Breeds with the BEST & WORST genetic diversity

By Carol Beuchat PhD

Carrying on from my blog post "What are we going to do about Terriers" (which you will need to read in order to understand the rest of this), I decided to post the pages for all of the groups that I had compiled.

From these, I have now gathered together a page of "best" and "worst" breeds that will make it easier to get a sense for which breeds are in the most trouble. Note that I did not include all breeds for which there are data on the MyDogDNA (MDD) website, and for some breeds there aren't enough data to display. So if you don't see your breed on the "worst" list, don't take that to mean that it's in robust genetic health. Go see if there are data available that I didn't include.

Here's a quick primer about how to interpret these data.

For each breed there is a scatter plot and a line graph. The scatter plot displays the data points for the individual dogs. They are usually color coded by country, and on the MDD website there are sometimes options to display these data by use (e.g., show, work, companion, etc).

Note that the scatter plot is not a typical graph: the x-axis is labeled "genetic relationships", and the y-axis has no label at all. This is because the graph was created by taking all of the data for a breed and asking the computer to find some characteristics in the DNA data that would separate all of the individuals along a gradient in the horizontal direction, then do the same thing in the vertical direction. This will separate all of the data in a way that displays the degree of genetic similarity (or difference) between individual dogs as proportional to the distance between their points. Then if you code the points by country of origin, you might see that the clusters correspond to different countries (like these data for Golden Retrievers). This is useful because it can tell you whether there are clusters of genetically distinct subpopulations in a breed, which would be handy if you wanted to outcross to improve genetic diversity in a litter. Because we don't really know what the actual differences are that separate these groups, the axes of the graph are usually labeled as "principal components"; in this case, "genetic relationships" is similarly vague.

You will also see plots like this of multiple breeds. For these, you need to be very careful about interpretation. A group of breeds that clusters one way in one analysis, might cluster very differently if a few breeds are added or removed. This means that you might think that breeds A and B are very similar genetically because they cluster near each other on a graph, but find them far apart when they are displayed in a different mix of breeds. So for example, among the sporting dogs, the Toller seems to be very similar to the Chesapeake Bay Retriever (upper figure), but in an analysis of breed relationships, the Toller doesn't even cluster with the retrievers, never mind the Chessie (lower figure).
​

Shannon et al 2016. Genetic structure in village dogs reveals a Central Asian domestication origin (supplement)

​The line graph below displays the data for percent heterozygosity, which is the proportion of loci that are heterozygous (the two alleles at the locus are different). Higher numbers are better (in the "green-is-good" zone), and lower numbers correspond to low heterozygosity (= higher homozygosity = higher inbreeding).

The median value for Goldens is low (32.0%), but there is a tail in the "good" direction (representing dogs that are less inbred) and, as you can see from the scatter plot above, there are genetically distinct subpopulations that could be exploited to improve heterozygosity.
​

Note that the orange line is not a depiction of "normal", or "good", or even adequate. It simply represents all of the dogs in the database, and you will note that the peak is down where green is fading into a moldy-looking yellow. A genetically healthy breed will have high heterozygosity and therefore a low incidence of disorders caused by recessive mutations because they must be homozygous to be expressed. The DNA testing we do is not "health" testing; it is mutation testing. If we want to have a health test, it should be the measurement of heterozygosity, which will suit for ANY recessive mutation, known or otherwise, and we could toss the mutation-specific tests. It would also reflect the genetic health of the immune system, which under natural selection is maintained as the most diverse part of the genome.

Bear in mind when you look at these data that they are not part of a formal study. The data are from whatever dogs are submitted for testing. It's tempting to say that they're biased, or not representative of your breed, or skewed towards show dogs, or too heavily influenced by dogs from Finland (which is where MyDogDNA is), or whatever. Don't do this. More data might shift some curves around a bit, but for most I suspect things won't change enough to matter. There are only a few breeds that can burnish their diversity credentials. For the rest, breeders should consider how they can minimize loss of the diversity they still have and make better use of it to improve the health of the next letter.

If you sailed through the explanation of the graphs, head over to the page of "best" and "worst" breeds.

And if you're not happy about the data for your breed, join one of the ICB courses that teach you the basics of population genetics and how to breed to produce healthier dogs by protecting the health of your gene pool. (The next course starts 4 April.)

---

Check out
ICB's online courses
​
*******************

Coming up NEXT -

Basic Population Genetics for Dog Breeders
Class starts 4 April 2016

***************************************

Visit our Facebook Groups

ICB Institute of Canine Biology
...the latest canine news and research

ICB Breeding for the Future
...the science of dog breeding
​

---