The sequence reads were processed by Stacks version 2.6[1].We used the process_radtags function to clean the tags, discarding reads of low quality (-q), removing reads with uncalled bases (-c) and rescuing barcodes and radtags (-r). SNP calling for each individual was done using the program denovo_map.pl of Stacks. The denovo_map.pl included six core components: building loci (ustacks), creating a catalog of loci (cstacks), and matching samples back against the catalog (sstacks), transposing the data (tsv2bam), adding paired-end reads to the analysis and calling genotypes (gstacks), and population genomics analysis (populations). 

-m 10 and -M 6 in the ustacks function

 -r 0.80 and --min-maf 0.05 in the populations function

other options set to default

The fsthet method [2] was used to identify loci with FST values that were excessively high or low compared with what was expected under neutrality. The analysis was performed using the fsthet package [2] on the R platform [3] by considering the loci below or above the 95% confidence intervals constructed with 1000 bootstraps for the expected relationship between HE and FST as outlier SNPs.

[1] N. Rochette, A. Rivera‐Colón, and J. Catchen. Stacks 2: Analytical methods for paired‐end sequencing improve RADseq‐based population genomics. Molecular Ecology, 28(21):4737-4754. 2019.

[2] Flanagan SP, Jones AG. Constraints on the FST–heterozygosity outlier approach. J Hered. 2017; 108(5):561–73. pmid:28486592.

[3] R Development Core Team. R: A language and environment for statistical computing. Vienna, Austria; 2018.