Explanations of Sex Considered Incomplete
We all had the same thought in high school: Pretty much everyone is doing it (…but me) Most multicellular creatures at least try at sexual reproduction every generation-in-a-while. Heck even bacteria have that horizontal-gene-transfer thing going on.
It’s kind of a stumper from the selfish gene perspective: Why go through all those dating apps and give up half your genes in the end when you could go solo and reproduce clonally? It’s 2x the fitness right off the bat! In a world shaped by 1% fitness differences of nearly neutral theory, that seems pretty tough to beat! Also: no need to get carpel tunnel from tinder swiping. Although, tbf there are plenty of ways to get carpel tunnel out there, especially for those of us who choose the more monastic path.
The current best explanation for recombination is probably correct. It’s laid out in full in Sara Otto’s fairly recent paper: “Selective Interference and the Evolution of Sex“ But it’s pretty dang complicated. Here it is
1.) Fitness variance determines the rate at which organisms can adapt and purge deleterious mutations.
2.) Linkage tends to make the average phenotype more neutral. This is a second order effect that is visible in simulations but pretty tough to derive.
3.) Recombination via sex restores linkage equilibrium (linkage equilibrium being a stupid phrase meaning a lack of correlation between traits) increasing fitness variance (but not the mean!)
Pithy plot from "Selective Interference and the Evolution of Sex" Sarah P. Otto. The ellipsoid represents the typical state of life. We really have no idea what sign “typical epistasis” should have, but zero seems reasonable, so it’s centered at zero. But due to some weird second order effects in the replicator equations, linkage is usually negative.
By the way Muller’s ratchet has nothing to do with why we do it. Mueller’s ratchet is about the entropic force of the infinite sites model and the limits of selection. Thats clearly for another blog post.
Anyway, my main thesis is that Otto’s paper is all great and correct, but theres a huge missing piece here: how does this second order effect accumulate and get you to that bare minimum 2X fitness?
And it has to do so fast too, inside coalescent times, because otherwise sex could never get established and would be unstable to the reestablishment of clonal reproduction. Short term, sex is really bad for gene fidelity, so the fitness benefit has to be immediate to guard against the greedy fixation of clonal subpopulations. Adding the absurdly dangerous game of recombination just makes it really tough to fathom.
Consider you have two working computer programs composed of hundreds of thousands of finely tuned, interacting functions and parameters. Now you do a git-merge, but randomly choose from each program the version you want. What do you get?
Answer: you get fired from your software engineering job. But if you are God, you get recombination and somehow the system is at least 2x as good as not doing that. Because of an increase in variance?? It’s very implausible to me!
The full explanation, it seems to me, does maybe recombine (see what I did there) some element of the Mueller’s ratchet idea after all, in that genotypic length is probably a crucial element of the picture. We have around 10,000 genes. A small benefit for all 10,000 of them gets a pretty big multiplier effect. Maybe sex seems weird, unsanitary, unnecessary, and endlessly fascinating, just because we aren’t good at thinking in 10,000 dimensions.
