D = c {1-[Sum (Xi Yi)]^a}^b,
where Xi and Yi are the frequencies of alleles of size i in populations X and Y respectively, and a, b, and c are constants. For Das, a, b, and c are all equal to one. For Da, a=0.5, while b and c are equal to one, and for Dc, a and b equal 0.5, and c equ
als 2 Sqrt(2)/Pi. For multiple loci, the average distance is taken over all loci. Although the expectations of these distances are clearly different, it has not been shown that there is substantial difference between the accuracy of Das and Da in reconstr
ucting phylogenetic trees. Dc is slightly less efficient than Da in phylogenetic reconstruction under the SSM model (Takezaki and Nei 1996). In the absence of range constraints these distances vary between some non-zero
positive number and c as time progresses. Latter's Fst distance and Nei's minimum genetic distance both separately incorporate the sum of the squared allele frequencies in each population in order to create distances which vary from zero to some positive
number, but these distances are generally less accurate than Da and/or Dc (Takezaki and Nei 1996).While the between locus variance of these measures is large, making it essential to bootstrap over loci to assess reliability, the degree of allele sharing may be affected by sampling, especially when the sample size is small. In such cases bootstrapping over individuals may provide useful information, although never as a substitute for bootstrapping over loci.