Introduction

The genus Copera Kirby, 1890 is a member of the subfamily Platycnemidinae in the family Platycnemididae1,2,3. It is represented by nine species worldwide4. Another genus of the subfamily is Platycnemis Burmeister, 1839, with 30 species4. In Malaysia, only the genus Copera is represented from the subfamily Platycnemidinae5,6. Three species occur in Peninsular Malaysia – C. ciliata (Selys, 1863), C. marginipes (Rambur, 1842) and C. vittata (Selys, 1863); two of these (C. marginipes and C. vittata) also occur in Sabah-Sarawak (Borneo).

Taxonomic uncertainty has been noted at the species and generic level1,2,3,7, particularly the generic status of the component genera (Copera and Platycnemis) from the subfamily Platycnemidinae3. In addition to the generic status, the taxonomic status of some species of Copera has been unclear. Before 1984, C. annulata and C. ciliata were considered the same species8. Recently, C. ciliata remains a synonym (as Psilocnemis ciliata Selys, 1863) of A. annulata in Korean Zygoptera9, indicating the occurrence of C. annulata in Malaysia. Taxonomic uncertainty of C. tokyoensis is reflected by the non-monophyly of its mitochondrial gene (COI-COII) genealogy with C. annulata10.

The most widespread species, C. vittata, is represented by no fewer than seven subspecies occurring in different parts of South-East Asia11. C. vittata is believed to contain several distinct species12. Red and black-legged forms occur in Borneo5. Such variants have been regarded as possibly representing separate species6.

The present study examined the DNA sequences of mitochondrial COI, COII, 16S rRNA and nuclear 28S rRNA genes in the three component species of the genus Copera in Malaysia. Additionally, members of the genus Coeliccia (Platycnemididae, Calicnemiinae) and members of the genus Prodasineura of the Protoneuridae were included for comparison.

Results

Sequence alignment and statistics

The COI and 16S rDNA nucleotide sequences appeared to be more variable and parsimony informative among all the data sets as shown by the statistics of the MP analyses. The consistency indices (CI) for COI, COII, 16S rDNA, 28S rDNA and COI + COII + 16S rDNA + 28S rDNA nucleotide sequences were 0.5685, 0.5042, 0.6880, 0.9157 and 0.7537, respectively; whereas the respective retention indices (RI) were 0.8736, 0.8290, 0.8696, 0.9721 and 0.8976.

Genetic divergence

The uncorrected p-distances of Copera and its related taxa based on COI, COII, 16S rDNA, 28S rDNA and COI + COII + 16S rDNA + 28S rDNA are summarised in Supplementary Table 1. Based on combined COI, COII, 16S rDNA and 28S rDNA sequences, the intraspecific p-distance varied from 0% to 1.08% (Supplementary Table 1e). The interspecific p-distance was many times larger: 9.41% to 12.82% for congeneric species of Copera; 13.90% to 15.39% between the genera Copera and Coeliccia; 13.66% to 14.79% between the genera Copera and Prodasineura; and 14.43% to 14.46% between the genera Coeliccia and Prodasineura (Supplementary Table 1e).

Phylogenetic relationships based on 28S rDNA nucleotide sequences

The annulata group of the genus Copera (C. ciliata) clustered with Platycnemis pennipes and was distinctly separated from the marginipes group (C. marginipes and C. vittata) (Fig. 1). The yellow-legged (CVIT3) and red-legged (CVIT1 and CIT2) morphs of C. vittata shared identical sequences.

Figure 1
figure 1

Phylogeny of the genus Copera and platycnemine subfamilies based on 28S rDNA nucleotide sequences.

(a) Numeric values at nodes are arranged in order of ML bootstrap support/Bayesian posterior probabilities. (b) Numeric values at nodes are arranged in order of MP bootstrap support/NJ bootstrap support.

Full size image

Phylogenetic relationships based on combined COI, COII and 16S rDNA nucleotide sequences

The annulata group (C. ciliata) of the genus Copera was distinctly separated from the marginipes group (C. marginipes and C. vittata) (Fig. 2). The yellow-legged (CVIT3) and red-legged (CVIT1 and CVIT2) morphs of C. vittata grouped in a highly supported clade in all analyses, indicating their genetic similarity. The Calicnemiinae (Coeliccia albicauda) appeared to be non-monophyletic with respect to the Platycnemidinae and the Disparoneurinae (Protoneuridae) showed a closer relationship with the Platycnemidinae.

