DNA assembly method
Sequence assembly methods (probe versus contig) yielded important differences across analyses. Notably, in the probe dataset, we observed systematically higher differences in heterozygous/homozygous allele calls between capture and RADseq replicates than between capture-based replicates (Table 2; Fig. 3) but we did not observe this pattern in the contig dataset. We also inferred higher rates of heterozygosity in the probe dataset (Fig. S4). Further, only by using the contig dataset pruned to SNPs within the coordinates of the original RADseq loci were we able to accurately place all four technical replicates together in phylogenetic analyses (Fig. 4; S6). As such, we present results in the main manuscript for only the contig dataset (pruned to SNPs within the coordinates of the original RADseq loci) and present results for the probe dataset in the Supporting Information. See the Discussion for more on this subject.