Researchers seeking to generate genomic data for non-model organisms are faced with a number of trade-offs when deciding which method to use. The selection of reduced representation approaches versus whole genome resequencing will ultimately affect the marker density, sequencing depth, and the number of individuals that can multiplexed. These factors can affect researchers' ability to accurately characterize certain genomic features, such as landscapes of divergence-how FST varies across the genomes. To provide insight into the effect of sequencing method on the estimation of divergence landscapes, we applied an identical bioinformatic pipeline to three generations of sequencing data (GBS, ddRAD, and WGS) produced for the same system, the yellow-rumped warbler species complex. We compare divergence landscapes generated using each method for the myrtle warbler (Setophaga coronata coronata) and the Audubon's warbler (S. c. auduboni), and for Audubon's warblers with deeply divergent mtDNA resulting from mitochondrial introgression. We found that most high-FST peaks were not detected in the ddRAD data set, and that while both GBS and WGS were able to identify the presence of large peaks, WGS was superior at a finer scale. Comparing Audubon's warblers with divergent mitochondrial haplotypes, only WGS allowed us to identify small (10-20kb) regions of elevated differentiation, one of which contained the nuclear-encoded mitochondrial gene NDUFAF3. We calculated the cost per base pair for each method and found it was comparable between GBS and WGS, but significantly higher for ddRAD. These comparisons highlight the advantages of WGS over reduced representation methods when characterizing landscapes of divergence.