1. Objective
Eurasian populations living during the Holocene (11,700 ybp. to present) have undergone considerable microevolutionary change. The rate of adaptive evolution during this period was 100 times greater than during the Pleistocene (Hawkes et al., 2007). Directional selection favoring cognitive variants has been predicted on the basis of culture-gene co-evolution stemming from the transition into agrarianism and urbanization (Cochran & Harpending, 2007) and also large-scale population replacements (Rindermann et al., 2013). To examine this question directly we compare a sample of ancient Eurasian genomes (ranging from 4.56 to 1.21 kyr BP) with a sample of modern European ones, using three different cognitive polygenic scores (PGS).
2. Methods
Sequence data from 99 ancient European genomes (from West Asia and Europe) were obtained from the European Nucleotide Archive. Sequence data on 503 modern European genomes were obtained from 1000 Genomes EUR Phase 3. The samples exhibited low levels of genetic differentiation relative to one another (Fst=0.016). Variant calling was used to construct three PGS from lists of nucleotides that were found to be genome wide significant ‘hits’ for educational attainment in recent large GWAS studies (e.g. Okbay et al., 2016). After excluding variants that were absent from the ancient genomes (due to modest coverage) and non-independent hits, the resultant PGS consisted of 130, 11 and 9 ‘hits’ respectively. The 9 hit PGS, constructed utilizing a technique developed by Piffer (2015), was comprised of nine GWAS hits from Okbay et al. (2016) that were in close linkage (linkage-window r>0.8) with ‘hits’ predicting educational attainment across two other large GWAS studies. The use of PGS comprised of small numbers of high-significance hits was intended to boost the signal of selection in the group comparison. Fisher’s Exact Test was used to compare the population means of these polygenic scores via the computation of an Odds Ratio statistic, which weights by genome coverage.
3. Results and Discussion
Significant differences were found between the ancient and modern European populations, favoring the latter, when all three PGS were used (ORs=.92 (130 SNP), .81 (11 SNP) and .81 (9 SNP), all p<.05). A correlation of .22 was also found between the 130 SNP PGS and genome year among a subsample of 66 ancient genomes (p(one tailed)=.04). This trend is presented in Figure 1.
[Figure 1 here]
Fig. 1. 130 SNP POLYCOG positive allele counts for 66 ancient genomes vs. year (scaled in terms of the BCE/CE calendar era). r=.22, p(one-tailed)=.04, N=66. Grey area around trend line = 95% CI.
Modern European genomes have higher frequencies of cognitive variants, relative to ancient ancestral ones, also PGS frequency increased in the ancient sample over 3.25 kyr – consistent with the prediction that the Holocene involved pronounced adaptive evolutionary change favoring higher cognitive ability.
Refs.
Cochran, G., & Harpending, H. (2009). The 10,000 year explosion. Basic Books.
Okbay, A., et al. (2016). Nature, 533, 539.
Piffer, D. (2015). Intelligence, 53, 43.
Rindermann, H., et al. (2012). Intelligence, 40, 362.
Hawks, J., et al. (2007). Proc. Natl. Acad. Sci. USA, 104, 20753.
