Figure 3.

(A) Violin plots representing the posterior distributions of the parameters that were inferred by simulation-MCMC. Each chart represents a phenotype and each violin plot a selection coefficient in each period (indicated). The two parameters at right are the initial allele frequency and the age of the derived allele. The empty spaces (e.g. Cream phenotype in the Middle Bronze Age) occur because some simulations had several periods merged due to the derived allele appeared very late in the sampling, making the selection at the earliest periods pointless. (B) Violin plots of the initial allele frequency (top) and age (bottom) of the derived alleles. The empty places correspond to the cases when the derived allele first appeared in the Pleistocene, the allele associated to the corresponding phenotype was not derived (Bay); or the introduction of alleles didn’t get an appreciable probability. The green line represents the time of domestication, and the numbers above the probabilities that the age of the derived allele was younger than domestication.