Table S1.

Sample size, mixed-effects model parameters, and estimated effects for data presented in Fig. 1C, ADH activity at ZH-86Fb

Sample size					Random effects				Fixed effects			Planned comparisons
Day reps	Construct	N (lines)	Vials per line	Total samples		SD		Unequal variance multiplier		Estimate	SE		t value	df	P
10	Single	3	3	72	Day	0.083	Dup	3.03	(Intercept)	0.436	0.027	(Dup – Single) vs. Single	8.17	5	0.00045
	Dup	4	3	90	Line	0.0087	Ident_Dup	2.61	Dup	0.676	0.016	Dup vs. Ident_Dup	0.49	5	0.64
	Ident_Dup	3	3	75	Residual	0.124	Single	1	Ident_dup	0.687	0.015

Model: lme(relative_ADH_activity ∼ construct, random = list(one1 = pdIdent(∼day-1), one2 = pdIdent(∼line-1)), weights = varIdent(form=∼1|construct)). Units for response variables are as follows: ADH activity, ΔAbs₃₄₀ per minute per milligram of protein; Adh mRNA, level relative to control gene RP49, i.e., 2^−(Adh Cq – RP49 Cq); β-galactosidase activity, ΔAbs₅₇₄ per hour * 1,000. “Estimate” denotes the estimate of the response variable using standard R style, such that (Intercept) refers to the mean of the first factor level (always set to Single) and the subsequent values (e.g., Dup) are the mean of the next factor level minus the intercept. For models with no intercept, means are shown. These parameter estimates were used to compute the planned t tests. Model formulas are given in nlme style. Experiments with a vial random effect (vial nested within line and day) measured multiple fly homogenates per culture vial; experiments without this term measured a single homogenate per vial. “Total samples” is often less than the product of the grouping factors due to (i) in some experiments, not all lines were available to measure on each day replicate, or (ii) occasional vials not producing flies.