Skip to content

QUANTIFYING THE DIFFERENCES OF GENOTYPIC INTELLIGENCE AND GENOTYPIC EDUCATION BETWEEN (SUPER)POPULATIONS IN 1000 GENOMES: TWO SIMPLE EMPIRICAL FORMULAS BASED ON THE NEW RESULTS OF DAVID HILL

Polygenic scores on intelligence do not reflect differences on genotypic intelligence between populations and superpopulations (Piffer, 2017; Sniekers, 2017). Polygenic scores on educational attainment (Piffer, 2016; Okbay, 2016) work slightly better, but British and Utah Whites have the same scores as South Asians, for example.

Based on the last revised study of David Hill (5 june 2017), that found MAF 0.001-0.01 account for 45% of genotypic IQ (G) and for 33% of genotypic EDU (E), I created two scores of populations and superpopulations of 1000 GENOMES for G and E:

G_POP = (POLY_IQ_POP : POLY_IQ_GBR x (1 + gdGBR) x 55% + MAF0.000-0.005_GBR : MAF0.000-0.005_POP x 45%) x IQ_GBR

E_POP = (POLY_EDU_POP : POLY_EDU_GBR x (1+ gdGBR) x 67% + MAF0.000-0.005_GBR : MAF0.000-0.005_POP x 33%) x IQ_GBR

I used MAF 0.000-0.005 (1000 GENOMES, 2015 – Extended Data, Figure 3b) as a proxy for MAF 0.001-0.01.

gdGBR is the genetic distance on SNP from British: 0 for Europeans, 0.060 for South Asians, 0.112 for East Asians and 0.176 for Africans. I used these genetic distances because not all SNP of found by GWAS on Europeans are IQ-increasing and EDU-increasing variants in other populations. Also, the other populations have IQ-increasing and EDU-increasing SNP that are neutral for Europeans. Higher is the genetic distance of a population, higher is this number of SNP that are not captured by POLY-IQ and POLY-EDU of Europeans.

These scores work much better than POLY-IQ and POLY-EDU, excepting populations that have Native American admixture. The explanation is simple. MAF 0.00-0.005 (1000 GENOMES, 2015) arose during the severe bottleneck (23,000-16,000 years ago) and the fast grow (since 16,000 years ago) of Amerindian populations. In Native Americans, many SNP that arose during this period reached high frequencies, hence they have higher than 0.005 frequency even in 1000 GENOMES total samples, and are not captured by my two formulas.

I used POLY-EDU and POLY-IQ found by Piffer (2016, 2017) using the results of GWAS of Okbay (2016) and Sniekers (2017), 161 SNP and 15 SNP respectively.

Here are the results for G, E, and measured IQ of populations and superpopulations. The correlations between G, E and measured IQ are much higher than for POLY-IQ and POLY-EDU:

 

 

POP        G           E         IQ

 

CHB – 105.27 – 104.56 – 105

CHS  – 104.98 – 105.17 – 105

JPT  –  104.42 – 103.29 – 105

KHV – 100.74 – 102.05 –  99

CDX – 100.26 – 102.24

EAS  – 103.06 – 103.42 – 103.50

 

FIN – 101.37 – 102.62 – 101

GBR   –  100 – 100 – 100

CEU  – 97.94 – 97.29 – 99

IBS   –  95.44 – 98.61 – 97

TSI   –  94.54 – 97.57 – 99

EUR  – 97.72 – 99.13 – 99.2

 

GIH – 91.72 – 94.98

PJL  – 86.91 – 93.89 – 84

STU – 85.05 – 91.91 – 79

BEB  – 84.59 – 93.02 – 81

ITU  – 83.74 – 92.60

SAS – 86.34 – 93.25 – 81.33

 

ASW – 81.98 – 87.53 – 85

ACB  – 81.64 – 86.95 – 83

YRI   – 81.27 – 86.71 – 71

ESN   – 80.70 – 86.02 – 71

MSL  – 80.51 – 82.91 – 64

GWD – 80.37 – 86.23 – 62

LWK  – 75.70 – 83.25 – 74

AFR   – 80.08 – 85.52 – 68.5

 

REFERENCES

1000 GENOMES (2015) https://www.nature.com/nature/journal/v526/n7571/full/nature15393.html

Hill, D.W. et al. (2017) Genomic analysis of family data reveals additional genetic effects on intelligence and personality. bioRxiv http://dx.doi.org/10.1101/106203

Okbay, A. et al. (2016). Genome-wide association study identifies 74 loci associated with educational attainment. Nature 533: 539-542.

