Genetic diversity of DNA microsatellites and the mtCOI gene
To test for null alleles, large allele dropout, and scoring errors, we used Microchecker version 2.2.3 (VanOosterhout, Hutchinson, Wills, & Shipley, 2004). Putative null alleles were detected in five loci (MAT2, MAT4, MAT10, MAT15, and MAT28) of five populations (DAG, GUA, DOV, QBY, and YOB; Table 2), which may have biased the estimates of population structure (Chapuis & Estoup, 2007). We accounted for this bias by comparing analyses after excluding these loci with those of the entire data set. Our main results and conclusions remained unchanged, however, regardless of the loci included in the analyses, and therefore only the results for the entire data set are presented here. Since A. cephalotes nests form family units and nestmates are closely related, only one individual per nest was used to evaluate the Hardy-Weinberg equilibrium (HWE) and linkage disequilibrium (LD). Tests were performed for each locus (HWE) and each locus pair (LD) for each population (Supplementary Table ST2) using GENEPOP (Rousset, 2008). A Bonferroni correction was applied with a global significance of α=0.05 for multiple comparisons.
Genetic diversity was characterized for each locus, region, and population as allele count (Na), allelic richness (AR), expected (He) and observed (Ho) heterozygosities, and the inbreeding coefficient (F IS), using the software GenAlex v. 6.5 (Peakall and Smouse 2012) and FSTAT v. 2.9.3.2 (Goudet, 1995; 2001) (Table 2, Supplementary tables ST2, ST3). Tests for significant differences from zero in the diversity parameters were performed using the ‘aov’function in R (Chambers, Freeny, & Heiberger, 1992). We used the software Bottleneck (Piry et al. 1999) to estimate recent changes in population sizes by implementing the IAM and SSM mutation models and a Wilcoxon signed-rank test to evaluate statistical significances. We also determined whether the allele frequency distribution differed from the expected L-shape. Where necessary, significant values were adjusted for multiple comparisons through Bonferroni correction.
The genetic diversity of mtCOI data was estimated through the number of haplotypes (h), nucleotide diversity (π), and haplotype diversity (hd), using DnaSP 5.19 (Librado & Rozas, 2009). Tajima’s D and Fu’s FS neutrality tests were performed to investigate signatures of recent population expansion (Ramos-Onsins & Rozas, 2002) when the null hypothesis of neutrality was rejected due to significant negative values (P < 0.05 for D, P< 0.02 for FS; Supplementary table ST3).