Neighbor Joining (NJ) trees for 16S (Fig. 2), COI and Cytb (not shown) were created for species delineation and to help elucidate species geographic ranges. The three molecular markers produced identical taxa groupings and similar tree topologies.
Here, we discuss our results with respect to only the 16S NJ tree (Fig. 2) since the topology was in agreement with our deeper phylogenetic analyses, and because analysis based on this gene showed a more conservative approach to species delineation. Within the in-group taxa, there were 15 major clades corresponding to four valid Hylarana species: H. aurantiaca (clade 1), H. temporalis (clade 12), H. gracilis (clade 13) and H. malabarica (clade 14); three named species currently considered as synonyms: H. intermedius (clade 2), H. flavescens (clade 9) and H. montanus (clade 10); and between seven (clades 3-8 and 11) to eight (clades 3-8, 11 and 14a) unnamed species. We herein describe these as — Hylarana doni sp. nov., Hylarana urbis sp. nov., Hylarana sreeni sp. nov., Hylarana indica sp. nov., Hylarana caesari sp. nov., Hylarana magna sp. nov. and Hylarana serendipi sp. nov., respectively (Appendix) and clade 14a is referred to as Hylarana malabarica haplogroup 1. For further discussion see ‘Morphological recognition of species’.
Genetic distance comparisons (Tables S2, S3) showed maximum intraspecific uncorrected pairwise distance of 1.5% (for 16S) and 3.6% (for COI) for Hylarana sreeni (clade 5). For Cytb, maximum intraspecific distance of 6.2% was observed for H. indica (clade 6) and 6.1% for H. sreeni. With respect to closely related species, the minimum interspecific uncorrected pairwise divergence was 16S: 2.0% between H. indica and H. sreeni, COI: 4.3% between H. sreeni and H. indica, Cytb: 8.0% between H. intermedius and H. doni. An analysis of the barcode gap showed no overlap between the maximum intraspecific pairwise distance within species and the minimum interspecific divergence between species (Fig. 3). The observed mean threshold values of sequence divergence for species delineation were: 2.6% for 16S, 5.2% for COI and 8.7% for Cytb. A high level of genetic variation was observed between H. malabarica typical and H. malabarica haplogroup 1 (clades 14a-b) (Fig. 2). Though there was considerable sequence divergence with an average of 3.7% for 16S, the COI and Cytb sequence divergences (3.2% and 6.2% respectively) were lower than the threshold values observed for other species pairs. Maximum divergence in our dataset was observed within the Hylarana temporalis group of Sri Lanka, with mean values of 10.5% (16S), 18.1% (COI) and 23.8% (Cytb) between the species H. serendipi and H. temporalis.
Overall, not only were our results concordant for all the three molecular markers but also in agreement with earlier studies that have proposed 3% and 4-8% as threshold values for 16S and COI, respectively (Fouquet et al., 2007a; Smith et al., 2008). Only two of the recognised species pairs showed slightly lower mean sequence divergence values for 16S (2.7% between Hylarana intermedius-H. doni, and 2.5% between H. indica-H. sreeni). However, morphologically these represented well-supported distinct lineages. For Cytb, the overall threshold value was slightly lower than the values shown by certain other studies (Kotaki et al., 2010; Hasan et al., 2012). However, there are not many comparable studies available for amphibian barcoding based on Cytb, and since species delineation based on Cytb sequences was in agreement with the other two molecular markers in the present study, an 8.7% threshold value for Cytb seems realistic. For detailed genetic comparisons, see the ‘Genetic divergence’ section of each species in the Appendix.
Our barcoding results, in the light of taxonomic interpretations, showed remarkable patterns of species level distribution. All of the Western Ghats and Sri Lankan taxa formed distinct clades. The Western Ghats groups showed north-south distribution trends among different species, while the Sri Lankan species showed a dry zone-wet zone (again essentially a north-south) distribution trend (Fig. 1). Hylarana malabarica (India) and H. gracilis (Sri Lanka), both with wider distributions, formed the basal clades to all of the remaining Western Ghats-Sri Lankan clades (Fig. 2).
Fig. 2. Neighbour Joining tree of Kimura-2-parameter (K2P) distances based on mtDNA 16S for 102 Hylarana samples from the Western Ghats-Sri Lanka biodiversity hotspot. Collection numbers are presented in Table S1.