Sequence Analysis
Short read processing was conducted using ipyrad v.0.7.1 (Eaton, 2014) on the Hopper HPC Cluster at Auburn University. The workflow for ipyrad involves the following steps: 1) demultiplexing raw reads; 2) quality filtering reads; 3) de novo clustering of data within samples using VSEARCH (Rognes et al., 2016) with clusters then aligned using MUSCLE (Edgar, 2004); 4) jointly estimating heterozygosity and error rate; 5) estimating consensus allele sequences from clustered reads; 6) de novo clustering of data and alignment across samples; 7) filtering/formatting output files for downstream analyses (see Table 1 for parameter settings used). Branched assemblies were created to assess different parameter settings (i.e. different clustering thresholds and minimum samples per locus) and differing number of individuals present in the matrix (i.e. all individuals and only A. unicolor B individuals), and the final data matrix including all individuals was chosen after comparing the efficacy of each built from the branched assemblies (see Results; Table 1).