Standard evolutionary theory should incorporate the complexity of adaptive evolving systems — including species, niches and environment — as dynamic relationship networks (see Nature 514, 161–164; 2014).

For example, epigenetic inheritance — which changes gene expression but not the DNA sequence — involves the storage of molecular information and its retrieval, transfer and processing at the supramolecular level. This involves transitory processes that are self-organized, self-assembled and dynamic.

DNA replication too is one of countless functional tasks of interest in the study of evolution: changes propagate through interlinked levels of organization, inducing connectivity and interaction at all scales of the multilevel system.

The process of natural selection is now being captured, by modelling fitness attractors that incorporate power laws and non-equilibrium steady states at the edge of chaos, with energy landscapes made of basins, valleys, floors, ridges and saddle points (see, for example, K. Friston J. R. Soc. Interface 10, 20130475; 2013).