Inbreeding vs. Outbreeding


There's a big controversy in the dog world today -- at least, for a vocal minority, it's a big controversy, and it shows some signs of wanting to spread into the mainstream of show breeders.  Everyone concerned with breeding probably needs to know enough about it to get around in the discussion.  The controversy concerns the systems of inbreeding used to fix and then to maintain the characteristics of today's purebred dogs.  It can be summed up briefly this way: 

Does inbreeding as universally practiced by today's dog fancy inevitably result in a negative impact on the health of purebred dogs?

I'm assuming that everyone interested in this question already understands the way the terms "inbreeding", "line breeding", and "outbreeding" are used by breeders.  I won't go into that here in any detail.  Just in case, I shall say briefly:  Inbreeding is considered to be the breeding of close relatives:  uncle-niece or closer.  Line breeding is inbreeding, but to a lesser degree:  grandfather-granddaughter or further away, and usually implies that the breeder is trying to maximize the genetic contribution of one particular ancestor.  Precisely where breeders draw the line between inbreeding and line breeding depends on the breeder and isn't important for the purposes of this discussion, which assumes that inbreeding and line breeding are essentially the same.

Outbreeding is frequently taken to mean "no repeated ancestors within the last five generations."  You will find statements all over the web, and in a lot of books and articles written by and for breeders, that ancestors further back in the pedigree than that have no important influence and don't count.  This is not true, as for example Jerold Bell shows with his analysis of his Gordan Setter's pedigree.  (The Ins and Outs of Pedigree Analysis; you can find a copy here: )  A standard 5-generation pedigree can conceal a heck of a lot of "background" inbreeding.  Really the only way to confirm you are outbreeding is to calculate a coefficient of inbreeding (COI) for your planned breeding, based on at least a fifteen-generation pedigree (more is better), and you must confirm that this COI is lower than the average inbreeding coefficient for your breed.  The COI for your "outbred" puppies may still be much, much higher than the average COI for most natural, non-domestic, populations of animals.

Technically, the only true "outcross" would be to a dog from another breed.  Many purebred dog breeds are descended from a very small number of ancestors.  All were created by closing the gene pool off from the general run of dogs -- that is why pure breeds have distinct characteristics.  That is why it is accurate to say that all modern breeds were established and are maintained via inbreeding.

So is inbreeding inherently harmful?

Various advocates of outbreeding will vehemently answer Yes to this question, and they have a plausible argument to support their position.  The majority, probably, of show breeders who have thought about this issue, will say No.  They, too, have plausible arguments to support this answer.

Actual evidence for either position is in surprisingly short supply.  Assumptions, on the other hand, litter the ground in abundance.  The following is meant to lay out clear hypotheses and to show how data could be collected that would allow fanciers to evaluate current breeding programs and decide whether fundamental changes should be made -- for the overall benefit of the breeds they profess to love.

First, please keep in mind that human beings have a terrible time assessing objective reality.  A teacher will tell you that young Timmy is ALWAYS out of his chair and ALWAYS disruptive in class -- then, when she makes actual notes about his behavior, will find that young Timmy was out of his chair and disruptive only an average of eight times per day, which is about normal for boys in his class.  A gardener will swear that THIS spring is the worst in memory -- the coldest and rainiest -- but checking actual records shows that in fact about one spring in three has had comparable weather for the past fifty years.  A neighborhood will become convinced that the power lines located near their houses are causing their kids to develop leukemia -- when a check of medical records nationwide reveal that the rates of leukemia for that neighborhood are precisely normal.  This sort of thing happens all the time.  A breeder who asserts that immune disorders have vastly increased in purebred dog populations in the past few decades is making a precisely similar assertion.

The ONLY way to know what is happening in populations of purebred dogs is to keep objective records for decades and then examine them, keeping in mind that changes in vaccination protocols, incidence of infectious disease, changes in diet, changes in diagnostic tools, etc, over the same time period, can seriously impact the data.  An emotional conviction that rates of hip dysplasia, or anything else, have increased in purebred dogs cannot be taken as proof that this has occurred if no actual data are available for earlier populations.  This is true for all health concerns.  Conviction is not an adequate substitute for facts.