Piffer, D. (2016). Polygenic selection on educational attainment: a replication. https://figshare.com/article/Polygenic_selection_on_educational_attainment_a_replication/3381439

Piffer, D. (2017)  2017 Intelligence GWAS: Group-level polygenic scores http://rpubs.com/Daxide/279148

Sniekers, S. et al (2017) Genome-wide association meta-analysis of 78,308 individuals identifies new loci and genes influencing human intelligence. Nature Genetics doi:10.1038/ng.3869

NEW EVIDENCE FOR THE DECREASE OF THE GENOTYPIC INTELLIGENCE DURING HOLOCENE: A RECENT STUDY OF DAVID HILL

David Hill republished, in 5 june 2017, an extremly interesting revised study in bioRxiv:

In Table 3 page 21, David Hill presents the variance of intelligence and education explained by minor alleles with different frequencies, that demonstrates different genetic architectures for the two complex traits, despite many authors use educational attainment as a proxy for intelligence.

The most interesting fact is that MAF 0.001-0.01 account for 44% of genotypic IQ and only for 33% of genotypic EDU. The MAF 0.001-0.01 have arose between 25,000 and 5,500 years ago (Zwick, 2001). There are much more MAF 0.001-0.01 that decrease than increase IQ and EDU. The very high percentage of variance explained by both of them could indicate a relaxed selection on the two traits since Last Glacial Maximum, and a relatively weaker selection on genotypic IQ than on genotypic EDU. This decrease of the selection parallels the decrease of brain size last 25,000 years (Ruff, 1997).

Another very interesting fact is that MAF 0.1-0.2 account for significantly less variance of IQ than all other MAF. MAF 0.1-0.2 arose between 325,000 and 200,000 years ago (Zwick, 2001), and their uniformly distribution in humans suggests this period was the period with the strongest selection on intelligence in humans. This is the period of emergence of Homo Sapiens. Also, brain size increased at 1450 cmc, larger than today brains. The penultimate place in explaining differences of IQ is occupied by MAF 0.01-0.1 (that arose between 200,000 and 25,000 years ago (Zwick, 2001)) suggesting during this period the selection on intelligence also was strong, and purifying selection uniformised the distribution of these MAF in humans.

Also, these MAF 0.01-0.1 and MAF 0.4-0.5 (that arose between 550,000 and 475,000 years ago (Zwick, 2001)) account for significantly less variance of EDU than all other MAF, suggesting between 200,000 and 25,000 years ago and between 550,000 and 475,000 years ago the selection for EDU was stronger than during other periods. Upper Palaeolithic selection on EDU is in line with the results of Racimo (2017), that found selection in East Asians before Holocene. Also, around 500,000 years ago emerged archaic humans, and brain sizes expanded from 900 cmc to 1300 cmc. Between 200,000 and 25,000 years ago there was strong selection on both EDU and IQ. During this period the brain size of humans peaked at 1600 cmc.

The lower decrease of selection on EDU than on IQ during Holocene could be explained if we admit that there are two types of variants that favor high-EDU: the first are shared variants that favor high-IQ, and the second are variants that favor high (self-)domestication of humans. Many of variants that favor high domestication favor low-IQ too. During Holocene variants that favor high-IQ decreased, and variants that favor low-IQ but high domestication increased. By this way, genotypic IQ decreased more than genotypic EDU during Holocene.

Concerning MAF younger than MAF studied by the article of David Hill, an older paper estimated that 81.4% of the rare protein-altering SNVs found in Americans with European origins originated in the last 5,000 years, and 14.4% of these SNVs are deleterious. 91.2% of the deleterious alleles originated in the last 5,000 years (Fu et al., 2013).

Although, Spain (2015) found fewer rare SNP in exome of people with very high IQ. Also, the IQ of superpopulations in 1000 GENOMES parallel the proportion of individuals that carry uncommon minor SNP (MAF frequency lower than 5%) in genes (Rao, 2017).

Probably future studies of David Hill will produce even stronger proofs for the decrease of genotypic intelligence since Palaeolithic period.

REFERENCES

Fu, W., O’Connor, T.D., Jun, G., Kang, H.M., Abecasis, G., Leal, S.M.,… & Akey, J.M. (2013). Analysis of 6,515 exomes reveals the recent origin of most human protein-coding variants. Nature 493: 216-220.

Hill, D.W. et al. (2017) Genomic analysis of family data reveals additional genetic effects on intelligence and personality. bioRxiv http://dx.doi.org/10.1101/106203

Racimo, F. et al (2017) Detecting polygenic adaptation in admixture graphs. bioRxiv doi: http://dx.doi.org/10.1101/146043

Rao, A.R. & Nelson, S.F. (2017) Calculating the statistical significance of rare variants causal for Mendelian and complex disorders. bioRxiv doi: http://dx.doi.org/10.1101/103218

Ruff, C.B., Trinkaus, E. & Holliday, T.V. (1997). Body mass and encephalisation in
Pleistocene Homo. Nature 387: 173-176.

Spain, S.L., Pedroso, I., Kadeva, N., Miller, M.B., Plomin, R. & Simpson, M.A. (2015). A genome-wide analysis of putative functional and exonic variation associated with extremely high intelligence. Molecular Psychiatry 21: 1145-1151.

