To best understand this article in the context of the visual breed identification literature, please see National Canine Research Council's complete analysis here.

Article Citation:
Voith, V. L., Ingram, E., Mitsouras, K., & Irizarry, K. (2009). Comparison of Adoption Agency Breed Identification and DNA Breed Identification of Dogs. Journal of Applied Animal Welfare Science12(3), 253-262. doi:10.1080/10888700902956151

National Canine Research Council Summary and Analysis:

This study compared adoption agencies’ visual breed identifications of 20 mixed-breed dogs against DNA identification. The results showed little agreement between reported breed (from visual identification) and actual breed as determined by DNA analysis. Of the 16 dogs that had been assigned a specific breed as predominant by their adopting agency, DNA analysis showed that only 4 were actually predominantly comprised of said breed. Furthermore, for 3 of the 4 that did have genetic markers matching their visually identified breed, that breed contributed at most 12.5% of their makeup. Finally, those 3 dogs also had other breeds of equal contribution that were not identified by their adopting agencies. One additional dog showed a less dominant breed that matched its visual identification. In all only 31% of the dogs showed any DNA evidence of the adoption agencies’ named breeds anywhere in the dogs’ ancestry. Over 90% of the dogs did not have their visually identified breeds as the predominant breed in their DNA analysis. This marked unreliability of visual breed identification presents a strong case for ending the common practice of using data based on visual breed identification as a basis for studies which are then used to justify public and private policies regarding whether and how dogs so identified may be kept.

The researchers used dogs adopted from shelters and rescues because they commonly label their dogs by breed for adopters. Owners (students, staff, and faculty at the College of Veterinary Medicine at Western University of Health Sciences) volunteered 50 dogs for the study, of which twenty from 17 different agencies were ultimately included. All dogs were adopted from a shelter, rescue, animal control, or similar agency. Owners completed surveys detailing the dogs’ demographic information including weight, age, sex, and breed. Researchers verified the weight and five dogs were randomly selected from each of four weight categories (<20 lb, 21–40 lb, 41–60 lb, and >60 lb) for a total of twenty subjects. Blood samples were taken and sent to MARS VETERINARYTM laboratory to be analyzed using The Mars Veterinary Wisdom Panel MXTM. DNA analysis was conducted and breed was determined using a statistical model that infers breed from a pattern of 300 genetic markers.

All dogs were determined to be mixed breed by their adopting agencies and 80% were assigned at least one predominant  breed. Four dogs were assigned a type rather than breed, and one was assigned both a type and a breed. DNA testing reported all breeds that contributed at least 12.5% of a dog’s makeup. Breeds were classified as either predominant or less dominant based on their representation in the DNA analysis, such that the breed(s) with the highest reported percentage was labeled as predominant.

The data are compelling evidence that visual breed identification is unreliable. There are additional troubling implications of these results, particularly in the realm of canine research and policy. Many studies have been conducted linking breed to behavior, and many of these studies have relied on visual breed identification as the basis for analysis.

Considering the potential severe implications for both the owners and dogs of specific breeds and/or types from Breed-Specific Legislation (e.g., bans, euthanasia, containment, and increased fees), agencies should exercise great caution when labeling a dog of unknown ancestry. If the lineage is unknown and DNA analysis unavailable, a “best guess” based on visual physical characteristics should not be made; it is reckless to do so given these data that demonstrate its unreliability. Moreover, researchers cannot responsibly conduct experiments nor cite studies that rely on visual identification for determining breed. Unless DNA analysis is conducted or the dog’s parentage is documented by pedigree, specific breeds or breed groups should not be indicated.

There is a potential sampling bias in this study that should be considered. It is plausible that participants who disagreed with the breed assigned to their dog were more likely to volunteer their dogs for DNA testing. Additional studies with larger random samples could be conducted to rule out this possibility. However, more recent studies by Voith et al. (2013) and Olson et al. (2014) do corroborate the present findings, so this weakness is unlikely to have affected the data.

Finally, the authors describe several limitations of DNA analysis. The technique involves comparing the target dog’s genetic markers to a large database of dogs with known parentage. Because of this reliance on a sample group, results may differ based on the composition of the comparison group as well as the specific statistical model used. This means that an individual dog could render different results over time or database. While this may seem like a flaw in the study, it only further supports the notion that breed is ambiguous at best and should not be used to categorize a dog’s behavior or to influence public policy.

Abstract and Link to Purchase Full Text of the Original Article: