Australian Shepherd Health & Genetics Institute

Australian Shepherd Health & Genetics Institute

Generic selectors
Exact matches only
Search in title
Search in content

Is Outcrossing Dangerous?

Italy-flagby C.A. Sharp

First published in the “Australian Shepherd Journal,” Mar/Apr 2005


Breeders’ lore has long maintained that outcrossing ought to be done only occasionally if at all because doing so dilutes the qualities of a line courts disaster.  There are numerous tales about breedings that went wrong because of an outcross.  But is this really the case?What is an outcross?Before deciding whether outcrossing is or is not a good idea, we need to have a good understanding of what it is.  Generally speaking, an outcross is the mating of two largely unrelated animals within the same breed.  Virtually every dog in a breed will be at least slightly related to all others if you search far enough back in the pedigrees.When people review printed pedigrees of prospective mates and see few if any common names behind both dog and bitch, the resulting litter is assumed to be an outcross.  However, if one does not have a thorough knowledge of the breed’s pedigree history, an apparent outcross may be nothing of the sort.In populous breeds like the Australian Shepherd it is very possible that two dogs will share no common ancestry on a three- to five-generation pedigree but will actually be closely related.  The breed average 10-generation coefficient of inbreeding (COI) for Australian Shepherds is somewhere around 14%, or slightly higher than that of half-siblings who are otherwise unrelated.  The COI is calculated over ten generations because fewer will not represent an accurate measure of background inbreeding.   Most Aussies are already closely related, though it is possible to find some with substantially different pedigrees.

The best way to gauge whether a cross is or is not an outcross is by calculating the COIs of the parents and the proposed litter.  If the result for the cross is low (under 6.25%) or at least substantially lower than that of the lowest parental COI, it can be termed an outcross.  Here are some examples:

Do the Numbers

For those who enjoy delving into the complexities of mathematics, here is some information on formulae that pertain to the discussion of linebreeding vs. outcrossing and their impact on the likelihood of producing undesirable traits.

An excellent explanation of Wright’s Coefficient of Inbreeding (COI) can be found in Willis, Genetics of the Dog (Howell 1989) p. 320-5. Discussion of the Hardy-Weinberg equilibrium formula can be found in Ackerman, The Genetic Connection (AAHA 1999) p. 18.

Under the Hardy-Weinberg law, the possible genotypes for a gene with alleles A and a, where p is the frequency of A and q is the frequency of a in the population:

AA pxp
Aa 2xpxq
aa qxq

If you factor in COI, represented here by F, you get:

AA pxpx(1 – F) + pxF = pxp + pxqxF
Aa 2xpxqx(1 – F) = 2xpxq – 2xpxqxF
aa qxqx(1 – F) + qxF = qxq + pxqxF

As the COI increases, the proportion of heterozygotes (Aa) decrease while the homozygotes (AA and aa) increased. If aa is the genotype for an unwanted trait, the number of dogs exhibiting that trait will have increased because of inbreeding. The higher the COI the more frequently you will see the unwanted trait.

For further reading:
Falconer & Mackay, Introduction to Quantitative Genetics (Longman 1997) p62

Hartl, A Primer of Population Genetics (Sinauer 1988) p54

The author would like to thank James Seltzer, PhD, mathematici The an and Dalmatian breeder, for his assistance with this sidebar.

  • Litter A: The dog has a COI of 16.4%; the bitch’s COI is 11.3%. Their pedigrees have nothing in common for at least five generations. The COI for their litter is 13.0%, therefore they actually have a great deal in common beyond the printed pedigree. Litter A is not an outcross.
  • Litter B: Both dog and bitch share a common great grandsire. The dog’s COI is 9% and the bitch’s is 6.3%. Except for that one great grandsire, their extended pedigrees share almost nothing. The COI of the cross is 1.8%, of which half comes from that common great grandsire. Litter B is an outcross.
  • Litter C: Both dog and bitch are tightly linebred though from different bloodlines, with COIs of 25% and 37% respectively. The cross between them would have a COI of only 2.3%. Though they are both rather inbred they have almost no common pedigree. Litter C is an outcross.
  • Litter D: The dog’s COI is 1.3% and bitch’s is 6.9%. Their litter is 13.2%. The dog’s sire and the bitch’s dam were from the same line. Litter D is not an outcross.

Now that we have established what an outcross is (and is not) we need to examine the question of how useful it is as a breeding tool.

But does it work?

There are volumes of breeder lore supporting the assumption that outcrossing is not a viable breeding strategy if done over multiple generations, though it might be done occasionally within a linebreeding program but not without some risk.  That assumption is not supported by genetic science or mathematical analysis, [see sidebar] so why is this belief so pervasive?

In large part it stems from the assumption that linebreeding is superior because it allows the breeder to concentrate desirable genes while at the same time eliminating those that are undesirable.  Linebreeding is very effective for fixing trait that are readily observed or measured.  A trait is genetically “fixed” when it is consistent throughout a population, like the color black in Schipperkes.  If linebreeding were not effective in this respect, we wouldn’t have so many distinct breeds of dog.

Genetically complex traits can be difficult to fix, though diligent selection for them over many generations can significantly increase their frequency in a breed population.  Herding ability is a genetically complex trait involving multiple genes plus environment.  Even in bloodlines where selection for a high level of herding ability has been strong for generation after generation, the breeder still will produce some pups that don’t measure up.

