Reference genome choice compromises population genetic analyses.

Abstract

Characterizing genetic variation in natural populations is vital to evolutionary biology; however, many non-model species lack genomic resources. Here, we demonstrate that reference bias significantly affects population genomic analyses by mapping whole-genome sequence data from gray foxes (Urocyon cinereoargenteus) to a conspecific reference and two heterospecific canid genomes (dog and Arctic fox). Mapping to the conspecific genome improved read pairing by ∼5% and detected 26%-32% more SNPs and 33%-35% more singletons. Nucleotide diversity estimates increased by over 30%, F_ST increased from 0.189 to 0.197, and effective population size estimates were 30%-60% higher with the conspecific reference. Recombination rates varied by up to 3-fold at chromosome ends with heterospecific references. Importantly, F_ST outlier detection differed markedly, with heterospecific genomes identifying twice as many unique outlier windows. These findings highlight the impact of reference genome choice and the importance of conspecific genomic resources for accurate evolutionary inference.