A stochastic backward-time framework in population genetics that models the genealogical process by tracing sampled alleles' lineages to common ancestors; coalescent events occur at rates determined by effective population size and demographic history.

Principle

By following lineages backward, pairwise coalescence rates encode the demographic and reproductive processes of the population; under neutral models and large population limits, genealogies converge to simple coalescent processes (e.g., Kingman's coalescent).

Demonstration

Sampling n gene copies in a constant-size neutral population: the rate at which any pair of lineages coalesces is (k choose 2)/(2N_e) when k lineages remain, giving expected times to the most recent common ancestor on the order of 2N_e generations and producing distributions of segregating sites and pairwise diversity consistent with observed sequence variation.

Misapplication

Applying the basic coalescent without accounting for selection, recombination, population structure, nonrandom mating, or recent demographic changes; equating effective population size directly with census size without acknowledging scaling and interpretation differences.

Consequence

Provides a principled link between sequence data and demographic/selection parameters, enabling inference of population size changes, migration, and timing of events; supplies null distributions for neutrality tests and parameter estimation frameworks.

Reversal

Forward-time models (Wright–Fisher, Moran) simulate reproduction directly; inverting the perspective loses some analytic simplifications of the backward approach but recovers full-population stochastic dynamics—both views are complementary limits of the same process.

Boundary

Valid under assumptions of neutrality (unless extended), random mating, and appropriate scaling limits; simple coalescent results break down under strong selection, high recombination within loci, complex population structure, or violations of model assumptions unless generalized.

Semantic Tension

Tension between coalescent 'trees as random objects' and phylogenetic approaches that treat trees as fixed estimands; also between different coalescent variants (multiple-merger coalescents vs Kingman) depending on variance in offspring number.

Synthesis

Coalescent theory is a backward-time stochastic mapping from sampled genetic variation to ancestral genealogies and demographic parameters: it condenses forward-time reproductive processes into coalescence rates that facilitate inference, but its conclusions depend critically on neutrality, recombination, and structure assumptions and require appropriate generalization when those fail.