Refer to the following publication for complete methodology details:

Goetze, E., Andrews, K., Peijnenburg, K. T. C. A., Portner, E., Norton, E. L. (2015) Temporal Stability of Genetic Structure in a Mesopelagic Copepod. PLoS One 10(8): e0136087. doi:10.1371/journal.pone.0136087

In summary (excerpted from above):

For H. longicornis species 1, deviations from Hardy-Weinberg equilibrium (HWE) and linkage disequilibrium were examined using ARLEQUIN v3.5.1.3 and GENEPOP v4.2 for all microsatellite loci [36–38]. We tested for the presence of null alleles in microsatellite data using MICROCHECKER v2.2.3 [39], and estimated null allele frequencies and calculated population pairwise FST values with correction for null alleles in FreeNA [40]. Microsatellite genetic diversity indices of observed and expected heterozygosity, average alleles per locus, and allele richness were calculated in GENETIX v4.05 and FSTAT [35,41]. Pairwise FST values were calculated among all sample sites using both microsatellite and mtCOII data, as a measure of population subdivision across samples (ARLEQUIN v3.5.1.3, [38]). Significance was assessed following correction for multiple comparisons using the false discovery rate (FDR, [42,43]). Pairwise ΦST values also were calculated for the mtCOII data. We identified the nucleotide substitution model that best fit our mtCOII data using the Akaike Information Criterion, as implemented in jModelTest v2.1.4 [44], and the K81 or three-parameter model was selected as the best model (TPM3uf+G). The Tamura and Nei substitution model, which was the closest available model in Arlequin, was used to calculate pairwise and global ΦST values, and to estimate genetic diversity at each site. Hierarchical Analyses of Molecular Variance (AMOVA) based on FST were carried out to partition the genetic variance across both space (ocean gyres) and time (sampling years), for both marker types. In these analyses, we tested for population structure under the following groupings: with samples stratified by (1) northern and southern subtropical gyres (2 gyres), and (2) across two sampling years (2010, 2012). Global FST values were estimated using non-hierarchical AMOVAs among all samples, as well as among subsets of the data across ocean gyres and sampling years. Significance was tested with 10,000 permutations of genotypes or haplotypes among populations. Principal coordinate analysis (PCA) plots of linearized pairwise FST values based on both mtCOII and microsatellite data were used to visualize spatial and temporal genetic differentiation among samples. Population structure was further examined using a Bayesian clustering method implemented in STRUCTURE [45,46] for microsatellite loci. We used admixture and correlated allele frequency models, with a burn-in of 105 steps followed by 106 steps, with and without using sampling location as a prior. We ran these analyses for each of the 2010 and 2012 datasets using K = 1 to K = 10, and for the dataset of combined years using K = 1 to K = 20. We ran three separate replicates for each K to investigate consistency of Pr(X|K). The true K was evaluated by visual inspection of barplots and comparing Pr(X|K) across K values.

HALO_population.csv