Model | |||
---|---|---|---|
Statistic | Panmixia | 4Nm = 1 | 4Nm = 0.7 |
D. melanogaster | |||
C_{hud} | 10^{−15} | 0.006 | 0.054 |
B' | 10^{−5} | 0.684 | 0.956^{a} |
D. simulans | |||
^{C}hud | 10^{−9} | 0.004 | —^{b} |
B' | 0.19^{e} | 0.996 | —^{b} |
We consider a symmetric two-island model; all individuals are sampled from one deme. 4Nm (3Nm for X-linked loci) is the population migration parameter and corresponds to the number of migrants per deme per generation. C_{hud} and B' are measures of linkage disequilibrium (see materials and methods). P(C_{hud}) and P(B') are one-tailed. Each cell lists the probability U that the distributions of P values for the statistic are uniformly distributed, as they should be under the null model. For C_{hud} and B', low U values correspond to excess linkage disequilibrium in the actual data relative to simulations.
↵^{a} For 4Nm = 0.7, there is significantly too little linkage disequilibrium in the actual data as measured by B'. (Note that this cannot be inferred from considering the value 1-U since B' is discrete.)
↵^{b} Simulations were not performed for D. simulans with this migration rate.
^{c} Even though this test does not reveal this feature, the distribution of B' is unusual when 4Nm = 1 (values of B' are lower in actual data than in simulated data; i.e., there is significantly less linkage disequilibrium in actual data than expected).