Figure 5

IM analysis of Panama H. melpomene and H. cydno data. The panels show approximate Bayesian posterior probability distributions for effective population size of H. melpomene (θ _mel ), H. cydno (θ _cyd ), and the ancestor of the two species (θ _A ). The time since divergence of the two species (t), and the locus-specific bidirectional introgression rates are also shown (m). The three datasets analysed are the basic IM dataset (blue), and IM reduced dataset 1 (pink) and IM reduced dataset 2 (green). Analysis of the basic IM dataset is compromised by recombination within Tpi and Ci, which is assumed not to occur in the IM algorithm. Reduced datasets containing apparently non-recombined segments of the genes were analyzed to overcome this difficulty. IM reduced dataset 1 differs only from IM reduced dataset 2 in that a different, shorter, part of the Ci locus is used; the low sequence information probably explains why there is little information in the former run in the last panel (probabilities for IM dataset 1 are enhanced 10-fold in this panel only, for clarity). The curves show useful parameter estimation , except in the case of ancestral population size (θ _A ), the upper tail of the time of divergence (t) and the introgression for Ci for IM reduced dataset 1.