Evolutionary dynamics of the most populated genotype on rugged fitness landscapes

We consider an asexual population evolving on rugged fitness landscapes which are defined on the multi-dimensional genotypic space and have many local optima. We track the most populated genotype as it changes when the population jumps from a fitness peak to a better one during the process of adaptation. This is done using the dynamics of the shell model which is a simplified version of the quasispecies model for infinite populations and standard Wright-Fisher dynamics for large finite populations. We show that the population fraction of a genotype obtained within the quasispecies model and the shell model match for fit genotypes and at short times, but the dynamics of the two models are identical for questions related to the most populated genotype. We calculate exactly several properties of the jumps in infinite populations some of which were obtained numerically in previous works. We also present our preliminary simulation results for finite populations. In particular, we measure the jump distribution in time and find that it decays as $t^{-2}$ as in the quasispecies problem.

💡 Research Summary

The paper investigates how an asexual population adapts on rugged fitness landscapes—high‑dimensional genotype spaces populated with many local fitness peaks—by focusing on the dynamics of the most populated genotype (MPG). Two complementary frameworks are employed. The first is a “shell model,” a reduced version of the infinite‑population quasispecies equations, in which all genotypes at the same Hamming distance from a reference sequence are grouped into a shell. The dynamics of each shell’s frequency (x_k(t)) obeys a set of coupled differential equations that incorporate the mutation rate (\mu), a transition matrix (M_{jk}) describing mutational jumps between shells, the shell’s average fitness (W_k), and the population‑wide mean fitness (\bar W). By initializing the system with a single central genotype, the authors show that the shell model reproduces the full quasispecies dynamics for fit genotypes and, crucially, predicts the times at which the MPG jumps from one fitness peak to a higher one.

The second framework is a Wright–Fisher simulation of large but finite populations, which adds stochastic sampling (genetic drift) to the deterministic mutation‑selection process. By varying the population size (N), the authors demonstrate that when (N) is sufficiently large (e.g., (N\ge10^5)), the finite‑population dynamics converge to the shell‑model predictions: the same sequence of MPG replacements occurs, and the statistical distribution of jump times follows a power law (P(t)\sim t^{-2}). For smaller (N), drift introduces variability in the intervals between jumps and delays the ultimate attainment of the global fitness optimum, highlighting the role of stochasticity in realistic evolutionary scenarios.

A central analytical result is the exact derivation of the jump‑time distribution for infinite populations. By solving the shell‑model equations, the authors obtain the probability density for the occurrence of the (j)‑th jump, showing it decays as (t^{-2}). This matches earlier numerical observations and confirms that adaptation on rugged landscapes proceeds via rare, large‑scale transitions rather than a smooth, incremental climb. The paper also quantifies the correspondence between the shell model and the full quasispecies model, establishing that for high‑fitness genotypes the two are virtually indistinguishable in terms of population fractions and timing of dominance shifts.

Preliminary finite‑population simulations further validate the analytical predictions. The authors measure the jump distribution, confirm the (t^{-2}) decay, and explore how mutation rate (\mu) and the statistical properties of the fitness landscape (e.g., distribution of peak heights) affect the frequency and magnitude of jumps. They note that higher mutation rates increase the likelihood of early jumps but can also generate a higher load of deleterious mutants, while landscapes with broader fitness variance produce more pronounced jumps.

In the discussion, the authors outline future directions: extending the analysis to time‑dependent (changing) fitness landscapes, incorporating recombination, and comparing model outcomes with experimental evolution data from microbial systems. Overall, the study provides a rigorous theoretical foundation for understanding how populations navigate complex adaptive landscapes, demonstrating that the shell model offers a tractable yet accurate description of the most populated genotype’s evolutionary trajectory, and that the characteristic (t^{-2}) jump‑time law is a robust signature of adaptation on rugged fitness topographies.