Species distribution modelling of Hylarana Species (Anura, Ranidae) and the problem of accurate species identification.

Species of Hylarana Tschudi, 1838 are among the most common amphibian species in the Oriental region. Their poorly known geographical distribution impacts species delimitation and conservation actions. We compiled geographical data from literature, online databases and museum collections for the four currently recognized species, Hylarana erythraea, H. taipehensis, H. tytleri and H. macrodactyla. We obtained 1014 occurrence records and allocated a quality rate to each of these points: A-species identification confirmed by ancillary data; B-species identification unconfirmed; and C-conflict between species name, geographic origin and morphological characters; resulted in exclusion of the latter in final analyses. Based on these different datasets we built species occurrence maps for each of the four species. These maps were compared to already available distribution maps, introducing a more precise distribution for the four species but with persistence of data gaps within less prospected areas. We then built Species Distribution Models (SDMs) based on climatic data. Discrepancies between observed distribution ranges and areas with suitable climatic niches are discussed in a taxonomical context. Hylarana species, although taxonomically still not fully resolved and occupying similar habitats, exhibit differences in distribution pattern. Models also fill data gaps, introducing relevant suitable climatic niches within close range of species' distribution areas. Differences were observed between A and B-quality data models, with high suitable habitats being less present in most cases for B-quality. Our study shows the importance of merging A and B-quality data, represented by A+B dataset, thus creating more accurate models, but also underlines the necessity to have access to correct data to limit bias.