Evolutionary correlations between phenotypic and environmental traits characterize adaptive radiations. However, the lizard genus Liolaemus, one of the most ecologically diverse terrestrial vertebrate radiations on earth, has so far shown limited or mixed evidence of adaptive diversification in phenotype. Restricted use of comprehensive environmental data, incomplete taxonomic representation and not considering phylogenetic uncertainty may have led to contradictory evidence. We compiled a 26 taxon dataset for the L. gracilis species group, representing much of the ecological diversity represented within Liolaemus, and used environmental data to characterize how environments occupied by species’ relate to phenotypic evolution. Our analyses, explicitly accounting for phylogenetic uncertainty, suggest diversification in phenotypic traits toward the present, with body shape evolution rapidly evolving in this group. Body shape evolution correlates with the occupation of different structural habitats indicated by vegetation axes suggesting species have adapted for maximal locomotory performance in these habitats. Our results also imply that the effects of phylogenetic uncertainty and model misspecification may be more extensive on univariate, relative to multivariate analyses of evolutionary correlations, which is an important consideration of analyzing data from rapidly radiating adaptive radiations.

Morphological data were collected by Dr. Avila's lab. Genetic data were generated by Dr. Martinez as a graduate student with Dr.'s Morando & Avila. Dr. Edwards compiled the environmental data from satelite image (vegetation data), digital elevation model (elevation), and climatic databases, into map layers. Spatial information of specimens was used to extract and compile environmental data. Dr. Edwards used the genetic data generated by Dr. Martinez to build a species tree of candidate lineages. Dr. Edwards used the morphological data, phylogeny and environmental data in a series of comparative phylogenetic analyses.

These are the cleaned data used in the analyses of the paper. Missing value information and processing of data for the paper can still be viewed in the code.