Phenotypic plasticity can facilitate adaptive evolution in gene regulatory circuits

Table 1 Mutations and perturbations in the initial condition s0 produce the same phenotypes more often than expected by chance

N	c	d	Mean C± S.E.	p-value	Sample size^a
8	0.4	0.125	0.528 ± 0.006	< 2.2 × 10^-16	5485
		0.25	0.496 ± 0.006	< 2.2 × 10^-16	5905
		0.5	0.557 ± 0.005	< 2.2 × 10^-16	7269
	0.3	0.125	0.442 ± 0.006	< 2.2 × 10^-16	5885
		0.25	0.398 ± 0.006	< 2.2 × 10^-16	6050
		0.5	0.349 ± 0.005	< 2.2 × 10^-16	7144
20	0.3	0.1	0.712 ± 0.006	< 2.2 × 10^-16	5247
		0.25	0.8 ± 0.005	< 2.2 × 10^-16	6432
		0.5	0.885 ± 0.003	< 2.2 × 10^-16	8055
	0.2	0.1	0.63 ± 0.006	< 2.2 × 10^-16	5318
		0.25	0.645 ± 0.006	< 2.2 × 10^-16	5699
		0.5	0.711 ± 0.005	< 2.2 × 10^-16	7102
	0.1	0.1	0.47 ± 0.006	< 2.2 × 10^-16	6036
		0.25	0.408 ± 0.006	< 2.2 × 10^-16	6058
		0.5	0.364 ± 0.005	< 2.2 × 10^-16	6776

C >C^rand, according to a Wilcoxon signed-rank test. In this and all other tables, N refers to the number of genes in a circuit, c refers to the circuit's interaction density, and d refers to the fraction of gene activity differences between the unperturbed initial condition s₀ and the native phenotype $s ∞ n a t i v e$ .
^aEven though we sampled 10⁴ genotypes for each genotype network, we discarded genotypes in which either P_μ or $P s 0$ was empty.