Evolution & Natural Selection

How life changes - one generation at a time!

Junior level — plain language, no maths

Picture a colony of beetles living on brown soil. Most are brown, but every so often a mutation throws up a bright green one. A hungry bird overhead picks off the green beetles at a glance and misses the brown ones - so the brown beetles live, breed, and pass their brownness on. Generation after generation, green fades out and the colony turns almost entirely brown. That's natural selection, the single most powerful idea in all of biology.

Charles Darwin saw in 1859 that this one plain process shapes everything alive: whoever carries traits that help them survive and breed leaves more offspring, who inherit those traits. Stack that up over millions of years and it blooms into staggering variety - bacteria to blue whales, mushrooms to maple trees. Every creature you've ever laid eyes on sits at the end of an unbroken chain of survivors stretching back about 3.8 billion years, all the way to the very first cell.

The evidence pours in from every direction: fossils record the slow reshaping of bodies over time, DNA reveals exactly how closely any two species are related, and antibiotic-resistant bacteria and pesticide-proof insects are evolution unfolding right now, in real time. In the simulation below, watch a population reshape itself before your eyes as the selection pressure shifts.

Population genetics of selection, drift, and speciation

Student level — the core equations

Strip evolution to its definition and it's just change in allele frequencies over time, driven by four forces: natural selection, genetic drift (chance sampling in finite populations), mutation (fresh variation) and gene flow (migration). Selection at a locus with allele frequencies \(p\) and \(q = 1-p\) and fitnesses \(w_{11}, w_{12}, w_{22}\) shifts \(p\) each generation by \(\Delta p = \dfrac{pq\,[\,p(w_{11}-w_{12}) + q(w_{12}-w_{22})\,]}{\bar{w}}\), scaled by the mean fitness \(\bar{w}\). Fisher's Fundamental Theorem makes the direction inevitable: mean fitness climbs at a rate equal to the additive genetic variance in fitness. Evolution grinds uphill.

Uphill isn't the whole story, though, because chance gets a vote. Genetic drift jostles allele frequencies purely by the luck of who happens to breed, with variance \(p(1-p)/2N\) per generation. In a small population that noise can drown the signal: once a variant's advantage \(s\) drops below \(1/2N\), drift overrules selection, and even mildly harmful mutations can drift all the way to fixation. Small populations, in effect, evolve partly at random.

Let two populations stop swapping genes and they drift and adapt apart until they can no longer interbreed - speciation. The usual route is allopatric: a mountain range rises or a sea floods in, splits a population, and isolation finishes the job. Sympatric speciation, splitting with no physical barrier at all, needs unusually fierce disruptive selection. Mayr's biological species concept draws the line at reproductive isolation - though it strains on asexual microbes and on fossils, which is exactly why ecologists and phylogeneticists keep rival definitions in their back pockets.

Phylogenetics, the modern synthesis, and evo-devo

Scholar level — full mathematical depth

01The Modern Synthesis

For decades Darwin's selection and Mendel's genetics sat uneasily side by side - one continuous, one discrete. The Modern Synthesis (1930s–50s; Fisher, Wright, Haldane, Dobzhansky, Mayr) welded them together, showing that smooth, continuous variation is simply what you get from many Mendelian genes each nudging a trait a little. It turned evolution from a compelling narrative into a quantitative science, with population genetics supplying the equations of how alleles rise and fall.

02The coalescent: running the film backwards

A powerful shift was to stop asking how genes spread forward and instead trace a sample's ancestry backward. Kingman's coalescent (1982) does exactly that: follow any set of gene copies back in time and they merge, pair by pair, until all reach a single most recent common ancestor, on a timescale of about \(4N_e\) generations for diploids. This genealogical view is the workhorse null model of modern population genetics, turning patterns in present-day DNA into inferences about population size, structure and history.

03Reconstructing the tree of life

Given sequences from many species, phylogenetics hunts for the tree that best explains them. Maximum-likelihood methods pick the topology and branch lengths maximising \(P(\text{data} \mid T, \text{model})\), typically under a GTR substitution model with gamma-distributed rate variation across sites. Bayesian approaches (MrBayes, BEAST) go further, sampling the whole posterior \(P(T \mid \text{data})\) by MCMC and returning honest uncertainty on both the branching order and the timing.

04The molecular clock and deep time

Because neutral substitutions accumulate at a roughly steady rate, genetic distance doubles as elapsed time - a molecular clock. Relaxed-clock models that let the rate vary across lineages, calibrated against fossils, push dates far beyond what bones alone allow: eukaryotes around 2 billion years ago, animals near 750 million, tetrapods about 375 million. DNA lets us read a rough timestamp off events that left no fossil at all.

05Evo-devo: it's the regulation, not the genes

The biggest surprise of the genome era was how few genes separate a fly from a human, and how much of the difference is in when and where genes switch on. Evolutionary developmental biology showed that morphological diversity springs mostly from changes in gene regulation, not gene content. The Hox genes, a conserved cluster that stamps identity along the body axis, are shared across all bilaterians; shift their expression boundaries and a segment grows a leg instead of an antenna, without touching the genes themselves.

06One toolkit, endlessly reused

Evolution turns out to be a relentless reuser of parts. The same master switches recur across wildly separate lineages - Pax6 orchestrating eyes, Dlx limbs, Nkx2.5 hearts - a shared developmental toolkit deployed again and again in new contexts. It explains why eyes, limbs and hearts evolved convergently so many times: nature wasn't reinventing them from scratch each time, but redialling an ancient set of genetic controls. The unity of life is written not just in shared genes, but in shared ways of using them.