. 2008 Oct 10;1(4):631–644. doi: 10.1111/j.1752-4571.2008.00048.x

Table 2.

Statistical permutation tests of population subdivision between STP populations and the continental population with smallest F_ST values between STP and continental population comparisons

	St21(60)	Pr22(62)	GH14M(6)	IC16M(3)	Sen27M(13)	CAR08(7)	Ben04M(41)
St20(57)	ns/^*/ns	^*/^/^**	^**/^/^**	^**/^/ns	^*/^/^**	^*/^/^**	^*/^/^**
St21	–	^*/^/^**	^**/^/^**	^***/ns/ns	^*/^/^**	^*/^/^***	^*/^/^**
Pr22	–	–	^*/^/^**	^*/^/^**	^*/^/^**	^*/^/^**	^*/^/^**

Three statistics are given χ², a haplotype frequency-based statistic, K_ST^*, a sequence-based statistic, and S_nn. Levels of significance are coded as follows: ns, not significant; ^*0.01 < P < 0.05; ^**0.001 < P < 0.01; ^***P < 0.001. For each population, comparison the results are given as χ²/K_ST^*/S_nn. Populations are coded as in Table 1 and Fig. 2A. Numbers in parenthesis are sample sizes.

One-thousand permutations were made per test. Here, we perform 18 pair-wise comparisons, with 18 tests there is a 60.28% probability of finding a significant result by chance. Calculation of a strict Bonferroni adjustment without correlation, something we view as overly conservative in our situation, would require an alpha of 0.05, a P-value of 0.0027.

	St21(60)	Pr22(62)	GH14M(6)	IC16M(3)	Sen27M(13)	CAR08(7)	Ben04M(41)
St20(57)	ns/^*/ns	^*/^/^**	^**/^/^**	^**/^/ns	^*/^/^**	^*/^/^**	^*/^/^**
St21	–	^*/^/^**	^**/^/^**	^***/ns/ns	^*/^/^**	^*/^/^***	^*/^/^**
Pr22	–	–	^*/^/^**	^*/^/^**	^*/^/^**	^*/^/^**	^*/^/^**