Table 2. Trait-associated variants with Bonferroni-corrected significant evidence of being under polygenic adaptation in the Lazaridis et al. (2014) dataset, using the $Q_X$ statistic: $P<0.05/n,$ where n is the number of GWAS tested, assuming a ${\mathit{\chi }}^2$ distribution.

Graph	Trait	P	$P_{rand}$	$P_{emp}$
Seven-leaf graph (w/EuropeA)	Height**	$1.506\text{*}{10}^{-8}$	$<0.001$	$<0.001$
	Photic sneeze reflex**	0.00087	$<0.001$	0.006
	Educational attainment**	$6.11\text{*}{10}^{-6}$	0.001	0.001
	Unibrow	$4.763\text{*}{10}^{-7}$	0.173	$<0.001$
Seven-leaf graph (w/EuropeB)	Height**	$2.448\ast {10}^{-8}$	$<0.001$	$<0.001$
	Educational attainment**	$6.838\ast {10}^{-6}$	$<0.001$	0.001
	Age at voice drop**	0.000953	$<0.001$	0.001
	Photic sneeze reflex**	0.00086	$<0.001$	0.001
	Unibrow	$2.784\text{*}{10}^{-6}$	0.166	$<0.001$
Seven-leaf graph (w/EuropeC)	Educational attainment**	$4.032\ast {10}^{-6}$	$<0.001$	$<0.001$
	Photic sneeze reflex	0.000653	0.004	0.004
	Unibrow	$2.957\ast {10}^{-6}$	0.199	$<0.001$

We tested three different graphs with different sets of European panels, containing either low (EuropeA), medium (EuropeB), or high (EuropeC) EEF ancestry. We also computed P-values from 1000 samples in which we randomly switched the sign of effect size estimates, to simulate neutrality while preserving the genetic architecture of the traits ($P_{rand}$). Additionally, we computed P-values from an empirical null distribution produced using 1000 samples, each containing SNPs that were frequency-matched to the trait-associated SNPs, using their allele frequency in Europeans, to account for each GWAS’s ascertainment scheme ($P_{emp}$).

^**Trait-associated variants for which $P_{rand}<0.05/n.$

^*Trait-associated variants for which $P_{rand}<0.05$.