摘要: How the small genetic changes measured within populations scale up to the abrupt origins of biodiversity across the fossil record and the tree of life remains unresolved. Standard theory is gradualist and struggles with stasis broken by rapid bursts, multiple modes of heritable variation, and adaptation in lineages carrying little genetic variation. Here, we set a cross-scale evolutionary framework. It recast a trait as the attractor of a gene-regulatory network, so genetic change is buffered as cryptic variation and released when environment changing fast. Read as informational nonequilibrium, the framework recovers population genetics as a limit, makes stasis an equilibrium, and yields an evolutionary speed limit set by accessible variance. One mechanism reproduces stasis, punctuation, crisis survival, and radiation; fits to public fossil and phylogenetic data reproduce known patterns from a single mechanism, and makes discriminating falsifiable predictions.