With perfect linkage disequilibrium ($D' = 1, R^2 = 1$), you might have the following table of counts for the alleles:
B b A 100 0 a 0 100
With "partial" linkage disequilibrium ($D' < 1, R^2 < 1$), you'll see something like this:
B b A 100 25 a 25 100
But I've sometimes seen the following "imbalanced" linkage disequilibrium ($D' = 1, R^2 < 1$):
B b A 100 0 a 50 100
What gives rise to this sort of situation, where aB occurs, but Ab never does? Is there a name for this phenomenon? Which is a better measure of linkage disequilibrium in this case, the normalized LD coefficient $D'$, or the squared-correlation $R^2$?