Skip to main content
Dryad logo

Data from: Environment predicts repeated body size shifts in a recent radiation of Australian mammals


Roycroft, Emily J.; Nations, Jonathan A.; Rowe, Kevin C. (2019), Data from: Environment predicts repeated body size shifts in a recent radiation of Australian mammals, Dryad, Dataset,


Closely related species that occur across steep environmental gradients often display clear body size differences, and examining this pattern is crucial to understanding how environmental variation shapes diversity. Australian endemic rodents in the Pseudomys Division (Muridae: Murinae) have repeatedly colonized the arid, monsoon, and mesic biomes over the last 5 million years. Using occurrence records, body mass data, and Bayesian phylogenetic models we test whether body mass of 31 species in the Pseudomys Division can be predicted by their biome association. We also model the effect of eight environmental variables on body mass. Despite high phylogenetic signal in body mass evolution across the phylogeny, we find that mass predictably increases in the mesic biome, and decreases in arid and monsoon biomes. As per Bergmann’s rule, temperature is strongly correlated with body mass, as well as several other variables. Our results highlight two important findings. First, body size in Australian rodents has tracked with climate through the Pleistocene, likely due to several environmental variables rather than a single factor. Second, support for both Brownian motion and predictable change at different taxonomic levels in the Pseudomys Division phylogeny demonstrates how the level at which we test hypotheses can alter interpretation of evolutionary processes.


Occurrence data and environmental variables for 31 species of murine rodents in the Pseudomys Division was obtained from the Atlas of Living Australia (ALA,

Body size data was obtained from Breed B., Ford F. 2007. Native Mice and Rats. CSIRO Publishing, Collingwood, VIC.

To account for phylogenetic uncertainty in the Pseudomys Division topology, we used 100 randomly sampled trees from the posterior distribution of trees generated by Smissen and Rowe (2018).