Genetic Diversity After Several Generations

[BittyMooPeeb]

I have been inspired by the importing and genetic diversity thread to ask a question about a breed that *starts* with a limited gene pool.

When the Havanese breed was resurrected in the US in the 1970's, 15 dogs were used at the initial breeding stock (this 15 included closely related dogs). From these original 15, there are now many tens (perhaps hundreds) of thousands of Havanese. For these dogs who have had no other genetic material introduced, how genetically diverse are they in reality?

I have investigated this using COI (Coefficient of Inbreeding), and the more generations you include in the COI calculations, the higher the score, as they are all going back to the same ancestors. So a COI calculated with 5 generations may give a 9% inbreeding score, but going back 18 generations can give 80+% . So ... which is the correct score? Can we assume that dogs developing independantly from the initial stock have, by genetic mutation, introduced enough diversity so that we can look at a 5 generation pedigree and get an accurate figure for level of inbreeding, or is it more likely that the dogs resulting from these initial 15 are still extremely genetically similar?

Edited January 28, 2010 by BittyMooPeeb

[sandgrubber]

Genetic diversity is generally good, but biogeography and evolution tell us it isn't the only factor in health. Think about the goat populations on various smallish tropical islands . . . all descended from a pregnant nanny left behind a century ago. Many such populations are full of rugged, very healthy individuals. Or take most of the species 'native' to Hawaii or the Galapagos.

Having a narrow gene pool can be ok if the unhealthy are consistently culled (or desexed) over many generations. Island populations with narrow gene pools do tend to get wiped out when islands get settled by people and the species they bring with them, but that's because they often evolve under weak competition, often without top predators, and they get undone by introduced species or introduced diseases.

Edited January 28, 2010 by sandgrubber

[Lowenhart]

Normally, COI is not calculated over more than 10 generations.

To confirm the state of the breed, one would need to do extensive DNA profiling of the breeds worldwide population. My feeling is that there is enough mutations going on, enough rumours of "suspect" breeding lines (enough to split the breed!) and the fact that at each roll of the genetic dice, a dog can have an assemblege of genes that it's siblings don't have. As long as there are limitations on bottlenecks, I don't believe that high generation COI calcuations are relevant.

Don't forget COI is just math, it's not biology.

[sandgrubber]

Normally, COI is not calculated over more than 10 generations.

. . .

Don't forget COI is just math, it's not biology.

Unfortunately, in my experience, most people doing COI's do the five generation version -- and conclude things are great when the lines are in fact, highly inbred.

The COI is a place where math and biology are close together. That is, it's an excellent and reliable predictor of the fraction of genes derived from a single source. What it doesn't tell you is which genes, what defects were present in the 'founder', and whether defective genes have been passed on. . . which is really what you want to know. Both biology and math say genetics is a crap shoot. You can only call the probabilities. Not the roll of the dice.

Ten years hence, I'll bet genetics is able to be -- at reasonable cost -- specific about which genes have been passed on and able to identify most of the problems. I sincerely hope pedigree breeders will work themselves into a position to take advantage of such scientific advance.

[Lowenhart]

Unfortunately, in my experience, most people doing COI's do the five generation version -- and conclude things are great when the lines are in fact, highly inbred.

The COI is a place where math and biology are close together. That is, it's an excellent and reliable predictor of the fraction of genes derived from a single source. What it doesn't tell you is which genes, what defects were present in the 'founder', and whether defective genes have been passed on. . . which is really what you want to know. Both biology and math say genetics is a crap shoot. You can only call the probabilities. Not the roll of the dice.

Ten years hence, I'll bet genetics is able to be -- at reasonable cost -- specific about which genes have been passed on and able to identify most of the problems. I sincerely hope pedigree breeders will work themselves into a position to take advantage of such scientific advance.

COI is not based in biology, it's based in math. It weights ancestors but does not truly reflect the influence of the particular genetic traits of the "names" in the pedigree but gives a weighting based an instance. Having a extremely (genetically) heterogeneous dog multiple times in a pedigree is not the same as having a extremely (genetically) homogeneous dog a couple of times. A sire is has a high COI is bred to an unrelated dam (who in herself may have a high COI) and the offspring have a COI of 0%.

