Fig. 6.

Constraints and consequences of multivariate selection. (A) Individual trait contribution toward genetic constraint is identified by dropping the $i\text{th}$ trait from $\mathit{Z}$ (actual domestication trajectory) and ${\mathit{g}}_{\mathit{max},\mathit{T}}$ (genetic lines of least resistance) and measuring the angle ${\theta }_{\mathit{dropone}}^i$ between the two vectors. If ${\theta }_{\mathit{dropone}}^i$ is smaller than ${\theta }_T$ = 66.3° (angle between $\mathit{Z}$ and ${\mathit{g}}_{\mathit{max},\mathit{T}}$), then the $i\text{th}$ trait is said to constrain evolution. If ${\theta }_{\mathit{dropone}}^i$ is larger than ${\theta }_T$, then the $i\text{th}$ trait is said to assist evolution. (B) Multivariate response (${\mathit{R}}_{\mathit{i}}$) from hypothetical selection on a single $i\text{th}$ trait. ${\mathit{R}}_{\mathit{i}}$ is compared to $\mathit{Z}$ through the angle (${{\theta }^i}_Z$), and scalar projection ($\mathit{pro}{j}_Z{R}^i$) of ${\mathit{R}}_{\mathit{i}}$ on $\mathit{Z}$. ${{\theta }^i}_Z$ measures the deviation from $\mathit{Z}$ by selecting on the $i\text{th}$ trait. $\mathit{pro}{j}_Z{R}^i$ measures the evolutionary gain along $\mathit{Z}$ by selecting on the $i\text{th}$ trait.