I’ve put up a bunch of posts relating to inbreeding recently (1, 2, 3, 4). But I haven’t really defined it. First, let’s stipulate what inbreeding is not: it is not the same as incest. Acts of incest can include individuals who have no blood relationship to each other (e.g., Hamlet). Additionally, there are instances of inbreeding which are not necessarily incestuous. If a population is highly inbred, then individuals who are not relations by social custom may still be so genetically similar to a point where the pairing can not credibly be stated as an outcross. But still, what do I mean? To refresh myself I re-read the section on inbreeding in Hartl & Clark. And I think that helped clarify one implicit assumption which I have which may not be clear to everyone, and I’ll get to that.
In any case, first, what’s the deal with inbreeding? The short answer is that inbreeding is a measure of the probability of identity by descent of two alleles at a given locus in a given individual. This concise definition itself is the problem. These are all abstract concepts, close to being human categorical fictions useful in an instrumental sense. Locus is the most concrete one, as it is basically a gene (though not necessarily a gene, and you are probably aware that gene itself is a term which is the subject of contention). It just refers to a position on the genetic map. An allele is a genetic variant. If there is variation in genetic type at a locus, then you have at least two alleles. But note that alleles are a pre-DNA abstraction. They’re not specific changes in base pairs, or variations in genomic architecture. They’re just variants in a generic sense.
Identity by descent is both straightforward and almost mystical. It simply means that the two alleles are the same because they derive from the same common ancestor, as if they share a Platonic essence through the meiotic replication process. Note that if two alleles are the same in state they are not necessarily identical by descent. This is easy to understand. There are four base pair states, and one can imagine a circumstance where two alleles exhibit the same state because one of them has mutated from the ancestral state to a derived one.
And so there you have it, inbreeding isn’t like adenine or the human heart, it’s not a concrete material object, but an abstract conceptual phenomenon at some remove from everyday experience in a process sense. Of course the concrete outcomes of inbreeding (e.g., recessive diseases) are well known to us from “folk genetics.” But knowing the result or outcome of a process does not equip someone to properly model it.
Above I implied one aspect of the inbreeding phenomenon which I don’t think I discussed earlier: the dimension of time. I referred to “ancestral” and “derived” states. And obviously identity by descent refers to descent from a common ancestor across generations. Inbreeding can not be understood without its proper historical-demographic context, which shapes the genetic state of a given population at a given time.
Obviously if you go far back enough in time all allelic lineages coalesce back to a common ancestor. On a deep level all alleles are identical by descent at a given locus (granting the confusions which occur due to orthology, etc.). Therefore when you ascertain identity by descent you have to set a cut-off date at which all alleles are not identical by descent. From this point, time = 0, some alleles will go extinct and some will increase in frequency, due to random genetic drift. In a purely stochastic system eventually all alleles will become identical by descent in reference to time = 0 as one allele fixes in the population at frequency ~100%.
What does all this have to do with inbreeding? Let’s make this concrete. Let’s set time = 0 at 2,000 years before the present. Every individual in the world alive today has a specific genealogy which stretches back to that point, and their genes have genealogies which go back to 2,000 years ago. Some ancestors show up many times in our genealogies, while others show up rarely. In other words we are all somewhat inbred when compared to an idealized Hardy-Weinberg equilibrium. That’s because there is structure in human populations. But there is inbreeding, and then there is inbreeding. Inbreeding basically is a measure of the tangled reticulation of your genealogy. Populations which have gone through bottlenecks, and have lower long term effective populations, exhibit more of this collapse in number of distinct ancestors. It doesn’t take a rocket scientist to intuit that fewer ancestors often means more alleles identical by descent (any given individual is going to show up a lot more prominently in a person’s genealogy, and so likely to donate an allele which is passed on from both mother and father). Compared to Africans non-Africans are inbred. People who are not African have fewer distinct ancestors at time = 50,000 years before the present, and this shows in their genomes. Populations which have gone through bottlenecks, and have been isolated (e.g., on islands) tend toward more inbreeding.
But as I said above, there is inbreeding, and there is inbreeding. People whose parents are siblings or first cousins are genuinely inbred in a way that Amerindians, who went through a bottleneck, are not. Though inbreeding coefficients apply across the whole genome, giving a measure of the genetic contribution of recent the same ancestors from both parents, one of the primary negative outcomes are the fitness hits due to relatively rare problematic alleles. All humans come with a complement of very bad allelic variants, but most of them have strongly recessive expression. But if you have many loci where the alleles are identical by descent, that is, you’re homozygous, then the problematic alleles will be unmasked.
In other words, the primary genetic reason that inbreeding is not optimal is the exposure of these rare large effect deleterious alleles. If you had two siblings who mated who were totally purged of mutational, load then the progeny would be far less problematic. In fact with plants selfing lineages achieve just this state of mutational perfection by exposing their recessive alleles and purging their genetic load (this seems less attainable with complex animals).*
With genomic methods a lot of this has become more concrete. As I note above ascertaining identity by descent at a base pair to a high degree of certainty is difficult because there are only four base pair variants. But if you look at long tracts of DNA then you see very specific independent sequences of genetic variations, which serve to tag specific ancestors. That’s why looking at “runs of homozygosity” is probably one of the better ways to get at inbreeding. If you still have a lot of runs of homozygosity that means that the common ancestry was recent enough that mutation and recombination was not able to eliminate blocks of visible descent. It is also probable that rare deleterious mutations are still shared across these blocks (i.e., selection hasn’t purged them from the population, or they haven’t mutated back to a functional form).
And there you have it, bringing inbreeding back to intelligibility. Distinctive genomic patterns which span loci have a particular shelf-life because of recombination. That gives you a extreme upper bound in terms of how far back in time you want to go in considering a population inbred due to common ancestry. Second, these methods are comparative. The load imposed by rare deleterious alleles can be assessed by comparing with similar populations. Comparing Ashkenazi Jews with Africans is not useful. Rather, compare them with other West Eurasians.
Readers with clarifications in English welcome!
* Evolutionarily selfing lineages seem to be a dead end though. They become clonal types without ay genetic variation to adapt to suboptimal conditions.