Evolution of G-Matrices
G-matrices (genetic variance-covariance matrices) describe the architecture of quantitative genetic variation that underlies continuous traits. The G-matrix determines a population's response to selection in the short term, but its role in longer-term evolutionary processes is somewhat controversial. G-matrices are expected to evolve in both stochastic and deterministic ways, and may also be influenced by the environment.
Continuing collaborations allow our lab to explore the contexts in which G evolves, and how and when it influences trajectories of phenotypic diversification. With Lynda Delph (Indiana University) and Janet Steven (Christopher Newport University), we are continuing to analyze the results of experimental manipulations of artificial selection that were designed to test whether strong genetic correlations could be broken in Silene latifolia. We showed that in just a few generations of multivariate selection, extremely strong between sex correlations could be completely erased.
With Joel McGlothlin and Jonathan Losos (Washington University, St. Louis), we are analyzing the results of a massive breeding design that estimated G-matrices in seven species of Caribbean Anolis lizards. The experiment compares the architecture of genetic variation within and among islands, and within and among ectomorphs of lizards. This is the first data set to examine the evolution of G through an adaptive radiation in which history (age of divergence) is not confounded with selection. The results suggest that axes of genetic variation do indeed influence directions of phenotypic evolution over tens of millions of years.