Zwick, M.E. et al (2001) Genetic Variation Analysis of Neuropsychiatric Traits. in Methods in Genomic Neuroscience Chin, H.R., Moldin, S.O. (editors) CRC Press LLC

 

THE GENOTYPIC INTELLIGENCE OF EUROPEANS DECREASED SINCE BRONZE AGE (III)

Between the 70 de novo mutations (DNM) 1.4% are in exome, 1% are nonsynonymous and 0.4% are synonymous. For 70 de novo SNP (and 5 de novo indels), an individual will have 75×1.4%=1.05 total de novo mutations in exome, 75×1%=0,75 de novo nonsynonymous mutations, and 0,30 de novo synonymous mutations.
The percentages of nonsynonymous, synonymous and total SNP in exome between total SNP in genome are around: 0.28%, 0,32% and 0,6% respectively (very slightly higher in non-Africans than in Africans), significantly lower than percentages of de novo mutations (1000 GENOMES, 2015). At an accumulation rate of 70 SNP per generation, the oldest SNP of human genome have at least 1.5 millions years.
It means 74% of de novo nonsynonymous mutations, 20% of de novo synonymous mutations and 47% of all de novo mutations in exome were eliminated each generation by natural selection. But it means that at least 47% of de novo mutations in exome (in one DNA base by two!) are deleterious enough to be eliminated by selection. At this strength of selection pressure on exome, only 53% of individuals in each generation could reproduce, and this individuals must have 4 surviving offspring to maintain the actual number of population. Clark & Hamilton (2006) found between 2.2 and 3.2 surviving children for all economic and social classes in pre-Industrial England, and this is far to strength of selection that operated during last 1.5 millions years.
Also, al least 8.2% of the human genome sequence is functionally significant and selectively constrained (Rands, 2014). Hence, each individual will have, on average, 8.2% x 75 =6.1 deleterious DNM.
According with The Omnigenic Theory of Boyle, Li & Pritchard (2017), it means that 35% of people could have one de novo mutation detrimental for intelligence in the exome, because 15,000 genes (75% of all genes) are expressed in the brain. Although, Spain (2015) found fewer rare SNP in exome of people with very high IQ. Also, the IQ of superpopulations in 1000 GENOMES parallel the proportion of individuals that carry uncommon minor SNP (MAF frequency lower than 5%) in genes (Rao, 2017). Furthermore, according to Omnigenic Theory, all humans have 6 IQ-decreasing DMN in non-coding genome. This signify that polygenic score on IQ/EDU must increase to maintain the genotypic IQ/EDU at the actual level, and even POLY IQ/EDU can increase and genotypic IQ/EDU can decrease, despite selection for high-IQ.
During 100 generations, the accumulation of the IQ-decreasing DMN in non-coding genome will be 600 SNP. If the average effect of an IQ-decreasing DMN equates the average effect of an IQ-increased SNP found by GWAS, the POLY IQ must increase with 600 SNP after 100 generations to maintain the actual level of genotypic IQ. If POLY IQ/EDU contains 30,000 SNP, this signifies a 2% increase of POLY IQ/EDU.
Woodley & Piffer (2017) found a 2% increase of POLY EDU during 114 generations, since Bronze Age.
REFERENCES

 

Boyle, E.A. et al (2017) An Expanded View of Complex Traits: From Polygenic to Omnigenic. Cell 169(7): 1177-1186 DOI: http://dx.doi.org/10.1016/j.cell.2017.05.038

Clark, G. & Hamilton, G. (2006) Survival of the Richest: The Malthusian Mechanism in Pre-Industrial England. The Journal of Economic History 66(3): 1-30

Rands, C.M. et al (2014) 8.2% of the Human Genome Is Constrained: Variation in Rates of Turnover across Functional Element Classes in the Human Lineage. PLoS Genet 10(7): e1004525. doi:10.1371/journal.pgen.1004525

Rao, A.R. & Nelson, S.F. (2017) Calculating the statistical significance of rare variants causal for Mendelian and complex disorders. bioRxiv doi: http://dx.doi.org/10.1101/103218

Spain, S.L., Pedroso, I., Kadeva, N., Miller, M.B., Plomin, R. & Simpson, M.A. (2015). A genome-wide analysis of putative functional and exonic variation associated with extremely high intelligence. Molecular Psychiatry 21: 1145-1151.

Woodley, M.A. et al. (2017) Holocene selection for variants associated with cognitive ability: Comparing ancient and modern genomes. bioRxiv http://dx.doi.org/10.1101/109678

 

THE EASTERN ASIAN CRO-MAGNON THAT LIVED DURING THE LAST GLACIAL MAXIMUM WAS THE MOST INTELLIGENT HUMAN

Davide Piffer published the polygenic scores resulted from 15 leading SNP found by the most recent GWAS on intelligence (Sniekers, 2017):  http://rpubs.com/Daxide/279148

These scores suggest no significant increase of the intelligence of Western Eurasians since Out of Africa. Today differences in intelligence between Africans and Europeans are due rather of a higher decrease of intelligence of Africans than Europeans since Out of Africa.