It's quite true that purebreds as they exist today are historically an anomaly.  The closed studbooks that limit the gene pools of today's pure breeds represent a new kind of breeding attitude.  Earlier than about a hundred years ago, dogs were bred to what might be called "working standards" and it was perfectly ordinary for a breeder to add a bit of this and a dab of that during the process of creating his idea of the perfect dog.  This is still how Alaskan huskies are bred -- the sled dogs used for the major endurance races are not "pure" husky.  Breeders of working Alaskan huskies add pointer and retriever, setter and greyhound, to their working huskies -- whatever their experience leads them to expect should add speed and endurance to their dogs.  And it works.  Their dogs do not greatly resemble the spectacularly beautiful Siberian huskies you will see in the conformation ring.  They are racy, long-legged, lean animals that have no peers for racing -- they beat Siberian huskies hands down.  There is no mammal on Earth that rivals Alaskan huskies for endurance racing (Coppinger 2001, Dogs).

Alaskan huskies are the sorts of animals that result from breeding for performance without closing stud books -- and this is the sort of breeding that historically was common.  Livestock guarding dogs, herding dogs, war dogs, hunting and vermin dogs, lurchers and curs (in the technical sense of the word) -- these are the original performance breeds.  The idea of closing off a breeding population and breeding an isolated population to a physical standard of perfection is the new idea in breeding.  You will not get that kind of conception of purebreds as isolated populations in horse breeding, where crossbreds are purposely produced in great numbers to perform specific functions (Morabs, Aztecas, lots of "warmbloods", etc.).  Nor will you see closed studbooks in cattle breeding, where large operations deliberately produce crossbred heifers which they then cross to a bull of still a third breed in order to maximize fertility and growth rates.

Does it harm dogs to breed them as we do today?  The gene pools are tiny, relatively speaking.  All purebred dogs today are inbred.  A mating that looks "outbred" on paper -- to five generations -- will probably look inbred if taken back eight or more generations.  Jerold Bell ( did a pedigree analysis of one of his Gordan setters.  This animal was the product of a first-cousin mating, giving her an inbreeding coefficient of 6.25%.  This is a low level of inbreeding for purebred dogs.  However, when her inbreeding coefficient was recalculated over twelve generations, it was 30.81%, which is higher than it would have been for a parent-offspring mating (that would be 25%, of course).  Most of this animal's inbreeding was invisible in a normal pedigree.  The dog who provided the single greatest genetic contribution to this setter (other than her parents) was an ancestor who appeared first in her pedigree six generations back, and then appeared in the pedigree 33 times.  Bell refers to inbreeding that has been hidden back in the pedigree as "background" inbreeding.  It's clear from his analysis that background inbreeding can be very high, even for animals that appear outbred to four or five generations.

What does this do to purebred dogs?

The traditional answer (which may be correct) is:  Not Much.  Inbreeding increases homozygosity of alleles at all genetic loci -- everyone agrees on this (the math is simple).  Close inbreeding increases homozygosity fast, whereas looser inbreeding (linebreeding) increases homozygosity more slowly.  Homozygosity per se may not be a huge concern, however.  Deleterious alleles are not created by inbreeding, but are revealed in the phenotype of the dog when they are present homozygously.   This allows deleterious alleles (the "genetic load" of a breed) to be removed from the population as they show up, resulting, eventually, in a healthier population -- at least with respect to those alleles.

The newer answer (which may be correct) is:  It Destroys Disease Resistance.  The major histocompatibility complex is a part of the immune system that is concerned with recognizing foreign antigens.  One reasonable hypothesis is that heterozygosity at MHC loci is essential to allow an organism to defend itself against the huge diversity of foreign antigens it will contact throughout its life.  Inbreeding, which inevitably decreases heterozygosity at all loci, is then expected to inevitably reduce disease resistance.  As a secondary effect, or set of effects, inbreeding depression is also expected to invariably, or almost invariably, reduce fertility, litter size, growth rates, longevity, and general vigor.  These effects, arising from an increase in homozygosity in more subtle deleterious alleles that do not cause actual overt disease, are what lead cattle people to crossbreed their livestock.  It seems to work for them, and they're not likely to be wrong -- market forces have a lot of oomph when it comes to forcing cattle breeders to make economically sound decisions.

