2025/10/13 by Brendon E. Boudinot, Ziv E. Lieberman · 1 voice
Biochemistry, Genetics and Molecular Biology · Agricultural and Biological Sciences · #Insect and Arachnid Ecology and Behavior #Plant and animal studies #Evolution and Genetic Dynamics
paper · pdf · doi:10.1038/s42003-025-07849-8
openalex created_date 2025/10/13 · openalex publication_date 2025/10/13 · openalex updated_date 2026/07/23
Systematic bias is a major concern in phylogenetics, as large datasets can favor mutually exclusive hypotheses depending on data processing strategies and the choice and fit of models. Here, we point out flaws in study design and reporting in the recent empirical study of Cai 1 (= C24) on the phylogeny of ants (Formicidae). In this work, the author recycled two genomic amino acid (AA) datasets from Romiguier et al. 2 (= REA) and one multilocus nucleotide dataset from Borowiec et al. 3 (= BEA), which is not discussed further here as these data are known to be insufficient to confidently resolve the phylogenetic problem at hand 2 , 4 . After aggressive downsampling of the AA data from REA 2 , C24 1 applied site-heterogeneous empirical (C n ) and generalized time-reversible (GTR) profile mixture models (Fig. 1 ), claiming that these—particularly CAT-GTR—reject the results of REA and resolve the backbone phylogeny of ants once and for all. Fig. 1: Overview of the study designs of REA and C24. Data treatment ( a , b ) and analysis pipelines ( c , d ) showing that most site-heterogeneous models favor the Leptanillomorpha hypothesis ( c ) and that it is only questionable application of the CAT-GTR model and analysis of the uniquely incongruous Matrix C5 that favor the Leptanillinae-sister hypothesis ( d ). Arrows from Matrices R# ( a ) to C# ( b ) indicate treatments of REA alignments to produce reduced C24 alignments. AA amino acids, NT nucleotides, UCEs ultraconserved elements; terminals = ingroups + outgroups. Alignment ‘Ants101T-F90’ from Branstetter et al. 14 , 15 ; outgroups for R2 from Peters et al. 15 . Variable site % calculated using AMAS 16 (Supplementary Data 2 ). Full size image