Population genetic structure and demographic analyses
The resultant sequences were aligned and analyzed using ClustalW embedded in MEGA (ver. 5.2) (Kumar, Nei, Dudley, & Tamura, 2008; Kumar, Tamura, & Nei, 1994; Thompson, Higgins, & Gibson, 1994). The aligned sequences were submitted to GenBank along with the translated amino acid sequences. GenBank accession numbers are shown in Table S1. Haplotypes were determined using DnaSP (ver. 5.10) (Librado & Rozas, 2009).
A good correlation has been reported between ground vegetation and ant community diversity (Andersen, 1995, 1997; Lubertazzi & Tschinkel, 2003). Hence, sequence data were grouped according to regions on a vegetation map of the Korean peninsula overlaid with isothermal lines (Yi, 2011). Within the range of deciduous broad-leaved forests (temperate zone), the central area was designated region A, the southwestern area was designated region B, and the southeastern area was designated region C. Region D represented the evergreen broad-leaved forest (subtropical-warm temperate zone) and region E represented Yeosu-si, a central spot on the southern coast, based on the unique characteristics of the sample collected at this site. Jeju Island, a volcanic island far from the mainland of Korea, was labeled region F. Regions G and H were the USA and Japan, respectively.
Molecular diversity indices were calculated for all eight regions and each gene. Analysis of molecular variance (AMOVA) among regions, including the overall fixation index statistics (F ST) and pairwise F ST, was performed with 1000 permutations. To test the model of evolution and demographic expansion for the COI gene, neutrality tests (Tajima’s D and Fu’s FS ) (Fu, 1997; Tajima, 1989) and mismatch distribution tests were performed with 1000 replicates using Arlequin (ver. 3.5.1.2) (Excoffier & Lischer, 2010). Based on the mismatch distribution, demographic expansion patterns for seven regions (excluding region G, i.e., the USA, which lacks variation) were determined using DnaSP and edited using Microsoft PowerPoint 2013.
Genetic distances among haplotypes were calculated after selecting the best fit substitution model in MEGA. The median joining algorithm was employed to infer phylogenetic relationships among the haplotypes using Network (ver. 4.6.10), with a fixed connection limit at 1000 steps between haplotypes (Bandelt, Forster, & Röhl, 1999). The haplotype network was edited manually and reconstructed with the regional distribution data using Adobe Illustrator CS6 (Adobe Inc., San Jose, CA, USA).