Table 4. Power analysis inferring minimum sample sizes (N_min) needed to detect effect sizes given increasing levels of S-linked inbreeding depression (δ_S-linked) for relative growth rate in the three parental backgrounds (PB1–3).

	Detectable δS-linkeda	Effect sizeb	Nmin (actual power)c
Selfed progeny PB1 (S1S39)	0.05	0.450	51 (0.83)
MSerror=0.016	0.10	0.899	18 (0.88)
Neffectived=9	0.15	1.349	12 (0.95)
	0.20	1.799	9 (0.96)
	0.25	2.249	9 (1.00)
	0.30	2.698	6 (0.90)
	0.35	3.148	6 (0.96)
Selfed progeny PB2 (S3S23)	0.05	0.327	96 (0.81)
MSerror=0.019	0.10	0.654	27 (0.82)
Neffectived=21	0.15	0.981	15 (0.86)
	0.20	1.308	12 (0.93)
	0.25	1.635	9 (0.92)
	0.30	1.962	9 (0.98)
	0.35	2.289	6 (0.80)
Selfed progeny PB3 (S1S23)	0.05	0.268	138 (0.80)
MSerror=0.040	0.10	0.536	39 (0.83)
Neffectived=24	0.15	0.804	21 (0.87)
	0.20	1.071	12 (0.80)
	0.25	1.339	12 (0.94)
	0.30	1.607	9 (0.91)
	0.35	1.875	9 (0.97)

Abbreviations: MS, mean square; PB, parental background.

For each PB, the level of variance (MS_error) underlying the effect size calculation and the effective sample size (N_effective) was obtained from the analysis of variance presented in Table 3.

^aDetectable δ_S-linked: δ_Sx-linked=1−(MeanRGR_SxSx)/(MeanRGR_SxSy).

^bEffect size (f) standardized by MS_error: .

^cMinimum sample size (N_min) needed (rounded up to the nearest product of three) to achieve at least 80% power to detect the given effect size.

^dThe effective sample size was determined conservatively based on the sample size in the smallest genotype class: N_effective=N_{smallest genotype class} × number of genotype classes.