Effects of gene choice, base composition and rate heterogeneity on inference and estimates of divergence times in cypriniform fishes

Gene choice, base compositional heterogeneity and rate heterogeneity are known to influence phylogenetic reconstruction but are often overlooked. Here, we investigate the impact of these factors in reconstructing the phylogenetic relationships and divergence times of cypriniform fishes. We compile a multilocus data set of newly sequenced and previously published nuclear protein-coding genes from species from all major lineages of Cypriniformes. We collect data on the oldest known cypriniform fossils and gather body size information. We demonstrate that body size correlates with rates of molecular evolution and contributes to rate heterogeneity not only in Cypriniformes but also in a number of other fish lineages. We find that miniaturized cypriniforms show exceptionally high rates of molecular evolution that may lead to improper phylogenetic placement of these taxa due to issues associated with branch attraction. We identify and correct for base compositional heterogeneity and find that this has a marked impact on topology and a corresponding impact on the estimation of divergence times within Cypriniformes. Using gene sequences, fossil calibrations and two different methods, we show strong evidence that Catostomidae is sister to all other cypriniform families and the major clades of cypriniform fishes diverged long before their fossil record indicates.