Early biogeography of Otophysi points to the Neotropics as the cradle of Characiphysan fishes

Freshwaters represent less than 1% of Earth's surface and 0.02% of available aquatic habitable volume, but host nearly half of the 35,500 known species of bony fishes. Ostariophysan fishes represent 70% of all freshwater fishes with ca. 12,000 species distributed in five speciose orders. They constitute a large radiation whose internal phylogenetic relationships are still intensively debated. To better understand their early evolutionary history and origin, we reconstructed their phylogenetic relationships through a dense taxonomic sampling and the combined use of complete mitochondrial genomes and sequences of four nuclear genes. Maximum Likelihood of phylogenetic reconstructions and time tree estimates were applied to a data set consisting of 687 ostariophysan species for a total of 21,701 aligned positions, including 15,707 variable sites, and a Maximum Likelihood of model-based estimation of ancestral areas was used to reconstruct the early evolution of Otophysi. We produced a highly supported hypothesis of phylogenetic relationships among Otophysi, which points to the importance of plate tectonics in triggering several divergence events jointly with several subsequent range rearrangements, further consistent with the contraction of the tropical belt that started at the end of the Cretaceous and lasted throughout the Paleogene. The divergence of Cypriniformes and Characiphysi results from the joint influence of the separation of Laurasia and Gondwana, while the origin of Characiphysi is located in West Gondwana, and the subsequent expansion of the group cannot be explained without considering transcontinental dispersal.

Methods
Two data sets were assembled and subsequently combined, including a set of complete mitochondrial genomes and four nuclear data sets, including myh6, rag1, rag2, and sh3px. The nuclear genes come from the study by Rabosky et al. (2018) and were already filtered for rogue sequences corresponding to misidentification and/or paralogs. Only four nuclear genes were selected out of 21, which corresponds to those with a taxonomic coverage above 20% for ostariophysan fishes (Rabosky et al., 2018). The mitogenomes analyzed here come from different sources. First, previously published mitogenomes already assembled and annotated were downloaded from NCBI GenBank. Second, we inspected libraries of raw Illumina sequencing reads from previous studies using Ultra Conserved Element (UCE) to extract mitogenomes whenever enough reads were present to assemble a scaffold of at least 12,000 base pairs. For this, SRA files were downloaded from NCBI GenBank or ENA. Third, new mitogenomes were produced for Characiform lineages, which are poorly covered in available mitochondrial genomes in international repositories, to complement mitogenomes coming from the two previous sources.