Using ddRAD-seq phylogeography to test for genetic effects of headwater river capture in suckermouth armored catfish (Loricariidae: Hypostomus) from the central Brazilian Shield

River capture is a geological process of potentially great importance in shaping the genetic diversity, distributions, and community composition of freshwater taxa. Using phylogeographic analyses of ddRAD-seq data from suckermouth armored catfish (Hypostomus sp. 2) populations, we tested for predicted genetic effects of headwater river capture events in central Brazil, previously supported by geological and community ecological data. We analyzed 227 ddRAD tags (3829 SNP loci) across 42 samples. Molecular results strongly supported six Hypostomus genetic clusters/lineages, with the deepest divergence [~]1.25 million years ago in the early Pleistocene between a clade from the Upper Parana and Upper Sao Francisco river basins versus all other lineages. Consistent with the Parana Capture Hypothesis, several lines of evidence supported mid-Pleistocene colonization and vicariant isolation of Hypostomus populations from an ancestral Upper Parana population, including: (1) significant phylogeographic structure, with predicted phylogenetic patterns, (2) higher Parana lineage diversity, (3) ancestral geographic locations reconstructed in the Parana basin, and (4) non-random interdrainage dispersal and vicariance events, indicating river captures primarily into the Tocantins and Upper Sao Francisco basins c. [~]220,000-145,500 years ago. Phylogeographic inference was complicated by lack of lineage monophyly across loci and lineages distributed in multiple basins, the latter of which lent support to the non-mutually exclusive Frequent Interdrainage Dispersal Hypothesis. However, species tree and demographic modeling results suggested these were artefacts of incomplete sorting of alleles in large ancestral populations over a geologically recent timeframe of divergence. Qualitative and quantitative sensitivity analyses demonstrated that our downstream genetic results were robust to effects of varying ddRAD-seq assembly parameters, which heavily influenced the number of output loci. We predict that codistributed freshwater taxa in Central Brazil may not exhibit phylogeographic patterns similar to Hypostomus sp. 2 due to complex patterns of superimposed river capture events, or if smaller ancestral population sizes have allowed more complete lineage sorting in other taxa.