Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Transcriptome profiling with focus on potential key genes for wing development and evolution in Megaloprepus caerulatus, the damselfly species with the world's largest wings

  • Wiebke Feindt ,

    Roles Conceptualization, Funding acquisition, Investigation, Methodology, Writing – original draft

    Affiliations University of Veterinary Medicine Hannover, ITZ—Division of Ecology and Evolution, Hannover, Germany, Leibniz University Hannover, Hannover, Germany

  • Sara J. Oppenheim,

    Roles Methodology, Writing – review & editing

    Affiliation American Museum of Natural History, Sackler Institute for Comparative Genomics, New York, NY, United States of America

  • Robert DeSalle,

    Roles Conceptualization, Funding acquisition, Writing – review & editing

    Affiliation American Museum of Natural History, Sackler Institute for Comparative Genomics, New York, NY, United States of America

  • Paul Z. Goldstein,

    Roles Writing – review & editing

    Affiliation Systematic Entomology Laboratory (USDA-ARS), National Museum of Natural History, Washington, DC, United States of America

  • Heike Hadrys

    Roles Conceptualization, Funding acquisition, Writing – review & editing

    Affiliations University of Veterinary Medicine Hannover, ITZ—Division of Ecology and Evolution, Hannover, Germany, American Museum of Natural History, Sackler Institute for Comparative Genomics, New York, NY, United States of America, Yale University, Department of Ecology & Evolutionary Biology, New Haven, Connecticut, United States of America


The evolution, development and coloration of insect wings remains a puzzling subject in evolutionary research. In basal flying insects such as Odonata, genomic research regarding bauplan evolution is still rare. Here we focus on the world’s largest odonate species—the “forest giant” Megaloprepus caerulatus, to explore its potential for looking deeper into the development and evolution of wings. A recently discovered cryptic species complex in this genus previously considered monotypic is characterized by morphological differences in wing shape and color patterns. As a first step toward understanding wing pattern divergence and pathways involved in adaptation and speciation at the genomic level, we present a transcriptome profiling of M. caerulatus using RNA-Seq and compare these data with two other odonate species. The de novo transcriptome assembly consists of 61,560 high quality transcripts and is approximately 93% complete. For almost 75% of the identified transcripts a possible function could be assigned: 48,104 transcripts had a hit to an InterPro protein family or domain, and 28,653 were mapped to a Gene Ontology term. In particular, we focused on genes related to wing development and coloration. The comparison with two other species revealed larva-specific genes and a conserved ‘core’ set of over 8,000 genes forming orthologous clusters with Ischnura elegans and Ladona fulva. This transcriptome may provide a first point of reference for future research in odonates addressing questions surrounding the evolution of wing development, wing coloration and their role in speciation.


The bauplan evolution of the Pterygota (flying insects) is one of the major challenging subjects of evolutionary research. Although the unique appearance of wings in Hexapods has led to the greatest adaptive radiations in the animal kingdom, the precise developmental mechanisms and their evolution are yet not fully understood [1]. A wide range of research is focusing on wing development, shape and coloration and their role in speciation, but so far most research has been limited to more derived model systems such as Drosophila sp., Tribolium sp. and some Lepidoptera [17].

Today progress in high throughout sequencing, advancing analytical methods and an easy access to next generation sequence data, makes integrative approaches achievable for non-model organisms [710]. Specifically, transcriptomics are suitable because they enable simultaneously the analysis of expression patterns of known developmental genes and the identification of new candidate genes [8, 11]. Moreover, interspecific transcriptome comparisons enhance our ability to infer the mechanisms underlying homologous structural and functional changes as well as allow to detect fundamental principles and conserved features [12, 13].

Among the oldest flying insects [1416], Odonata (dragonflies and damselflies) with their exclusive set of bioindicator traits hold a key role as “non-model” organisms in ecological and evolutionary research [17, 18]. However, in the evolutionary-developmental context, the molecular basis of wing development and the evolution of morphological variation in odonate wings, so far received little attention. One particular species promising more insights into wing evolution is the Neotropical damselfly Megaloprepus caerulatus (Odonata: Zygoptera, Pseudostigmatidae), because a recent study of Megaloprepus revealed a radiation into at least three geographically separated cryptic species ([19], Fig 1A). These species show differences in wing shape, i.e. in wing width, the curvature of the lower wing margin and width of the blue wing band (Feindt et al. in prep). In addition, only the nominal species M. caerulatus shows sexual dimorphism in wing coloration. It has been described that modified expression patterns or signaling cascades are responsible for the variation in wing morphology, since such changes are associated with downstream responses to supposedly conserved wing-pattern genes [1, 2, 7, 20].

Fig 1. Thorax transcriptome of Megaloprepus caerulatus.

A) Phylogeny based on the 16S rRNA gene showing the position of Megaloprepus caerulatus within the Pseudostigmatidae using Teinobasis ariel as outgroup (cf. [19]). The NCBI accession numbers are: KF895223, DQ642987, KF895193, JQ966660, KF895130, KF895162, JQ966657, DQ642983, JQ966662. B) Exemplary illustration of a M. caerulatus larva–here of about 2.5–3 cm in length. The section between the two lines indicates the tissue used for RNA extraction. C) Difference in the number of transcripts over transcript length between the raw assembly and the filtered assembly. The filtering reduced redundancy and the amount of shorter transcripts. D) Length distribution of the final predicted open reading frames. Note whereas in plot B both axes are logarithmic; in C, only the x-axis is logarithmic.

Integrative research on the origin of morphological variation associated with diversification in odonates is rare and hampered by a shortage of primary data. ‘Omic’ studies are still at their beginning and have focused so far on three species: Enallagma hageni [21], Ischnura elegans [2225] and Calopteryx splendens [26]. Only one study addressed the importance of transcriptional information across embryogenesis to highlight gene sets involved in morphogenesis [24]. Undoubtedly there is a need to integrate developmental data into evolutionary research to—for example—obtain a broader knowledge of species and tissue specific expression patterns.

