNA genes), Nephila and Trichonephila form a clade distinctly separated from the other araneid subfamilies/genera. T. antipodiana and T. vitiana are closer related to each other than to another member T. clavata of the same lineage, and this lineage is separated distinctly from N. pilipes, supporting the placement of vitiana as a member and valid species of Trichonephila. The present study on the mitogenomes of limited taxonomic sampling reveals similar genetic distance between Nephila, Trichonephila and six other araneid genera, lending support for consideration of Nephila and Trichonephila as members of the family Araneidae. A more extensive taxonomic sampling of Nephila and Trichonephila species and related taxa is needed to reconstruct a robust phylogeny based on complete mitogenomes.

Materials and methods

Sample collection

Adult female spiders were collected from their webs with an insect sweep net. They were preserved in absolute ethanol and stored in − 20 °C freezer until use for DNA extraction. N. pilipes and T. antipodiana were collected in Kelantan, Peninsular Malaysia (6.1254° N, 102.4253° E), and T. vitiana from Lombok, Indonesia (8.6510° S, 116.3249° E). The Nephila and Trichonephila spiders are not endangered or protected by law. No permits are needed to study these spiders.

Mitochondrial DNA extraction, sample preparation and genome sequencing

The extraction of mitochondrial DNA was performed as previously described41. The purified mitochondrial DNA was quantified using Qubit dsDNA High Sensitivity Assay Kit (Life Technologies, USA) and normalized to a final concentration of 50 ng for sample and library preparation using Nextera DNA Sample Preparation Kit. Size estimation of the library was performed on a 2100 Bioanalyzer using High Sensitivity DNA analysis kit (Agilent Technologies) and a real-time quantification of the library was carried out in an Eco Real-Time PCR System using KAPA Library Quantification Kit. The library was sequenced using the Illumina MiSeq Desktop Sequencer (2 × 150 bp paired-end reads) (Illumina, USA)42.

Analysis of mitogenome

Raw sequence reads were obtained from the MiSeq system in FASTQ format. The overall quality of the sequences was assessed from their Phred scores using FastQC software43. Ambiguous nucleotides and raw sequence reads with lower than Q20 Phred score were trimmed and removed using CLC genomic workbench v.7.0.4 (Qiagen, Germany). Quality-filtered DNA sequences were mapped against the reference mitogenome T. clavata (NC_008063), before a de novo assembly was performed on the mapped DNA sequences. Contigs larger than 13 kbp were extracted for a BLAST search against NCBI nucleotide database to identify the mitochondrial genome of the spider species41. On the other hand, demultiplexed raw sequence reads that were free of sequencing adapter were subjected for de novo assembly using NOVOplasty with different lengths of k-mer44. The assembled genomes from both softwares were aligned and examined for terminal repeats to evaluate their circularity and completeness. The mitogenome sequences of N. pilipes, T. antipodiana and T. vitiana (previously N. vitiana) have been deposited in GenBank under the accession numbers MW178204, MW178205 and MW178206, respectively.

Gene annotation, visualization and comparative analysis

The assembled mitogenomes were submitted to MITOS web-server (http://mitos.bioinf.uni-leipzig.de/index.py) for an initial gene annotation45. The coding regions of protein coding genes (PCGs), transfer RNAs (tRNAs) and ribosomal RNAs (rRNAs) were further validated using nucleotide-nucleotide BLAST (BLASTn) and protein–protein BLAST (BLASTp)46 against the reference mitogenome of T. clavata (NC_008063). For tRNA genes that were not identified, we extracted the DNA sequences of their putative coding regions for an additional Infernal prediction with maximum overlap increased to 5026. The gene boundaries as well as the start and stop codons of PCGs were determined following multiple sequence alignment using ClustalW47. The overlapping and intergenic spacer regions were curated manually21. The nucleotide composition, amino acid frequency and relative synonymous codon usage (RSCU) in the complete mitogenomes were calculated in MEGA X48. The ratios of non-synonymous substitutions (Ka) and synonymous (Ks) substitutions for all PCGs were estimated in DnaSP6.049. The skewness of the mitogenomes was determined from formulae: AT skew = (A − T)/(A + T) and GC skew = (G − C)/(G + C)50. Inverted repeats or palindromes in the control region were checked using Tandem Repeats Finder (http://tandem.bu.edu/trf/trf.html)51. The circular mitogenomes of the spiders were visualized using Blast Ring Image Generator (BRIG)52.

Phylogenetic analysis

The complete mitogenomes of T. clavata and Araneidae available from GenBank (Table S1) were used for phylogenetic comparison. Tetragnatha maxillosa NC_025775 and Tetragnatha nitens NC_02806816 were used as outgroup taxa. In addition to T. clavata (NC_008063), 16 araneid mitogenomes available in the GenBank were retrieved for phylogenetic analysis (Supplementary Table S1). The mitogenomes of Tetragnatha maxillosa (NC_025775) and T. nitens (NC_028068) were selected as outgroups. The nucleotide and amino acid sequences of 13 PCGs and the nucleotide sequences of 2 rRNA genes of all mitogenomes were extracted for analysis. MAFFT was used for alignment of the individual nucleotide and amino acid sequences of PCG and rRNA gene sequences53. The poorly aligned and highly divergent regions were trimmed.

Alignments of individual genes were concatenated into five datasets: (1) nucleotide sequences of 13 PCGs; (2) nucleotide sequences of two rRNA genes; (3) nucleotide sequences of 15 mt-genes (13 PCGs, 2 rRNA genes); (4) amino acid sequences of 13 PCGs; (5) 13 PCGs with the third codon position excluded. The datasets were imported into PhyloSuite54 for phylogenetic analysis. The best-fit nucleotide substitution models for maximum likelihood (ML) analysis were determined using ModelFinder55 based on the Bayesian information criterion56. A ML analysis was performed using IQ-tree57 incorporated in PhyloSuite under ultrafast bootstrap algorithm with 10,000 replicates. The phylogenetic trees constructed were visualized in MEGA X48.

Kakusan v.358 was used to determine the best-fit nucleotide substitution models for Bayesian Inference (BI) analyses using the Bayesian Information Criterion56. Bayesian analyses were conducted using the Markov chain Monte Carlo (MCMC) method via MrBayes v.3.1.259, with two independent runs of 2 × 106 generations with four chains, and with trees sampled every 200th generation. Likelihood values for all post-analysis trees and parameters were evaluated for convergence and burn-in using the “sump” command in MrBayes and the computer program Tracer v.1.5 (http://tree.bio.ed.ac.uk/software/tracer/). The first 200 trees from each run were discarded as burn-in (where the likelihood values were stabilized prior to the burn-in), and the remaining trees were used for the construction of a 50% majority-rule consensus tree. Phylogenetic trees were viewed and edited by FigTree v.1.460.