Simulation lab
Genetic Drift Versus Selection Lab
When does chance beat selection? In small populations, random sampling of gametes (genetic drift) can fix even slightly harmful alleles, swamping weak selection. This panel simulates PURE drift: there is no selection term in it at all. Your experiment is to measure how violently allele frequencies swing by chance alone at small N, and then reason: if random swings per generation dwarf what weak selection could push, drift wins. Compare with the directional, repeatable change you would see in the natural-selection panel.
Genetic Drift Simulator
Random sampling alone changes allele frequencies — no selection involved.
Shrink N to 5–10 and drift dominates: alleles fix or vanish within a few dozen generations purely by chance. Raise N to 200 and frequencies barely move — drift is weak in large populations. Every replicate starts identically yet ends differently: that randomness is genetic drift.
Wright–Fisher ideal population: discrete generations, random mating, constant size N, no selection, mutation or migration. Each generation's 2N alleles are a binomial sample of the previous generation's gene pool. Real populations add all the complications this model leaves out.
How to use it
Set N = 10, p0 = 0.5, 8 replicates, 100 generations, and press 'Run again' several times; count how often alleles fix or are lost by pure chance.
Raise N to 200 with identical settings: trajectories hug p0, showing drift is weak in large populations.
Reason the comparison: weak selection (s around 0.01) shifts p by roughly s per generation, far smaller than the random jumps you saw at N = 10, so drift dominates there.
Try N = 5 with p0 = 0.9 and watch a common allele get lost entirely by chance.
Record the fixed, lost and still-segregating counts across several runs to feel the randomness quantitatively.
Honest framing matters here: this panel contains NO selection term; it is pure Wright-Fisher neutral drift, so the drift-versus-selection contest is made by your reasoning and by comparing with the separate natural-selection panel, not inside one simulation. Real populations experience drift and selection simultaneously. The model also omits mutation, migration and non-random mating, and assumes constant population size.
What you will learn
- Explain why the strength of drift scales approximately with 1/N.
- Explain how drift can fix or eliminate alleles despite weak selection in small populations.
- Predict when selection dominates (large N, strong selection) versus when drift dominates.
- Interpret replicate-to-replicate variation as the signature of randomness, not selection.