The success of linebreeding as a technique for consistently producing desirable traits has fed the myth that undesirable traits can be totally purged from a line or breed.  While an undesirable trait—whether it is something cosmetic like color or a serious health concern like epilepsy—can be reduced in frequency through diligent selection against it, in most cases it will not be possible to eliminate the trait entirely unless it is the result of a single gene dominant.  Most such traits were weeded out of modern breeds a long time ago.

If the trait is a single-gene recessive, careful and consistent selection against it will eventually reduce its frequency to low levels.  The trait will not be produced often, but it will still crop up from time to time.  Rare breeds may actually lose the gene through a process called genetic drift simply because their populations are so small that there is a fair chance the unwanted gene won’t get passed along.  In populous breeds, however, this is unlikely to happen.  The key to reducing the frequency of unwanted genes as much as possible is consistent negative selection no matter whether linebreeding or outcrossing is employed.

As we saw with herding ability, desirable traits with complex inheritance are impossible to fix.   Total elimination of such traits is equally impossible outside of very small breeding populations.  Hip dysplasia (HD)  is but one example.  The key to reducing the frequency of something like HD is diligent and consistent selection.

Though traits like HD can’t easily be eliminated, it may be possible to “clear” a line by elimination of some portion of genetic puzzle that produces them.  A single line is often a small breeding population within the breed as a whole.  Here is how it works:  If BINGO represents the combination of genes necessary to produce HD, a combination of linebreeding and diligent negative selection may produce a line lacking one or more elements.  This line would never produce that trait so long as it is kept closed.

But he’s such a nice dog…

A serious impediment to significantly reducing the frequency of unwanted traits is the set of priorities every breeder must establish for her breeding program:    Which traits are vital, which can be tolerated, and which are unacceptable.  It is impossible to achieve every breeding goal in a single litter, so the breeder will rearrange that priority list somewhat from one mating to the next.  This can interfere with the effort to eliminate unwanted traits because unless most or all breeders are consistent in viewing the trait as intolerable, it will persist.  Even if the gene frequency is significantly reduced over time, all it takes is one popular sire who is a carrier to make the trait common once again as people linebreed on him and his near kin.  Too often one hears the refrain, “but he’s such a nice dog!” even though that dog has or produces a serious defect or disease.

The complexities of breeder priorities are not the only reason unwanted genes persist.  Breeders may not know those genes are there  until the trait crops up in a litter.  The days of huge kennels with meticulous records on generation after generation of dogs are long gone.  Most breeders have only a very few dogs and regularly need to make use of outside studs or purchase new stock from another breeder.  Unless everyone keeps all their cards on the table, linebreeding cannot be fully effective in clearing genes for unwanted traits.

If the chain of information is broken anywhere in the breeder network, someone will make a disastrous mating choice no matter whether they linebreed or outcross.  When it happens in an outcross, people tend to ssume the outcross was the cause.  This attitude prevails because provides an easy answer hat allows the owners of both stud and brood bitch to assume it must have come from the other side.  In truth, both need to acknowledge the fact that the genes are there and it could happen again.

There is actually a greater risk of producing unwanted traits through linebreeding than outcrossing.  If a trait has occurred, the genes are present in the line.  Continued linebreeding will inevitably bring them together again.  COIs can be useful here.  The COI is a measure of how likely it is that both copies of a gene will be identical by decent from each side of the pedigree.  The higher the COI, the more likely you are to double up on genes both good and bad.

The COI is a mathematical demonstration of why people linebreed.  The tighter the linebreeding, the more likely desirabale traits will be produced.  Unfortunately, the same is true for unwanted traits.  There is not a dog in the world that does not have a few undesirable genes

Assortative mating

Too often people equate outcrossing with mating two dogs that are as different as can be.  Certainly if there is a huge disparity between them, the breeder is unlikely to produce animals of acceptable quality.  In Aussies, this might mean taking something from the show lines, all of which are heavily based on the Flintridge line of the 1960s and early 70s, to a working dog whose background is derives largely from the old Woods line.  Such dogs would be different in structure, coat, color, and behavior.  Their offspring would be unlikely to please the owners of either parent because of their significant differences will produce a littler with little uniformity. However, this is not because the litter is an outcross, but because the particular individuals used were so dissimilar and therefore unsuited to each other.

For outcrossing to work effectively, assortative mating should be employed.  It allows the breeder to maintain the desirable qualities and the same time reduce the risk of producing unwanted outcomes.

Assortative mating is the selection of pairs based on phenotype—what you see or know about the dog and its family—rather than the pedigree.  To be successful in any breeding endeavor, the breeder must have a clear idea of what traits he wants and which he does not, as well as how much potential his bitch has for carrying those genes.  Studs should be evaluated based on what they have produced and the phenotypes of the members of their extended families, including full- and half-siblings plus parents and grandparents and their full and half siblings.  If that family of dogs is consistent for the desired traits and lacks those the breeder wishes to avoid, the mating has a strong probability of success no matter what the pedigrees involved.

Risk for unwanted traits, like health issues, are maximally reduced by giving preference to suitable studs are the least related among the group you are considering.   Is outcrossing dangerous?  Not if done with suitable animals in conjunction with knowledge of both desired and unwanted traits in the family background of both sides of the equation.