A canine identity crisis: Genetic breed heritage testing of shelter dogs

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:
Gunter, L. M., Barber, R. T., & Wynne, C. D. (2018). A canine identity crisis: Genetic breed heritage testing of shelter dogs. PloS one, 13(8), e0202633. https://doi.org/10.1371/journal.pone.0202633
 

National Canine Research Council Summary & Analysis:

This paper is included because it is the largest study to date that relates shelter breed labels to canine genetics. The purpose of the study was to report the breed heritages of dogs housed in animal shelters and compare these objective findings to the labels given by shelter staff. The findings led the authors to conclude that, “when we consider the complexity of shelter dog breed heritage and the failure to identify multiple breeds based on visual identification coupled with our inability to predict how these breeds then interact within an individual dog, we believe that focusing resources on communicating the physical and behavioral characteristics of shelter dogs would best support adoption efforts.”
 
Two limited admission shelters—one in San Diego, CA (April 2015-April 2016) and the other in Phoenix, AZ (December 2014-August 2015)—participated in the study. All new intakes at both shelters were enrolled in the study. Wisdom Panel Canine DNA tests were used to collect buccal cells for DNA analysis.
 
Three generations of heritage were reported (to the great-grandparent level; 2 parents, 4 grandparents, and 8 great-grandparents). If a single breed signature could not be identified at the great-grandparent level, the dog was labeled “mixed.” The DNA report included the two most predominant breeds so that they may be compared to the primary and secondary breeds listed by shelter staff.
 
For the shelter breed labels, staff at the San Diego shelter (San Diego Humane Society) used existing protocol of assigning breed based on visual inspection. Staff were allowed to enter up to two breeds (primary and secondary), but only one was required. Shelter staff were provided with an AKC breed book and encouraged to label dogs with recognized breeds. In this study, a match was determined if the breed indicated by shelter staff was the prevalent breed (the greatest number of same breed signatures identified with two or more great-grandparents) or if both primary and secondary breeds were correctly indicated. Accuracy of visual breed identification was not assessed at the Arizona shelter.
 
Only 4.9% of the total sample was determined to be purebred by DNA analysis, and only 1.3% were identified as having 2 purebred parents of different breeds. Thus the vast majority of the dogs had 3 or more breeds in their heritage. The most common breed signature across the total sample was American Staffordshire Terrier, with 24% (Arizona Animal Welfare League) to 27.8% (San Diego Humane Society) of the dogs having this heritage at the level of at least 1 great-grandparent.
 
Oddly, the authors singled out the most common breed signature (American Staffordshire Terriers), and combined them with three other morphologically similar breeds which they dubbed “pit bull-types” and then compared this group to the entire remaining study population of 121 breeds (87 common to both shelters), in terms of length of stay at the shelters. The “pit bull-types” length of stay was twice as long. However, such a comparison does not take into account any number of factors that may affect length of stay. Size is one example. If the dogs with at least one great-grandparent from among the group of breeds designated by the authors were on average larger than the average across the rest of the shelter population and size was positively correlated with length of stay, this alone could account for the difference between the two groups. This group was also analyzed separately from the rest of the population in terms of staff visual breed identification accuracy. The larger the percentage of these breed signatures in a dog’s DNA, the more likely it was that one of the breeds would be visually identified. Without creating analogous groups of nearly equally common breeds (e.g., Chihuahuas and poodles) grouped with morphologically similar breeds, it cannot be assumed that their results would be any different.
 
Overall, compared to other studies on the topic, visual breed identification was better than previously reported. The primary breed labels assigned by shelter staff matched the prevalent breed by DNA analysis 57% of the time. However, 33% of the time, the label given by staff was not in the dog’s DNA, even at the great-grandparent level. That is, for 1 out of 3 dogs, the label given was completely incorrect. These statistics are for the San Diego Humane Society only, as accuracy of visual breed identification was not assessed at the Arizona Animal Welfare League.
 
Shelter staff were not allowed to label dogs as “pit bulls.” They were obligated to choose a specific breed, with the option to indicate 2 breeds. A dog labeled as one of the four breeds selected in this study as “pit bull-type” (e.g., American Staffordshire Terrier, American Bulldog, Bull Terrier, and Staffordshire Bull Terrier) was considered a match to the “pit bull-type” group if the DNA analysis revealed only one of these breeds to be present at the great-grandparent level (12.5%). The result is that this grouping, which, it is important to emphasize, is not a breed grouping recognized by either of the major breed registry organizations, resulted in a much larger target for an “accurate” visual breed identification than for any other dogs. Interestingly, if the staff making the identifications designated an actual breed group (e.g., “Terrier”) this was not considered a legitimate identification and was thrown out.
 
As with other papers in this realm (e.g., Hoffman et al., 2014 and Olson et al., 2015), there is a flaw in this study that cannot be ignored: the arbitrary classification of “pit bull-type” dogs. The authors define “pit bull-types” as any dog with one of four identified breeds: American Staffordshire Terrier, Bull Terrier, American Bulldog, and Staffordshire Bull Terrier. While these breeds were defined in this sample as “pit bull-types,” in reality there is no agreed upon definition for the classification—not in science, the law, any kennel clubs, nor animal shelters. The researchers created their own definition, but it is meaningless in any context other than this individual study; the breeds included with this label vary widely by shelter, municipality, region, and state. In fact, two different definitions are used within this paper: the introduction notes that, “The aforementioned US jurisdictions that ban these dogs use the preponderance of physical characteristics associated with the breeds of American Pit Bull Terrier, American Staffordshire Terrier, or Staffordshire Bull Terrier as their means of positive identification.” But in the paper, the researchers describe their own grouping as: “Broadening the analysis to include all the breeds typically classed as ‘pit bull-type’ (dogs with at least one GGP from American Staffordshire Terrier, American Bulldog, Bull Terrier, and Staffordshire Bull Terriers) [emphasis ours].” This further illustrates that the label used here is arbitrary—an opinion—and the importance of this cannot be overstated.
 
The authors acknowledge that their sample may be unintentionally skewed because they included only two limited-admission shelters from the southwest region of the country. Another caveat to consider is the effect of being enrolled in this study on the staff’s labeling. It is possible that their labeling habits changed because they knew that DNA analysis would reveal the dog’s true heritage. To account for this, researchers could have compared the number of breed varieties reported in the months prior to the study, to see if there were changes. Moreover, the provision of an AKC breed identification book may have impacted their labeling. These conditions could affect the generalizability of visual identification accuracy to other shelters.
 
Link to Full Text of the Original Article:
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0202633
 
Additional references:
 
 
 
Page last updated July 15, 2019
 
Share
Share on facebook
Share on twitter
Share on linkedin

Latest Research

Trending News