Despite massive reductions in the cost of DNA sequencing over the past decades, researchers remain faced with decisions about how to distribute sequencing effort along three dimensions: (a) how much of the genome to sequence (breadth of coverage), (b) how deeply to sequence each sample (depth of coverage), and (c) how many samples to sequence. Until recently, reduced-representation sequencing (e.g., RAD-seq [restriction site-associated DNA sequencing]), through which a small random portion of the genome can be sequenced deeply in many individuals to allow for simultaneous variant discovery and high-confidence genotyping, has been the most popular approach for population genomics of nonmodel organisms