Supermatrix phylogeny resolves goby lineages and reveals unstable root of Gobiaria

Gobies, sleepers, and cardinalfishes represent major clades of a species rich radiation of small bodied, ecologically diverse percomorphs (Gobiaria). Molecular phylogenetics has been crucial to resolving broad relationships of sleepers and gobies (Gobioidei), but the phylogenetic placements of cardinalfishes and nurseryfishes, as reciprocal or sequential sister clades to Gobioidei, are uncertain. In order to evaluate relationships among and within families we used a phylogenetic data mining approach to generate densely sampled trees inclusive of all higher taxa. We utilized conspecific amino acid homology to improve alignment accuracy, included ambiguously identified taxa to increase taxon sampling density, and resampled individual gene alignments to filter rogue sequences before concatenation. This approach yielded the most comprehensive tree yet of Gobiaria, inferred from a sparse (17 percent-complete) supermatrix of one ribosomal and 22 protein coding loci (18,065 characters), comprised of 50 outgroup and 777 ingroup taxa, representing 32 percent of species and 68 percent of genera. Our analyses confirmed the lineage-based classification of gobies with strong support, identified sleeper clades with unforeseen levels of systematic uncertainty, and quantified competing phylogenetic signals that confound resolution of the root topology. We also discovered that multilocus data completeness was related to maximum likelihood branch support, and verified that the phylogenetic uncertainty of shallow relationships observed within goby lineages could largely be explained by supermatrix sparseness. These results demonstrate the potential and limits of publicly available sequence data for producing densely-sampled phylogenetic trees of exceptionally biodiverse groups.