Thus, we here present a de novo transcriptome assembly from the larval thorax of M. caerulatus with the overall goal of detecting expressed genes related to wing patterning that might be relevant to the interplay between genomics, development and morphological variation. Specifically, we first focus on a high completeness of the transcriptome and catalogue the candidate wing genes in odonates found in M. caerulatus. Secondly, to portray larva-specific genes, we compared the transcriptome of M. caerulatus with that of two adult odonate species: I. elegans (Odonata: Coenagrionidae) and Ladona fulva (Odonata: Libellulidae).

Material and methods

Sample collection, RNA isolation and sequencing

One individual M. caerulatus larva (Fig 1A and 1B) was collected from a natural tree hole [27] in a lowland rain forest at the La Selva Biological Station (OTS, Organization for tropical Studies) in Costa Rica (10° 26’ N, 83° 59 W). The larva (total length = 1.96 cm) was immediately euthanized and stored in RNAlater (Thermo Fisher Scientific Inc., USA). Prior to RNA isolation, the larva was dissected on ice to isolate the thorax (including the dorso-lateral wing buds) from the head and abdomen (Fig 1B). The tissue was frozen in liquid nitrogen and ground with a pestle. Total RNA was extracted from the thorax using TRIzol reagent (Invitrogen, USA) in combination with RNeasy Micro kit (Qiagen Inc., USA) for subsequent RNA purification. Overall quality and quantity of the isolated RNA were assessed with the BioAnalyzer 2100 (Agilent Inc., USA). Although the larva was collected into RNAlater, some RNA degradation had occurred. Therefore, we used a TruSeq Stranded Total Library Preparation kit (Illumina, Inc., USA) for library preparation, with Ribo-Zero treatment to select preferentially for mRNA transcripts. The cDNA libraries were paired-end sequenced (2x125bp) on an Illumina HiSeq 2500 (Illumina, Inc.).

Read cleaning and de novo assembly

Raw sequence reads were first checked for overall quality using a Phred-like score in FastQC [28] and, based on these results, adapters and low quality reads were removed with Trimmomatic 0.33 [29] at the Q20 level. Reads containing ribosomal RNA (rRNA) sequences were erased from the dataset to avoid mis-annotation of rRNAs as putative proteins [30] using SortMeRNA version 2.0 [31]. The Kraken taxonomic sequence classification system version 0.10.5 [32] was applied to filter out prokaryotic sequences. Those reads belong potentially to microorganisms co-inhabiting tree holes or to the microbiome of the larva. Singleton reads (where only 1 member of a read pair remained after the previous cleanup steps) were further removed before assembly.

The de novo assembly was conducted using Trinity version 2.0.6 [33, 34] with default parameters except for setting the strand specific flag (RF), a read normalization, and a lower limit of 300 bp on contig size. Assembly quality and completeness were evaluated in several steps. General assembly summary statistics were calculated via [34]. As a more reliable estimator of assembly completeness we also calculated additionally the ExN50 statistic. Reads were mapped back to the assembly [35] and following Haas et al. [34] the Ex90N50 was determined. This represents the N50-value at 90% of the total normalized contigs, which is excluding contigs with a low read coverage. For an evaluation of completeness BUSCO-Benchmarking Universal Single-Copy Orthologs [36] version 1.1 was used and the RSEM-EVAL package distributed with DETONATE [37] represented our reference-free evaluation method to calculate assembly scores. Because CD-Hit [38] reduced our BUSCO scores, we finally filtered the raw assembly by applying RSEM-EVAL’s contig impact score [37]. Contigs with impact scores less or equal than zero were removed from the assembly using an in-house R script in RStudio [39] and the Bioconductor R package [40].

The cleaned raw reads are available under BioProject: PRJNA336267, BioSample: SAMN05507136 and the sequence read archive (SRA) SRR3997526. Our Trinity assembly used for all subsequent analyses is available in NCBI’s Transcriptome Shotgun Assembly database under the TSA GEXY00000000.

Gene prediction and functional annotation

Open reading frames (ORFs) from start to stop codon on a six-frame translation were identified using TransDecoder ( [34]. To further improve the ORF identification, the filtered assembly was first blasted against the arthropod data base (e-value cutoff: 1e-5) downloaded from UniProtKB [41]. This was followed by HMM (hidden Markov models) searches against the Pfam-A protein domain database [42] via Hmmer version 3.1 [43]. To maximize sensitivity, these results were retained as a basis for informing protein prediction in a second TransDecoder step (2-step prediction). The final predicted protein completeness was evaluated using BUSCO [36].

For functional annotation, initial sequence homology searches were performed with BLASTp (e-value cutoff: 1e-7) against an individually designed “insect reference data base”. This customized data base contained the arthropod protein database from UniProtKB (including SWISS-PROT and TrEMBL, [41]) and protein databases for 4 Hemiptera, 21 Hymenoptera, 3 Lepidoptera and Coleoptera, and 26 Diptera species as the closest relatives to Odonata available from NCBI (data downloaded August 3rd, 2015). Sequences without a hit were additionally blasted against the non-redundant database nr—RefSeq: NCBI Reference Sequence Database (downloaded June 14th, 2016) using BLASTp and an e-value threshold of 1e−7. Putative protein sequences and BLAST results were uploaded to Blast2GO [44, 45], where InterProScan [46] searches were carried out. The InterProScan and BLAST results were used for Gene Ontology (GO) term mapping ( [47].

Identification of key genes

The annotated transcriptome was screened for genes related to stress response, housekeeping genes, developmental genes, and genes responsible for wing development and coloration. Stress response and housekeeping genes were extracted from the annotated M. caerulatus transcriptome searching for keywords via Blast2GO [44].

To detect genes involved in insect development, reference sequences were downloaded from the Homeobox database (HomeoDB;, [48, 49]). Hereby we focused on the HOXL subclass (Hox genes and Hox-derived genes) and NKL subclass (ParaHox gene cluster), both are fractions of the largest gene class Antennapedia (ANTP class) within the homeobox genes. The HOXL subclass and NKL subclass reference sequences were blasted against the M. caerulatus transcriptome and hits were verified via local BLAST searches. In order to identify additional differences of gene expression between adults and larvae, the Hox gene and ParaHox gene cluster reference sequences were also blasted against the I. elegans (SRR1265958) and L. fulva (SRR1850403) transcriptomes (see section comparison with other Odonata).

