Malagasy frogs of the subgenus Brygoomantis in the mantellid frog genus Mantidactylus currently comprise 14 described species of mostly brown, riparian frogs. Data from DNA barcoding suggested that the diversity of this subgenus is dramatically underestimated by current taxonomy. We here provide a comprehensive revision of this subgenus. We use hybrid-enrichment based DNA barcode fishing to obtain mitochondrial DNA fragments from the name-bearing type material of 16 of the 20 available names for members of this subgenus, and integrate these into a genetic dataset consisting of 1305 individuals sampled across Madagascar. By thus assigning the nomina to genetic lineages, we can confidently establish synonyms, revalidate old names, and describe the remaining diversity. We take an integrative approach to our descriptions, drawing together genetics, morphometrics and morphology, and bioacoustics for assignment. We also provide a robust phylogenomic hypothesis for the subgenus, based on 12,818 nuclear-encoded markers (almost 10 million base pairs) for 58 representative samples, sequenced using a hybrid-enrichment bait set for amphibians. Those data suggest a division of the subgenus into eight major clades and show that morphological species complexes are often paraphyletic or polyphyletic. Lectotypes are designated for Rana betsileana Boulenger, 1882; Rana biporus Boulenger, 1889; Rana curta Boulenger, 1882; Mantidactylus ambohimitombi Boulenger, 1918; Mantidactylus tripunctatus Angel, 1930; and Rana inaudax Peracca, 1893. For several other nomina, previous authors had considered a certain syntype as holotype; this has been seen as lectotype designation by implication, which, however, is ambiguous according to the provisions of the International Code of Zoological Nomenclature. Hence, we validate a previous lectotype designation by implication for Limnodytes ulcerosus Boettger, 1880 by explicitly designating the same individual as lectotype. In one other such case, that of Mantidactylus brauni Ahl, 1929, we deviate from previous authors and designate a different specimen as lectotype. We revalidate Rana inaudax Peracca, 1893 as Mantidactylus inaudax (Peracca, 1893) bona species, and Mantidactylus tripunctatus Angel, 1930 bona species. The identities of three further species (M. ambohimitombi, M. biporus, M. tricinctus) are largely redefined based on new genetic data. By designating the lectotype of Rana aluta (MZUT An725.1) as the neotype of Mantidactylus laevis Angel, 1929 we also stabilize the latter nomen (as junior synonym of M. alutus) whose original type material is lost. Based on DNA sequences of its lectotype, we consider Mantidactylus brauni Ahl, 1929 as junior synonym of M. ulcerosus (rather than M. biporus). We formally name 20 new species and four new subspecies: M. ambohimitombi marefo ssp. nov., M. ambohimitombi miloko ssp. nov., M. mahery sp. nov., M. steinfartzi sp. nov., M. incognitus sp. nov., M. jonasi sp. nov., M. katae sp. nov., M. kortei sp. nov., M. riparius sp. nov., M. fergusoni sp. nov., M. georgei sp. nov., M. jahnarum sp. nov., M. marintsoai sp. nov., M. grubenmanni sp. nov., M. gudrunae sp. nov., M. augustini sp. nov., M. bletzae sp. nov., M. brevirostris sp. nov., M. eulenbergeri sp. nov., M. glosi sp. nov., M. stelliger sp. nov., M. manerana sp. nov., M. manerana fotaka ssp. nov., and M. manerana antsanga ssp. nov. This leaves Mantidactylus subgenus Brygoomantis with 35 described species and six subspecies (including nominate subspecies). Based on our taxonomic revision, we discuss (i) the importance of definitive assignment of historical names via archival DNA analysis; (ii) the relevance of the subspecies category to name geographic variation within species; (iii) the value of molecular characters in formal species diagnoses in taxa with substantial individual variation of morphology; (iv) the use of phylogenomic approaches for taxonomy, by confirming that some morphologically similar taxa are not each other’s closest relatives, and in several cases belong to entirely different major subclades within Brygoomantis, thus facilitating lineage diagnosis; and (v) the need to interpret genetic distances in a probabilistic framework rather than using fixed thresholds, where higher distances confer a higher likelihood of genetic incompatibilities across the genome and thus completion of speciation.
Read full abstract