Prehistoric dog DNA sheds light on 11,000 years of canine evolution

In the past decade, a revolution in extracting and sequencing ancient DNA has meant that it is now possible to directly study evolutionary change at the genetic level in the context of the geographical and temporal information provided by archaeology. Pontus Skoglund and colleagues have now used this approach to shed new light on humanity’s common history with our oldest domestic companion, the dog (Bergström et al, 2020).

The first dogs most likely originated in Eurasia, when a wolf population was domesticated in the process of scavenging food from Paleolithic hunter-gatherers. These ancestors of modern dogs formed a distinct lineage, genetically separate from all present day wolves, before the peak of the last ice age, but there is little consensus regarding when, where, and how many times domestication took place, owing to the scarcity of Paleolithic dogs in the archaeological record; prior to this study, only six ancient dog and wolf genomes were available. Making any inferences as to ancestry and geographical origin from modern dog breeds has also proved to be problematic, as analysis of their genomes reveals a complex population structure.

The extent to which dog evolution has been affected by their relationship with humans was also unclear. Previous mitochondrial DNA and genomic studies suggested a link between the genetic signatures of ancient dogs and when and where they lived, but the degree to which their dispersals and adaptations were coupled to those of humans was unknown, as was the extent to which domesticated dogs and their wild wolf relatives exchanged genetic material.

To address all these issues, Skoglund, his postdoc Anders Bergström and colleagues — comprising archaeologists from around the world, and members of two other ancient DNA labs — sequenced 27 ancient dog genomes going back 11,000 years, using material taken from animals from Europe, the near East and Siberia. These were then matched in age, geographic location and cultural context with 17 sets of human genome-wide data.

Bergström and Skoglund began their analysis of these data by asking which ancient wolves might have contributed to the origins of dogs, and whether there had been multiple contributions, perhaps from different domestication centres. After their split, wolves and dogs clearly continued to interbreed, generating regional admixtures — for example, Iberian wolves are genetically closer to European dogs than to Asian dogs — but gene flow was seemingly all one way: from dog to wolf. If wolf-to-dog gene flow had occurred, one would expect to find that all wolves would be more similar to wolf-like dogs than to non-admixture dogs. In fact, some wolves are equally distant from all dogs, past and present, suggesting the opposite. Gene flow exclusively from dogs into their wild relatives is intriguing as it is contrary to that discovered for other domesticated species, where wild introgression is common. This result is also consistent with a scenario in which all dogs derive from an ancient now-extinct wolf population; domestication perhaps only happened once. 

Having so many samples allowed the group to ask questions about when and where distinct dog lineages first arose. By 11,000 years ago, at the Pleistocene-Holocene boundary, it was clear that there were already five major dog ancestry lineages in existence: Neolithic Levantine, Mesolithic Karelian, Mesolithic Baikalian, ancient American, and the forebears of the New Guinea Singing Dog and Australian Dingo. This diversity means that the first dogs from which they all descend must have arisen several thousand years before that, well before the agricultural revolution and therefore predating all other domesticated species. The earliest Neolithic European dogs, living some 7,000 years ago, were a mixture of the Karelian and Levantine branches.

Analysis of genetic drift in ancient European and near Eastern dogs revealed a diagonal, linear gradient — a cline — that fitted well with previous studies suggesting an eastwest axis of dog ancestry. Modern European dogs all sit at the midpoint of the cline, indicating their dual origins, but also showing the homogenisation of the diversity present 5,000 - 10,000 years ago. Today, this modern European ancestry has dispersed globally, with only small pockets of precolonial ancestry surviving in breeds such as the chihuahua and the Rhodesian ridgeback.

The existence of the diagonal cline is hard to explain with a standard family tree-like history or by long-standing continuous gene flow. Rather, it is consistent with a major admixture episode occurring around 7000 to 10,000 years ago, which recapitulates that seen in human history, and likely corresponds to the incursion from the east of inwardly migrating farmers and their dogs mixing with the resident hunter-gatherer human and dog populations.

Using the matched sets of human and dog genomes, it was possible to show that broadly speaking, dog and human remains found in the same places and times were similar in terms of where they mapped onto the spectrum of each species’ population variation, with broad features of east-west division matching well. For example, the mixed ancestry of modern European dogs, midway between animals living with Mesolithic hunter-gatherers or with incoming Neolithic farmers, matches that of modern day Europeans, who are a mixture between Neolithic huntergatherers and Neolithic farmers from Anatolia. There were some decouplings, however. The expansion of steppe pastoralists into Late Neolithic and Bronze Age Europe transformed the ancestry of human populations, but the incomers apparently adopted the endogenous dogs, rather than bringing their own. Conversely, the eastward migrations of steppe pastoralists into East Asia had a more substantial impact on the ancestry of dogs than humans. 

Having complete genome sequences for these ancient dogs opens up the fascinating prospect of being able to track the parallel evolution of dogs and their human masters. Bergström and coworkers discuss one such shared adaptation: variation in the copy number of the AMY2B pancreatic amylase gene, which encodes an enzyme involved in starch breakdown. Starch consumption is a characteristic of agricultural societies and of hunter-gatherers in arid environments, and copy numbers of AMY1, the human version of AMY2B, are on average higher in ancient populations with high-starch diets. This is mirrored in dogs: the hunter-gatherer dogs had generally retained lower copy numbers of AMY2B than agriculturalists’ dogs, although there was some variation across the whole set of samples.

While this paper settles a number of longstanding controversies about the origin of dogs, the central question of where and when dogs first arose has yet to be answered. The solution to this puzzle likely lies in identifying the now-extinct Paleolithic wolf source population, something Skoglund is currently engaged upon.

This work is part of a wider endeavour in the Skoglund lab, as it is also acting as a model system for the study of mammalian evolution using ancient genomes. This has great biomedical relevance, as all human biology and disease is the outcome of evolution, and many current health challenges have appeared or radically changed their dynamics recently with changes such as agriculture, urbanisation, and industrialisation. Linking evolution and population history to historical factors such as epidemics, environmental and societal changes, will uncover new insights about human biology and its mechanisms.