The POLY IQ of super-populations from 1000 GENOMES are: AFR-0.4714, SAS-0.4439, EUR-0.4627, EAS-0.5129, AMR-0.4515.

All of these 15 SNP increase the intelligence of Europeans. It is expected fewer of them increase the intelligence of SAS, even fewer of EAS and the fewest of AFR, because genetic distances and times since divergence between these populations.

We can suppose that in the moment of Out of Africa, the Africans and the future non-Africans have roughly the same polygenic scores on these 15 SNP. We can suppose the polygenic score increase during last 60,000 years in all populations. If non-Africans ancestors are also the ancestors of LWK (today LWK have the lowest polygenic score of today Africans: 0.4567), the increase of the polygenic score in EUR is only 0.01 higher than the increase of LWK during last 60,000 years. But if the ancestors of the non-Africans are not the ancestors of LWK, the frequency of these 15 variants increased more in Africans than in Europeans since the Out of Africa, because all the other today Africans (excepting LWK) have higher polygenic scores than the average of Europeans. Also, the frequency increased more in Africans than in South Asians. This results are in line with polygenic scores on EDU, obtained with the 74 SNP found by the GWAS of Okbay (2016): ASW have the same score with IBS and TSI, and higher score than CEU and GBR. Also, CEU have lower score than all Africans, excepting LWK. (The GWAS on EDU find not only IQ-increasing SNP, but also SNP that favor „domestication”, and these common variants spread due of civilization. For example, peoples that entered faster in Neolithic and complex civilization, like IBS and TSI, have 0.01-0.02 higher scores on EDU-SNP and 0.02 lower scores on IQ-SNP than peoples that had latter entrance in sedentary civilizations, like CEU, GBR, FIN.)

Althought, Racimo (2017) found selection on polygenic score on educational attainment only for EAS, and not for other populations or super-populations of 1000 GENOMES, and this selection in EAS produced before 10,000 years ago. If there was not positive selection on EDU, the genotypic EDU decreased due of the increase of rare variants detrimental to EDU, produced by de novo mutations. It is necessary a positive selection on a complex trait (and an increase of polygenic score) to maintain this trait at the actual level. Hence, the result of Racimo (2017) demonstrates that genotypic EDU decreased in all populations during Holocene, and decreased in all (super)populations excepting EAS between Out of Africa and Holocene. But it is probable there were some periods of selection on EDU in Europeans during last 60,000 years, because all GWAS did not find any common SNP that decreases barely 1 or to 2 IQ points the genotypic EDU, hence the IQ-decreasing SNP did not reach the frequency of common polymorphism in Europeans. The alternation of periods of positive selection and periods of negative selection could explain too the fact, noticed by Piffer (2016), that there are 45% ancestral alleles between IQ-increasing SNP found by the GWAS of Okbay (2016). During periods of selection against high intelligence, IQ-decreasing variants could reach the frequency of common polymorphism. Also, EUR have fewer IQ-increasing ancestral alleles than EAS, and this fact could be due of the spread of IQ-decreasing derived alleles during longer (or stronger) periods of negative selection on IQ in EUR than in EAS.

Although, the selection against high-IQ did not produce only after Industrial Revolution. De la Croix (2017) found Upper class (presumed also the most intelligent social class) had the lowest fertility in pre-industrial England.

It would be very strange a higher increase of the frequency of some neutral alleles in Africans, and a lower increase of the same beneficial alleles in Europeans during 2,000 generations. It is more probable these 15 SNP (and the other IQ-increasing SNP) were under a soft selection in Europeans, and partially in other populations, the most of last 60,000 years. It is more probable too these IQ-increasing common SNP were selected against in all the populations (mostly in Europeans, South Asians and Native Americans) during some periods that favored selection against high-intelligence: Neolithic transition, entrance in complex civilizations and Industrial Revolution. This selection against high intelligence decreased the polygenic scores more in Europeans, South Asians and Amerindians than in Africans, because many of these alleles are not related to intelligence in Africans and are not influenced by the selection against high-IQ. During periods of soft selection for high-IQ, like in European Metal Ages and Medieval Age (Woodley & Piffer, 2017), the polygenic score could increase even if genotypic intelligence remains constant or even decreases, because the detrimental effect of de novo mutations is partially compensated by the increase of polygenic score, and only partially compensated by elimination of deleterious de novo mutations and other rare alleles. The frequency of these common SNP could increase even if the selection on intelligence is zero, due of pleiotropy of some of them, and of selection on other complex traits.

