TABLE 1

Statistical tests comparing the configurations of mutations

AbbreviationStatistical testFrequency classesReference
fd2MCH2 × 2 Monte Carlo test of homogeneityr = 1,1 < r < mSawyer et al. (1987)
fdMCH2 × (m – 1) Monte Carlo test of homogeneityr = 1, r = 2,. . ., r = m – 1
fdMWUMann-Whitney U-testr= 1, r= 2,...,r= m–1Akashi and Schaeffer (1997)
TDTajima's D-kestNot applicableTajima (1989)
pdMCH2 × 2 Monte Carlo test of homogeneity1 ≦ r < m, r = mMcDonald and Kreitman (1991)
sidMCH2 × 3 Monte Carlo test of homogeneityr = 1,1 < r < m, r = mTempleton (1996)
fddMCH2 × m Monte Carlo test of homogeneityr= 1, r= 2,...,r= m
fddMWUMann-Whitney U-testr= 1, r= 2,...,r= mAkashi (1997a)

Frequency classes assume that ancestral and derived states are inferred at variable nucleotide sites (unfolded distributions). The original tests suggested by Sawyer et al. (1987), McDonald and Kreitman (1991), and Templeton (1996) were performed on folded distributions. Monte Carlo tests of homogeneity were employed in lieu of G-tests, Fisher's exact tests, and χ2 tests employed in the references.