Background & Summary

Stag beetles (Family: Lucanidae), comprise over 1,800 species and subspecies1, noted for enlarged allometry mandibles and male polymorphism2. As a holometabolous insect, they undergo a complete metamorphosis with four life stages: egg, larva, pupa, and adult3,4. The larvae primarily feed on decaying wood, while the adults are mostly nocturnal and feed on plant juices, fruits, or other decaying organic matter5,6,7. Due to their saprophagous nature, stag beetles play essential roles in the carbon and nitrogen cycles, which also makes stag beetles an important indicator species for evaluating forest ecosystems1,8,9. Based on observations of male lucanid beetles, Darwin (1871)10 noted that “The great mandibles of the male Lucanidae are extremely variable both in size and structure…and are used as efficient weapons for fighting”. It implies that individuals with larger mandibles have a better chance of defeating their rivals and winning mating rights11,12. Due to its unique and variable appearances, as well as interesting behavioral phenomena, this group has garnered the affection of many collectors, entomologists and evolutionary biologists8,10.

Stag beetles are widely distributed across various biogeographic regions and represent a group with significant nodal importance in the process of evolution13. The research on the Lucanidae family is predominantly concentrated on molecular taxonomic studies8,14,15, based on the data from nuclear gene fragments16, mitochondrial multi-gene fragments17 and mitochondrial genomes18,19. However, these data are insufficient to provide more insights into the formation and differentiation of the stag beetles’ mandibles20. Decoding high-quality reference genomes has been proven to be the cornerstone of inferring phylogeny and exploring the molecular basis behind phenotypic innovation21, e.g., the antlers of cervids22, the long tail feathers of birds23,24, and the horns of some scarabs25,26. The limited availability of genomic data hindered our research on Lucanidae family.

Dorcus hopei (Saunders, 1854), distributing from central and northeastern China, is a well-known species notable for its sword-shaped mandibles2,27,28 (Fig. 1). Comparatively, its mandibles are simpler to observe, with a large sharp bump and a relatively small inner tooth. Owing to the restriction of insect allometry and scaling relationship29,30,31, its male trimorphism in mandibles and body sizes is a very rare type. Based on these characteristics, D. hopei is a good choice for performing long-term studies in stag beetles.

Fig. 1
figure 1

Sexual dimorphism and male trimorphism in Dorcus hopei. The scale bar is 1 cm.

In this study, we successfully assembled the first chromosome-level reference genome of D. hopei using Illumina, Nanopore and Hi-C sequencing, the information from which could enhance our understanding of stag beetle survival and evolution. Furthermore, it provides a novel clue for uncovering the molecular basis of extreme mandibles development and male trimorphism formation in the future.


Sample information

Adult male D. hopei specimens were collected from Shou County, Huainan City, Anhui Province, China, during May and June from 2017 to 2021. The beetles were subsequently reared in the laboratory (23 °C, 14 h:10 h light/dark cycle, and 45% relative humidity) and provided with brown sugar jelly and bananas as food. Two adult males were selected for next-generation genomic sequencing using the Illumina platform, one adult male was used for long-read genomic sequencing with the Oxford Nanopore platform, two adult males were selected for Hi-C sequencing, and one adult male was used for transcriptomic sequencing.

Illumina, nanopore, Hi-C, and RNA sequencing

Genomic DNA was isolated from the leg muscles using a Trelief Animal Genomic DNA Kit (TsingKe, China). Paired-end libraries (insert size: 350 bp) were generated using a NEBNext Ultra DNA Library Prep Kit (New England Biolabs, USA) with the Illumina HiSeq 4000 platform at Novogene (Tianjin, China). After filtering the bases in the raw reads of quality <Q20, we obtained 55.13 Gb (113x) clean Illumina data.

For Oxford Nanopore long-read sequencing, DNA from thorax muscles were extracted using a Qiagen DNAeasy Kit (Qiagen, German). Subsequently, the extracted DNA was treated with the NEBNext Ultra End Repair/dA-Tailing module (New England Biolabs, USA) to incorporate adapters for priming sequencing reactions (NextOmics, China). The library was constructed using a 1D DNA Ligation Sequencing Kit (SQK-LSK109) (Oxford Nanopore Technologies, England) and sequencing was performed on a PromethION flow cell (NextOmics, China) to obtain 46.34 Gb (95x) Nanopore data.

