Biological and taxonomic sampling biases strongly distort reconstructions of mammal trait evolution, according to a study published Monday in Nature Communications.
A research team led by Graciela Sotelo simulated a future fossil record using living land mammals to measure how much evolutionary pattern survives incomplete preservation. Fossilization is non-uniform and depends on sedimentation rates, species ranges, body sizes, and scientific collection effort. Incomplete records frequently risk skewing analyses of deep-time biodiversity dynamics.
To evaluate these effects, the researchers filtered present-day species by sediment availability, body mass, geographic distribution, and spatial and taxonomic sampling patterns across three preservation rates. The filtered species were systematically pruned from a baseline mammal phylogenetic tree. The team then analyzed body mass and diet evolution using phylogenetic comparative methods.
The study found that biological and taxonomic filters strongly skew evolutionary inferences. Spatial filters, in contrast, do not cause significant distortion in evolutionary models.
The Galicia Supercomputing Center provided the computational resources required for the simulations. Research collaborators included scientists from Universidade de Vigo, University College London, Martin Luther University Halle-Wittenberg, Trinity College Dublin, and Axencia Galega de Innovación.
Funding for the work included grants from the European Research Council, the UK Natural Environment Research Council, the Royal Society, the Spanish Ministerio de Ciencia, Innovación y Universidades, the Agencia Estatal de Investigación, and Xunta de Galicia.