Figure 2
figure 2

Phylogeny of the genus Copera and platycnemine subfamilies based on combined COI + COII + 16S rDNA nucleotide sequences.

Numeric values at nodes are arranged in order of ML bootstrap support/MP bootstrap support/NJ bootstrap support/Bayesian posterior probabilities.

Full size image

Phylogenetic relationships based on combined COI, COII, 16S and 28S nucleotide sequences

The annulata group (C. ciliata) of the genus Copera was distinctly separated from the marginipes group (C. marginipes and C. vittata) (Fig. 3). The yellow-legged morph (CVIT3) of C. vittata was genetically similar to the red-legged morph (CVIT1 and CVIT2) and they were highly supported as monophyletic in all analyses. Calicnemiinae (Coeliccia albicauda) was not monophyletic with respect to the Platycnemidinae and the Disparoneurinae (Protoneuridae) showed a closer relationship with the Platycnemidinae.

Figure 3
figure 3

Phylogeny of the genus Copera and platycnemine subfamilies based on combined COI + COII + 16S rDNA + 28S rDNA nucleotide sequences.

Numeric values at nodes are arranged in order of ML bootstrap support/MP bootstrap support/NJ bootstrap support/Bayesian posterior probabilities.

Full size image

Discussion

Species identification based on morphological characters has proven to be problematic in sibling and polymorphic odonate taxa and in other organisms. Morphological characters have also posed problems at higher taxonomic levels. Currently, molecular sequence data are used for determining the systematic status and phylogenetic relationship at various taxonomic levels. The mitochondrial COI, COII and 16S rRNA genes have been commonly used to study the phylogenetics of odonate species10,13. Additionally, the slower-evolving nuclear 28S rRNA gene has been used for determining odonate phylogeny13.

Here, the genetic similarity of the yellow-legged and red-legged forms of C. vittata indicates that these two morphs are conspecific rather than members of a species complex. Female-limited colour polymorphism is common in adult odonates14. By contrast, male-limited polymorphisms are not as common15. As in the larvae of Ceriagrion chaoi16, the colour morphs of C. vittata are not sex-limited.

It can be expected that the topology of the phylogenetic trees that were generated by different methods may vary (Fig. 1 and Supplementary Fig. 1). Such variation has been reported for other organisms, e.g., in filarial parasites17 and libellulid dragonflies18. However, the phylogenetic trees produced from the combined analyses of the mitochondrial encoded COI + COII + 16S rDNA and the COI + COII + 16S rDNA + nuclear encoded 28S rDNA generated by ML, MP, NJ or BI methods are concordant (Figs. 2, 3). Among the four markers used, COI appears to be the most variable (based on the genetic p-distance and support values of the phylogenetic trees) and is thus suitable as the single marker of choice for species differentiation and phylogenetic analysis of Odonata.

The present molecular data (COI, COII, 16S rDNA and 28S rDNA nucleotide sequences) clearly separate the marginipes group (C. marginipes and C. vittata), with coloured legs and the tibiae only moderately distended, from the annulata group (including C. ciliata), with white legs and greatly dilated tibiae (Figs. 1–3). A genetic difference between C. marginipes and C. annulata has also been reported based on the DNA sequences of the large and small subunit nuclear and mitochondrial ribosomal RNAs and part of the nuclear EF-1α19. Additionally, the larvae of C. marginipes and C. vittata possess fringes of long filaments at the margins of the caudal lamellae; this structure is not present in C. ciliata6. Additionally, ghost forms occur in the immatures of C. marginipes and C. vittata. All lines of evidence support the conclusion that the present component species of the genus Copera belong to two distinct genetic lineages that most likely warrant separate generic status.

