Natural selection

Look closely at any group of animals or plants and you will see they are not identical. Some are a little bigger, faster, darker, taller. That everyday variation is where the whole story starts.

Here is the second fact: more get born than can possibly survive. A single moth lays hundreds of eggs, but the world only has room for a few. So there is a squeeze, and who makes it through is not completely random. If a trait happens to help you survive and have young in the place you live, you leave behind more offspring, and they tend to carry that same trait.

Do that over and over, generation after generation, and the helpful trait becomes common while the unhelpful one fades. The population has changed, without anyone planning it.

A real example: peppered moths in England. They rest on tree bark, and birds eat the ones they can spot. When soot from factories darkened the bark, pale moths stood out and got eaten, while dark moths blended in and survived. Within a few decades most of the moths were dark. The trees got cleaned up later, and the pale ones came back. The moths did not choose any of this. The surroundings did the sorting.

Natural selection needs three ingredients, and if all three are present it happens on its own. There has to be variation (individuals differ). That variation has to be heritable (offspring resemble their parents). And it has to affect reproduction (some variants leave more surviving young than others). Give it those three and the mix of traits in the population will shift.

The word "fitness" trips people up. In biology it does not mean strong or athletic. Fitness is just the expected number of offspring an individual leaves. A frail moth that survives and breeds is fitter, in this exact sense, than a sturdy one the birds picked off first.

Selection works on phenotypes, the actual visible traits, and it can only sort variation that already exists. It does not create anything new. New variation comes from mutation, random changes in the DNA, plus the reshuffling of sexual reproduction. Selection is the filter; mutation is what refills the tray.

Depending on which variants win, selection takes different shapes. Directional selection pushes a trait one way (moths get darker). Stabilising selection favours the middle and trims the extremes (very large and very small newborns both fare worse than average). Disruptive selection favours both extremes over the middle.

Two measured cases make it concrete. Antibiotic resistance in bacteria is natural selection you can watch in a week: the few cells that happen to carry a resistance trait survive the drug and repopulate. And on the Galápagos, Peter and Rosemary Grant measured finch beaks year by year; after a drought that left only tough, large seeds, the average beak size in the next generation was measurably bigger, because big-beaked birds ate and survived better.

One thing to hold onto: nothing is trying to evolve. There is no goal and no designer. Selection is a filter that runs automatically whenever those three ingredients are present.

Evolution is a change in allele frequencies. Zoom in from traits to genes and evolution has a precise definition: it is a change, across generations, in how common the different versions of a gene (the alleles) are in a population. The peppered-moth story is really the frequency of the dark allele rising from near zero toward one. To know whether selection is acting at all, you need a baseline for what happens when it is not. That baseline is Hardy–Weinberg equilibrium: for one locus with alleles at frequencies \(p\) and \(q = 1 - p\), a large, randomly mating population with no selection, mutation, migration or drift keeps genotype frequencies at \[ p^2 + 2pq + q^2 = 1, \] generation after generation. Nothing changes. Real populations depart from this null, and the departures are what we call evolution.

Selection as a force on frequency. Attach a relative fitness \(w\) to each genotype and let \(s\) be the selection coefficient (the fractional disadvantage of the disfavoured type, so its fitness is \(1 - s\)). For a favoured allele the per-generation change works out to roughly \[ \Delta p \approx \frac{s\,p(1-p)}{\bar{w}}, \] where \(\bar{w}\) is the mean fitness of the population. The factor \(p(1-p)\) is the catch: change is slowest when an allele is very rare or nearly universal, and fastest in the middle, which is exactly the S-shaped sweep the interactive above traces out. That is the whole machine in one line.

Drift competes with selection. Selection is not the only thing moving allele frequencies. In a finite population, frequencies also wander at random from one generation to the next simply because of sampling; this is genetic drift. Its strength scales with the effective population size \(N_e\): drift dominates when \(N_e\,s\) is small, selection dominates when it is large. So the same tiny advantage that sweeps to fixation in a huge population can be lost to chance in a small one.

Fitness is about reproduction, not survival. Selection also works through mating and through relatives. Sexual selection favours traits that improve mating success even at a survival cost (the peacock's tail). Kin selection favours helping relatives, because they share your genes; Hamilton's inclusive fitness counts offspring you raise yourself and the extra relatives that survive because you helped. This is also the answer to the tired charge that "survival of the fittest" is circular. It would be circular if fitness meant "whoever survives", but it does not. Fitness is defined by expected reproduction, which can be measured independently of who happens to live, so the theory makes real, testable predictions.

What selection is not. It is not progress toward some higher form; a parasite that loses its gut is as evolved as anything else. It is not for the good of the species; it acts on individuals (and genes), and traits that help the group but hurt the individual usually lose. And it has no foresight. It cannot save a trait for later or plan around a coming change. It only ever rewards what pays off in the current environment, which is why a shift in that environment, like soot on the bark or a new antibiotic, can reverse a trend that once looked permanent.