Note that this second idea produces one very clear hypothesis:  the reduction of overall disease resistance should be inevitable and should always correlate with the degree of homozygosity within a population.  This is a very rigid hypothesis.  Note that one counter-example is sufficient to disprove this hypothesis.  The mechanism proposed does not readily allow for counter-examples to exist, so this hypothesis cannot easily be re-conceptualized in a less rigid form, such as usually correlates with or tends to correlate with.

In addition, the "inbreeding-always-bad" idea produces a second hypothesis:   that the degree of inbreeding should correlate significantly with reduced fertility, litter sizes, growth rates, longevity, and vigor.  This second hypothesis does not mandate the strength of the expected correlation; however, it does anticipate always or almost always getting negative correlations of COI with some or all of these characteristics.

These hypotheses are distinct.  One could be true while the other is false.  Data that support one hypothesis should not be taken as also supporting the other.

You can calculate the degree of inbreeding as the COI -- the coefficient of inbreeding.  Theoretically, it should be possible to get records from scrupulously-honest long-time breeders who have produced lots of puppies and tracked their health carefully in objective ways, calculate the COI for each of the animals they have produced, and check for correlations.

Lacking such data sets, is it possible to evaluate these hypotheses?

Well, it is, sort of.  For the first hypothesis, what we need to look for particularly is evidence that tight inbreeding does not inevitably lead to the failure of a population, preferably in dog, or at least canid, populations, but any population will do because the MHC hypothesis is so clear-cut.

What does a literature survey show us?  The lack of a large university library in my neighborhood is a problem, but here's what I've found so far that seems to address this problem:


1.  Population bottlenecks in cheetahs have resulted in the creation of an extremely inbred species, as is well-known.  Effective population sizes in the wild are even smaller than they would appear from the number of animals actually in the populations.  Although reproductive success in captive cheetah populations is very poor, and this has been attributed to the extreme lack of genetic variability in the species (and this attribution has been made in the popular press, so the idea has gained more visibility than is, perhaps, ideal), reproductive success in the wild is actually very good.  Wild cheetahs do not differ significantly from captive animals in terms of how closely they are inbred, so the inbreeding hypothesis cannot be the correct explanation for lack of captive-breeding success.  A high fraction of sperm in cheetahs is malformed, and this may be due to inbreeding, but the good reproductive success in the wild again indicates this is not important.  Wild cheetahs do not show other effects considered typical of inbreeding, such as reduced litter sizes.  MHC loci retain some heterozygosity in cheetahs and there have not been any major disease epidemics in wild populations in recent history. 

Kelly, MJ.  2001.  Lineage loss in Serengeti cheetahs:  consequences of high reproductive variance and     heritability of fitness on effective population sizes.  Cons Biol:  137 -147.         

O'Brian et al.  1985.  Genetic basis for species vulnerability in the cheetah.  Science 227: 1428-1434.

O'Brian et al.  1986.  The cheetah in genetic peril.  Scientific American: 84-92.

O'Brian et al.  1987.  East African cheetahs: evidence for two population bottlenecks.  Proc Natl Acad Sci  USA 84: 508-511.

Merola, M.  1994.  A reassessment of homozygosity and the case for inbreeding in the cheetah, Acinonyx jubatus:  implications for conservation.  Cons Biol 8: 961-971.

Wielebnowski, N.  1996.  Reassessing the relationship between juvenile mortality and genetic monomorphism in captive cheetahs.  Zoo Biol 15: 353-369.


2.  Ellegren et al. describe a severely bottlenecked population of beaver, which recovered with no or very little apparent effect from inbreeding.  These beaver are, very interestingly, monomorphic (homozygous) for MHC loci.  Even a comparison population with lots of genetic variability in general was still homozygous for MHC loci.  Ellegren et al. suggest that beaver may be tolerant of inbreeding because of their population dynamics, which require dispersal along waterways and thus restrict chances for outbreeding.  The authors also note that beaver appear to lack behavioral mechanisms that would serve to reduce inbreeding -- that is, they do not hesitate to mate with close relatives.

