Genomic exploration of the journey of Plasmodium vivax in Latin America
Fig 5
Most likely P. vivax colonization history scenarios of Latin America inferred using the Approximate Bayesian Computation Random Forest approach.
(a) Twelve colonization scenarios were tested. Distinct solid colored lines correspond to populations with distinct effective size parameters (labels starting with “N”) and horizontal dashed lines represent admixture events with a contribution ra and 1-ra from each of the two contributors. Time parameters of the different events are displayed along the vertical line on the right of each scenario (labels starting with “t”, not to scale). The parameters of each scenario are characterized by probabilistic distributions detailed in S3 Table. Under each scenario is the mean percentage of RF classification votes ± standard deviation (SD) and the mean type II error ± SD. The three most supported scenarios (scenarios 6, 4 and 8, with 46.0% of the total votes) are highlighted by dotted-line rectangles. (b) The best-supported scenario (scenario 6) in the DIYABC-RF analysis, scaled to relative time-parameter estimates (converted to years assuming a generation time of 5.5 generations/year). On the left, each time parameter estimate is indicated by the mean (circle), median (diamond), and 90% confidence intervals (colored bars). Here, the results are only shown considering the Colombian population as a Latin America representative population. Results considering the French Guiana population were very similar (see S4 Table).