TABLE 1

Demographic modeling of the European population

Model parametersP
(F0 ≤ 13)
P
(F0 = 13)
P
(F0 ≥ 13)
P
( ≥ 0.045)
ModelTbNbTmN0/Ni0Average
Constant1.261<10-4<10-4-0.0770.0847
Bot 1100,0001,0003,6000.52.601<10-4<10-4-0.967<10-4
Bot 2100,0001,0007,5000.250.601<10-4<10-4-1.050<10-4
Bot 3100,0005001,7500.52.501<10-4<10-4-0.955<10-4
Bot 4100,0005004,1500.250.551<10-4<10-4-1.049<10-4
Bot 550,0001,0002,9000.59.140.93360.0512a0.1176-0.672<10-4
Bot 650,0001,0004,4000.253.131<10-4<10-4-1.028<10-4
Bot 750,0005001,5000.59.080.93140.0484a0.1167-0.712<10-4
Bot 850,0005002,2500.252.941<10-4<10-4-1.049<10-4
Bot 925,0001,0002,7500.522.400.01320.00700.9938-0.355<10-4
Bot 1025,0001,0003,8500.2512.510.63330.1153a0.4820-0.790<10-4
Bot 1125,0005001,3000.520.210.04400.02100.9770-0.3350.0013
Bot 1225,0005002,0000.2511.560.74070.1093a0.3696-0.850<10-4
  • The models are denoted as follows: Constant, constant population size without recombination; Bot 1-12, bottleneck models without recombination for 12 different sets of values of Tb, Nb, and N0/Ni. A severe bottleneck of size Nb was introduced Tb generations ago in a population of initial size Ni and maintained for Tm generations. After that time, the population was allowed to grow exponentially to the current population size N0. Ni = 106 was assumed. The value of the population mutation parameter was 0.0127, which is equal to the observed average value of θ for the African sample. For the constant-size simulations, the corresponding θ value of the European sample was used. The values of Tm were chosen such that the simulated and observed total numbers of segregating sites across all 105 fragments are in close agreement. F0 is the number of fragments with no variation; P (F0 ≤ 13), P (F0 = 13), and P (F0 ≥ 13) are the probabilities of obtaining at most, exactly, or at least 13 fragments with no polymorphism, respectively. Average is the value of Tajima’s D across all fragments averaged over all 10,000 simulation runs, and P ( ≥ 0.045) is the probability of observing a value of Tajima’s D across fragments equal to or larger than the value observed in the European sample.

  • a Likelihood-ratio test, two-tailed, P < 0.05 (i.e., the respective bottleneck model fits the observation of F0 = 13 better than the Constant).