Ellegren et al.  1993.  Major histocompatibility complex (MHC) and low levels of DNA fingerprinting variability  in a reintroduced and rapidly expanding population of beavers.  Proc Natl Acad Sci USA 90: 8150-8153.


3.  Mates in a wild population of shrews were more related than chance would have dictated for the population, implying that these animals prefer or were compelled by ecological factors to choose mates that were related to themselves.  Relatedness had no effect on fecundity and offspring homozygosity had no effect on their ensuing reproductive success.  This one is from an Orals Poster I got off the internet, so it's not based on a massive, detailed collection of data; however, the result reported is definitely suggestive.

Duarte et al. Does inbreeding drive dispersal in the greater white-toothed shrew, Crosula?


4.  In captive Indian rhinos, inbreeding vs. outbreeding effects were examined for correlation with gestation period, birth mass, infant mortality, and growth of the offspring.  Inbred calves produced by matings between rhinos from a strongly inbred and highly homozygous population grew more slowly but suffered less mortality than non-inbred or outbred (mates, in this case, from very distinct populations) calves.  Inbreeding did not effect gestation period or birth mass.

Zschokke, S. and B. Baur.  2002.  Inbreeding, outbreeding, infant growth and size dimorphism in captive Indian rhinoceros (Rhinoceros unicornis).  Can J Zool 80:  2014-2023.


5.  There is no apparent inbreeding depression in quite highly inbred Estonian native ponies with regards to work ability or conformation, although the most inbred animals appear to mature more slowly, according to


6.  A very highly inbred herd of cows ( ) in the north of England seems to be both genetically uniform and perfectly fine for traits of vigor and health.


6.  There is (surprise!) a vertebrate species which self-fertilizes, normally the province of plants!  Who knew?  This animal, a killifish (Rivulus marmoratus), is a very plentiful and widely dispersed species which is found in coastal mangrove swamps, in marine or brackish environments.  It's an interesting little critter ecologically, but most interesting is its reproductive biology.  Most individuals of R. marmoratus are hermaphrodites (a few are male in some populations) and obligately self-fertilizing.  Three histocompatibility clones appear to exist.  Individuals are highly homozygous, of course.  Fitness traits of lab-produced heterozygotes have not been compared to natural homozygotes because of the considerable difficulty in getting clonal individuals to outcross.  There is no evidence for inbreeding depression in this species, which is by all accounts very successful.  It seems clear, also, that MHC monomorphism cannot be a problem in this species.

Turner, BJ.


How relevant is all this to dogs?  Good question.  It seems very likely that at least some species do okay even under very strong inbreeding regimes.  Other areas to look for data would be in studies of pond fishes, plateau animals, island populations, and other natural populations that must be subject to fairly strong inbreeding.  Meanwhile:


7The very strongly inbred population of Isle Royale wolves was thought, up until 1993, to be succumbing to inbreeding depression and bad luck.  However, they have bounced back dramatically since then, assisted by a bad year for moose that allowed them to kill prey animals more easily.  There are no signs of disease or genetic problems in the Isle Royale wolves, although there is some evidence of skeletal asymmetry, which is thought to be an effect of inbreeding.  Nor have signs of inbreeding depression been found in a captive population of Mexican grey wolves arising from seven founders.  Moreover, wolves have been observed to naturally inbreed (prefer closely related mates) in at least some natural populations.  Further, severely inbred populations that may have been suffering from reduced reproductive success appear to need very little immigration to rebound strongly.

Haber, GC.

Line, L.  1996.  The long-running wolf-moose drama:  wolves recover from disaster.  New York Times, March 19.

Hendrick, P.  1995.  Genetic evaluation of three captive Mexican grey wolf lineages and consequent recommendations.  Unpublished.

Beuler, LE.  1973.  Wild Dogs of the World.

Mech, L.D. 1970. The Wolf:  The Ecology and Behavior of an Endangered Species. The Natural History Press, Garden City, NY. 384 p.

Shields, W.M. 1983. Genetic considerations in the management of the wolf and other large vertebrates: an alternative view. Pp. 90-92. In: Carbyn, L.N. (ed.). Wolves in Canada and Alaska. Canadian Wildlife Service Report Series No 45.