It would be quite interesting to do genetic diversity analysis on breeds such as the Havanese (15 founders), Lowchen (6 founders, though there are inaccuracies in pedigrees and unrecorded outcrossing to other non-Lowchen), or PWD.

[sandgrubber]

Unfortunately, in my experience, most people doing COI's do the five generation version -- and conclude things are great when the lines are in fact, highly inbred.

The COI is a place where math and biology are close together. That is, it's an excellent and reliable predictor of the fraction of genes derived from a single source. What it doesn't tell you is which genes, what defects were present in the 'founder', and whether defective genes have been passed on. . . which is really what you want to know. Both biology and math say genetics is a crap shoot. You can only call the probabilities. Not the roll of the dice.

Ten years hence, I'll bet genetics is able to be -- at reasonable cost -- specific about which genes have been passed on and able to identify most of the problems. I sincerely hope pedigree breeders will work themselves into a position to take advantage of such scientific advance.

COI is not based in biology, it's based in math. It weights ancestors but does not truly reflect the influence of the particular genetic traits of the "names" in the pedigree but gives a weighting based an instance. Having a extremely (genetically) heterogeneous dog multiple times in a pedigree is not the same as having a extremely (genetically) homogeneous dog a couple of times. A sire is has a high COI is bred to an unrelated dam (who in herself may have a high COI) and the offspring have a COI of 0%.

It would be quite interesting to do genetic diversity analysis on breeds such as the Havanese (15 founders), Lowchen (6 founders, though there are inaccuracies in pedigrees and unrecorded outcrossing to other non-Lowchen), or PWD.

COI is used by zoo biologists to reduce inbreeding. It is based on the biology of gene recombination. Many biologists are poor at math and avoid it wherever possible. That doesn't reduce the validity of the math. In theory, the index incorporates the COI of the ancestor, so your high COI sire bred to an unrelated dam should end out with his COI getting halved . . . not end up with a 0. I agree that the COI is overrated. And I'm sure, were you to travel around South Pacific islands you could find some very very hearty goats and rats with extremely high COI's. But it is a well documented and straightforward index

[BittyMooPeeb]

Thanks for your posts Sandgrubber and Lowenhart. Gives me a lot more to think about.

[gundoglover]

Just to add a bit to the excellent posts above, the small number of dogs involved in establishing a breed is called the "founder effect" and varies from breed to breed. It is related to another concept called the effective population size. For example, even though humans number in the billions, the effective population size has been estimated as low as 7,500 (African humans) and 3,100 (non-African humans) due to the "bottlenecks" that human populations have passed through, particularly the "out of Africa" populations approximately 20,000 years ago (genome.cshlp.org/content/17/4/520.full.pdf).

The COI is more relevant to the deviation from assortative mating that is presumed in the mathematical models. Assortative mating assumes that each individual has the same likelihood of passing on its genes to the next generation. However, in practice this is not the case, some individuals through either potency (ie, evolution through survival of the fittest) or human intervention (ie, selective breeding) contribute more genes to the next generation. These may mate with their close relatives, and their descendents may mate with close relatives. The COI is way of measuring this mathematically & can be helpful to track whether populations are losing genetic diversity over time. (The reason that only 5 generations is used to calculate COI's is that the degree of relationship is very small for doublings further back in the generations, ie, less than 1/2^9 = 1/512 = 0.001953125 = 0.1953125%).

So, evolutionary pressures (and artificial selection) may result in a population with a high COI but also a very high fitness for their niche. The survival of the fittest mechanism in evolution contributes to the elimination of unfit individuals from the gene pool, either through early death or inability to attract sexual partners. In selective breeding, it is the breeder's task to ensure that only the fittest individuals pass on their genes to the next generation. In addition, the founder effect may be large and the effective population size small but if selection has been towards greater and greater fitness, then this may still result in a relatively fit population.

There is a good explanation of effective population size and genetic diversity at www.nature.com/scitable/topicpage/Genetic-Drift-and-Effective-Population-Size-772523

[BittyMooPeeb]

Thanks gundoglover. :D

It seems (from posts) that a high COI is not necessarily a bad thing, so does anyone know why some canine councils are moving towards banning close matings?

[Jed]

Because PDE said it was bad?

Might be worth asking them.