Phylogenetic analyses based on COI, COII and 16S rDNA nucleotide sequences invariably indicate that the annulata group containing C. ciliata is more closely related to the genus Platycnemis than the marginipes group (Fig. 2, Supplementary Figs. 1–3). This relationship is reflected by the separation of C. marginipes from the grouping of C. annulata with Platycnemis pennipes based on the large and small subunit nuclear and mitochondrial ribosomal RNAs and part of the nuclear EF-1α19. Based on molecular evidence, the annulata group of the genus Copera should perhaps be placed in the genus Platycnemis as originally indicated1.

The family Platycnemididae, as currently delimited, is not monophyletic. Based on mitochondrial genes (COI, COII and 16S rRNA), the genus Coeliccia, a member of the subfamily Calicnemiinae (Platycnemididae), is not recovered with the Platycnemidinae. The Disparoneurinae of the ‘Protoneuridae’ (represented by Prodasineura spp.) shows a closer relationship to the Platycnemidinae than to the Calicnemiinae. However, the phylogenetic relationships of Platycnemidinae and Calicnemiinae are not concordant based on mitochondrial genes (COI, COII and 16S rRNA) (Fig. 2, Supplementary Figs. 1–3) or the nuclear 28S rRNA gene (Fig. 1). This discrepancy reflects the need to use multiple genes and extensive taxon sampling to arrive at a more accurate phylogenetic relationship.

Nucleotide sequences of the nuclear ribosomal genes 5.8S, 18S and ITS1 and 2 indicate that the Platycnemididae is non-monophyletic because of several representatives of uncertain placement, but the true Platycnemididae and the non-American protoneurids are closely related20. An earlier study from 122 phylogenetically informative characters (skeletal morphology and wing venation of adults and a few larval characters) also indicates that the family Platycnemididae is not demonstrably monophyletic21.

The Paleotropical component of the Protoneuridae appears to be more closely related to the Platycnemididae and the Isostictidae22. Based on the finding of the Old World disparoneurine protoneurids nesting within the Platycnemididae and well-separated from the New World protoneurine Neoneura aaroni, it has been suggested that the Disparoneurinae be regarded as a subfamily of the Platycnemididae19. The present analysis, involving Prodasineura spp., Nososticta solida and Phylloneura westermanni of the Disparoneurinae (‘Protoneuridae’), concurs with previous findings of their close relationship with the Platycnemididae.

In summary, the yellow-legged and red-legged forms of C. vittata are most likely conspecific. The present dataset supports the inclusion of the annulata group of the genus Copera (C. ciliata and C. annulata/C. tokyoensis) in the genus Platycnemis and Disparoneurinae of the Old World ‘protoneurids’ as a subfamily of Platycnemididae. The Disparoneurinae appear to be more closely related to the Platycnemidinae than to the Calicnemiinae. The inclusion of the Disparoneurinae as a subfamily of the Platycnemididae renders Platycnemididae monophyletic.

Methods

Ethics statement

No specific permits were required for the described field studies. The damselflies were collected in open ditches and ponds and not from any national parks or protected areas. No specific permissions were required and the damselflies are not endangered or protected species.

Specimens

Specimens were collected using sweep nets or plastic bags. All three Copera species inhabit sluggish channels and shallow pools in swampy areas. They were identified with established literature5,12. Additionally, Coeliccia albicauda (Förster, 1907), a member of the Calicnemiinae (Platycnemididae) and three species of Prodasineura (Protoneuridae, Disparoneurinae) were included for comparison. Two species of Orthetrum (Anisoptera) were used as an outgroup. Details of the species studied are listed in Table 1.

Table 1 Nucleotide sequences of COI, COII, 16S rRNA and/or 28S rRNA genes for 55 taxa of odonates used in the present study. Orthetrum testaceum and Orthetrum glaucum were used as outgroups. NA, not available
Full size table

