Conditions for Darwinian evolution in compartmentalized autocatalytic reaction networks

The first forms of life likely consisted of protocells endowed with metabolism, growth, and division. Such systems may have evolved due to variation and heredity in their chemical composition, even before the advent of genetics. However, whether compositional heredity is robust enough to sustain evolution by natural selection remains unknown, especially given that early compartmentalization cycles were likely imperfect, potentially disrupting stable inheritance across generations. Here, we show that multistable autocatalytic reaction networks can maintain heritable compositional states across a broad class of growth-division regimes, including continuous, serial, symmetric division, and multi-fragmentation cycles. We further identify parameter domains that preserve stable inheritance in the presence of stochastic variation, such that selection can operate efficiently. We finally demonstrate rudimentary forms of evolution by natural selection in populations of protocells with two heritable states, which we illustrate in an experimentally feasible setting. Our findings establish conditions for natural selection in compartmentalized autocatalytic systems and set the stage for understanding the minimal requirements for open-ended evolution.