Figure 2

Hyper-posterior and Hyper-prior samples given Marquesas DNA sequence data (seven taxon-pairs). Dashed lines depict hyper-prior distributions and solid lines depict hyper-posterior distributions. Stage 1 hyper-parameter estimates: (A) Average effective population size E((θ _τ )₂) of the putatively colonized endemic populations at the seven isolation times (T _C and T _V ). (B) The number of descendent populations per seven (Y) taxon-pairs that arise by colonization at times $T_C=\{{\tau }_C^1,\mathrm{...},{\tau }_C^Z\}$. Stage 2 hyper-parameter estimates where hyper-prior is conditional given constant value of Z = 7: (C) ΩC dispersion index of Z = 7 colonization times; Ω C = Var(τ _C )/E(τ _C ) where τ _C is colonization time. (D) E(τ _C ), average colonization time across Z = 7 colonization times. For each estimate, tolerance was 0.001 (2,000 accepted draws) using the local regression algorithm.