. 2008 Dec;180(4):2151–2161. doi: 10.1534/genetics.108.092452

TABLE 1.

Proportionality coefficients (ρ) between among- (D) and within- (G) population G matrices vs. their neutral expectation (Γ = F/(1 − F)) from genetic markers

		Value of F_st/(1 − F_st)	Value of ρ_st
Scale of comparison	d.f.	Value of F_st/(1 − F_st)		Early traits	Late traits
One-level test
Among populations uncorrected for habitat	D_P: d.f. = 15	Γ_P = 3.14	ρ_P = 0.19^a	ρ_P = 0.24^a
Among populations uncorrected for habitat		(2.73–3.66)	(0.15–0.28)	(0.18–0.35)
Two-level hierarchical test
Between habitats	D_H: d.f. = 1	Γ_H = −0.10	ρ_H = 0.18	ρ_H = 1.28^a
		(−0.22–0.03)	Randomization P = 0.48	Randomization P = 0.004
Among populations within habitats	D_P/H: d.f. = 14	Γ_P/H = 3.19	ρ_P/H = 0.20^a	ρ_P/H = 0.18^a
Among populations within habitats		(2.80–3.71)	(0.15–0.29)	(0.14–0.26)

Measures of phenotypic (ρ) and genetic (Γ) divergence are given for both early and late traits separately and for two alternative MANOVA models (see materials and methods for details). In the one-level test, G is compared either to D among populations without correcting for habitats (D_P, ρ_P). In the two-level test, G is compared both to D among populations nested within habitats (D_P/H, ρ_P/H) and to D between habitats and (D_H, ρ_H). When available, the C.I. is given in parentheses. The degrees of freedom (d.f.) of the corresponding matrix estimates are given, and the corresponding within-population G matrix was estimated with d.f. = 244.

A significant difference, i.e., nonoverlapping C.I., or a significant randomization test for habitat effects.