Skip to main content
. 2022 Jun 29;607(7918):313–320. doi: 10.1038/s41586-022-04824-9

Extended Data Fig. 4. A schematic model of how deep population structure could explain why dogs require ancestry from an outgroup population in qpAdm analyses.

Extended Data Fig. 4

Under this model, there is deep population structure between different wolf populations, including the wolf population that becomes the progenitor of dogs. High rates of gene flow over time largely homogenises the ancestry of all populations, but it does not completely erase the deep structure. If the true dog progenitor population is not sampled, a single-source qpAdm model involving one of the sampled wolf populations will not fit dog ancestry, because dogs do not share all of the genetic drift that has occurred in the history of the sampled population. But if an outgroup population is included as a source in qpAdm, this can account for the ‘missing’ deep ancestry in dogs, and therefore result in a model that fits dog ancestry.