The researchers searched GenBank for ant sequences and retained 31 specimens that covered the cox1–tRNALeu–cox2 region, including leafcutting ants from the three genera and several myrmicine outgroups. Some records were complete mitochondrial genomes, while others were partial mitochondrial sequences or conventional Sanger-sequencing data. Alignment against a reference mitochondrial genome produced a matrix of 2,607 nucleotide sites. The targeted region comprises the cox1 protein-coding gene, a first intergenic spacer, the mitochondrial tRNALeu gene, a second intergenic spacer and cox2. Because more than 11,000 database records mentioning cytochrome c oxidase were found but relatively few covered the full target region, the complete structural analysis involved fewer taxa than the broader comparison of barcode sequences.

One of the clearest signals involved the end of the cox1 gene. Leafcutting ants had a complete TAA stop codon, whereas the outgroup Wasmannia auropunctata showed duplication of the final isoleucine codon and stop codon. Atta species consistently contained a three-nucleotide insertion immediately before the cox1 stop codon. This insertion added one codon and translated to an extra amino acid in the affected Atta sequences. Amoimyrmex and Acromyrmex lacked the insertion and instead matched the relevant outgroup state in that respect. Cox1 was also three nucleotides longer in Atta than in Amoimyrmex and Acromyrmex. The shared Atta feature was interpreted as a candidate genus-level character, but the authors emphasized that its evolutionary history and diagnostic reliability require further testing.

The second protein-coding gene revealed a different arrangement. Among the 11 taxa for which complete cox2 sequences were available, Amoimyrmex carried three additional codons before the stop codon compared with Atta and most sampled Acromyrmex. Acromyrmex versicolor was an exception, with two extra codons rather than the three-codon extension seen in Amoimyrmex. The same TAA stop codon occurred in every evaluated species. These contrasting extensions in cox1 and cox2 produced genus-level patterns consistent with the recognized phylogeny: Atta was distinguished by the cox1 insertion, while Amoimyrmex had the longer cox2 terminus. The reduced taxon count for cox2 means that this pattern was observed in a smaller subset than the overall alignment.

A particularly conspicuous feature appeared in the intergenic region between cox1 and tRNALeu. All sampled leafcutting ants had a spacer, unlike the no-spacer state in the Solenopsis outgroup and the four-nucleotide spacer in W. auropunctata. Atta and Acromyrmex shared a much longer spacer, reaching as many as 285 nucleotides, whereas the spacer in Amoimyrmex was as short as those of the outgroups. The distribution of this feature resembles a possible shared derived trait for the Atta-plus-Acrymyrmex clade, which is sister to Amoimyrmex in the study’s taxonomic framework. No comparable long spacer was found between tRNALeu and cox2, indicating that the extra sequence was not simply distributed throughout the mitochondrial genome.

The transfer RNA itself also displayed structural variation. Acromyrmex tRNALeu sequences were slightly longer than those of the other leafcutting-ant genera, and species within Acromyrmex possessed insertions in the TψC and DHU loops that were interpreted as a potentially novel feature. Predicted secondary structures differed in the lengths of the D-loop and TψC-loop by small amounts, ranging from one to eight nucleotides. The acceptor stem and anticodon loop were conserved, while the anticodon stem was shorter in Amoimyrmex and some Acromyrmex species than in Atta, where it was consistently longer. The authors did not claim that these small changes affect transfer-RNA function, but the variation in sequence length and fold contributed to the proposal that mitochondrial structure can convey systematic information.

Base composition supplied a further comparative trait rather than a clear genus by itself. The cox1 and cox2 genes were adenine-thymine rich in all three genera, as expected for insect mitochondrial DNA. Mean AT content for cox1 ranged from 65.44% in Amoimyrmex to 70.47% in Atta, while Acromyrmex fell between them at 68.88%. For cox2, the mean AT values were 69.12% in Amoimyrmex, 76.66% in Atta and 74.02% in Acromyrmex. Thus, the sister genera Atta and Acromyrmex had higher AT proportions than Amoimyrmex in both genes. Differences were also pronounced at the third codon position, reaching 86.24% AT in cox1 and 85.84% in cox2 for Atta. Mean uncorrected pairwise distances among the genera were low, generally between about 2% and 3% under the comparisons reported, underscoring the challenge of finding unequivocal sequence-based boundaries among these closely related groups.

The exploratory analysis is best understood as an extension of character-based barcoding into structural traits, not as a new phylogenetic analysis or a formal revision of the ants. Mitochondrial markers offer practical advantages because they mutate relatively rapidly, are maternally inherited and can be amplified and sequenced comparatively efficiently, but they represent only one genealogical history. A single mitochondrial tree can therefore fail to capture the full complexity of species relationships, and nuclear paralogues can complicate the interpretation of mitochondrial data. The sample was also modest, complete cox2 data were available for only 11 taxa, several records were partial, and the researchers did not test the candidate characters in populations broad enough to establish universality. Insertions and deletions were observed across leafcutting-ant lineages relative to outgroups, so their origins and stability require additional analysis.

Even with those limitations, the results add molecular support to an integrative classification built from phylogeny, morphology, karyotypes, colony characteristics and other biological evidence. Previously, Amoimyrmex species were placed in Acromyrmex before an integrative assessment moved them into a separate genus. The mitochondrial characters examined here broadly reproduced the resulting pattern: Amoimyrmex had a short cox1-to-tRNALeu spacer and a longer cox2 terminus, Acromyrmex had a long spacer and distinctive tRNALeu loop insertions, and Atta had the cox1 insertion and consistently long anticodon stem. The study therefore presents structural mitochondrial traits as an additional resource for circumscribing leafcutting-ant genera, comparable in purpose to cytogenetic evidence. Extending the approach to other ant groups would determine whether the same kinds of characters are informative beyond this model clade and may clarify the genomic processes responsible for the observed variation.

Subject of Research: Exploration of mitochondrial structural traits for genus-level taxonomy of leafcutting ants

Article Title: Exploratory analysis of molecular traits of the mitochondrial DNA of leafcutting ants to infer taxonomic characters towards an integrative taxonomy

Article References: Onyinyechi, P. N., Travenzoli, N. M., Cristiano, M. P., & Cardoso, D. C. (2025). Exploratory analysis of molecular traits of the mitochondrial DNA of leafcutting ants to infer taxonomic characters towards an integrative taxonomy. Journal of Hymenoptera Research, 98, 147-163. https://doi.org/10.3897/jhr.98.133204

Image Credits: AI Generated

DOI: 10.3897/jhr.98.133204

Keywords: leafcutting ants, mitochondrial DNA, integrative taxonomy, molecular systematics, DNA barcoding, Atta, Acromyrmex, Amoimyrmex, mitogenome, ant evolution, Exploratory, analysis