Fig. 2

Seven different scenarios inferred for the introduction history of Xenopsylla cheopis in Madagascar and Mayotte. Three populations were considered: population from Longoni in Mayotte (LON), population from Marolaka (MAR) and population from Amparafaravola (AMP) in Madagascar where N1, N2, N3 are their respective effective population sizes. Nb1, Nb2 and Nb3 correspond to the numbers of founders in the introduced population. Time scales corresponding to generations back in time from the sampling date (time 0) are shown at the left (t1, t2 and t3 generations ago); db, is the duration of the initial bottleneck. For all the scenarios, LON, MAR and AMP derived from an unsampled ancestral population having N4 effective population size. Particularly for each of the scenarios 1–4, derivation from the ancestral population was independent. Scenario 1: AMP derived from the ancestral population at t3 followed by LON at t2 then, MAR at t1. Scenario 2: LON derived from ancestral population at t3 followed by MAR at t2 then, AMP at t1. Scenario 3: LON derived from ancestral population at t3 followed by AMP at t2 then, MAR at t1. Scenario 4: AMP derived from ancestral population at t3 followed by MAR at t2 then, LON at t1. The remaining scenarios (5–7) assumed that two parental populations had diverged from an ancestral population at t2 before they would admix and gave the third population at t1. Scenario 5: the parental populations were LON and MAR and their admixture at a rate r3 gave AMP. Scenario 6: the parental populations were LON and AMP and their admixture at a rate r2 gave MAR. Scenario 7: the parental populations were MAR and AMP and their admixture at a rate r1 gave LON