Even if not all these 15 SNP increase the IQ of East Asians, their polygenic score is 0.05 higher than in Europeans. This is in line with all polygenic scores on educational attainment, resulting after the counts of SNP found by the GWAS of Rietveld (2013), Davies (2016) and Okbay (2016). All of these counts found 0.05 higher POLY EDU of East Asians than of Europeans and other Eurasians. Probable many (or even mostly) of the IQ-increasing mutations in linkage with SNP found by different GWAS are older than the divergence of Europeans and East Asians, and these mutations increase the intelligence of both populations. The 0.05 higher polygenic score of all East Asians (in 1000 GENOMES and in ALFRED too) than Europeans demonstrates a strong selection on intelligence of East Asians. It is near zero probability that the same very strong selection pressure operated on Eastern Siberian hunter-gatherers and on Vietnamese and Han farmers during Holocene. It is more probable this high selection on intelligence operated before Holocene, before the separation of different populations of East Asians, but after the split of Native Americans, that have 0.06 lower polygenic score than East Asians, and that diverged from East Asians 23,000 years ago (Raghavan, 2015). Also, Northern East Asians have higher POLY IQ than Southern East Asians, due of the dilution of Northern Han Chinese during Southward migration. But Japanese have the highest POLY IQ of EAS, even higher than Northern Han Chinese, and it means the increase of the strong selection on IQ of East Asians produced before the divergence of Jomons, estimated 22,000-23,000 years ago (Kanazawa-Kiriyama, 2017). The most probable this strong selection on intelligence of East Asians produced during Last Glacial Maximum.

After this high increase of the genotypic intelligence of East Asians, the selection pressure relaxed and probably their intelligence decreased during Holocene, due of warmer climate and of civilizations, mostly by accumulation of IQ-decreasing rare variants.

The Eastern Asian Cro-Magnon living during Last Glacial Maximum had the highest genotypic intelligence of all humans ever living on the Earth.

Finally, I would like to draw attention to some proof, which I find indisputable,
of the superiority of Palaeolithic man to modern man in terms of visual-spatial
intelligence and memory. A study that analyzed how accurately quadrupedal
walking was rendered in 1000 works of art from the Palaeolithic and modern times
found an error rate of 46.2% in Palaeolithic artists, of 83.5% in artists before 1887,
and of 57.9% in artists after 1887, the year when Eadweard Muybridge published
a series of 20,000 photographs investigating the stages of animal locomotion
(Horvath et al., 2012). Furthermore, Cro-Magnons had an error rate even lower
than the rate of those illustrating animal anatomy books, of 63.6%, and they were
very close to the error rate of taxidermists of natural history museums, of 41.1-
43.1% (Horvath et al., 2009).

The loss of (visuo-spatial) intelligence of modern man compared to Paleolithic man should not surprise us. Today, untrained and even trained adult man has a working memory for numbers lower than that of a young trained chimpanzee (Inoue, 2007; Matsuzawa, 2009; Cook, 2010). Also, chimpanzee choice rates in competitive games match equilibrium game theory predictions. Chimpanzee’s choices are close to the equilibrium predictions and are more responsive than human choices to past history and to payoff changes (Martin, 2014). Also, baboons but not modern humans break cognitive set in a visuo-motor task, indicating greater mental flexibility (Pope, 2015). Furthermore, Koko the gorilla, at 43 to 65 month, scored between 85.2 and 91.7 IQ points on the Stanford-Binet intelligence test (Patterson, 1993), surpassing many living humans who already benefited by Lynn-Flynn effect at that time, in 1975-1976.

 

REFERENCES

Cook, Peter & Wilson, Margaret (2010) Do young chimpanzees have extraordinary working memory?Psychonomic Bulletin & Review 17(4): 599-600

Davies, G., Marioni, R.E., Liewald, D.C., Hill, W.D., Hagenaars, S.P.,… & Deary, I.J. (2016). Genome-wide association study of cognitive functions and educational attainment
in UK Biobank (N=112,151). Molecular Psychiatry 1-10.

De la Croix, D. et al (2017) “Decessit sine prole” – Childlessness, Celibacy, and Survival of the Richest in Pre-Industrial England. CEPR Discussion Paper No. DP11752. Available at SSRN: https://ssrn.com/abstract=2896042

Horvath, G., Csapo, A., Nyeste, A., Gerics, B., Csorba, G. et al. (2009). Erroneous quadruped walking depictions in natural history museums. Current Biology 19: 61-62.

Horvath, G., Farkas, E., Boncz, I., Blaho, M. & Kriska, G. (2012). Cavemen were better at depicting quadruped walking than modern artists: Erroneous walking illustrations in the fine arts from prehistory to today. PLoS ONE 7(12): e49786

Inoue, Sana & Matsuzawa, Tetsuro (2007) Working memory of numerals in chimpanzees. Current Biology 17(23): 1004-1005

Kanazawa-Kyryiama, H. (2017) A partial nuclear genome of the Jomons who lived
3000 years ago in Fukushima, Japan. Journal of Human Genetics 62: 213–221

Martin, Cristopher F. et al (2014) Chimpanzee choice rates in competitive games match equilibrium game theory predictions. Scientific Reports 4: 5182 doi: 10.1038/srep05182

Matsuzawa, Tetsuro (2009) Symbolic representation of number in chimpanzees. Current Opinion in Neurobiology 19: 92–98

Okbay, A., Beauchamp, J.P., Fontana, M.A., Lee, J.J., Pers, T.H., Rietveld, C., & Benjamin, D.J. (2016). Genome-wide association study identifies 74 loci associated with educational attainment. Nature 533: 539-542.