Genes responsible for wing pigmentation and wing development including wing shape such as the wing gene regulatory network (wing-patterning network) and the four major wing developmental signaling pathways (Hedgehog: Hh, Decapentaplegic: Dpp, wingless: wg and Notch: N) were identified within the M. caerulatus transcriptome via reciprocal BLASTp searches. Thus reference sequences were downloaded from Swiss-Prot [41] or NCBI and blasted against the transcriptome. All potential positive hits were verified via a local BLAST search or inside B2GO [44].

Comparison with other Odonata

The predicted open reading frames of M. caerulatus were compared to the damselfly I. elegans (SRR1265958) and to the dragonfly L. fulva (SRR1850403). Raw reads for both species were downloaded from the NCBI’s sequence read archive (, assembled de novo [33, 34] and open reading frames predicted ( following the steps described above, again under the strict completeness control. The Trinity assemblies for I. elegans and L. fulva are available upon request.

Overlaps were determined via comparative sequence similarity applying a reciprocal BLAST search using an in-house Perl script that reverts to BLASTp with a significant e-value of 1e-7. OrthoVenn [50] was further applied to categorize the transcripts into orthologous clusters. It simultaneously annotates the clusters, which were extracted to compare among the three transcriptomes.


Sequencing and de novo transcriptome assembly

Sequencing generated more than 14.4 Gbp of raw data consisting of ~115 million 125 bp paired-end reads. The cleanup steps used to filter the raw reads reduced their number by ~2%, for a final set of 112 million high-quality reads (see Table 1 for a detailed trimming report).

Table 1. Trimming statistics using three different filtering steps.

Using Trinity [33, 34] raw reads were assembled de novo into a transcriptome containing 567,572 contigs longer than 300 bp, with an N50 value of 1,956 bp (Table 2). Using Bowtie 2 [35] read support was assessed by mapping the reads back to the assembly and found that 73% of the reads mapped back in proper pairs. The Ex90N50 statistic was 2,478 bp and therefore higher than the traditional N50 measure. To evaluate the quality of the individual contigs, we used RSEM-EVAL [37], which is displaying impact scores as an estimate of read support for each contig and its contribution to the assembly. Some 84,000 low scoring contigs were removed from the assembly, reducing the assembly size to 382,606 contigs (Fig 1C). These initial assembly evaluation steps are critical in de novo transcriptome studies, because false positives (the inclusion of misassembled contigs) will lead to errors in gene prediction, annotation, and further downstream analyses such as expression profiling. However, false negatives (the elimination of legitimate contigs) can reduce the completeness of the transcriptome; thus, evaluations should be repeated after each filter step (Table 2). The final assembly was ~93% complete based on BUSCO’s [36] arthropod reference database of 2,675 single-copy orthologs present in >90% of the species (Table 2), which is consistent with results from other recently published insect transcriptomes (e.g. [22]).

Table 2. Assembly statistics during final assembly evaluation steps.

Gene prediction and functional annotation

TransDecoder.LongORFs [34] identified about 93,000 potential open reading frames (ORFs) in the final M. caerulatus assembly. The homology-based second step retaining BLAST [51] and Pfam [52] search results in TransDecoder.Predict [34] resulted in a final set of 61,560 predicted proteins longer than 100 amino acids (Fig 1D, Table 2).

The continuous BLASTp search against our custom ‘RefSeq’ database allowed the determination of gene functions of about 73.04% of our sequences. However, the 27% that had no hit to this database and were additionally blasted against the entire non-redundant database, which produced hits for another 1%. The top hit species distribution shows the highest number of hits against the basal Hymenopterans (Symphyta) Athalia rosae and Orussus abietinus (see Fig 2A). Accuracy of our assembly and the predicted protein coding genes were supported by consistently high e-values (42% of the blast hits had e-value >1e-100; see e-value distribution in S1 Fig).

Fig 2. Functional annotation of the M. caerulatus transcriptome.

A) Distribution of top hits shows all species to which M. caerulatus had at least 100 hits to. B) Classification of the functional annotation into the three gene ontology (GO) categories: molecular function (MF), cellular component (CC), and biological process (BP) at GO level 5. Displayed are the distribution of the top 20 GO terms and the number of sequences with the corresponding assignment.

Our gene ontology (GO) term assignment [47] via Blast2GO [44] revealed 78% of the putative genes had an InterPro hit, and 46% had a GO annotation (Fig 2B). Longer sequences were more likely to be annotated than shorter ones (see S2 Fig): approximately 50% of the sequences >200 amino acids were annotated, and almost all of those >500 AA.

Identification of candidate genes

Stress response genes.

Environmental studies on insects frequently focus on heat shock proteins (HSPs), a large and highly conserved gene family involved in protein metabolism and insect survival through their roles in protein folding and repair (e.g. [53, 54]). Under cellular stress, HSP expression levels increase and their assessment in natural environments can help identify stress adaptation under climate change or habitat fragmentation [25]. We identified 23 HSP genes and 3 general stress response genes (see S1 File for the AA sequences and gene names).

Housekeeping genes.

Basic cell functions are controlled by housekeeping genes, expressed in every tissue under most experimental conditions (e.g. [55]) and these serve as a baseline for normalizing quantitative real-time PCR or RNA-Seq gene expression experiments. We identified the majority of housekeeping genes commonly found in insects, including the ribosomal proteins S18 and L13a as well as ATP and actin genes (S2 File).

Developmental genes.

Hox genes are of particular interest as they encode transcription factors that modulate bauplan development during early embryogenesis and determiners of cell fate (e.g. [56]). With the focus on the Antp-class genes, we identified M. caerulatus orthologs for only three Hox genes, including: Antennapedia (Antp, Hox6-8), Ultrabithorax (Ubx, Hox6-8) and Sex combs reduced (Scr, Hox5); and one ParaHox gene (Nedx). Hox genes were also identified in the I. elegans and L. fulva transcriptomes. An alignment of these first full-length homeodomain amino acid sequences for our 16 detected Hox and ParaHox genes in Odonata is shown in S3 File.

