Mean parameter estimates (SD) | ||||
---|---|---|---|---|

Simulated β | Simulated N_{3}E(s) | N_{1} assumed | β | N_{3}E(s) |

a. Mild bottleneck followed by population expansion | ||||

0.2 | 10 | 20 | 0.20 (0.057) | 15.2 (14.6) |

100 | 0.20 (0.058) | 15.0 (14.8) | ||

1 | 20 | 1.16 (0.31) | 9.8 (3.13) | |

100 | 1.17 (0.31) | 9.7 (2.97) | ||

5 | 20 | 5.4 (1.14) | 9.8 (0.64) | |

100 | 5.6 (1.56) | 11.6 (2.60) | ||

b. Severe bottleneck followed by population expansion | ||||

0.2 | 10 | 20 | 0.25 (0.083) | 8.2 (3.12) |

100 | 0.27 (0.091) | 7.7 (3.05) | ||

1 | 20 | 1.08 (0.28) | 9.7 (2.71) | |

100 | 1.11 (0.31) | 9.7 (2.43) | ||

5 | 20 | 5.6 (2.40) | 9.6 (0.69) | |

100 | 6.4 (3.25) | 9.3 (0.67) | ||

c. Two phases of accelerating population expansion | ||||

0.2 | 10 | 20 | 0.22 (0.081) | 10.6 (7.61) |

100 | 0.24 (0.10) | 10.3 (8.52) | ||

1 | 20 | 1.1 (0.22) | 10.1 (1.84) | |

100 | 1.2 (0.31) | 9.4 (1.83) | ||

5 | 20 | 5.6 (2.00) | 9.7 (0.88) | |

100 | 7.0 (3.09) | 9.2 (0.99) |

Replicate data sets were generated by Monte Carlo simulation for three values of β and

*E*(*s*) = 0.1 under demographic models involving three epochs that violated the assumptions of the model simulated: (a)*N*_{1}= 100,*N*_{2}= 25,*N*_{3}= 100,*t*_{2}= 5,*t*_{3}= 100; (b)*N*_{1}= 100,*N*_{2}= 10,*N*_{3}= 100,*t*_{2}= 10,*t*_{3}= 100; and (c)*N*_{1}= 25,*N*_{2}= 50,*N*_{3}= 100,*t*_{2}= 100,*t*_{3}= 100. The parameter*f*_{0}= 0.95. The data sets consisted of 40 alleles sampled for 10,000 neutral sites and 2000 selected sites. Each data set was analyzed separately assuming two different*N*_{1}'s (20 and 100), and estimates of*N*_{2},*t*, β,*E*(*s*), and*f*_{0}were obtained. There were 5 replicates for β = 1 and β = 5 and 10 replicates for β = 0.2.