Imagine a mutation arises in a single individual—a tiny genetic change that, against all odds, would make its carrier slightly better at surviving. You might expect natural selection to seize this advantage and propagate it through the population. The reality is far stranger.

Most beneficial mutations vanish within a handful of generations. Not because they fail to confer advantage, but because their carriers happen to die young, fail to reproduce, or simply roll the genetic dice poorly. Chance, not selection, governs the earliest moments of nearly every adaptive change.

This is one of evolution's quieter truths: the visible patterns of adaptation we observe in nature represent the rare survivors of a vast statistical slaughter. Understanding why most mutations die quickly—and which ones manage to spread—reveals how stochastic the foundations of evolutionary change really are.

When a new mutation appears in a diploid population of size N, it exists in exactly one copy among 2N gene copies. Its initial frequency is 1/2N—a number that, for any realistic population, is vanishingly small. In a population of 10,000 individuals, a new mutation begins life at a frequency of 0.00005.

Population genetic theory shows that, even with no selection at all, the probability of eventual fixation for a neutral allele equals its current frequency. So a new neutral mutation has a 1/2N chance of ever sweeping the population, and a (2N-1)/2N chance of disappearing. The mathematics are unforgiving.

Reproductive variance compounds the problem. Even if a carrier is perfectly healthy, they might have no offspring, or only one. Each generation, the mutation passes through a stochastic bottleneck where Mendelian segregation, mating chance, and demographic luck all conspire to erase it.

This is why most evolutionary action happens not in single dramatic mutations but in the slow shuffling of frequencies among variants that have already survived this brutal early gauntlet. The newly arisen mutation is the underdog of evolutionary biology.

Takeaway

Evolution is filtered first by chance, then by selection. A mutation must survive luck before fitness ever has a chance to matter.

In 1927, J.B.S. Haldane derived a result that still shapes how we think about adaptive evolution. For a new beneficial mutation with selective advantage s, its probability of ultimately fixing in the population is approximately 2s. A mutation conferring a 1% fitness advantage has only a 2% chance of spreading.

Read that again. Even substantially beneficial mutations—the kind we imagine selection should embrace—are lost 98 times out of 100. The intuition that natural selection efficiently captures every advantage is simply wrong at the level of individual genetic variants.

The formula emerges from branching process mathematics. Early in its history, a mutation exists in so few copies that demographic stochasticity dominates selection. Only once the mutation reaches a frequency where it appears in many individuals does selection reliably take over and push it toward fixation.

This has a profound implication: the same beneficial mutation may arise repeatedly in a population, fail repeatedly, and only eventually—on perhaps the hundredth try—catch a lucky demographic wave and spread. Adaptation is partly a story of persistence against statistical headwinds.

Takeaway

Selection coefficients translate into probabilities, not certainties. A mutation that is twice as good doesn't always win—it just wins twice as often.

Because new mutations face such terrible odds, much of observable adaptation draws on a different reservoir: standing genetic variation—alleles already present in the population at non-trivial frequencies. These variants have already survived the lottery.

When environments change, populations that adapt rapidly often do so by selecting on variation that was previously neutral or even slightly deleterious. The classic Italian wall lizards introduced to Pod Mrčaru, the peppered moths of industrial England, and threespine sticklebacks colonizing freshwater lakes all adapted primarily from standing variation, not from waiting for new mutations to arise.

Standing variation occupies a privileged position. An allele present at 5% frequency in a population of 10,000 exists in roughly 1,000 gene copies. Demographic chance cannot easily erase it. Selection, however weak, can grip it immediately and pull frequencies upward without first surviving the stochastic gauntlet.

This reframes how we think about evolutionary speed. Rapid adaptation isn't necessarily evidence of strong selection on novel mutations—it's often evidence that the right variants were already waiting, hidden in the genetic background, for conditions that would make them valuable.

Takeaway

Populations don't adapt by inventing solutions from scratch. They adapt by promoting solutions they already had, quietly preserved at low frequency.

The fate of new mutations reveals evolution as a process governed as much by chance as by selection. Beneficial variants disappear routinely. Adaptation often draws on variation already accumulated, not on new genetic innovations.

This stochastic foundation doesn't weaken evolutionary theory—it sharpens it. Selection works on probabilities and populations, not on individuals and certainties. Patterns emerge from countless trials, most of them failed.

When we observe an adaptation in nature, we're seeing the lucky few that survived a vast statistical winnowing. The mutations that won the lottery. The variants that were already present when the world changed. Evolution is, in this sense, a record of survivors—nothing more, nothing less.