This site is not maintained. Click here for the new website of Richard Dawkins.


← A misguided attack on kin selection

s.k.graham's Avatar Jump to comment 19 by s.k.graham

@ Comment 17 by Anaximander:

(r can never be greater than 1 since you can't be more related to

someone else than you are to yourself - thus "devalued")

Why not? What if somebody has two copies of my "altruism gene"?


Relatedness is not measured one allele at a time. We do not say "for allele A I have a relatedness to you of 1 because I have one copy and you have one copy, but for allele B our r-value is 0, because I have it and you don't."

Relatedness is based on the probability (without having sequenced our genomes) that any given allele in my genome is also in yours. The r-value is determined by knowing the chain(s) of reproductive relationships that connect two organisms, and the mechanism(s) of inheritance (diploid, haploid, clone...). You get half of your genes, at random, from each of your parents. So does your sibling. If you crunch the math that means that each of your alleles also has a 50% chance of being in your sibling. If you & I had all of our great grandparents in common (but none of our grandparents or parents) then our relatedness would be 1/8 (unless I made an arithmetic error). This definition of r-value is based on "new mutations" and on locations in the genome where there is wide variety of competing alleles in the population. This is where "the action is" in evolution. Obviously if a at a particular location on the genome, a particular allele has come to dominate the population (say 99% of individuals have it), then the probability of it being shared between any two individuals is going to be about 99% regardless of kinship connections -- we don't care about those alleles in calculating relatedness. We do care about the rare alleles that make up the other 1%. If you have one of those alleles, then your relatedness to another person tells you the chance that one of your relatives also has it.

There is a fuzzy area for alleles which have come to be present in a large fraction of the population. Basically the larger the fraction of population that already has a specific allele, the less accurate the relatedness calculation is in predicting probability that others have that allele (it will give a lower probability lower than actual -- like the 99% case mentioned above). The evolutionary biologists (the mathematically inclined ones anyway) know about these complications (or should) and take them into account (or should).

Tue, 31 Aug 2010 19:49:18 UTC | #508809