This gives me a chance to read more of Nowak, Tarnita and Wilson (NTW) and in particular the Supplementary Information. This explains in devastating detail why "kin selection" simply doesn't work, and why measures of "relatedness" and "Hamilton's Rule" are highly misleading.
In fact they demonstrate that:
- Hamilton's Rule almost never holds. Even if you recast it to redefine "relatedness" it is still wrong, and if you follow suggestions in (Gardner & al 2007) to redefine it as B(R)R>C(R) it becomes "very complicated to separate any effects and it generally provides no intuition whatsoever"
- Relatedness does not reliably predict evolutionary outcomes. You can have populations with identical levels of "relatedness" which lead to completely different evolutionary behaviour, depending on the evolutionary dynamics involved.
And towards the end of the SI they say "The epicycles of inclusive fitness calculations are not needed, given that we can formulate precise descriptions of how natural selection acts in structured populations."
It seems to me that we can build on these ideas to provide direct refutations of the "Selfish Gene" nonsense. Since it is clear that relatedness is a poor predictor of evolutionary outcome it should be possible to find:
a. Situations where genes that lead to the promotion of a trait are reduced or eliminated from the gene pool. (I have in mind when there are 3 traits A,B,C and B will dominate provided C invades A to a certain extent but not otherwise).
b. Situations (which I think actually may be common in eusocial insects and others) where it is non-genetic modifications (either learned behaviour or epigenetics) that lead to the evolution of a trait, and not genes at all.