Phylogenetic Systematics of Dart-Poison Frogs and their Relatives Revisited (Anura: Dendrobatoidea)

Despite the impressive growth of knowledge on the phylogenetic systematics of dart-poison frogs and their relatives (Dendrobatoidea) over the past decade, many problems remain to be addressed We analyzed up to 189 phenomic characters (morphology, behavior, defensive chemicals) and 15 mitochondrial and nuclear loci scored for 564 dendrobatoid and outgroup terminals, including 76 newly sequenced terminals and > 20 previously unanalyzed species, using tree-alignment and the parsimony optimality criterion in the program POY v. 5. 1.1 and additional analyses of the implied alignment using TNT v. 1. 5. Even though data coverage was highly heterogeneous, the strict consensus of 639 optimal trees is highly resolved and we detected only one instance of wildcard behavior involving a small clade of outgroup species The monophyly of the median lingual process (MLP) possessing genus Anomaloglossus is decisively refuted, with the cis-Andean species being sister to Rheobates within Aromobatidae and the trans-Andean species nested within Hyloxalinae, implying two independent origins of the structure in Dendrobatoidea. Although this result was unexpected, it is not surprising given that the MLP evolved at least five times in Asian and African ranoids, including Arthroleptidae, Dicroglossidae, Mantellidae, and Rhacophoridae and either once in the most recent common ancestor of the massive clade Victoranura followed by independent losses or multiple times within component lineages. We restrict Anomaloglossus to the cis-Andean MLP-possessing species, describe a new genus for the trans-Andean MLP-possessing species, and resurrect Paruwrobates for its sister group, which includes Dendrobates andinus (formerly Ameerega), D. erythromos (formerly Hyloxalus and, until recently, Ameerega), and Prostherapis whymperi (formerly Hyloxalus). We also transfer Dendrobates maculatus from Ameerega to Epipedobates, making Ameerega an exclusively cis-Andean group We describe two new species of the trans-Andean MLP-possessmg genus-one from Cerro Tacarcuna, near the Colombo-Panamanian border, and the other from 800-900 m elevation on the western versant of the Colombian Cordillera Occidental (Cauca Department-bringing the total number of species in the genus to seven The discrete, round, white to yellowish-brown dots found on the venter of the new species from Cerro Tacarcuna and at least one other trans-Andean MLP-possessmg species are formed by large, ellipsoid, densely distributed (up to 80 glands/mm(2)) granular glands Although specimens of the new species from Cerro Tacarcuna exuded a noxious milky substance when handled, lipophilic alkaloids were not detected In addition to the unexpected placement of the trans-Andean MLP-possessmg species, major findings include the unexpected placement of Colostethus ruthvem and its undescribed sister species (the ``C.{''} ruthveni group) within Dendrobatmae as sister of the newly recognized tribe Dendrobatini (all dendrobatines except Phyllobates and the ``C.{''} ruthveni group) We describe a new genus for C. argyrogaster and C. fugax to remedy the paraphyly of Colostethus caused by the placement of those species as sister to Ameerega Our evidence rejects the sister group relationship of Dendrobates + Oophaga in favor of Dendrobates + Adelphobates, which is consistent with their uniquely low diploid chromosome number of 2n = 18 (2n = 20 in Oophaga). With the exception of Anomaloglossus and Colostethus, all other genera are monophyletic We recognize several monophyletic species groups-including the Atlantic Forest, trans-Andean, and 22-chromosome groups within Allobates, the An stepheni, An megacephalus, and An beebei groups in Anomaloglossus, the C. latinasus (formed by the C. inguinalis and C. latinasus clades) and C. fraterdameli groups within Colostethus, and the Am braccata and Am rubriventris groups within Ameerega-identify unambiguously optimized phenomic synapomorphies, and summarize patterns in the evolution of the diploid chromosome number, swelling of Finger IV in males, relative length of Fingers II and III, length of Finger V, and testicular and intestinal pigmentation Finally, we address criticisms of the current taxonomy of Neotropical poison frogs and their relatives, concluding that they are either overstated, misguided, or false, and that the current system of names better communicates knowledge of the diversity of these frogs Our results highlight the importance of increased taxon sampling, and we conclude by identifying key species to include in future phylogenetic analyses. (AU)