Lookit that, green wins! It’s good! It’s great! It’s better! Natural selection selected it! … Actually, not necessarily. I want to talk about my favorite evolutionary process, which sadly, is often confounded with other things and inadequately explained. But its first stages are happening in your gonads right now, sooo … I figure you might want to know.
Here’s a bit of discipline spat-starting terminology for you, guaranteed to make the biologist you know twitch and mutter. I’m a bad person for enjoying it so much, yes I am. Microevolution concerns changes in single-species populations. The typical time-unit for causes is whatever that species’ generation time is, related to cohort, meaning the sequential groups of individuals who are born, mature, reproduce, and die more-or-less together. The typical units of effect concern genetic, physiological, and ecological variables, the content which usually leads anyone to say “this species is blah-de-blah” or “does blah-de-blah.” Again, though, the focus with microevolution is typically intrinsic and interactive causes at that generational time-scale, and their effects on that particular group at that particular time.
Allele frequencies are almost always discussed in terms of selection, and don’t get me wrong, there’s a pretty good historical reason. Exactly a century ago, biology as a discipline struggled in the grip of an apparently insurmountable intellectual crisis, because Darwin’s and Mendel’s ideas were perceived to clash. During the first decades of the century, culminating in 1926, William Castle, Wilhelm Weinberg,and G. H. Hardy independently reconciled them by representing two alleles for one gene in this way.
- p + q = 1
- p2 + 2pq + q2 = 1
Which quite rightly means probably zilch to you, so I’ll do it in plain language: simply reproducing genes into the next generation, over and over, does nothing to their alleles’ frequencies (percentages) in the population. Mendel’s ideas apply to individual pairings, not to whole-population frequencies. Don’t expect a gene’s percentages to shake out into, say, half A and half a for a given gene, by itself, over time, or into one-quarter AA, one-half Aa, and one-quarter aa. Doesn’t happen, because by itself, the frequencies that go into the reproductive pool, are what comes out next generation. (Biologists call this “equilibrium,” which is actually a terrible word for it, as no counter-acting forces/effects are occurring.)
This was a landmark in the history of biology; it’s fair to say that only then did it become a unified discipline. The trouble is that the principle has routinely been over-stated to say that the primary way in which allelic frequencies change is always selection, such that the other ways are mis-represented as exceptional or edge cases. This is merely an artifact that selection was the discussion topic at that time regarding altering allelic frequencies.
How selection changes allelic frequencies, generation to generation, is very direct – individuals with certain alleles or combinations thereof are reproducting disproportionately more, or as I’d sort of confusingly prefer to say it, “less poorly,” than other individuals. This biases the reproductive pool from the start, so, done. But what else can be happening? My main squeeze, genetic drift.
The interaction between the Drunkard’s Walk principle and the Gambler’s Fallacy is enormously important here. It’s not like coin-flipping, which returns to no-history baseline with each flip; each “step” means your starting point is where you landed last time. By definition that means the 50-50 chance for each instance means nothing about “returning to the middle” over time, and in practice, rather to the contrary. Look at the lead image for this post. Each colored line shows ups and downs at the designated time scale. Increase the time-scale’s units, and green swoops up while the others stay flat or swoop down – because each 50-50 “bounce” has to start from where it landed last time.
I’ll say it again: just because there’s a 50-50 chance that an offspring gets one allele or the other from a given parent, doesn’t mean the alleles’ existing population percentages are perfectly recapitulated every generation. Minor deviation is not only possible, but expected, just the same way you don’t expect every person at the table to get exactly one red and one black suit every time you shuffle a deck of cards and each person draws two.
I like the card analogy because making gametes is a lot like separating your two cards and throwing them back into the deck for a new shuffle, generation by generation. People are also correct to bring in sampling error, perhaps even give primary importance to it, because not all gametes actually go into making new offspring – far from it, in fact – and which are collectively “lost” along the way has nothing to do with maintaining the parental population’s allele frequencies.
Biologists who’ve been trained specifically in the New Synthesis model of evolution don’t like to talk about drift. It’s kind of insidious, as it looks exactly like selection both genetically and phenotypically, and offends the aesthetic sense that changing populations are nicely integrated with environmental features and species’ “needs” and “roles.” When you look across organisms, you’re supposed to bask in the notion that the features you see represent an adaptive landscape, such that current diversity represents a kind of catalogue of these creatures’ historical high points of reproductive performance. Thinking about drift means the diversity landscape may well not reflect an adaptive landscape, and so, to these biologists, it’s cheating.
Here’s a good bit about bottlenecks:
Easier, right? You can find them in any bio textbook under “genetic drift,” but they’re aren’t really drift at all, they’re population-size effects. Although they definitely alter allelic frequencies without selection being involved, they have nothing to do with segregated alleles and probabilities of recombination. Lumping them with drift and using that term collectively is an artifact of highlighting selection in the pedagog and minimizing drift’s potential importance, and as I said, because these things are a bit easier to explain than drift proper.
Here’s the real Limburger-smelling heresy in today’s post. I suggest it’s better to think of drift as the baseline process, the ongoing, minor, scattershot deviation from theoretical (and nonexistent) equilibrium. Then one considers population history such as gene migration, bottlenecks, and founder effects, the latter two potentially including inbreeding, as the baseline physical context for what’s been happening. Selection therefore occurs in this cauldron of ongoing processes and historical impacts, and does not by default outweigh or reverse them. It is theoretically sound to suggest that much selection is so consequential – and in some cases, muzzled, if you will – because a particular change in frequencies was jump-started by one of these processes, thus presenting that specific bias in the population already.
Links: (about those funky orchids) Why are orchid flowers so diverse? Reduction of evolutionary constraints by paralogues of class B floral homeotic genes
Next: Graven images