Wing genes.

Insect wing development is controlled by the wing-patterning network (wing gene regulatory network) in which the Hox genes Scr and Ubx act jointly with cell signaling molecules, selector genes and transcription factors to modulate wing morphogenesis, differentiation and growth [2, 57]. These signaling molecules are further grouped into four main signaling pathways: Hedgehog (Hh), Decapentaplegic (Dpp), wingless (wg) and Notch (N) constituting overarching structures [5, 58, 59]. In addition, the wing-patterning network influences wing coloration, as developmental gene expression determines the activity of subsequent pigment genes (e.g. [2]).

We were able to identify most representatives of the pigmentation genes, 14 of the 21 genes described so far for the wing-patterning network, and the four main developmental signaling pathways. For the latter, we could discover 6 genes related to accurate cell differentiation and growth in the Hedgehog pathway, 8 genes that are included in the cell fate determination by the Notch pathway, 9 genes associated with the development of wings in the wingless pathway and 3 genes could be connected to dorsal/ventral patterning and development of the wing epithelia in the Decapentaplegic pathway (see Table 3 for a complete overview of wing genes and their related pathways and S4 File for the corresponding amino acid sequences). The reciprocal BLASTp searches against well-annotated patterning genes revealed an additional 19 genes described in wing coloration and general pigmentation studies [2, 4, 6064]. We detected pigmentation genes from the melanin pathway (yellow, black, tan, pale), the pteridine pathway (henna, rosy, prat), the ommochrome pathway (vermillion, white, scarlet) and pigment granule genes (dor, garnet). We also found phenol oxidases (PO), which contribute to melanization among other functions [6, 64] and the Ecdysone receptor (EcR), a hormone involved in wing growth [65, 66].

Table 3. Selection of genes responsible for wing development in insects.

Genes of the four major signaling pathways and functionally related genes (“Others”) are arranged in the according columns. Members of the wing-patterning network [2, 57] known to be associated with a specific signaling pathway are shown in the grey row [5, 9, 58, 59, 6772]. Genes identified in the M. caerulatus transcriptome are shown by bold; while the corresponding amino acid sequences and unigene IDs are given in S4 File and a description of primary gene functions are in the S1 Table. Genes that could not be identified in the M. caerulatus transcriptome are shown in black font.

Comparison with other Odonata

Although direct comparative transcriptome analyses struggle with differences in sample preparation (e.g. different tissue collection, developmental stages, etc.) and by the difficulty of accurate ortholog detection, first comparisons amongst well-annotated sequences are appropriate for a selected set of questions (e.g. [13, 73]). Here, we compared our findings with the transcriptomes of I. elegans and L. fulva to search for unique gene expression. The damselfly I. elegans belongs to the Coenagrionidae—a sister family to the Pseudostigmatidae to which M. caerulatus is associated—while the more distantly related dragonfly L. fulva belongs to the suborder Anisoptera.

Our results reflect these relationships in that the highest overall sequence similarity is represented by the 12,569 reciprocal best hits between M. caerulatus and I. elegans. In the M. caerulatus / L. fulva search, 11,136 reciprocal best hits were obtained, similar to the results for I. elegans / L. fulva (Fig 3A). The comparison of both overlapping and unique orthologous clusters for each species and species pair showed a similar result (Fig 3B). Using OrthoVenn [50] we retrieved a total of 29,464 clusters, with 8,196 clusters containing genes from all three species. Of these, a functional annotation was available for 4,810 clusters.

Fig 3. Comparisons among three odonate transcriptomes based on the open reading frames: Ladona fulva, Ischnura elegans, and M. caerulatus.

A) The overall sequence similarity identified via reciprocal blast search among transcriptomes presented in a Venn diagram shows a greater number of overlapping genes between M. caerulatus and I. elegans than between the dragonfly L. fulva and the two damselflies. B) Overlapping orthologous gene clusters (OrthoVenn). Both analyses show a similar sized overlap among species.

To gain insights into larva-specific genes we focused on the unique orthologous clusters of M. caerulatus. In total 4,589 were detected, but only 1,168 clusters had a usable annotation. Those clusters were related primarily to general cell functions such as phosphorylation (serine/threonine-protein kinases, cytochrome oxidases) and signal transduction (receptor tyrosine kinases). Among potential larva-specific transcripts, we identified the following related to wings and general development: (i) encore regulates dorso-ventral polarity in embryos and larvae; (ii) flightless-1 plays a structural role in indirect flight muscle; and (iii) krueppel is involved in gap class segmentation. Other interesting findings were the O-mannosyl-transferase 2 that is responsible for somatic muscle development and the Ryanodine receptor 44F which is involved in proper muscle function, i.e., in larval body wall muscles, and is therefore essential for larval development [74, 75].


At the base of flying insects, Odonata have a long-standing record in ecological and evolutionary research. This head start should encourage their future role as non-model systems in integrative genomic research. The first transcriptome profiling of a larval tissue from Megaloprepus caerulatus represents a step towards this direction to study wing development and evolution, and speciation.

Transcriptome assembly and functional annotation

Odonate genomes are among the larger known genomes within winged insects [76], and dragonfly and damselfly transcriptomes likewise appear to be larger than those of prominent model species such as Diptera or Lepidoptera. Our assembly of raw sequences resulted in over 500,000 putative transcripts, which were reduced via strict evaluation to 61,560 high quality protein-coding genes. This number and the ExN50 expression value reflect some redundancy, but the transcriptome size is comparable to that of other odonates (e.g. [21, 22]). Furthermore, in comparison to previous odonate studies, function could be assigned to a greater number of genes which may reflect growing resources in genomics and the use of customized and frequently updated reference databases [77].

