ddRAD SNP calling
The sequences were quality assessed using FastQC (Andrews, 2010) and the reads were demultiplexed by barcode using the “process_radtags ” module (default parameters) of the Stacks v1.48 bioinformatics pipeline (Catchen et al. , 2013). This module also filtered out low-quality reads. The retained reads, now missing variable length barcodes, were then trimmed to a standard 148 bases in length. For each individual, matching forward and reverse reads were then concatenated into a single longer “artificial” read using a custom Perl script (Genbank BioProject Accession number: PRJNA1173582). Individuals that had <100,000 reads were removed from further processing.
The individual data were then processed using the “denovo_map.pl ” module of stacks (-m 10 -M 2 -n 0) to assemble and create a catalogue of genetic loci contained in the data. The Stacks scripts “export_sql.pl ” and “populations ” and filtering steps were used to retain all loci that fulfilled the following criteria: a) contained exactly one SNP (in the concatenated forward and reverse reads) to avoid physically linked markers and ensure conserved sequence surrounding the target SNP to facilitate primer design; b) contained exactly two alleles c) had a read depth of ≥10 reads per individual to maximize the likelihood of the SNP being real.