Patterson, Francine & Gordon, Wendy (1993) The Case for the Personhood of Gorillas.In Paola Cavalieri& Peter Singer (eds.) The Great Ape Project. New York: St. Martin’s Griffin, pp. 58-77

Piffer, D. (2017)  2017 Intelligence GWAS: Group-level polygenic scores http://rpubs.com/Daxide/279148

Pope, Sarah M. et al (2015) Baboons (Papio papio), but not humans, break cognitive set in a visuomotor task. Animal Cognition 18:1339–1346

Raghavan, M. (2015) Genomic evidence for the Pleistocene and recent population history of Native Americans. Science Vol. 349, Issue 6250, aab3884 DOI: 10.1126/science.aab3884

Racimo, F. et al (2017) Detecting polygenic adaptation in admixture graphs. bioRxiv doi: http://dx.doi.org/10.1101/146043

Rietveld, C.A., Medland, S.E., Derringer, J., Yang, J., Esko, T., & Koellinger, P.D. (2013). GWAS of 126,559 individuals identifies genetic variants associated with educational
attainment. Science 340: 1467-1471.

Sniekers, S. et al (2017) Genome-wide association meta-analysis of 78,308 individuals identifies new loci and genes influencing human intelligence. Nature Genetics doi:10.1038/ng.3869

Woodley, M.A. et al. (2017) Holocene selection for variants associated with cognitive ability: Comparing ancient and modern genomes. bioRxiv http://dx.doi.org/10.1101/109678

THE GENOTYPIC INTELLIGENCE OF EUROPEANS DECREASED SINCE BRONZE AGE (II)

Woodley & Piffer (2017) found an increase of nearly 0.02 (from 0.4521 to 0.4717) of polygenic score on EDU since 3,500 years ago.
Counting the SNP discovered by the GWAS on IQ of Sniekers (2017), Davide Piffer (2017) found (http://rpubs.com/Daxide/279148) the lowest polygenic score between Europeans for Iberians (0.4535) and Tuscans (0.4579). Britsh score (0.4654) nearly 0.01 higher, and Utah-Whites (0.4747) score nearly 0.02 (exactly the increase of POLY EDU since Bronze Age, from 0.45 to 0.47) higher than the average of Iberians and Tuscans. But the polygenic count based on GWAS on EDU of Okbay (2016) found 0.01 higher scores for Iberians (0.513) and Tuscans (0.513) than for Utah-Whites (0.503) and British (0.506). Also, the count based on SNP discovered by the GWAS on EDU of Davies (2016) found 0.02 and 0.01 higher polygenic scores for Iberians (512) and Tuscans (0.501), respectively, than for Utah-Whites (0.493). It means Iberians and Tuscans could have 0.03 at 0.04 more common SNP related to EDU, but not to IQ, than Utah-Whites.
These 0.03-0.04 SNP are probably variants that favored the (auto)domestication of humans. The variants that favor domestication could have higher frequencies in Iberians and Tuscans because the higher Neolithic ancestry of these peoples than more Northern peoples, and because their faster entrance in a complex civilization.
Also, it is possible the common variants related to domestication increased even more than 0.02 since Bronze Age, but common SNP related on IQ could decrease during last 3,500 years.

EVIDENCE FOR 13,000 YEARS OF RELAXED SELECTION (ON INTELLIGENCE) IN WEST EURASIANS

Harris & Pritchard (2016) found an increased 5′-TCC-3′ to 5′-TTC-3′ mutation rate in Europeans from about 15,000 to 2,000 years ago.
Mathieson & Reich (2017) confirmed this increase for all West Eurasians. The highest increase for this type of mutation was found in a 7,000 years old European farmer, and it was found also in a 8,000 years old European hunter-gatherer, but this effect is not strongly driven by ancestry (farmer/hunter-gatherer) or by latitude, but is predicted by longitude (increasing east to west). This is in line with the increase of the mutation rate in sexual populations during range expansion found by Cobben & Kubisch (2014) and with the relaxed selection on the wave front of expansion found by Peischl (2016) in French Canadians, 6-9 generations ago. Also, Clark & Hamilton (2007) found higher fertility of Lower Class than Upper Class in French Canadians.
We know Near Easterner hunter-gatherers migrated in Europe 13,000 years ago, followed by Anatolian Neolithic farmers 9,000 years ago, and Steppe pastoralists 5,000 years ago. The relaxed selection on the front waves of these migrations could lead to a relaxed selection pressure that could favor not only the accumulation of mutations, but the increase of the mutation rate too. This is in line with the gradient east to west of the increase of the mutation rate.
 