Beyond this, increasing tissue diversity should in turn increase the number of genes sequenced and annotated, but difficulties with the assembly of heterozygous sequences can limit the quality of the reconstructed transcripts and thereby impair the reliability of BLAST results. In spite of the comparatively successful assignment of function, some 25% of the putative proteins lack annotation. Some of these genes are probably misassembled transcripts that do not actually exist or, alternatively, represent odonate- or Megaloprepus-specific proteins that simply lack homologous sequences in current databases.

Candidate genes

Little is known about developmental genes such as Hox genes or those responsible for the pathways of wing development and coloration in odonates (but see [56]). In accordance with our expectations we found three Hox genes in the larval thorax. Interestingly, six Hox representatives could be detected in the adult I. elegans (S3 File), suggesting that Hox expression may be of functional importance in adults as well as larvae. Two of the Hox genes identified in M. caerulatus are involved in important wing traits: Scr suppresses wing development in the prothorax [1], while Ubx controls hind wing identity [72] and is an important modulator in the wing-patterning gene regulatory network [2]. It acts as a selector gene, influencing morphological characters such as wing venation and regulates wingless (wg), splat (sal) and vestigial (vg) in opposing mechanisms [71]. In Drosophila it facilitates the development of halteres and in Tribolium the sclerotized fore wings [72].

The development of wings and their shape is controlled by the wing-patterning network through the modulation of gene expression [6, 78]. It was originally described in Drosphila melanogaster and is supposedly largely conserved across holometabolous insects [2, 57]. However, in hemimetabolous insect orders information on wing differentiation across larval stages is limited [79]. In the Megaloprepus transcriptome, we identified 14 genes from the wing-patterning network. So far Dpp has been described to inhibit Dll (Distal-less) in an early stage of the signaling cascade within the wing-patterning network, but later in the development of imaginal wing discs it activates omb (Optomotor-blind), sal (spalt) and vg (vestigial) to shape cell growth, vein positioning and intervein cell differentiation [2, 57]. In the pupal stage of holometabolous insects, a significant reorganization of tissues and organs takes place, while hemimetabolous insects undergo a more gradual developmental transition. Thus some of the mechanisms of wing development in Odonata most likely differ from those of holometabolous insects, and further investigation of the timing and related genes may shed light onto the developmental changes that characterize the bauplan transition to holometaboly [1].

Wing coloration in odonates is highly variable across species. Some have only a colored pterostigma or different sized wing spots, while some other species show entirely colored wings. We identified 19 genes related to insect pigmentation. Furthermore, our data showed a higher relative expression of both, the phenol oxidases and yellow in comparison to the house keeping genes (S5 File). This could be an indication of polymerization of cuticular pigments following larval molt. However, since the larva was collected from its natural environment, this remains an assumption pending controlled experiments under laboratory settings to identify the genes and pathways responsible for coloration. Targeted RNA sequencing in parallel with in situ hybridization studies would thus provide deeper insights into gene expression during the course of odonate development.

Finally, some of the wing and developmental genes that were not identified in our analysis may simply lack expression at the time of collection, but most likely may indicate modified signaling pathways. However, we suggest that the genes identified here reflect an early stage of wing development, also because the Megaloprepus larva used bore visible wing buds on its thorax. In our comparison between the larvae and the two adult odonates, we found aside of the wing genes, proteins known as essential for larval development (O-mannosyltransferase, Ryanodine receptor 44F). However, many of those ~4,500 transcripts found only in M. caerulatus may be species-specific rather than larva-specific.


Megaloprepus caerulatus has a longstanding record in ecological and evolutionary research [19, 8084]. The de novo transcriptome presented here is the first genomic resource for Neotropical odonates and may hopefully enhance future genomic research in odonates. For M. caerulatus comparative studies at different developmental stages involving the newly discovered species might reveal mechanisms of wing shape divergence, demographic patterns, micro-evolutionary changes and genomic regions under selection in changing environments.

Because of its close ecological association with Neotropical old growth rainforests, high vulnerability to climate shifts and forest disturbances, Megaloprepus is an effective bioindicator of the history (and future) of old growth rainforests. Genomic monitoring of key genes combined with ecological data could provide early insights into the effects of environmental changes.

Supporting information

S1 Table. Genes related to wing development and coloration in insects.

Included are the descriptions of the main functions for genes involved in pigmentation (S1A) and genes of the wing-patterning gene network (GRN; S1A) found in M. caerulatus including their Unigene IDs and FPKMs.


S1 Fig. E-value distribution for BLAST hits against Megaloprepus' predicted proteins.


S2 Fig. Length distribution of annotated sequences within the M. caerulatus transcriptome.


S1 File. Amino acid sequences of M. caerulatus stress response genes.


S2 File. Amino acid sequences of M. caerulatus house keeping genes.


S3 File. Homeobox genes (HOXL subclass and NKL subclass) homeodomain amino acid alignment for I. elegans, L. fulva and M. caerulatus.


S4 File. Amino acid sequences of wing genes found in M. caerulatus.

Included are representatives of the wing-patterning gene network and the four signaling pathways: Notch (N), wingless (wg), Decapentaplegic (Dpp) and Hedgehog (Hh), and pigmentation genes.


S5 File. In silico quantification of Megaloprepus expression levels.



We are grateful to the Ministerio de Ambiente y Energía, Costa Rica (MINAE) for processing a collecting permit and to the Comisión Nacional para la Gestión de la Biodiversidad, Costa Rica (CONAGEBIO) for permitting the use of genetic material. We thank A. Narechania, S. Mehr and K. Kamm for helpful comments. Sequencing was conducted at the New York Genome Center ( All computations were performed on resources provided by AMNH, the ITZ in Hannover and the Institute for Bioinformatics in Leipzig.

Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the USDA; USDA is an equal opportunity provider and employer.


