A simple model reveals why complex evolutionary innovations follow predictable paths.

Determining the principles underlying the origin of novel characters has been a fundamental goal of evolutionary biology. Yet, key mechanisms remain poorly understood, hindered by the lack of a general, mechanistic model that unites genotypic and phenotypic change and predicts outcomes. Here, we present a model founded in developmental biology, where phenotype is controlled by a hierarchical gene regulatory network (GRN) consisting of regulators specifying character identity and effectors producing specific states. While our model is simplified, evolutionary simulations for divergence between repeated characters and switching between alternative identities easily recreated empirical patterns. While observations of the emergence of novel identities are lacking due to their rarity or multi-step nature, our simulations reveal that the most complex characters exhibit the strongest convergence in regulatory pathways (deep homology). Our model provides insights into the mechanisms underlying evolutionary novelties and offers a framework for the developmental evolution of a variety of traits.