Comparative Quantitative Genetics of the Pelvis in Four-Species of Rodents and the Conservation of Genetic Covariance and Correlation Structure

Quantitative genetics is a powerful tool for predicting phenotypic evolution on a microevolutionary scale. This predictive power primarily comes from the Lande equation (Δz̅=Gβ), a multivariate expansion of the breeder's equation, where phenotypic change (Δz̅) is predicted from the genetic covariances (G) and selection (β). Typically restricted to generational change, evolutionary biologists have proposed that quantitative genetics could bridge micro- and macroevolutionary patterns if predictions were expanded to longer timescales. While mathematically possible, making quantitative genetic predictions across generations or species is contentiously debated, principally in assuming long-term stability of the G-matrix. Here we tested stability at a macroevolutionary timescale by conducting full-sib breeding programs in two species of sigmodontine rodents from South America, the leaf-eared mice Phyllotis xanthopygus and Phyllotis darwini and estimated the G-matrices for eight pelvic traits. To expand our phylogenetic breadth, we incorporated two additional G-matrices measured for the same traits from Kohn & Atchley's 1988 study of the murine rodents Mus musculus and Rattus norvegicus. Using a phylogenetic comparative framework and four separate metrics of matrix divergence or similarity, we found no significant association between evolutionary divergence among species G-matrices and time, supporting the assumption of stability for at least some structures. However, the phylogenetic sample size is necessarily small. We suggest that small fluctuations in covariance structure can occur rapidly, but underlying developmental regulation prevents significant divergence at macroevolutionary scales, analogous to an Ornstein-Uhlenbeck pattern. Expanded taxonomic sampling will be needed to test this suggestion.