  1. 1. Engel MS, Davis SR, Prokop J. Insect wings: the evolutionary development of nature’s first flyers. Arthropod Biology and Evolution: Springer; 2013. p. 269–98.
  2. 2. Connahs H, Rhen T, Simmons RB. Transcriptome analysis of the painted lady butterfly, Vanessa cardui during wing color pattern development. BMC genomics. 2016;17(1):1.
  3. 3. Ferguson LC, Maroja L, Jiggins CD. Convergent, modular expression of ebony and tan in the mimetic wing patterns of Heliconius butterflies. Development genes and evolution. 2011;221(5–6):297–308. pmid:22139062
  4. 4. Kronforst MR, Papa R. The functional basis of wing patterning in Heliconius butterflies: the molecules behind mimicry. Genetics. 2015;200(1):1–19. pmid:25953905
  5. 5. Liu S, Wei W, Chu Y, Zhang L, Shen J, An C. De novo transcriptome analysis of wing development-related signaling pathways in Locusta migratoria manilensis and Ostrinia furnacalis (Guenée). PloS one. 2014;9(9):e106770. pmid:25207539
  6. 6. Massey J, Wittkopp PJ. The genetic basis of pigmentation differences within and between Drosophila species. Current topics in developmental biology. 2016;119:27–61. pmid:27282023; PubMed Central PMCID: PMCPMC5002358.
  7. 7. Roux J, Rosikiewicz M, Robinson‐Rechavi M. What to compare and how: Comparative transcriptomics for Evo-Devo. Journal of Experimental Zoology Part B: Molecular and Developmental Evolution. 2015;324(4):372–82.
  8. 8. Oppenheim SJ, Baker RH, Simon S, DeSalle R. We can't all be supermodels: the value of comparative transcriptomics to the study of non‐model insects. Insect molecular biology. 2015;24(2):139–54. pmid:25524309
  9. 9. Abouheif E, Favé M-J, Ibarrarán-Viniegra AS, Lesoway MP, Rafiqi AM, Rajakumar R. Eco-evo-devo: The time has come. Ecological genomics: Springer; 2014. p. 107–25.
  10. 10. Gilbert SF, Bosch TC, Ledón-Rettig C. Eco-Evo-Devo: developmental symbiosis and developmental plasticity as evolutionary agents. Nature Reviews Genetics. 2015;16(10):611–22. pmid:26370902
  11. 11. Alvarez M, Schrey AW, Richards CL. Ten years of transcriptomics in wild populations: what have we learned about their ecology and evolution? Molecular ecology. 2015;24(4):710–25. pmid:25604587
  12. 12. Parikh A, Miranda ER, Katoh-Kurasawa M, Fuller D, Rot G, Zagar L, et al. Conserved developmental transcriptomes in evolutionarily divergent species. Genome biology. 2010;11(3):R35. pmid:20236529
  13. 13. Gerstein MB, Rozowsky J, Yan K-K, Wang D, Cheng C, Brown JB, et al. Comparative analysis of the transcriptome across distant species. Nature. 2014;512(7515):445–8. pmid:25164755
  14. 14. Misof B, Liu S, Meusemann K, Peters RS, Donath A, Mayer C, et al. Phylogenomics resolves the timing and pattern of insect evolution. Science. 2014;346(6210):763–7. pmid:25378627
  15. 15. Simon S. A Brief Guide to the evolution of flying insects. BioEssyas. 2013.
  16. 16. Simon S, Strauss S, von Haeseler A, Hadrys H. A phylogenomic approach to resolve the basal pterygote divergence. Molecular biology and evolution. 2009;26(12):2719–30. pmid:19713325
  17. 17. Córdoba-Aguilar A. Dragonflies and Damselflies: Model Organisms for Ecological and Evolutionary Research. Oxford: Oxford University Press; 2008.
  18. 18. Bybee S, Córdoba-Aguilar A, Duryea MC, Futahashi R, Hansson B, Lorenzo-Carballa MO, et al. Odonata (dragonflies and damselflies) as a bridge between ecology and evolutionary genomics. Frontiers in zoology. 2016;13(1):46.
  19. 19. Feindt W, Fincke O, Hadrys H. Still a one species genus? Strong genetic diversification in the world`s largest living odonate, the Neotropical damselfly Megaloprepus caerulatus. Conservation Genetics. 2014;15(2):469–81.
  20. 20. van’t Hof AE, Campagne P, Rigden DJ, Yung CJ, Lingley J, Quail MA, et al. The industrial melanism mutation in British peppered moths is a transposable element. Nature. 2016;534(7605):102–5. pmid:27251284
  21. 21. Shanku AG, McPeek MA, Kern AD. Functional Annotation and Comparative Analysis of a Zygopteran Transcriptome. G3: Genes| Genomes| Genetics. 2013;3(4):763–70. pmid:23550132
  22. 22. Chauhan P, Hansson B, Kraaijeveld K, de Knijff P, Svensson EI, Wellenreuther M. De novo transcriptome of Ischnura elegans provides insights into sensory biology, colour and vision genes. BMC genomics. 2014;15(1):808.
  23. 23. Chauhan P, Wellenreuther M, Hansson B. Transcriptome profiling in the damselfly Ischnura elegans identifies genes with sex-biased expression. BMC genomics. 2016;17(1):985. pmid:27905879
  24. 24. Simon S, Sagasser S, Saccenti E, Brugler MR, Schranz ME, Hadrys H, et al. Comparative transcriptomics reveals developmental turning points during embryogenesis of a hemimetabolous insect, the damselfly Ischnura elegans. Scientific Reports. 2017;7(1):2045–322.
  25. 25. Lancaster LT, Dudaniec RY, Chauhan P, Wellenreuther M, Svensson EI, Hansson B. Gene expression under thermal stress varies across a geographic range expansion front. Molecular ecology. 2016.
  26. 26. Ioannidis P, Simao FA, Waterhouse RM, Manni M, Seppey M, Robertson HM, et al. Genomic Features of the Damselfly Calopteryx splendens: Representing a Sister Clade to Most Insect Orders. Genome biology and evolution. 2017;9(2):415–30. pmid:28137743
  27. 27. Yanoviak S, Fincke O. Sampling methods for water-filled tree holes and their artificial analogues. Insect sampling in forest ecosystems Blackwell Publishing, London. 2005:168–85.
