We used maximum likelihood and Bayesian methods to analyze the molecular data. Based on the findings by Macey et al. (2000), Otocryptis wiegmanii and Calotes calotes were used as outgroups. ND2 sequences for these two genera and one sequence of Sitana bahiri were downloaded from GenBank (S1). Sequences were aligned using ClustalW and uncorrected genetic distances were calculated using MEGA 5 (Tamura et al., 2011). The combined dataset was partitioned into four genes (ND2, RAG1, R35, PDC). PartitionFinder v1.1.1 (Lanfear et al., 2012) was used to find the best partition scheme and model of sequence evolution for each partition. The optimal partitioning scheme included seven partitions (see S2). Likelihood analysis in RaxML takes only one model of sequence evolution, therefore we used GTR+G for all seven partitions. In case of Bayesian analysis seven partitions were assigned to different models (see S2). We employed the RAxML GUI (Silvestro and Michalak, 2012) to conduct a maximum likelihood phylogenetic analysis. This GUI implementation uses RAxML 1.3.1 (Stamatakis et al., 2005). We used the ML+ rapid bootstrap method to search for best scoring maximum likelihood tree, and assessed branch support using 1000 non-parametric bootstrap replicates. A Bayesian tree was also generated using the program MrBayes 3.2 (Ronquist et al., 2012). For this analysis, two Markov chains were initiated from random trees and allowed to run for 1,000,000 generations sampling every 100 generations. The analysis was terminated when the standard deviation of split frequencies was less than 0.005, and the first 250,000 generations were discarded as “burn-in”.