Researchers have assembled nine complete mitochondrial genomes from 10 beetles representing three salt-marsh specialist ground beetles: Bembidion minimum, Dicheirotrichus gustavii and Pogonus chalceus. The work produced a near-complete set of sequences for the sampled animals, but one sample, labelled NS-PC2, did not yield a complete assembly. Its longest calculated contig was 11,568 base pairs. The beetles were collected along the Wadden Sea coast in Schleswig-Holstein, Germany, in 2024.
A common layout, with local variation
The nine successful assemblies ranged from 16,204 to 17,398 base pairs. Each contained the same broad package of mitochondrial genes: 13 protein-coding genes, 22 tRNA genes and two rRNA genes, arranged in a conserved insect gene order. These were observed assembly features, rather than estimates accompanied by inferential confidence intervals.
Composition was similarly consistent. All assembled genomes were strongly biased toward adenine and thymine, with GC content of about 20 percent. The study also examined codon usage, meaning how often alternative three-letter genetic codes appeared in protein-coding genes. Frequently used codons, defined in the analysis as those with relative synonymous codon usage greater than 2, often ended in A or T. A correspondence analysis, which summarizes patterns across variables, separated the three species into distinct clusters; its first two dimensions accounted for 97.8 percent of the variation.
The clearest differences appeared in repeated DNA in the control region. Tandem repeats, stretches of sequence copied next to themselves, varied between species and between individuals. In Dicheirotrichus gustavii, an approximately 376-base-pair repeat fragment appeared in samples FK-DG2 and LK-DG6, but no comparable repeats were found in its other samples. The study found that these repeats explained only part of the differences in genome length, and noted that short-read sequencing makes such regions difficult to assemble.
A sharp split at species level
To check how the sequences grouped, the researchers aligned the complete mitochondrial genomes without dividing them into separate regions. They then used a maximum-likelihood analysis, a tree-building method, with the PHYML program and the GTR model. The analysis was tested with 1,000 bootstrap replicates, repeated resamplings used to assess how consistently a branch is recovered.
At the species level, the result was strong. Bembidion minimum, Dicheirotrichus gustavii and Pogonus chalceus each formed a monophyletic mitochondrial clade, meaning the sampled sequences for each species grouped together on the tree. Every one of those clades had a bootstrap support score of 100. That is mitochondrial evidence of consistent species-level grouping, not a complete estimate of each species' entire evolutionary history.
The wider family tree was less settled. Relationships among Carabidae species received bootstrap support of roughly 27 to 66, a weak-to-moderate range that the study treated as limited resolution. The data separated the three focal species cleanly but did not provide equally firm answers about their deeper relationships within the beetle family.
What the sequences can and cannot answer
The researchers also used COX1 sequence comparisons as an identification check. COX1 BLASTn searches, which compare a DNA sequence with reference sequences, found at least 99.6 percent identity to the reference for the corresponding species in every newly assembled genome. That sequence similarity supports consistency between the assemblies and their species labels, but it does not independently establish a complete species taxonomy.
The laboratory workflow started with DNA extracted from whole specimens using the DNeasy Blood and Tissue Kit, after brief homogenization and overnight digestion with proteinase K. Whole-genome sequencing used a single Illumina NextSeq1000 lane with 300 cycles. Before assembly, adapters and bases below Phred quality 20 were removed. The main reconstruction used NOVOPlasty with species-specific COI seed sequences, a baited de novo strategy that builds the mitochondrial sequence from the sequencing reads. GetOrganelle was used to complete sample FK-DG9 after NOVOPlasty failed.
The study covers 10 whole specimens, three species and a limited geographic sampling area, so it was not set up to test population structure, geographic differentiation or sex-specific genomic differences. One sample's failure and the difficulty of assembling repetitive control regions leave open whether longer-read sequencing would recover those regions and genome-length variation more completely. Broader taxon sampling or different phylogenetic designs may likewise be needed to clarify Carabidae relationships.
Sequence data were reported as available in NCBI under accessions PZ095925 to PZ095933, except NS-LK2, which the authors make available on request. The article and supporting information contain the generated and analyzed data, and the specimens are deposited at the Zoological Museum Kiel. The work was supported by DFG Research Infrastructure NGS_CC through the Next Generation Sequencing Competence Network, with analyses conducted at the Competence Centre for Genomic Analysis in Kiel; Open Access funding was organized by Projekt DEAL.
Paper data and sources
Original title: De Novo Assembly and Comparative Mitochondrial Genome Analysis of Three Salt Marsh Specialist Ground Beetle Species (Coleoptera: Carabidae).
Authors: Anne Mack, Michael Kuhlmann, Christine Ewers
Journal/Repository: Ecology and evolution
Status: Peer-reviewed
First online: 2026-08-20
DOI: 10.1002/ece3.74195
Original paper