Fig. 5: Dimensions with the most developmental variation do not evolve the fastest.
From: Macroevolution along developmental lines of least resistance in fly wings
a, If developmental bias constrains evolution, morphological evolution is expected to be faster if it aligns well with the main axes of the D matrix. To test this hypothesis, we first quantified the vector of shape change along the edges of the phylogeny and computed the developmental variance captured by shape evolution. If evolution occurs mainly along the main axes of D, this variance will be large and close to the variance explained by the first eigenvector of D (dmax), indicating an alignment between shape evolution and developmental bias. b, We then correlated the strength of this alignment with the rate of evolution (shape evolution in Procrustes distance per million years) along the branch, expecting a positive relationship if fast rates of evolution are constrained along wing dimensions with high variability. In contrast to this expectation, the rate of evolutionary change does not depend on its alignment with D (or M; Extended Data Fig. 6). c, This observed correlation (dashed vertical lines) is also much lower than what is expected under pure Brownian motion. d, The increase in wing shape divergence with phylogenetic distance. The data in grey (below 40 million years) are from Houle et al.18.
