Three selection schemes — same N, same s, same start — different outcomes. Why?
The Moran process is a finite-population model with overlapping generations — one birth and one death per timestep, keeping population size constant. This tool simulates 2 alleles (A and a) with relative fitness wA = 1 + s, wa = 1.
The three schemes are mathematically identical for neutral evolution (s = 0) but diverge under selection because selection placement changes the Markov chain's stationary distribution. Scheme III retains reversibility and a closed-form stationary distribution for any number of alleles; Schemes I and II become nonreversible when fitnesses are unequal and require computational methods.
Based on Braha & de Aguiar (2026), Selection Mechanisms, Stationary Distributions, and Reversibility in Multiallelic Moran Models, arXiv:2607.06732. 📝 Concept note