Also, Mallick (2016) demonstrated an accumulation of mutations 5% higher in non-Africans than in Africans since the Out of Africa. The highest accumulation in Eurasians was found in West Eurasians. Most of this accumulation could produce after Last Glacial Maximum. The warming of the climate and a more sedentary life-style in Near East could relax the selection in this region. This is evidence of more permanent settlements and even of farming in Near East since 13,000 (Wilcox, 2012), 19,000 (Ramsey, 2016) and even 23,000 years ago (Snir, 2015). Also, the expansion of Near Easterners in Eurasia was not „Malthusian”, but rather resembled with the expansion of French Canadians. Furthermore, the decrease of human body size since Upper Palaeolithic is in line with the increase of the mutation rate, that is negatively correlated with the body size in mammals (Martin & Palumbi, 1993). Also, the mutation rate in mammals is higher in warmer climates (Gillman, 2009).
 
The evidence of a relaxed „general” selection is not a direct proof for a decrease of selection on intelligence. But a strong selection on intelligence could not lead to an increase of the mutation rate during 13,000 years (15,000 – 2,000 years ago), because the mutational target on intelligence is enormous. Probably genotypic intelligence decreased during this period in West Eurasia, and this is in line with Cold Winters Theory of Richard Lynn.
 
REFERENCES
Clark, G. & Hamilton, G. (2007) ECONOMIC STATUS AND REPRODUCTIVE SUCCESS IN NEW FRANCE. https://econ.ucalgary.ca/sites/econ.ucalgary.ca/files/seminars/clark%20and%20hamilton%20new%20france%202.pdf
Cobben, M.M.P. & Kubisch, A. (2014) The evolution of mutation rate in sexual populations during range expansion. bioRxiv doi: http://dx.doi.org/10.1101/008979
Gillman, L.N., et al (2009). Latitude, elevation and the tempo of molecular evolution in mammals. Proceedings of the Royal Society B 276: 3353-3359
Harris, K. & Pritchard, J.K. (2016) Rapid evolution of the human mutation spectrum. bioRxiv doi: http://dx.doi.org/10.1101/084343
Mallick, S. et al (2016) The Simons Genome Diversity Project: 300 genomes from 142 diverse populations. Nature 538(7624): 201-206. doi: 10.1038/nature18964
Martin, A.P. & Palumbi, S.R. (1993) Body size, metabolic rate, generation time and the molecular clock. PNAS 90(9): 4087–4091
Mathieson, I. & Reich, D. (2017) Differences in the rare variant spectrum among human populations. PLoS Genet 13(2): e1006581. doi:10.1371/journal.pgen.1006581
Peischl, S. et al (2016) Relaxed selection during a recent human expansion. bioRxiv doi: http://dx.doi.org/10.1101/064691
Ramsey, M.N. et al (2016) Risk, Reliability and Resilience: Phytolith Evidence for Alternative ‘Neolithization’ Pathways at Kharaneh IV in the Azraq Basin, Jordan. PLoS ONE 11(10): e0164081. doi:10.1371/journal.pone.0164081
Snir, A. et al (2015) The Origin of Cultivation and Proto-Weeds, Long Before Neolithic Farming. PLoS ONE 10(7): e0131422. doi:10.1371/journal.pone.0131422
Wilcox, G. (2012) Pre-Domestic Cultivation during the Late Pleistocene and Early Holocene in the Northern Levant. in Biodiversity in Agriculture: Domestication, Evolution, and Sustainability, editors Gepts, P. et al. Cambridge University Press

THE GENOTYPIC INTELLIGENCE OF EUROPEANS DECREASED SINCE BRONZE AGE (I)

UPDATE 03.05.2017

Michael Woodley & Davide Piffer (2017) published in bioRxiv a very interesting study, comparing polygenic scores on educational attainment of today and Bronze Age samples. They demonstrated a positive selection on intelligence, but they believe they demonstrated the increase of the genotypic intelligence of Europeans since Bronze Age. Their results rather demonstrate a decrease of the genotypic intelligence during last 4,000 years.

The most powerful GWAS on educational attainment (Okbay, 2016) found 74 common SNP that favor high-IQ (equating the educational attainment with the intelligence). The average effect of each of these SNP is 0.02% of variance and 0.02 SD of educational attainment (Okbay, 2016, Extended Data Figure 2). But 1 SD of EA equates with 3.8 IQ points (Kong, 2017), hence the effect of each SNP is 0.02 x 3.8 = 0.076 IQ points. In aggregate, these 74 SNPs explain 0.43% of the variation in educational attainment (Okbay, 2016).