  28. 28. Andrews S. FastQC: A quality control tool for high throughput sequence data. Reference Source. 2010.
  29. 29. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. 2014:btu170.
  30. 30. Tripp HJ, Hewson I, Boyarsky S, Stuart JM, Zehr JP. Misannotations of rRNA can now generate 90% false positive protein matches in metatranscriptomic studies. Nucleic acids research. 2011;39(20):8792–802. PubMed PMID: PMC3203614. pmid:21771858
  31. 31. Kopylova E, Noé L, Touzet H. SortMeRNA: fast and accurate filtering of ribosomal RNAs in metatranscriptomic data. Bioinformatics. 2012;28(24):3211–7. pmid:23071270
  32. 32. Wood DE, Salzberg SL. Kraken: ultrafast metagenomic sequence classification using exact alignments. Genome Biol. 2014;15(3):R46. pmid:24580807
  33. 33. Grabherr MG, Haas BJ, Yassour M, Levin JZ, Thompson DA, Amit I, et al. Trinity: reconstructing a full-length transcriptome without a genome from RNA-Seq data. Nature biotechnology. 2013;29(7):644–52. PubMed PMID: 21572440; PubMed Central PMCID: PMCPmc3571712.
  34. 34. Haas BJ, Papanicolaou A, Yassour M, Grabherr M, Blood PD, Bowden J, et al. De novo transcript sequence reconstruction from RNA-seq using the Trinity platform for reference generation and analysis. Nat Protocols. 2013;8(8). Epub 1512.
  35. 35. Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nature methods. 2012;9(4):357–9. pmid:22388286
  36. 36. Simão FA, Waterhouse RM, Ioannidis P, Kriventseva EV, Zdobnov EM. BUSCO: assessing genome assembly and annotation completeness with single-copy orthologs. Bioinformatics. 2015:btv351.
  37. 37. Li B, Fillmore N, Bai Y, Collins M, Thomson JA, Stewart R, et al. Evaluation of de novo transcriptome assemblies from RNA-Seq data. Genome biology. 2014;15(12):553. pmid:25608678
  38. 38. Li W, Godzik A. Cd-hit: a fast program for clustering and comparing large sets of protein or nucleotide sequences. Bioinformatics. 2006;22(13):1658–9. pmid:16731699
  39. 39. Racine JS. RStudio: A Platform‐Independent IDE for R and Sweave. Journal of Applied Econometrics. 2012;27(1):167–72.
  40. 40. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, et al. Bioconductor: open software development for computational biology and bioinformatics. Genome biology. 2004;5(10):R80. pmid:15461798
  41. 41. Boutet E, Lieberherr D, Tognolli M, Schneider M, Bansal P, Bridge AJ, et al. UniProtKB/Swiss-Prot, the manually annotated section of the UniProt KnowledgeBase: how to use the entry view. Plant Bioinformatics: Methods and Protocols. 2016:23–54.
  42. 42. Finn RD, Coggill P, Eberhardt RY, Eddy SR, Mistry J, Mitchell AL, et al. The Pfam protein families database: towards a more sustainable future. Nucleic Acids Research. 2016;44(D1):D279–D85. pmid:26673716
  43. 43. Wheeler T, Eddy S. nhmmer: DNA homology search with profile HMMs. Bioinformatics. 2013;29(19):2487–9. pmid:23842809; PubMed Central PMCID: PMCPmc3777106.
  44. 44. Conesa A, Götz S, García-Gómez JM, Terol J, Talón M, Robles M. Blast2GO: a universal tool for annotation, visualization and analysis in functional genomics research. Bioinformatics. 2005;21(18):3674–6. pmid:16081474
  45. 45. Götz S, García-Gómez JM, Terol J, Williams TD, Nagaraj SH, Nueda MJ, et al. High-throughput functional annotation and data mining with the Blast2GO suite. Nucleic acids research. 2008;36(10):3420–35. pmid:18445632
  46. 46. Zdobnov EM, Apweiler R. InterProScan–an integration platform for the signature-recognition methods in InterPro. Bioinformatics. 2001;17(9):847–8. pmid:11590104
  47. 47. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. Gene Ontology: tool for the unification of biology. Nature genetics. 2000;25(1):25–9. pmid:10802651
  48. 48. Zhong YF, Butts T, Holland PW. HomeoDB: a database of homeobox gene diversity. Evolution & development. 2008;10(5):516–8.
  49. 49. Yf Zhong, Holland PW. HomeoDB2: functional expansion of a comparative homeobox gene database for evolutionary developmental biology. Evolution & development. 2011;13(6):567–8.
  50. 50. Wang Y, Coleman-Derr D, Chen G, Gu YQ. OrthoVenn: a web server for genome wide comparison and annotation of orthologous clusters across multiple species. Nucleic acids research. 2015;43(W1):W78–W84. pmid:25964301
  51. 51. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. Journal of molecular biology. 1990;215(3):403–10. pmid:2231712
  52. 52. Punta M, Coggill PC, Eberhardt RY, Mistry J, Tate J, Boursnell C, et al. The Pfam protein families database. Nucleic acids research. 2011:gkr1065.
  53. 53. Hendrick JP, Hartl F-U. Molecular chaperone functions of heat-shock proteins. Annual review of biochemistry. 1993;62(1):349–84.
  54. 54. Zhao L, Jones W. Expression of heat shock protein genes in insect stress responses. Invertebrate Surviv J. 2012;90:93–101.
  55. 55. Yang C, Pan H, Liu Y, Zhou X. Stably expressed housekeeping genes across developmental stages in the two-spotted spider mite, Tetranychus urticae. PloS one. 2015;10(3):e0120833. pmid:25822495
  56. 56. Hadrys H, Simon S, Kaune B, Schmitt O, Schöner A, Jakob W, et al. Isolation of hox cluster genes from insects reveals an accelerated sequence evolution rate. PloS one. 2012;7(6):e34682. pmid:22685537
  57. 57. Abouheif E, Wray GA. Evolution of the gene network underlying wing polyphenism in ants. Science. 2002;297(5579):249–52. pmid:12114626
