Molecular phylogenetics of Oestroidea (Diptera: Calyptratae) with emphasis on Calliphoridae: insights into the inter-familial relationships and additional evidence for paraphyly among blowflies.

The superfamily Oestroidea, comprising ∼15,000 species, is a large and ecologically diverse clade within the order Diptera. Among its six commonly recognized families, Calliphoridae seems to be crucial for understanding evolutionary relationships in the group, as it is recognized as a controversial paraphyletic grouping. To further investigate this matter, the ITS2, 28S, COI and 16S regions were used to infer phylogenetic relationships in Oestroidea with maximum-parsimony (MP), maximum-likelihood (ML) and Bayesian inference (BI) methods. For the BI analyses, a deep evaluation of different data partitioning strategies was conducted, including consideration of structural conformation (ITS2 and 16S) and codon position (COI) information. Results suggest the existence of two main clades in Oestroidea: (Tachinidae+Mesembrinellinae) and (Rhiniinae, (Sarcophagidae+Calliphoridae sensu stricto)). Oestridae was recovered as sister group of the remaining Oestroidea in the MP trees while it was placed closer to the (Rhiniinae+Sarcophagidae+Calliphoridae sensu stricto) group in the ML and BI trees. A paraphyletic Calliphoridae was recovered, confirming the exclusion of Rhiniinae, a clade recently promoted to family status and therefore already excluded. Mesembrinellinae could also be considered a distinct group apart from Calliphoridae, although further studies are required. Consideration of structural and codon position information led to a significant increase in the log-likelihoods of the analyses, which were accompanied by small changes in the inferred topologies, branch lengths and posterior probability support values. However, as model complexity increases, so does uncertainty across the estimated parameters, including tree topologies, and phylogenies inferred under very parameter-rich models may be less reliable even when possessing higher log-likelihoods.