Woodley & Piffer used a polygenic score using 130 common SNP resulted from the same GWAS (Okbay, 2016). They found a polygenic score of 3,298.5 : (3,298.5 + 3,997.5) = 45.21% for Bronze Age samples and 61666 : (61,666 + 69,114) = 47.15% for today Europeans. We can assume that for only 74 SNP Woodley & Piffer could find the same values of polygenic scores of today Europeans (47.15%) and Bronze Age Eurasians (45.21%). It means an average difference (47.15 – 45.21)% x 74 = 1.44 more IQ-increasing SNP in each today sample, equating with 1.44 x 0.076 = 0.11 IQ points.

Jointly, the variance explained by the 74 SNP is 0.43%, but the variance explained by all the common SNP is 15.6% (Hill, 2017). We can assume that polygenic score increased with (47.15% – 45.21%) = 1.94% for all IQ-increasing common SNP of entire genome. In this case, the average total increase of the genotypic IQ due of common SNP since Bronze Age is (15.6% : 0.43%) x 0.11 IQ points = 3.99 IQ points.

It means the selection pressure on intelligence was strong enough to increase the frequency of common SNP that favor high-IQ with the  the equivalent of 3.99 IQ points.

But common SNP account for 15.6 : (15.6 + 28.1) = 35.67% and rare variants account for 64.33% of genotypic IQ (Hill, 2017). It means the same selection pressure that increased the IQ with 3.99 points on common polymorphism SNP will increase the IQ on rare variants with (64.33% : 35.67%) x 3.99 IQ points = 7.20 IQ points. The total increase of the genotypic IQ will be 3.99 + 7.20 = 11.19 IQ points. In fact, since Bronze Age, the selection eliminated IQ-decreasing common SNP and IQ-decreasing rare variants that equate with 11.19 IQ points.

The average age of Bronze Age samples is 3,440 years, equating with 3,440 : 30 = 114.66 generations. Hence, the selection eliminated IQ-decreasing (common and rare) variants of 11.16 : 114.66 = 0.098 IQ points by generation.

If the average decrease of genotypic intelligence by de novo mutations is higher than 0.098 IQ points by generation, the genotypic IQ of Europeans decreased since Bronze Age.

If we estimate a decrease of 8 IQ points of the genotypic intelligence, due of dysgenic fertility after the demographic transition (started 8 generations ago), the selection could eliminate IQ-decreasing (common and rare) variants (11.6 + 8) : (114.66 – 8) = 0.184 IQ points by generation before the demographic transition. If the average decrease of genotypic IQ by de novo mutations was higher than 0.184 IQ points by generation, the genotypic intelligence of Europeans decreased even before the demographic transition.

PS. The polygenic scores for 9 SNP and 11 SNP are 5% higher for today Europeans than for Bronze Age Europeans. If we assume that for all IQ-increasing common SNP the increase is of 5%, the selection eliminated IQ-decreasing (common and rare) variants equating with (5% : 1.94%) x 0.098 = 0.253 IQ points by generation since Bronze Age, and equating with (5% : 1.94%) x 0.184 = 0.474 IQ points by generation (before Industrial Revolution).

PPS. (26.05.2017) Prevalence of ADHD is 7%. Heritability of ADHD is 75%, hence sporadic cases are 85% of all cases. 36% of sporadic cases of ADHD are due of de novo mutations (Kim, 2017). Paternal age higher than 45 years increases 13 fold the risk for ADHD (D’Onofrio, 2014). The prevalence of ADHD due of de novo mutations is 0.07 x 0.85 x 0.36 = 0.0214

A meta-analysis (Frazier, 2004) found 7 at 11 points lower than average IQ of those with ADHD. In all psychiatric diseases, cases due of de novo mutations have the lowest IQ. We can consider those with ADHD by de novo mutations have at least 11 points lower IQ than average.
The generational lost of genotypic intelligence due of ADHD by de novo mutations is 0.0215 x 11 = 0.236 IQ points.
Also, people with ADHD have higher fertility than average (Williams, 2006).
Even only the generational loss of IQ (0.236 points) due of de novo mutations producing ADHD overcompensated the positive selection on intelligence between Bronze Age and Industrial Revolution found by Woodley & Piffer, that had a strength of 0.184 IQ points by generation.

REFERENCES

Hill, D.W. et al. (2017) Genomic analysis of family data reveals additional genetic effects on intelligence and personality. bioRxiv http://dx.doi.org/10.1101/106203
Kong, A. et al. (2017) Selection against variants in the genome associated
with educational attainment. PNAS 114(5): E727-E732. doi: 10.1073/pnas.1612113114.
Okbay, A. et al. (2016). Genome-wide association study identifies 74 loci associated with educational attainment. Nature 533: 539-542.
Woodley, M.A. et al. (2017) Holocene selection for variants associated with cognitive ability: Comparing ancient and modern genomes. bioRxiv http://dx.doi.org/10.1101/109678