DNA extraction, polymerase chain reaction and sequencing

Genomic DNA was extracted and PCR amplification was performed as described in Lim et al23. except with variations in annealing temperature for different primers. The primers and annealing temperature for PCR were: COI – COS2265 (forward): GCACAAGAAAGAGGGAAAAAAGA, COA3625 (reverse): GCCCCACAAATTTCGGAACATTG, at 50°C10,24; COII – C2-J-3102: AAATGGCAACATGAGCACAAYT, TK-N-3773: GAGACCAGTACTTGCTTTCAGTCATC, at 50°C25 16S – LR-J-13756 (16S-F): TAGTTTTTTTAGAAATAAATTTAATTTA, LR-N-13308 (16S-R): GCCTTCAATTAAAAGACTAA, at 42°C26 and Hym_16S_F: TTGACTGTACAAAGGTAGC, Hym_16S_R: GATATTACGCTGTTATCCC, at 50°C27 and 28S rDNA – 28sf, 5′-AAGGTAGCCAAATGCCTCATC-3′; 28sr, 5′-AGTAGGGTAAAACTAACCT-3′ at 50°C28.

The PCR amplicons were assayed by electrophoresis on 1.0% agarose mini gels stained with SYBR Safe DNA gel stain (Invitrogen, USA) and visualised under UV light. The amplicons were isolated and purified using the LaboPassTM PCR purification kit (Cosmo Genetech, South Korea). The purified PCR products were sent to a commercial company for sequencing. Samples were sequenced using BigDyeH Terminator v3.1 Sequencing Kit and analysed on an ABI PRISMH 377 Genetic Analyser.

DNA sequences from GenBank

To elucidate the phylogenetic relationship among the different species of Copera and related taxa, sequences generated from this study were combined with GenBank sequences (Table 1) to construct phylogenetic trees.

Genetic divergence

To assess the species level variation of Copera and related taxa, selected specimens were used to measure the uncorrected (p) pairwise genetic distances using PAUP* 4.0b10 software29. All individual markers, combined mitochondrial markers COI + COII + 16S rDNA and combined COI + COII + 16S rDNA + 28S rDNA were used to estimate uncorrected (p) pairwise genetic distances.

Sequence alignment and phylogenetic analysis

The COI, COII, 16S rDNA and 28S rDNA nucleotide sequences were initially aligned using the CLUSTAL X program30 and subsequently manually aligned. The combined COI + COII + 16S rDNA and COI + COII + 16S rDNA + 28S rDNA nucleotide sequences were also analysed to better understand the systematic relationships among different Copera species and related taxa. To investigate the utility of combining sequences from different molecular markers, statistical congruence was tested using a partition homogeneity test (PHT)31,32. The PHT was performed in PAUP* 4.0b1029 using 100 replicates and the heuristic standard search options.

Maximum likelihood (ML) analysis was performed via Treefinder version October 200833. Bayesian (BI) analysis was performed using MrBayes 3.1.234. The best fit nucleotide substitution model was determined using KAKUSAN v.335, which also generated input files for ML and BI. Best fit models were evaluated using the corrected Akaike Information Criterion36,37 for ML and the Bayesian Information Criterion (BIC) with significance determined by Chi-square analysis.

The best model for COI, combined COI + COII + 16S rDNA and combined COI + COII + 16S rDNA + 28S rDNA was the general time-reversible (GTR) model of DNA evolution with a gamma shape parameter (G); the best model for COII was J2 with a gamma shape parameter (G); the best model for mitochondrial 16S rDNA and nuclear 28S rDNA was the HKY model with a gamma-shaped parameter (G).

ML analyses were performed with 1000 bootstrap replicates. Two parallel runs were performed in MrBayes using four Markov chain Monte Carlo (MCMC) chains. One million MCMC generations were run, with convergence diagnostics calculated every 1000th generation for monitoring the stabilisation of log likelihood scores. Trees in each chain were sampled every 100th generation. A 50% majority rule consensus tree was generated from the sampled trees after discarding the first 20%.

Maximum Parsimony (MP) analyses were performed using PAUP* 4.0b1029 using a heuristic search with 100 random sequence addition replicates and a tree bisection reconnection (TBR) branch-swapping algorithm. Gaps in the alignment were treated as missing data. All characters were treated as unordered and equally weighted, the Multrees option active and branches with a maximum length of zero collapsed to yield polytomies. To assess support for the resulting nodes, bootstrap percentage (BP) was computed with 1000 replications using one random taxon addition under the heuristic search method with TBR swapping. For the datasets that yielded more than one tree, the trees sharing the same topology with ML and BI analyses were chosen.