ABSTRACT
Despite the diversity of life, studies of variation across animals often remind us of our deep evolutionary past. Abundant genome sequencing over the last ∼25 years reveals remarkable conservation of genes and recent analyses of gene regulatory networks illustrate that not only genes but entire pathways are conserved, reused, and elaborated in the evolution of diversity. Predating these discoveries, 19th-century embryologists observed that though morphology at birth varies tremendously, certain stages of embryogenesis appear remarkably similar across vertebrates. Specifically, while early and late stages are variable across species, anatomy of mid-stages embryos (the phylotypic stage) is conserved. This model of vertebrate development and diversification has found mixed support in recent analyses comparing gene expression across species possibly owing to differences across studies in species, embryonic stages, and gene sets compared. Here we perform a meta-analysis of 186 microarray and RNA-seq expression data sets covering embryogenesis in six vertebrate species spanning ∼420 million years of evolution. We use an unbiased clustering approach to group stages of embryogenesis by transcriptomic similarity and ask whether gene expression similarity of clustered embryonic stages deviates from the null hypothesis of no relationship between timing and diversification. We use a phylogenetic comparative approach to characterize expression conservation pattern (i.e., early conservation, hourglass, inverse hourglass, late conservation, or no relationship) of each gene at each evolutionary node. We find an enrichment of early conservation and hourglass patterns and a large depletion of genes exhibiting no distinguishable pattern of conservation. Using this approach, we ask whether the proportions of genes following distinct evolutionary conservation patterns change through evolutionary time and whether genes consistently follow the same pattern across nodes of the vertebrate phylogeny. We find that genes exhibiting an hourglass pattern at one node of the phylogeny are more likely to show an hourglass pattern at other nodes with 89 hourglass genes shared in at least three of the four nodes compared to only six early conservation genes. Consistent with the hourglass hypothesis, this finding suggests that genes following an hourglass pattern are more conserved over evolutionary time.