For Hi-C sequencing, cells isolated from head tissues were fixed with formaldehyde and subsequently digested using the restriction enzyme MboI. The DNA was purified and then sheared into 300–600 bp fragments using a Covaris M220 device (Covaris, USA). After DNA size selection using AMPure XP beads, point ligation junctions were pulled down using Dynabeads MyOne Streptavidin C1 (ThermoFisher, USA). Then the Hi-C library was sequenced on the Illumina NovaSeq sequencing platform at Novogene (China), and we got 48.46 Gb (100x) Hi-C data.

Transcriptomic sequencing was used to assist in gene structure annotation. RNA was extracted from the head tissue of one adult male using TRIzol. RNA quality was assessed using an RNA Nano 6000 Assay Kit for 2100 Bioanalyzer Systems Kit (Agilent Technologies, China). The libraries were generated using a NEBNext Ultra RNA Library Prep Kit (New England Biolabs, USA) and sequenced on the Illumina Hiseq platform at Novogene (Tianjin, China). And 8.07 Gb were obtained for assisting genomic annotation.

Chromosome-level genome assembly

Illumina data was used to estimate genome size based on 17 k-mer size analysis using KmerFreq v5.032. The estimated genome size of D. hopei was 487.15 Mb, with heterozygosity of 0.021 based on the frequency distribution of 17-mers (Fig. 2a). The Oxford Nanopore long reads were used to assemble and polish the primary genome with NextDenovo v2.5.0 ( (parameters: -k 0 -p 15) and purge_dups v1.0.033. The Illumina short reads were used to correct errors at the base level in the above-polished genome using NextPolish v1.4.034. The Nanopore assembly was 496.47 Mb (N50 = 3.94 Mb), comprised 232 contigs, and achieved a BUSCO completeness score of 99.90% (Table 1).

Fig. 2
figure 2

Assembly of chromosome-level genome of Dorcus hopei. (a) 17-mer analysis of the D. hopei genome based on Illumina reads, X-axis represented depth (x); Y-axis represented the proportion of the frequency of that depth to the total frequency of all depths. (b) Heatmap of Hi-C data showing nine chromosome boundaries (Chr1 to Chr9). The comparison of (c) genome size and N50 length, (d) BUSCO scores, and (e) repeat elements in D. hopei and six other species. (f) Circos tracks showing chromosome length, GC content, density of protein-coding genes, and repetitive elements (SINE, short interspersed elements; LINE, long interspersed elements; LTR, long terminal repeat elements).

Table 1 Statistics for Dorcus hopei genome assembly and gene annotation.

The Hi-C paired-end reads were iteratively mapped to the Nanopore assembly using HiC-Pro v2.9.035. The paired tags were then filtered using restriction enzyme digesting fragments with Juicer v1.60 and contigs were ordered and orientated using 3D de novo assembly software (3D-DNA) v18092236. Finally, JuiceBox v1.11.0837 was applied to correct contig orientation and move suspicious fragments into unanchored groups by visual exploration of the Hi-C heatmap. After Hi-C assembly, the resulting 496.58 Mb genome was assembled into 18 chromosomes (2n = 8AA + XY) (Table 1; Fig. 2b). Notably, 96.18% of the contigs from the “Nanopore assembly” were successfully anchored to nine chromosomes, with a scaffold N50 of 54.61 Mb and 99.80% BUSCO completeness (1.7% duplicated genes) (Table 1), indicating relatively high assembly integrity (Fig. 2c,d).

