Category: Pop. Genetics

Summary: Finding the leakage rate where a dormant seed bank stops protecting modular quasispecies localization under strong recombination.

Dormant reservoirs can buffer evolving populations against harmful mixing, while recombination can either help adaptation or wash out structured genotype concentrations. This experiment asks when leakage out of a protective seed bank becomes large enough that modular localization can no longer survive under fixed high recombination.

The model builds a dense symmetric mutation-selection-recombination operator with active and dormant copies of each genotype, then uses GPU eigensolves and iterative deepening to bisect the seed-bank leakage rate. The threshold is defined by loss of the localized leading state rather than by a simple average-fitness summary.

That makes the experiment a direct test of dormancy as a structural buffer in quasispecies dynamics. It measures the point where the reservoir is still present in the model but no longer functions as an effective protection mechanism.

Method: GPU dense symmetric eigensolve with iterative deepening and bisection on seed-bank leakage in a mutation-selection-recombination operator.

What is measured: Critical leakage threshold, leading-state localization, spectral growth behavior, and threshold bracket width.