Table 1 Simulation results
True Embedded ImagenMethodMean (SD)2.5%97.5%Mean C.I. widthCoverage
Two samples (sample at t = 0, 8)
1000100Embedded Image1,059.7 (253.5)699.81,657.8
MLNE1,080.7 (260.7)711.31,695.41,283.3960
Embedded Image1,033.2 (247.3)684.11,604.81,195.5956
5000500Embedded Image5,272.4 (1,164.5)3,534.18,056.8
MLNE5,276.7 (1,166.7)3,539.98,083.96,046.3970
Embedded Image5,217.1 (1,149.6)3,501.67,958.15,957.4967
Three samples (sample at t = 0, 4, 8)
1000100Embedded Image1,107.8 (638.8)661.82,050.7
MLNE1,076.6 (243.9)734.91,704.61,134.2957
Embedded Image1,030.9 (226.8)709.41,605.41,054.0960
5000500Embedded Image5,567.7 (2,038.2)3,165.910708.0
MLNE5,254.0 (1,153.4)3,530.28,198.15,427.4950
Embedded Image5,202.0 (1,138.5)3,495.98,008.45,352.2953
  • For each parameter setting, 1000 replicate populations were simulated and all three methods are used to estimate Embedded Image. The true Embedded Image, sample size per generation, and number of temporal samples are shown. A total of 500 unlinked loci are used in each run and the initial allele frequencies are sampled from the uniform distribution. The mean, standard deviation, 2.5 and 97.5 percentiles of the 1000 runs are reported. For MLNE and Embedded Image, the mean width of the 95% confidence interval (C.I.) is also computed. The last column shows the number of C.I.’s (of 1000 simulations) that cover the true value Embedded Image.