  58. 58. Shimmi O, Matsuda S, Hatakeyama M, editors. Insights into the molecular mechanisms underlying diversified wing venation among insects. Proc R Soc B; 2014: The Royal Society.
  59. 59. Yan S-J, Gu Y, Li WX, Fleming RJ. Multiple signaling pathways and a selector protein sequentially regulate Drosophila wing development. Development. 2004;131(2):285–98. pmid:14701680
  60. 60. Kronforst MR, Barsh GS, Kopp A, Mallet J, Monteiro A, Mullen SP, et al. Unraveling the thread of nature’s tapestry: the genetics of diversity and convergence in animal pigmentation. Pigment cell & melanoma research. 2012;25(4):411–33.
  61. 61. Nishikawa H, Iga M, Yamaguchi J, Saito K, Kataoka H, Suzuki Y, et al. Molecular basis of wing coloration in a Batesian mimic butterfly, Papilio polytes. Scientific reports. 2013;3:3184. pmid:24212474
  62. 62. Reed RD, Nagy LM. Evolutionary redeployment of a biosynthetic module: expression of eye pigment genes vermilion, cinnabar, and white in butterfly wing development. Evolution & development. 2005;7(4):301–11.
  63. 63. True JR. Insect melanism: the molecules matter. Trends in ecology & evolution. 2003;18(12):640–7.
  64. 64. Wittkopp PJ, Carroll SB, Kopp A. Evolution in black and white: genetic control of pigment patterns in Drosophila. TRENDS in Genetics. 2003;19(9):495–504. pmid:12957543
  65. 65. Herboso L, Oliveira MM, Talamillo A, Pérez C, González M, Martín D, et al. Ecdysone promotes growth of imaginal discs through the regulation of Thor in D. melanogaster. Scientific reports. 2015;5:12383. pmid:26198204
  66. 66. Monteiro A, Tong X, Bear A, Liew SF, Bhardwaj S, Wasik BR, et al. Differential expression of ecdysone receptor leads to variation in phenotypic plasticity across serial homologs. PLoS Genet. 2015;11(9):e1005529. pmid:26405828
  67. 67. Bray S. Drosophila development: Scalloped and Vestigial take wing. Current biology. 1999;9(7):R245–R7. pmid:10209108
  68. 68. Kubota K, Goto S, Eto K, Hayashi S. EGF receptor attenuates Dpp signaling and helps to distinguish the wing and leg cell fates in Drosophila. Development. 2000;127(17):3769–76. pmid:10934021
  69. 69. Ou J, Deng H-M, Zheng S-C, Huang L-H, Feng Q-L, Liu L. Transcriptomic analysis of developmental features of Bombyx mori wing disc during metamorphosis. BMC genomics. 2014;15(1):820.
  70. 70. Swarup S, Verheyen EM. Wnt/wingless signaling in Drosophila. Cold Spring Harbor perspectives in biology. 2012;4(6):a007930. pmid:22535229
  71. 71. Tomoyasu Y. Ultrabithorax and the evolution of insect forewing/hindwing differentiation. Current Opinion in Insect Science. 2017;19:8–15. pmid:28521947
  72. 72. Tomoyasu Y, Wheeler SR, Denell RE. Ultrabithorax is required for membranous wing identity in the beetle Tribolium castaneum. Nature. 2005;433(7026):643–7. pmid:15703749
  73. 73. Berens AJ, Hunt JH, Toth AL. Comparative transcriptomics of convergent evolution: different genes but conserved pathways underlie caste phenotypes across lineages of eusocial insects. Molecular biology and evolution. 2014;32(3):690–703. pmid:25492498
  74. 74. Lyalin D, Koles K, Roosendaal SD, Repnikova E, Van Wechel L, Panin VM. The twisted gene encodes Drosophila protein O-mannosyltransferase 2 and genetically interacts with the rotated abdomen gene encoding Drosophila protein O-mannosyltransferase 1. Genetics. 2006;172(1):343–53. pmid:16219785
  75. 75. Sullivan KM, Scott K, Zuker CS, Rubin GM. The ryanodine receptor is essential for larval development in Drosophila melanogaster. Proceedings of the National Academy of Sciences. 2000;97(11):5942–7.
  76. 76. Ardila-Garcia A, Gregory T. An exploration of genome size diversity in dragonflies and damselflies (Insecta: Odonata). Journal of Zoology. 2009;278(3):163–73.
  77. 77. Wadi L, Meyer M, Weiser J, Stein LD, Reimand J. Impact of knowledge accumulation on pathway enrichment analysis. bioRxiv. 2016:049288.
  78. 78. Erwin DH, Davidson EH. The evolution of hierarchical gene regulatory networks. Nature reviews Genetics. 2009;10(2):141. pmid:19139764
  79. 79. Medved V, Marden JH, Fescemyer HW, Der JP, Liu J, Mahfooz N, et al. Origin and diversification of wings: Insights from a neopteran insect. Proceedings of the National Academy of Sciences. 2015;112(52):15946–51.
  80. 80. Longchamps Selys Ed. Synopsis des Agrionines. Première Légion.–Pseudostigma–. Bulletin de l’Académie Royale des Sciences, des Lettres et des Beaux—Arts de Belgique. 1860;2(10):9–27.
  81. 81. Fincke OM, Hedström I. Differences in forest use and colonization by Neotropical tree‐hole damselflies (Odonata: Pseudostigmatidae): Implications for forest conversion. Studies on Neotropical Fauna and Environment. 2008;43(1):35–45.
  82. 82. Fincke OM. The population ecology of Megaloprepus coerulatus and its effect on species assemblages in water-filled tree holes. Insect populations in theory and in practice: Springer; 1998. p. 391–416.
  83. 83. Fincke O. Behavioral ecology of the giant damselflies of Barro Colorado Island, Panama (Odonata: Zygoptera: Pseudostigmatidae). In: Quintero D, Aiello A, editors. Insects of Panama and Mesoamerica: selected studies. Oxford, UK: Oxford University Press; 1992. p. 102–13.
  84. 84. Fincke OM, editor Use of forest and tree species, and dispersal by giant damselflies (Pseudostigmatidae): their prospects in fragmented forests. Forest and dragonflies 4th WDA International Symposium of Odonatology Pensoft, Sofia; 2006.