[sandgrubber]

All else equal, I'd rather have a low COI than a high COI .. . but all else is never equal. Most statistical models including the COI presume everything is independent (the IID assumption), when in reality 'everything is connected to everything else' may be a safer assumption. I don't think it's a bad thing to restrict close inbreeding, but I think it's WRONG to take an absolutist perspective on it.

[gundoglover]

I agree with sandgrubber's comments. I think that the CC's are looking at banning close matings because (a) they are rare, and (b) it is seen to be politically correct and possibly expedient to do so.

[angelsun]

And because all too many 'humanize' linebreeding/inbreeding and refer or beleive it to be incest failing to take into account the different numbers of human genes versus canine genes.

I'm sure we have all heard the wives tale about the reason the dog was "cross" was because he was inbred.

As for COE's, in the end as stated by one, it's the frequency of a name or series of names, it's got nothing to do with WHAT that name was and how it affected the progeny and future generations....in my opinion, COE's are for those that want to breed to paper (pedigrees) without taking into account the dogs behind the name.

I linebreed, but I know the dogs I"m linebreeding to....all too often breeders will line breed to a dog they know only by reputation and that isn't always accurate.

[BittyMooPeeb]

And because all too many 'humanize' linebreeding/inbreeding and refer or beleive it to be incest failing to take into account the different numbers of human genes versus canine genes.

Hi Angelsun,

thanks for your post . Could you explain the bolded bit above please? What are the numbers of genes and how does this affect outcomes for close breedings?

[gundoglover]

I think that angelsun is referring to dogs having 78 pairs of chromosomes and humans having 46 pairs of chromosomes, the theory being that breeding closely related dogs is potentially less problematic than breeding closely related humans together.

The concept of linkage supports this view. Linkage refers to genes that are on the same chromosome and therefore tend to be assorted together during reproduction. The closer together that the genes are on the chromosome the more likely that they will be passed on together to the next generation. This is because when the chromosomes come together in sexual reproduction they cross over each other and recombine, hence the study of recombinant molecular genetics.

However, what this means for inbreeding is that with many more chromosomes than humans, there is a reduced likelihood of selecting for one gene and doubling up on an unwanted gene just because they happen to be on the same chromosome. The greater the number of chromosomes the less linkage for the genes. It still happens but is less frequent. So, in theory it should be possible to inbreed more closely with dogs than humans before defects occur in the offspring.

[sandgrubber]

I think that angelsun is referring to dogs having 78 pairs of chromosomes and humans having 46 pairs of chromosomes, the theory being that breeding closely related dogs is potentially less problematic than breeding closely related humans together.

The concept of linkage supports this view. Linkage refers to genes that are on the same chromosome and therefore tend to be assorted together during reproduction. The closer together that the genes are on the chromosome the more likely that they will be passed on together to the next generation. This is because when the chromosomes come together in sexual reproduction they cross over each other and recombine, hence the study of recombinant molecular genetics.

However, what this means for inbreeding is that with many more chromosomes than humans, there is a reduced likelihood of selecting for one gene and doubling up on an unwanted gene just because they happen to be on the same chromosome. The greater the number of chromosomes the less linkage for the genes. It still happens but is less frequent. So, in theory it should be possible to inbreed more closely with dogs than humans before defects occur in the offspring.

Kangaroo breeders beware. They only have 12 chromosomes

I'd be interested to know if there's any empirical evidence to support this theory . . . eg, do fruitflies (six chromosomes) fall apart under inbreeding? Do carp, with 104, stand up to it well?

Edited February 8, 2010 by sandgrubber

[gundoglover]

I'd be interested to know if there's any empirical evidence to support this theory . . . eg, do fruitflies (six chromosomes) fall apart under inbreeding? Do carp, with 104, stand up to it well?

There is some evidence from recombinant experiments, eg, www.ncbi.nlm.nih.gov/pmc/articles/PMC1461880/

This research with plants shows that inbreeding depression is not random, but is increased for linked genes (ie, genes on the same chromosome). The more closely linked the lower the recombination rate for genes and the greater the depression associated with inbreeding. So, from this we would expect stronger effects of inbreeding depression for kangaroos (fewer chromosomes, therefore higher degree of linkage, lower rate of recombination for most genes) than carp.