Abstract
Accurately delimiting species boundaries is a nontrivial undertaking that can have significant effects on downstream inferences. We compared the efficacy of commonly used species delimitation methods (SDMs) and a population genomics approach based on genomewide single-nucleotide polymorphisms (SNPs) to assess lineage separation in the Malaysian Torrent Frog Complex currently recognized as a single species (Amolops larutensis). First, we used morphological, mitochondrial DNA and genomewide SNPs to identify putative species boundaries by implementing noncoalescent and coalescent-based SDMs (mPTP, iBPP, BFD*). We then tested the validity of putative boundaries by estimating spatiotemporal gene flow (FASTSM-COAL2, ABBA-BABA) to assess the extent of genetic isolation among putative species. Our results show that the A.larutensis complex runs the gamut of the speciation continuum from highly divergent, genetically isolated lineages (mean F-st=0.9) to differentiating populations involving recent gene flow (mean F-st=0.05; N-m>5). As expected, SDMs were effective at delimiting divergent lineages in the absence of gene flow but overestimated species in the presence of marked population structure and gene flow. However, using a population genomics approach and the concept of species as separately evolving metapopulation lineages as the only necessary property of a species, we were able to objectively elucidate cryptic species boundaries in the presence of past and present gene flow. This study does not discount the utility of SDMs but highlights the danger of violating model assumptions and the importance of carefully considering methods that appropriately fit the diversification history of a particular system.