Theberge, J.B. 1983. Considerations in wolf management related to genetic variability and adaptive change. Pp. 86-89. In: Carbyn, L.N. (ed.). Wolves in Canada and Alaska: their status, biology and management. Canad. Wildl. Serv., Rep. Ser. No. 45.

Vila et al.  2002.  Rescue of a severely bottlenecked wolf population by a single immigrant.  Proc R Soc Lond.


8.  Willis, in his book Genetics of the Dog, has a chapter about inbreeding.  In this chapter, he mentions (with citations, of course) several populations of rats and cattle that did fine with very intense inbreeding.  One population of rats was taken through 25 generations of brother-sister matings without ever compromising its vigor or fertility -- or clearly, its disease resistance.  Somebody would probably have noticed if all the rats had died of infectious diseases halfway through the experiment.  Following is a citation for the rat study, which I haven't seen personally, plus a more extensive description:


9.  King, H. D.  1919.  Studies on inbreeding.  175 pp.  Reprinted from the May, July, and October (1918) and August (1919) issues of the J. of Experimental Zoeology.  [That's not my typo -- it's really spelled that way in the reference I found]  King examined the effects of inbreeding on growth, body weight, fertility, "constitutional vigor," and sex ratio of the standard albino lab rat.  The intensely inbred line (and it was intensely inbred) prospered in every way -- King evidently practiced strong selection for desirable traits (large size, large litter size, health).  Note that it doesn't matter how much culling King had to do to achieve her results -- homozygosity was the inevitable result of the very intense inbreeding practiced and homozygosity alone did not destroy the line.  Despite the date on her work, it deserves respect, as King evidently received very high honors indeed for a female scientist of that era ( ).


So, what's the answer to the questions about inbreeding in purebred dogs?

Good question.

Inbreeding cannot possibly lead invariably to disease susceptibility, via an increase in homozygosity of the MHC.  If this chain of causal events was correct, then the results we observe in some of the reports cited above (especially the beaver and killifish studies) could not occur.  The beaver populations, monomorphic (homozygous) for MHC loci, specifically drive a stake through the heart of the idea that MHC diversity is always crucial to maintaining a healthy population.  This hypothesis must be oversimplified at best and wrong at worst.  We do not know what effect MHC homozygosity will have or is having on purebred dogs.  We will not and cannot answer this more specific question until actual data on dog populations are collected in a sufficiently rigorous manner to allow such an answer.

Inbreeding cannot possibly lead invariably to a decrease in litter size, fertility, longevity, birth weight, and general vigor.  If it did, we could not see a failure of these expectations for cheetahs, beaver, rhinos, shrews, captive wolves, etc.  On the other hand, frequently inbreeding depression does correlate with at least one of these expectations, as we see in cattle and in innumerable studies of natural and managed populations.

There are many observations dog breeders have made that may support the idea that inbreeding depression may sometimes negatively influence various reproductive and other characteristics of dogs.

Mary Roslin Williams, in her book Reaching for the Stars, commented that litter sizes in labs had declined dramatically during the years she had been breeding (she did not suggest, and would almost certainly disagree with, the idea that inbreeding is fundamentally a flawed breeding system).

Grossman & Grossman, in their book Winning with Purebred Dogs, noted that they never had problems with puppy mortality (fading puppy syndrome) during their early years of breeding.  They hypothesized that inbreeding had led to their eventual problems with mortality and reported an outcross which yielded a litter that did not experience mortality.  Their sample size is far too small to allow reliable conclusions.  Their comments that other breeders had reported the same pattern certainly suffers from the unreliablity of human beings in perceiving what is actually going on, whether the observation is in fact accurate or not.  That is, once a breeder personally experiences a particular problem, than he is certainly more likely to note other instances of that problem in the future, regardless of how frequently the problem actually occurs.  This is precisely similar, for example, to Oliver Sacks becoming interested in Tourette's syndrome after finding one patient with the syndrome and then immediately spotting numerous cases of Tourette's "on the street".  This sort of problem is why somebody's impression of a phenomenon is no substitute for rigorously-collected data.

