Data from: Testing convergence versus history: convergence dominates phenotypic evolution for over 150 million years in frogs

Striking evolutionary convergence can lead to similar sets of species in different locations, such as in cichlid fishes and Anolis lizards, and suggests that evolution can be repeatable and predictable across clades. Yet, most examples of convergence involve relatively small temporal and/or spatial scales. Some authors have speculated that at larger scales (e.g., across continents), differing evolutionary histories will prevent convergence. However, few studies have compared the contrasting roles of convergence and history, and none have done so at large scales. Here we develop a two-part approach to test the scale over which convergence can occur, comparing the relative importance of convergence and history in macroevolution using phylogenetic models of adaptive evolution. We apply this approach to data from morphology, ecology, and phylogeny from 167 species of anuran amphibians (frogs) from 10 local sites across the world, spanning ∼160 myr of evolution. Mapping ecology on the phylogeny revealed that similar microhabitat specialists (e.g., aquatic, arboreal) have evolved repeatedly across clades and regions, producing many evolutionary replicates for testing for morphological convergence. By comparing morphological optima for clades and microhabitat types (our first test), we find that convergence associated with microhabitat use dominates frog morphological evolution, producing recurrent ecomorphs that together encompass all sampled species in each community in each region. However, our second test, which examines whether and how much species differ from their inferred optima, shows that convergence is incomplete: that is, phenotypes of most species are still somewhat distant from the estimated optimum for each microhabitat, seemingly because of insufficient time for more complete adaptation (an effect of history). Yet, these effects of history are related to past ecologies, and not clade membership. Overall, our study elucidates the dominant drivers of morphological evolution across a major vertebrate clade and shows that evolution can be repeatable at much greater temporal and spatial scales than commonly thought. It also provides an analytical framework for testing other potential examples of large-scale convergence.PhylogeniesThe two phylogenies used for data analyses in this paper. See Appendix S2 for details on phylogenies.MoenEA2015SystBiol.nexusSupplementary AppendicesAll five supplementary appendices, with supplementary figures and tables (including data from the paper).2015-09-21_SupMat_SystBiol.docx2nd.history.test.MoenEAR function for the second history test (i.e. breaking down variance around optimum into variance due to history and due to random deviation). Details for use of the function are found as comments in the initial lines of the file. Executing this code will produce a table like Table 4 in the published paper.ou.optima.MoenEATable of estimated OU optima for the preferred OU model of evolution (lowest AICc). This can be used to do an example analysis with the R code for the second history test.sample.morph.data.MoenEATable of principal component data and microhabitat state for the preferred OU model of evolution (lowest AICc). This can be used to do an example analysis with the R code for the second history test