Genome annotation

We choose several reference species to assist annotation, including five other coleopteran species (Scarabaeoidea: Onthophagus taurus (GCA_000648695.2), Oryctes borbonicus38 (GCA_902654985.2), Trypoxylus dichotomus39 (GCA_023509865.1); Staphylinidea: Nicrophorus vespilloides40 (GCA_001412225.1); Tenebrionoidea: Tribolium castaneum41 (GCA_000002335.3)), and one dipteran species, Drosophila melanogaster42 (GCA_000001215.4). We uploaded the detailed species information table to figshare43. Initially, we annotated repetitive sequences in the D. hopei genome by identifying LTRs and tandem repeats using LTR_Finder v1.0544 and Tandem Repeat Finder v4.07b45, respectively. Transposable elements (TEs), including DNA elements, long interspersed nuclear elements (LINEs), short interspersed nuclear elements (SINEs), and long terminal repeats (LTRs), were next identified using RepeatMasker v4.0.546 against a de novo repeat library constructed with RepeatModeler v1.0.447 and Repbase TE library v16.0248 separately at the DNA level. Finally, TE-relevant proteins were identified using RepeatProteinMask v4.0.947 at the protein level. The final genome assembly (Hi-C assembly) of D. hopei comprised 57.45% repetitive sequences, totaling approximately 285.27 Mb, which is almost twice that of T. castaneum (31.15%) (Fig. 2e). Among the repetitive sequences in the D. hopei genome, the major categories included unclassified sequences (32.39%), DNA elements (11.36%) with maximum density in each chromosome, LINEs (7.95%), and LTRs (7.55%) (Fig. 2f).

Protein-coding genes were predicted using a combination of de novo-, homology-, and transcriptome-based approaches. We utilized the repeat-masked genome and applied the de novo-based gene prediction software Augustus v3.4.049, using models trained on protein sequences from the O. borbonicus genome38, with default parameters. TBLASTN v2.12.050 and GeneWise v2.4.151 were used for homology prediction. The transcriptome data were then aligned to the genome using HISAT2 v2.0.0-beta52. Based on the resulting BAM files and reference genome, the transcriptomic sequences were assembled using StringTie v2.1.453. To form a comprehensive, non-redundant set of genes, we performed several integrations using EVidenceModeler (EVM) v1.1.154, assigning different weight values to the seven genomes based on their BUSCO scores and gene structure components (gene length, coding sequence length, exon number and length, and intron length). The EVM gene set with the best BUSCO value and gene structure components was then selected as the final gene prediction. Finally, resulting in the annotation of 11,231 protein-coding genes in the D. hopei genome. We uploaded the complete gene annotation tables to figshare43. Compared to the different gene features of other six species, the D. hopei genome annotations were comprehensive (Fig. 3), further validating the quality and accuracy of the genome annotation.

Fig. 3
figure 3

Distribution statistics of gene features among the seven species. The comparison of (a) mRNA length, (b) CDS length, (c) exon number, (d) exon length and (e) intron length in D. hopei and other six species.

Finally, we performed functional annotation of the genome. The protein sequences of the genome were searched for homology-based function assignments against the KEGG, NR, TrEMBL, and SwissProt databases using BLASTP v2.2.2655 with an e-value cut-off of 1e-5. Domains in the D. hopei genome using InterProScan v5.54–87.056 with InterPro and GO database. And combined above results, 88.52% of the predicted genes were functionally annotated using six functional protein databases (Table 2).

Table 2 Statistics of functional annotation of the Dorcus hopei protein-coding genes.

Data Records

The chromosome-level assembly and annotation file of D. hopei has been deposited in figshare database57. Raw sequencing data (Illumina reads, Nanopore reads, Hi-C reads, RNA-seq reads) and sample information are available at NCBI, which can be found under identification number SRP44076458. The assembly also has been deposited in NCBI with the accession number GCA_033060865.159. More detailed information about selected species, the results of genomic annotation (repeated sequences and gene structure), orthologs, and synteny has been deposited in figshare database43.

Technical Validation

Quality assessment of the assembled genome was performed using the following methods. Firstly, BWA v0.7.1760 was used to map the Illumina reads to the D. hopei assembly and Samtools v1.3.161 was used to calculate the mapping ratio. The Illumina short reads with a 99.10% accuracy ratio were mapped to the final assembly (Table 1). Secondly, compared N50 length/number with other six selected species. The D. hopei genome displayed a longer N50 (54.61 Mb) and better continuity compared to the chromosome-level genomes of T. castaneum and T. dichotomus (Fig. 2c). Thirdly, insecta_odb10 with 1,367 genes in BUSCO v5.2.262 was used to evaluate genome assembly and annotation completeness. The final assembly had 99.8% BUSCO scores with 0.1% fragmented and 0.1% missing sequences (Fig. 2d). Additionally, we got nine pairs of chromosomes based on Hi-C data, mirroring that of congeneric species Dorcus parallelipipedus63. All these results suggest that we got a high-quality assembly of D. hopei with high integrity, continuity and accuracy.