This web page was produced as an assignment for an undergraduate course at Davidson College.

Complete Mitochondrial Genomes of Ancient Canids Suggest a European Origin of Domestic Dogs

Paper Review by Ellen Johnson

Previous mitochondrial and Y chromosome data indicated domestic dogs were of East Asian and Middle Eastern origin dating 15,000 years ago while fossilized remains indicated domestic dogs were of European and Siberian origin dating 36,000 years ago. Researchers sought to resolve this conflict regarding the origin of domestic dogs. Using mtDNA analysis of modern dog, modern wolf, and ancient canid sequences, they concluded that modern dogs are of European origin. The origin of the domestic dog ranges from 18,800 to 32,100 years ago.

Previous research also indicated that modern dogs underwent two population bottlenecks. Using Bayesian Skygrid analysis, researchers came to support this conclusion. Initial domestication efforts may have contributed to steady increase in dog population size from 18,800 years ago to 5,000 years ago. More recent dog breed formation may have contributed to the sharp population size increase since 2,500 years ago.

Overall, the paper was easy to follow and the figures were easy to interpret. The phylogenetic tree generated with mtDNA sequences is well supported with the majority of the branches meeting bootstrap values above 90%. The number and scope of ancient genomes compared is impressive.

The methods used to generate the figures were the hardest to understand. Bayesian analysis was used to generate both Figures 1 and 2, but never explained in the text. This proved especially problematic for Figure 2 where researchers jump from comparative mtDNA analysis to population size analysis. Readers are not provided a clear idea as to how researchers used sequence information to track population size over the past 20,000 years.

I found their determination of the origin of the domestic dog much more convincing than their determination of population bottlenecks. Much of that stems from my inability to detect the logical transition from mtDNA sequences to population size trajectories. Without that transition, readers are left to take the researchers’ claims at face value. This is a problem for several statements made throughout the paper. For example, readers are expected to follow the logic that since Belgian canids have distinct mtDNA sequences from modern dogs, Belgian canids are likely the result of abandoned domestication or a newly discovered population of gray wolves. I did not find this or several of their other conclusions to be self-evident.

I found their critique of the study refreshing since many papers do not explicitly point out potential shortcomings. They mainly focused on potential problems when using only a single genetic locus. Given that they analyzed complete mtDNA sequences, I presume they define the entire mitochondrial genome as one loci. This point could use clarification. Closer examination of the critique revealed self-promoting undertones. Researchers suggest idealized rather than practical ways to improve upon their study. For example, analysis of multiple loci from nuclear genomes could provide more statistical power but performing such analysis is unlikely given the old age and poor preservation of many of the DNA samples.


Table 1:

Table 1 shows statistics regarding the complete mtDNA analysis of 18 prehistoric canids. All canids were of European and American origin. DNA capture and high-throughput sequencing resulted in median 12-fold coverage of the 18 prehistoric canid mtDNA genomes. An average 15,014 nucleotides are supported by at least twofold coverage.

Figure 1:

Figure 1 shows a phylogenetic tree generated by comparing mtDNA genomes of the 18 prehistoric canids to mtDNA genomes of 49 modern wolves and 77 modern dogs. The mtDNA genomes of four coyotes served as the outgroup. Labels identify each sequence by country of origin and approximate age in years before present. Orange lines and labels indicate wolf sequences. Blue lines and labels indicate dog sequences. Gray line and labels indicate ambiguous classification. Asterisks indicate bootstrap values greater than 90%. 



Belgian dogs contain the most distinct mitochondrial DNA sequences. The mtDNA analysis indicates they are not direct ancestors of modern dogs. Researchers speculate the Belgian canids are either part of aborted domestication or a newly discovered population of gray wolf. While I appreciate they are speculating, researchers provide insufficient explanation to persuade readers that these are likely conclusions.


Modern dogs separate into four clades labeled A-D. Clade A is the most diverse and contains 64% of all the modern dog sequences studied. Clade A shares a most recent common ancestor (MRCA) with pre-Columbian New World dogs that lived 18,800 years ago (green box). Clade A and New World dogs share a MRCA with a wolf from Switzerland that existed 32,100 years ago (blue box). Researchers do not provide readers with a satisfying transition from this data to their conclusion that dogs likely arrived in the New World with the first humans. It is placed in the text without explanation and does not seem particularly relevant to their main point regarding the origin of dog domestication.



Clade D contains sequences from two Scandinavian dog breeds. These breeds share a MRCA with a wolf-like canid from Switzerland that existed approximately 18,300 years ago (green box).


Clade C has a MRCA with ancient German canids that existed from 16,000 and 24,000 years ago (red boxes).


Clade B is most closely associated with European wolves from Sweden and the Ukraine. The MRCA existed about 9,200 years ago (green box).

Figure 1 Conclusions: Clades A, C, and D appear most closely associated with ancient European canids. Clade B appears most closely associated with European wolves. Asian and Middle Eastern wolf sequences do not show close associations to any of the modern dog clades. Researchers conclude that modern dogs are of European origin rather than East Asian or Middle Eastern origin as previous genetic data suggested. The origin of the domestic dog ranges from 18,800 (MRCA of clade A and New World dogs) to 32,100 years ago (MRCA of clade A/New World dogs and European wolf from Switzerland). Given this date range, researchers further conclude that dog domestication began in European hunter-gatherer societies. This is yet another statement extrapolated from the data with seemingly insufficient support.

Figure 2:

Figure 2 depicts clade A and pre-Columbian dog population size trajectories. The solid line indicated population size (median logNe). Dates are shown in years before present. The plot shows continuous population growth from the MRCA to about 5,000 years ago. Population growth declines from 5,000 to 2,500 years ago. Population growth increases most dramatically from 2,500 years ago to present day.

Figure 2 Conclusions: Researchers conclude the steady growth may be the result of the first domestication efforts while the sharp increase may be the result of more recent dog breed formation. They also note that human population sizes follow a similar pattern indicating dog populations are likely dependent on human populations. They do not provide any of the human population data used for this comparison.



Thalmann, O et al. 2013. Complete Mitochondrial Genomes of Ancient Canids Suggest a European Origin of Domestic Dogs. Science 342: 871-4.


Return to Home



Genomics Page
Biology Home Page

Email Questions or Comments.

© Copyright 2014 Department of Biology, Davidson College, Davidson, NC 28035