Fig. 5. Integrating theoretical modeling with experiment shows the maintenance of selfish mitochondrial genomes in a heteroplasmic state through multiple distinct evolutionary mechanisms.

Maximum-likelihood estimates of intra- and inter-organismal fitness effects (left and center columns, respectively), each as a function of mutant frequency, together with the resulting most evolutionarily stable mutant frequency distribution (right column). Similar to bottom row of Fig. 3, but for the 4 mitochondrial genotypes shown in Fig. 4: mptDf2 (a–c, Supplementary Data 2), mpt4 (d–f, Supplementary Data 3), mpt2 (g–i, Supplementary Data 4), and mptDf3 (j–l, Supplementary Data 5). Each plot shows maximum-likelihood estimates corresponding to the empirical data shown in Fig. 4 (solid black lines) and 100 parametric bootstrap data sets (colored lines) to visualize confidence estimates. Model parameters specifying intra-organismal selection, organismal selection, and intra-organismal drift (left column, insets), were collectively estimated using a joint maximum-likelihood approach that combines all 3 corresponding empirical data sets in Fig. 4. Error bars in genetic drift plots (left column, insets) indicate mean, minimum and maximum bootstrap values. See Supplementary Data 2–5 for empirical and bootstrap model parameters.