Rch scheme (all). A summary in the analyses performed with each dataset, which includes the various partitions and models is shown in Table S2. Phylogenetic analyses were performed making use of maximum likelihood (ML) and Bayesian inference (BI) approaches. In both ML and BI analyses, alignment gaps were treated as missing data along with the nuclear gene sequences were not phased. Partitions and models for each and every dataset, with priors, specifications and parameters for every single Acid Blue 9 analysis are specified in Table S2. Maximum likelihood trees have been estimated in RAxML v.7.4.two (Stamatakis, 2006) as implemented in raxmlGUI v.1.3 (Silvestro Michalak, 2012). All ML analyses had been performed with a GTR + G model of sequence evolution and 100 replicates. Each inference was initiated having a random starting tree and nodal help was assessed with 1,000 bootstrap pseudoreplicates (Felsenstein, 1985). Bayesian analyses were performed with BEAST v.1.eight.two (Drummond et al., 2012). All BEAST analyses have been carried out in CIPRES Science Gateway (Miller, Pfeiffer Schwartz, 2010), and the .xml file was manually modified to “Ambiguities = true” for the nuclear partition (c-mos) to account for variability in the heterozygote positions, instead of treating them as missing information. For all analyses implemented in BEAST, the convergence of runs was assessed by the productive sample size (ESS) values of parameters (>200) applying TRACER v.1.6 (Rambaut et al., 2014). LogCombiner and TreeAnnotator (both offered in BEAST package) have been utilized to infer the ultrametric tree immediately after discarding 10 as burn-in. Nodes have been viewed as strongly supported if they received ML bootstrap values 70 and posterior probability (pp) assistance values 0.95 (Wilcox et al., 2002; Huelsenbeck Rannala, 2004). To determine divergent lineages inside Rhynchocalamus, putative species boundaries had been tested employing the multi-rate Poisson Tree Processes (mPTP; Kapli et al., 2016) model, utilizing the webserver (http://mptp.h-its.org/). This is an improved process of the previously published species delimitation method PTP (Zhang et al., 2013). As this analysis relies on single locus information, we reconstructed a ML haplotype concatenated mitochondrial phylogenetic tree as specified above for dataset four. A nuclear network was constructed for the nuclear gene c-mos. SeqPHASE (Flot, 2010) was employed to convert the input files plus the software PHASE v.2.1.1 to resolve phased haplotypes (Stephens, Smith Donnelly, 2001; Stephens Scheet, 2005). Default settings ofTamar et al. (2016), PeerJ, DOI ten.7717/peerj.7/PHASE had been employed, except for phase probabilities, which have been set as 0.6. The phased nuclear sequences had been applied to create a median-joining network making use of NETWORKS v.five (Bandelt, Forster R l, 1999). As a way to test alternative topologies, topological constraints had been constructed. We enforced alternative topologies and compared to the unconstrained ideal ML tree, with the Approximately-Unbiased (AU; Shimodaira, 2002) and Shimodaira-Hasegawa (SH; Shimodaira Hasegawa, 1999) tests. Per-site log likelihoods had been estimated utilizing raxmlGUI v.1.3 (Silvestro Michalak, PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/20008976 2012) and p-values have been calculated making use of CONSEL (Shimodaira Hasegawa, 2001).Divergence time estimatesUnfortunately, no calibration information for Rhynchocalamus are presently identified, precluding the usage of internal calibration points and preventing a direct estimation on the time in our phylogeny. Therefore, for any temporal framework we made use of calibrations of other members of the Western P.