Longevity was negatively correlated with increasing COI in one analysis done on Rhodesian ridgebacks ( ), but I don't know about the sample size or methodology for that analysis.  A similar correlation was found by Armstrong for standard poodles ( ), Armstrong being one of the more high-profile proponents for outbreeding, and even out-crossing, purebred dogs.  Comparisons between standard poodles and Clumber spaniels found that the (more inbred) Clumbers live on average shorter lives than the (less inbred) Standards.  Of course, Clumber spaniels are physically very, very different from standard poodles!  If the physical characteristics of Clumber spaniels turn out to generally decrease lifespan, which is very plausible, then where does this leave that result?  It would be interesting, and far more realistic, to see a comparison between more physically similar breeds, or between inbred and outbred animals within one breed.

Many breeders, not to mention random people you meet around and about, will tell you that health problems are on the rise and will go on to blame these problems on inbreeding.  Data supporting the link between inbreeding and disease susceptibility are lacking in purebred dogs, however common opinions may be.  Data from natural populations of vertebrate species indicate that the inevitableness of this link is, at best, highly questionable.  My best guess is that we will find, during the next decade or so, that the connection between MHC heterozygosity and disease resistance is more complex than is currently assumed.

On the other hand, links between inbreeding, high COI, and problems with fertility, litter size, longevity, birth weight, and / or general vigor may be more strongly supported (although not universally so; see the rats, above).  Conviction alone does not equal support.  When breeders ( conformation.htm ) offer examples of animals with low COI that are nevertheless outstanding in the show ring and as producers, that is nice, but it does not provide evidence to support any position whatsoever.  You need to know what proportion of top winners and producers have low vs. high COIs, not whether it's possible to get a top winner / producer who has a low COI.

An interested breeder might be able to generate such evidence (or counter-evidence -- you don't know till you look what you're going to get) by taking a random list of 100 top winners vs. 100 mediocre animals and seeing how many of each group had low vs. high COI -- keeping in mind that COI for a five-generation pedigree will be less than for a ten-generation pedigree.  This data set could be confounded by breeding systems if most breeders who produce top winners also inbreed heavily and most who don't, don't.  Winners may not be good producers, so that's a separate analysis.  Or, you could get a list from several top breeders of animals with high COI vs. animals with low COI and see what proportion of each group went on to be top winners and producers.  This data set could be confounded if breeder expectations for and investment in high-COI animals tended to be greater than for low-COI animals.  There are ways to work around problems with data sets, but thinking through possible problems with your data is crucial, preferably before you draw conclusions.

Easier to generate are data relevant to whether low COI correlates with longevity, litter size, fertility, and health. It would perhaps be wise to separate health out into separate categories once you get your data set in hand:  autoimmune problems.  Polygenic problems.  Simple genetic problems.  Structural defects.  Susceptibility to infectious disease.  Etc.  You get this kind of data with comprehensive breed surveys, valuable for many reasons.  Looking at longevity via surveys is somewhat difficult because people who have owned dogs for a long time may forget about dogs that died young when generating lists of dogs and their ages at death.  Also, a data set which excludes still-living dogs while including their dead age-mates will underestimate longevity, perhaps seriously.  Even harder to examine is a possible link between high COI and reduced vigor.  (Define "vigor".)

Notice, by the way, that if you look at twenty different possible correlations between COI and canine characteristics, that you are likely to get one result that is significant at the p = .05 level by accident (that is, it is not really significant).  That is what "significant at the p = .05 level" means -- that there is a five percent (1 in 20) probability that the "significance" you think you see occurred by chance alone.


If all this seems complicated, I can't help that.  The subject is rife with complicating factors.  The take-home message, I suppose, ought to include the understanding that MHC homozygosity definitely cannot inextricably be linked to immune dysfunction.  And that you need to be careful to look at the methodology underlying assertions before accepting them.  And that subjective impressions about canine health and genetics, no matter how strongly believed, are sometimes a desperately poor guide to what really is going on.  That one is always a problem, in all aspects of life.  And -- most important -- that a careful breeder who keeps meticulous records may be able to add a heck of a lot to this discussion.



HWE and the practical breeder

Case Studies