2.3 Data analysis
2.3.1 Genetic diversity
Forward and reverse splicing of all sequences using SeqMan in Larsergene v7.1.0 (Swindell & Plasterer, 1997), and then compare and edit the sequences using the Clustal W method in MEGA v7.0 (Kumar et al., 2016). The COI and Cyt b sequences were matched one by one for Multi-locus Sequence Analysis (MLSA) using PhyloSuite v1.2.2 (Zhang et al., 2020). Count base composition and content, polymorphic sites (S ), and parsimony informative sites (P i) using MEGA v7.0 (Kumar et al., 2016). The haplotype numbers (H ), haplotype diversity (H d), nucleotide diversity (π ), and mean pairwise difference (K ) were counted using DnaSP v6.0 (Rozas et al., 2017).
2.3.2 Genetic structure
The Bayesian information criterion (BIC) in jModelTest v2.1.10 (Darriba et al., 2012) was used to establish a substitution model for the haplotype datasets prior to phylogenetic analysis. Subsequently, the mitochondrial COI and Cyt b gene datasets haplotypes were used to reconstruct the phylogenetic tree using the Bayesian inference (BI). The congeneric species Lagocephalus laevigatus was chosen as an out group, from NCBI access number 10400364 (COI ) and 10400369 (Cyt b ). Bayesian inference study was carried out using MrBayes v3.2.7 (Ronquist et al., 2012), and one set of four chains was permitted to run concurrently for 20 million generations. Every 1000 generations, a sample of the tree was taken, with the first 25% being eliminated as burn-in. As the sampled generations increased, the log-likelihood maintained a constant level, and stationarity was attained when the split average frequencies’ average standard deviation was less than 0.01 (Hall, 2016). Phylogenetic tree editing was done with FigTree v1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/), and median-joining haplotype network was produced using PopART v1.7 (Leigh & Bryant, 2015).
Genetic distance within and between populations were calculated using MEGA v7.0 (Kumar et al., 2016). Subsequently, AMOVA was used to quantify genetic variation using F -statistics at two geographically distinct levels of subdivision: among and within populations. To test for statistical significance, 10, 000 permutations of the fixation indexF ST were performed between pairs of populations using Arlequin v3.5 (Excoffier & Lischer, 2010).
2.3.3 Demographic history
The Tajima’s D (Tajima, 1989) and Fu’s F s (Fu & Li, 1993) tests were utilized to check for neutral evolution. To examine population growth, the mismatch distribution (Rogers & Harpending, 1992) between the sum of squared deviations (SSD) and Harpending’s raggedness index (Rg) was analyzed with Arlequin v3.5 (Excoffier & Lischer, 2010). Changes in effective population size (Ne ) over time were deduced using Bayesian skyline plot analysis in BEAST v2.6.3 (Bouckaert et al., 2019). To account for possible site-specific variations, the rate of clock mutation was fixed at 1×10−8 per year, as recommended for reef fishes (Delrieu-Trottin et al., 2017). With a sample every 1000 iterations, 100 million generations of separate independent Markov chain Monte Carlo (MCMC) studies were carried out. The molecular clock was calibrated using an average divergence rate of 2% per million years for mtDNA (Schubart et al., 1998). ESS values were detected until they reached 200, and these parameter values were displayed in Tracer v1.7.1 (Rambaut et al., 2018).